GNAT Reference Manual
*********************

   GNAT Reference Manual

   GNAT, The GNU Ada 95 Compiler

   Version 3.12p 

   Date: $Date: 1999/06/28 22:42:46 $

   Ada Core Technologies, Inc.

   (C) Copyright 1995-1999, Ada Core Technologies - All Rights Reserved

   This document may be copied, in whole or in part, in any form or by
any means, as is or with alterations, provided that (1) alterations are
clearly marked as alterations and (2) this copyright notice is included
unmodified in any copy.

   Silicon Graphics and IRIS are registered trademarks and IRIX is a
trademark of Silicon Graphics, Inc.

   IBM PC is a trademark of International Business Machines Corporation.

   UNIX is a registered trademark of AT&T Bell Laboratories.  DIGITAL

   VADS is a registered trademark of Rational Software Inc.

   The following are trademarks of Digital Equipment Corporation: DEC,
DEC Ada, DECthreads, Digital, OpenVMS, and VAX.

About This Guide
****************

This manual contains useful information in writing programs using the
GNAT compiler. It includes information on implementation dependent
characteristics of GNAT, including all the information required by Annex
M of the standard.

   Ada 95 is designed to be highly portable,and guarantees that, for
most programs, Ada 95 compilers behave in exactly the same manner on
different machines. However, since Ada 95 is designed to be used in a
wide variety of applications, it also contains a number of system
dependent features to be used in interfacing to the external world.

   Note: Any program that makes use of implementation-dependent features
may be non-portable.  You should follow good programming practice and
isolate and clearly document any sections of your program that make use
of these features in a non-portable manner.

What This Reference Manual Contains
===================================

   This reference manual contains the following chapters:

   * *Note Implementation Defined Pragmas:: lists GNAT
     implementation-dependent pragmas, which can be used to extend and
     enhance the functionality of the compiler.

   * *Note Implementation Defined Attributes:: lists GNAT
     implementation-dependent attributes which can be used to extend and
     enhance the functionality of the compiler.

   * *Note Implementation Advice:: provides information on generally
     desirable behavior which are not requirements that all compilers
     must follow since it cannot be provided on all systems, or which
     may be undesirable on some systems.

   * *Note Implementation Defined Characteristics:: provides a guide to
     minimizing implementation dependent features.

   * *Note Standard Library Routines:: provides a listing of packages
     and a brief description of the functionality that is provided by
     Ada's extensive set of standard library routines as implemented by
     GNAT.

   * *Note The Implementation of Standard I/O:: details how the GNAT
     implementation of the input-output facilities.

   * *Note Interfacing to Other Languages:: describes how programs
     written in Ada using GNAT can be interfaced to other programming
     languages.

   * *Note Specialized Needs Annexes:: describes the GNAT
     implementation of all of the special needs annexes.

   * *Note Compatibility Guide:: includes sections on compatibility of
     GNAT with other Ada 83 and Ada 95 compilation systems, to assist
     in porting code from other environments.

   This reference manual assumes that you are familiar with Ada 95
language, as described in the International Standard
ANSI/ISO/IEC-8652:1995, Jan 1995.

Conventions
===========

Following are examples of the typographical and graphic conventions used
in this guide:

   * `Functions', `utility program names', `standard names', and
     `classes'.

   * `Option flags'

   * `File Names', `button names', and `field names'.

   * VARIABLES.

   * *Emphasis*.

   * [optional information or parameters]

   * Examples are described by text
          and then shown this way.

Commands that are entered by the user are preceded in this manual by the
characters "$ " (dollar sign followed by space). If your system uses
this sequence as a prompt, then the commands will appear exactly as you
see them in the manual. If your system uses some other prompt, then the
command will appear with the $ replaced by whatever prompt character
you are using.

Related Information
===================

   See the following documents for further information on GNAT

   * `GNAT User's Guide', which provides information on how to use the
     GNAT compiler system.

   * `Ada 95 Reference Manual', which contains all reference material
     for the Ada 95 programming language.

   * `Ada 95 Annotated Reference Manual', which is an annotated version
     of the standard reference manual cited above. The annotations
     describe detailed aspects of the design decision, and in
     particular contain useful sections on Ada 83 compatibility.

   * `DEC Ada, Technical Overview and Comparison on DIGITAL Platforms',
     which contains specific information on compatibility between GNAT
     and DEC Ada 83 systems.

   * `DEC Ada, Language Reference Manual, part number AA-PYZAB-TK' which
     describes in detail the pragmas and attributes provided by the DEC
     Ada 83 compiler system.

Implementation Defined Pragmas
******************************

Ada 95 defines a set of pragmas that can be used to supply additional
information to the compiler. These language defined pragmas are
implemented in GNAT and work as described in the Ada 95 Reference
Manual.

   In addition, Ada 95 allows implementations to define additional
pragmas whose meaning is defined by the implementation. GNAT provides a
number of these implementation-dependent pragmas which can be used to
extend and enhance the functionality of the compiler. This section of
the GNAT Reference Manual describes these additional pragmas.

   Note that any program using these pragmas may not be portable to
other compilers (although GNAT implements this set of pragmas on all
platforms). Therefore if portability to other compilers is an important
consideration, the use of these pragmas should be minimized.

`pragma Abort_Defer'
     Syntax:

          pragma Abort_Defer;

     This pragma must appear at the start of the statement sequence of a
     handled sequence of statements (right after the `begin'). It has
     the effect of deferring aborts for the sequence of statements (but
     not for the declarations or handlers, if any, associated with this
     statement sequence).

`pragma Ada_83'
     Syntax:

          pragma Ada_83;

     A configuration pragma that establishes Ada 83 mode for the unit to
     which it applies, regardless of the mode set by the command line
     switches. In Ada 83 mode, GNAT attempts to be as compatible with
     the syntax and semantics of Ada 83, as defined in the original Ada
     83 Reference Manual as possible. In particular, the new Ada 95
     keywords are not recognized, optional package bodies are allowed,
     and generics may name types with unknown discriminants without
     using the (<>) notation. In addition, some but not all of the
     additional restrictions of Ada 83 are enforced.

     Ada 83 mode is intended for two purposes. Firstly, it allows
     existing legacy Ada 83 code to be compiled and adapted to GNAT
     with less effort.  Secondly, it aids in keeping code backwards
     compatible with Ada 83.  However, there is no guarantee that code
     that is processed correctly by GNAT in Ada 83 mode will in fact
     compile and execute with an Ada 83 compiler, since GNAT does not
     enforce all the additional checks required by Ada 83.

`pragma Ada_95'
     Syntax:

          pragma Ada_95;

     A configuration pragma that establishes Ada 95 mode for the unit
     to which it applies, regardless of the mode set by the command
     line switches.  This mode is set automatically for the `Ada' and
     `System' packages and their children, so you need not specify it
     in these contexts. This pragma is useful when writing a reusable
     component that itself uses Ada 95 features, but which is intended
     to be usable from either Ada 83 or Ada 95 programs.

`pragma Annotate'
     Syntax:

          pragma Annotate (IDENTIFIER {, ARG});
          
          ARG ::= NAME | EXPRESSION

     This pragma is used to annotate programs. IDENTIFIER identifies
     the type of annotation. GNAT verifies this is an identifier, but
     does not otherwise analyze it. The ARG argument can be either a
     string literal or an expression. String literals are assumed to be
     of type `Standard.String'. Names of entities are simply analyzed
     as entity names. All other expressions are analyzed as
     expressions, and must be unambiguous.

     The analyzed pragma is retained in the tree, but not otherwise
     processed by any part of the GNAT compiler. This pragma is
     intended for use by external tools, including ASIS.

`pragma Assert'
     Syntax:

          pragma Assert (
            boolean_EXPRESSION
            [, static_string_EXPRESSION])

     The effect of this pragma depends on whether the corresponding
     command line switch is set to activate assertions. If assertions
     are inactive, the pragma has no effect. If assertions are enabled,
     then the semantics of the pragma is exactly equivalent to:

          if not boolean_EXPRESSION then
             System.Assertions.Raise_Assert_Failure (string_EXPRESSION);
          end if;

     The effect of the call is to raise
     `System.Assertions.Assert_Failure'. The string argument, if given,
     is the message associated with the exception occurrence. If no
     second argument is given, the default message is `FILE:NNN', where
     FILE is the name of the source file containing the assert, and NNN
     is the line number of the assert. A pragma is not a statement, so
     if a statement sequence contains nothing but a pragma assert, then
     a null statement is required in addition, as in:

     Note that, as with the if statement to which it is equivalent, the
     type of the expression is either Standard.Boolean, or any type
     derived from this standard type.

          ...
          if J > 3 then
             pragma (Assert (K > 3, "Bad value for K"));
             null;
          end if;

     If the boolean expression has side effects, these side effects
     will turn on and off with the setting of the assertions mode,
     resulting in assertions that have an effect on the program. You
     should generally avoid side effects in the expression arguments of
     this pragma. However, the expressions are analyzed for semantic
     correctness whether or not assertions are enabled, so turning
     assertions on and off cannot affect the legality of a program.

`pragma Ast_Entry'
     Syntax:

          pragma AST_Entry (entry_IDENTIFIER);

     This pragma is implemented only in the OpenVMS implementation of
     GNAT. The argument is the simple name of a single entry; at most
     one `AST_Entry' pragma is allowed for any given entry. This pragma
     must be used in conjunction with the `AST_Entry' attribute, and is
     only allowed after the entry declaration and in the same task type
     specification or single task as the entry to which it applies.
     This pragma specifies that the given entry may be used to handle
     an OpenVMS asynchronous system trap (`AST') resulting from an
     OpenVMS system service call. The pragma does not affect normal use
     of the entry. For further details on this pragma, see the DEC Ada
     Language Reference Manual, section 9.12a.

`pragma C_Pass_By_Copy'
     Syntax:

          pragma C_Pass_By_Copy
            ([Max_Size =>] static_integer_EXPRESSION);

     Normally the default mechanism for passing C convention records to
     C convention subprograms is to pass them by reference, as
     suggested by RM B.3(69). Use the configuration pragma
     `C_Pass_By_Copy' to change this default, by requiring that record
     formal parameters be passed by copy if all of the following
     conditions are met:

        * The size of the record type does not exceed
          STATIC_INTEGER_EXPRESSION.

        * The record type has `Convention C'.

        * The formal parameter has this record type, and the subprogram
          has a foreign (non-Ada) convention.

     If these conditions are met the argument is passed by copy, i.e.
     in a manner consistent with what C expects if the corresponding
     formal in the C prototype is a struct (rather than a pointer to a
     struct).

     You can also pass records by copy by specifying the convention
     `C_Pass_By_Copy' for the record type, or by using the extended
     `Import' and `Export' pragmas, which allow specification of
     passing mechanisms on a parameter by parameter basis.

`pragma Comment'
     Syntax:

          pragma Comment (static_string_EXPRESSION);

     Synonymous for pragma Ident.

`pragma Common_Object'
     Syntax:

          pragma Common_Object
               [Internal =>] LOCAL_NAME,
            [, [External =>] EXTERNAL_SYMBOL,
            [, [Size     =>] EXTERNAL_SYMBOL]
          
          EXTERNAL_SYMBOL ::=
            IDENTIFIER
          | static_string_EXPRESSION

     This pragma enables the shared use of variables stored in overlaid
     linker areas corresponding to the use of `COMMON' in Fortran.  The
     single object LOCAL_NAME is assigned to the area designated by the
     EXTERNAL argument.  You may define a record to correspond to a
     series of fields. The SIZE argument is syntax checked in GNAT, but
     otherwise ignored.

`pragma Complex_Representation'
     Syntax:

          pragma Complex_Representation ([Entity =>] LOCAL_NAME);

     The ENTITY argument must be the name of a record type which has
     two fields of the same floating-point type. The effect of this
     pragma is to force gcc to use the special internal complex
     representation form for this record, which may be more efficient.
     Note that this may result in the code for this type not conforming
     to standard ABI (application binary interface) requirements for
     the handling of record types. For example, in some environments,
     there is a requirement for passing records by pointer, and the use
     of this pragma may result in passing this type in floating-point
     registers.

`pragma Component_Alignment'
     Syntax:

          pragma Component_Alignment (
               [Form =>] ALIGNMENT_CHOICE
            [, [Name =>] type_LOCAL_NAME]);
          
          ALIGNMENT_CHOICE ::=
            Component_Size
          | Component_Size_4
          | Storage_Unit
          | Default

     Specifies the alignment of components in array or record types.
     The meaning of the FORM argument is as follows:

    `Component_Size'
          Aligns scalar components and subcomponents of the array or
          record type on boundaries appropriate to their inherent size
          (naturally aligned). For example, 1-byte components are
          aligned on byte boundaries, 2-byte integer components are
          aligned on 2-byte boundaries, 4-byte integer components are
          aligned on 4-byte boundaries and so on. These alignment rules
          correspond to the normal rules for C compilers on all
          machines except the VAX.

    `Component_Size_4'
          Naturally aligns components with a size of four or fewer
          bytes. Components that are larger than 4 bytes are placed on
          the next 4-byte boundary.

    `Storage_Unit'
          Specifies that array or record components are byte aligned,
          i.e.  aligned on boundaries determined by the value of the
          constant `System.Storage_Unit'.

    `Default'
          Specifies that array or record components are aligned on
          default boundaries, appropriate to the underlying hardware or
          operating system or both. For OpenVMS VAX systems, the
          `Default' choice is the same as the `Storage_Unit' choice
          (byte alignment). For all other systems, the `Default' choice
          is the same as `Component_Size' (natural alignment).

     If the `Name' parameter is present, TYPE_LOCAL_NAME must refer to
     a local record or array type, and the specified alignment choice
     applies to the specified type. The use of `Component_Alignment'
     together with a pragma `Pack' causes the `Component_Alignment'
     pragma to be ignored. The use of `Component_Alignment' together
     with a record representation clause is only effective for fields
     not specified by the representation clause.

     If the `Name' parameter is absent, the pragma can be used as either
     a configuration pragma, in which case it applies to one or more
     units in accordance with the normal rules for configuration
     pragmas, or it can be used within a declarative part, in which
     case it applies to types that are declared within this declarative
     part, or within any nested scope within this declarative part. In
     either case it specifies the alignment to be applied to any record
     or array type which has otherwise standard representation.

     If the alignment for a record or array type is not specified (using
     pragma `Pack', pragma `Component_Alignment', or a record rep
     clause), the GNAT uses the default alignment as described
     previously.

`pragma CPP_Class'
     Syntax:

          pragma CPP_Class ([Entity =>] LOCAL_NAME);

     The argument denotes an entity in the current declarative region
     that is declared as a tagged or untagged record type. It indicates
     that the type corresponds to an externally declared C++ class
     type, and is to be laid out the same way that C++ would lay out
     the type.

     If (and only if) the type is tagged, at least one component in the
     record must be of type `Interfaces.CPP.Vtable_Ptr', corresponding
     to the C++ Vtable (or Vtables in the case of multiple inheritance)
     used for dispatching.

     Types for which `CPP_Class' is specified do not have assignment or
     equality operators defined (such operations can be imported or
     declared as subprograms as required). Initialization is allowed
     only by constructor functions (see pragma `CPP_Constructor').

     Pragma `CPP_Class' is intended primarily for automatic generation
     using an automatic binding generator tool. Ada Core Technologies
     does not currently supply such a tool; *Note Interfacing to C++::
     for more details.

`pragma CPP_Constructor'
     Syntax:

          pragma CPP_Constructor ([Entity =>] LOCAL_NAME);

     This pragma identifies an imported function (imported in the usual
     way with pragma Import) as corresponding to a C++ constructor. The
     argument is a name that must have been previously mentioned in a
     pragma Import with CONVENTION CPP, and must be of one of the
     following forms:

        * `function FNAME return T'Class'

        * `function FNAME (...) return T'Class'

     where T is a tagged type to which the pragma `CPP_Class' applies.

     The first form is the default constructor, used when an object of
     type T is created on the Ada side with no explicit constructor.
     Other constructors (including the copy constructor, which is
     simply a special case of the second form in which the one and only
     argument is of type T), can only appear in two contexts:

        * On the right side of an initialization of an object of type T.

        * In an extension aggregate for an object of a type derived
          from T.

     Although the constructor is described as a function that returns a
     value on the Ada side, it is typically a procedure with an extra
     implicit argument (the object being initialized) at the
     implementation level. GNAT issues the appropriate call, whatever
     it is, to get the object properly initialized.

     In the case of derived objects, you may use one of two possible
     forms for declaring and creating an object:

        * `New_Object : Derived_T'

        * `New_Object : Derived_T := (CONSTRUCTOR-FUNCTION-CALL WITH
          ...)'

     In the first case the default constructor is called and extension
     fields if any are initialized according to the default
     initialization expressions in the Ada declaration. In the second
     case, the given constructor is called and the extension aggregate
     indicates the explicit values of the extension fields.

     If no constructors are imported, it is impossible to create any
     objects on the Ada side. If no default constructor is imported,
     only the initialization forms using an explicit call to a
     constructor are permitted.

     Pragma `CPP_Constructor' is intended primarily for automatic
     generation using an automatic binding generator tool. Ada Core
     Technologies does not currently supply such a tool; *Note
     Interfacing to C++:: for more details.

`pragma CPP_Destructor ([Entity =>] LOCAL_NAME);'
     This pragma identifies an imported procedure (imported in the
     usual way with pragma Import) as corresponding to a C++
     destructor. LOCAL_NAME must be previously mentioned in a pragma
     `Import' with `Convention CPP', and be of the following form:

          procedure FNAME (obj : in out T'Class);

     where T is a tagged type to which pragma `CPP_Class' applies. This
     procedure will be called automatically on scope exit if any
     objects of T are created on the Ada side.

     Pragma `CPP_Destructor' is intended primarily for automatic
     generation using an automatic binding generator tool. Ada Core
     Technologies does not currently supply such a tool; *Note
     Interfacing to C++:: for more details.

`pragma CPP_Virtual'
     Syntax:

          pragma CPP_Virtual
               [Entity     =>] ENTITY,
            [, [Vtable_Ptr =>] vtable_ENTITY,]
            [, [Position   =>] static_integer_EXPRESSION])

     This pragma serves the same function as pragma `Import' in that
     case of a virtual function imported from C++. The ENTITY argument
     must be a primitive subprogram of a tagged type to which pragma
     `CPP_Class' applies. The VTABLE_PTR argument specifies the
     Vtable_Ptr component which contains the entry for this virtual
     function. The POSITION argument is the sequential number counting
     virtual functions for this Vtable starting at 1.

     The `Vtable_Ptr' and `Position' arguments may be omitted if there
     is one Vtable_Ptr present (single inheritance case) and all
     virtual functions are imported.  In that case the compiler can
     deduce both these values.

     No `External_Name' or `Link_Name' arguments are required for a
     virtual function, since it is always accessed indirectly via the
     appropriate Vtable entry.

     Pragma `CPP_Virtual' is intended primarily for automatic generation
     using an automatic binding generator tool. Ada Core Technologies
     does not currently supply such a tool; *Note Interfacing to C++::
     for more details.

`pragma CPP_Vtable'
     Syntax:

          pragma CPP_Vtable (
            [Entity      =>] ENTITY,
            [Vtable_Ptr  =>] vtable_ENTITY,
            [Entry_Count =>] static_integer_EXPRESSION);

     Given a record to which the pragma `CPP_Class' applies, this
     pragma can be specified for each component of type
     `CPP.Interfaces.Vtable_Ptr'.  ENTITY is the tagged type, VTABLE_PTR
     is the record field of type `Vtable_Ptr', and ENTRY_COUNT is the
     number of virtual functions on the C++ side. Not all of these
     functions need to be imported on the Ada side.

     You may omit the `CPP_Vtable' pragma if there is only one
     `Vtable_Ptr' component in the record and all virtual functions are
     imported on the Ada side (the default value for the entry count in
     this case is simply the total number of virtual functions).

     Pragma `CPP_Vtable' is intended primarily for automatic generation
     using an automatic binding generator tool. Ada Core Technologies
     does not currently supply such a tool; *Note Interfacing to C++::
     for more details.

`pragma Debug'
     Syntax:

          pragma Debug (PROCEDURE_CALL_STATEMENT);

     If assertions are not enabled on the command line, this pragma has
     no effect. If asserts are enabled, the semantics of the pragma is
     exactly equivalent to the procedure call. Pragmas are permitted in
     sequences of declarations, so you can use pragma `Debug' to
     intersperse calls to debug procedures in the middle of
     declarations.

`pragma Eliminate'
     Syntax:

          pragma Eliminate (
              [Unit_Name       =>] IDENTIFIER |
                                   SELECTED_COMPONENT
            [,[Entity          =>] IDENTIFIER |
                                   SELECTED_COMPONENT |
                                   STRING_LITERAL]
            [,[Parameter_Types =>] PARAMETER_TYPES]
            [,[Result_Type     =>] result_SUBTYPE_MARK]);
          
          
          PARAMETER_TYPES ::=
            null
          | SUBTYPE_MARK {, SUBTYPE_MARK}

     This pragma indicates that the given entity is unused in a
     program. The entity may be either a subprogram or a variable. If
     the entity to be eliminated is a library level subprogram, then
     only the first argument, specifying the corresponding unit name,
     is required. If the item is an entity of a library package, then
     the first argument specifies the unit name, and the second
     argument specifies the particular entity. If the second argument
     is in string form, it must correspond to the internal manner in
     which GNAT stores entity names (see compilation unit Namet in the
     compiler sources for details). The third and fourth parameters are
     optionally used to distinguish between overloaded subprograms, in
     the same manner as is used for pragma IMPORT_PROCEDURE.

     The effect of the pragma is to allow the compiler to optionally
     eliminate the code or data associated with the named entity. If
     the declaration of the entity would have resulted in side effects,
     these side effects may or may not occur in the resulting program.
     Any reference to an eliminated entity may cause a compile time
     error, link time error, or incorrect results at runtime.

     The intention of pragma Eliminate is to allow a program to be
     compiled in a system independent manner, with unused entities
     eliminated, without the requirement of modifying the source text.
     Normally the required set of Eliminate pragmas is constructed
     automatically using the gnatelim tool.

`pragma Export_Exception'
     Syntax:

          pragma Export_Exception (
               [Internal =>] LOCAL_NAME,
            [, [External =>] EXTERNAL_SYMBOL,]
            [, [Form     =>] Ada | VMS]
            [, [Code     =>] static_integer_EXPRESSION]);
          
          EXTERNAL_SYMBOL ::=
            IDENTIFIER
          | static_string_EXPRESSION

     This pragma is implemented only in the OpenVMS implementation of
     GNAT. It causes the specified exception to be propagated outside
     of the Ada program, so that it can be handled by programs written
     in other OpenVMS languages.  This pragma establishes an external
     name for an Ada exception and makes the name available to the
     OpenVMS Linker as a global symbol. For further details on this
     pragma, see the DEC Ada Language Reference Manual, section
     13.9a3.2.

`pragma Export_Function ...'
     Syntax:

          pragma Export_Function (
               [Internal         =>] LOCAL_NAME,
            [, [External         =>] EXTERNAL_SYMBOL]
            [, [Parameter_Types  =>] PARAMETER_TYPES]
            [, [Result_Type      =>] result_SUBTYPE_MARK]
            [, [Mechanism        =>] MECHANISM]
            [, [Result_Mechanism =>] MECHANISM_NAME]);
          
          EXTERNAL_SYMBOL ::=
            IDENTIFIER
          | static_string_EXPRESSION
          
          PARAMETER_TYPES ::=
            null
          | SUBTYPE_MARK {, SUBTYPE_MARK}
          
          MECHANISM ::=
            MECHANISM_NAME
          | (MECHANISM_ASSOCIATION {, MECHANISM_ASSOCIATION})
          
          MECHANISM_ASSOCIATION ::=
            [formal_parameter_NAME =>] MECHANISM_NAME
          
          MECHANISM_NAME ::=
            Value
          | Reference
          | Descriptor [([Class =>] CLASS_NAME)]
          
          CLASS_NAME ::= ubs | ubsb | uba | s | sb | a | nca

     Use this pragma to make a function externally callable and
     optionally provide information on mechanisms to be used for
     passing parameter and result values. We recommend, for the
     purposes of improving portability, this pragma always be used in
     conjunction with a separate pragma `Export', which must precede
     the pragma `Export_Function'.  GNAT does not require a separate
     pragma `Export', but if none is present, it assumes `Convention
     C'. Pragma `Export_Function' (and `Export', if present) must
     appear in the same declarative region as the function to which
     they apply.

     INTERNAL_NAME must uniquely designate the function to which the
     pragma applies. If more than one function name exists of this name
     in the declarative part you must use the `Parameter_Types' and
     `Result_Type' parameters is mandatory to achieve the required
     unique designation. SUBTYPE_ MARKs in these parameters must
     exactly match the subtypes in the corresponding function
     specification, using positional notation to match parameters with
     subtype marks.  Passing by descriptor is supported only on the
     OpenVMS ports of GNAT.

`pragma Export_Object ...'
     Syntax:

          pragma Export_Object
                [Internal =>] LOCAL_NAME,
             [, [External =>] EXTERNAL_SYMBOL]
             [, [Size     =>] EXTERNAL_SYMBOL]
          
          EXTERNAL_SYMBOL ::=
            IDENTIFIER
          | static_string_EXPRESSION

     This pragma designates an object as exported, and apart from the
     extended rules for external symbols, is identical in effect to the
     use of the normal `Export' pragma applied to an object. You may
     use a separate Export pragma (and you probably should from the
     point of view of portability), but it is not required.  SIZE is
     syntax checked, but otherwise ignored by GNAT.

`pragma Export_Procedure ...'
     Syntax:

          pragma Export_Procedure (
               [Internal        =>] LOCAL_NAME
            [, [External        =>] EXTERNAL_SYMBOL]
            [, [Parameter_Types =>] PARAMETER_TYPES]
            [, [Mechanism       =>] MECHANISM]);
          
          EXTERNAL_SYMBOL ::=
            IDENTIFIER
          | static_string_EXPRESSION
          
          PARAMETER_TYPES ::=
            null
          | SUBTYPE_MARK {, SUBTYPE_MARK}
          
          MECHANISM ::=
            MECHANISM_NAME
          | (MECHANISM_ASSOCIATION {, MECHANISM_ASSOCIATION})
          
          MECHANISM_ASSOCIATION ::=
            [formal_parameter_NAME =>] MECHANISM_NAME
          
          MECHANISM_NAME ::=
            Value
          | Reference
          | Descriptor [([Class =>] CLASS_NAME)]
          
          CLASS_NAME ::= ubs | ubsb | uba | s | sb | a | nca

     This pragma is identical to `Export_Function' except that it
     applies to a procedure rather than a function and the parameters
     `Result_Type' and `Result_Mechanism' are not permitted.

`pragma Export_Valued_Procedure'
     Syntax:

          pragma Export_Valued_Procedure (
               [Internal        =>] LOCAL_NAME
            [, [External        =>] EXTERNAL_SYMBOL]
            [, [Parameter_Types =>] PARAMETER_TYPES]
            [, [Mechanism       =>] MECHANISM]);
          
          EXTERNAL_SYMBOL ::=
            IDENTIFIER
          | static_string_EXPRESSION
          
          PARAMETER_TYPES ::=
            null
          | SUBTYPE_MARK {, SUBTYPE_MARK}
          
          MECHANISM ::=
            MECHANISM_NAME
          | (MECHANISM_ASSOCIATION {, MECHANISM_ASSOCIATION})
          
          MECHANISM_ASSOCIATION ::=
            [formal_parameter_NAME =>] MECHANISM_NAME
          
          MECHANISM_NAME ::=
            Value
          | Reference
          | Descriptor [([Class =>] CLASS_NAME)]
          
          CLASS_NAME ::= ubs | ubsb | uba | s | sb | a | nca

     This pragma is identical to `Export_Procedure' except that the
     first parameter of LOCAL_NAME, which must be present, must be of
     mode `OUT', and externally the subprogram is treated as a function
     with this parameter as the result of the function. GNAT provides
     for this capability to allow the use of `OUT' and `IN OUT'
     parameters in interfacing to external functions (which are not
     permitted in Ada functions).

`pragma Extend_System'
     Syntax:

          pragma Extend_System ([Name =>] IDENTIFIER);

     This pragma is used to provide backwards compatibility with other
     implementations that extend the facilities of package `System'. In
     GNAT, `System' contains only the definitions that are present in
     the Ada 95 RM. However, other implementations, notably the DEC Ada
     83 implementation, provide many extensions to package `System'.

     For each such implementation accommodated by this pragma, GNAT
     provides a package `Aux_XXX', e.g. `Aux_DEC' for the DEC Ada 83
     implementation, which provides the required additional
     definitions. You can use this package in two ways.  You can `with'
     it in the normal way and access entities either by selection or
     using a `use' clause. In this case no special processing is
     required.

     However, if existing code contains references such as `System.XXX'
     where XXX is an entity in the extended definitions provided in
     package `System', you may use this pragma to extend visibility in
     `System' in a non-standard way that provides greater compatibility
     with the existing code. Pragma `Extend_System' is a configuration
     pragma whose single argument is the name of the package containing
     the extended definition (e.g. `Aux_DEC' for the DEC Ada case). A
     unit compiled under control of this pragma will be processed using
     special visibility processing that looks in package
     `System.Aux_XXX' where `Aux_XXX' is the pragma argument for any
     entity referenced in package `System', but not found in package
     `System'.

`pragma Finalize_Storage_Only'
     Syntax:

          pragma Finalize_Storage_Only (first_subtype_LOCAL_NAME);

     This pragma allows the compiler not to emit a Finalize call for
     objects defined at the library level. This is mostly useful for
     types where finalization is only used to deal with storage
     reclamation since in most environments it is not necessary to
     reclaim memory just before terminating execution, hence the name.

`pragma Float_Representation'
     Syntax:

          pragma Float_Representation (FLOAT_REP);
          
          FLOAT_REP ::= VAX_Float | IEEE_Float

     This pragma is implemented only in the OpenVMS implementation of
     GNAT.  It allows control over the internal representation chosen
     for the predefined floating point types declared in the packages
     `Standard' and `System'. For further details on this pragma, see
     the DEC Ada Language Reference Manual, section 3.5.7a. Note that
     to use this pragma, the standard runtime libraries must be
     recompiled. See the description of the `GNAT LIBRARY' command in
     the OpenVMS version of the GNAT Users Guide for details on the use
     of this command.

`pragma Ident'
     Syntax:

          pragma Ident (static_string_EXPRESSION);

     This pragma provides a string identification in the generated
     object file, if the system supports the concept of this kind of
     identification string.  The maximum permitted length of the string
     literal is 31 characters.  This pragma is allowed only in the
     outermost declarative part or declarative items of a compilation
     unit.  On OpenVMS systems, the effect of the pragma is identical
     to the effect of the DEC Ada 83 pragma of the same name.

`pragma Import_Exception'
     Syntax:

          pragma Import_Exception (
               [Internal =>] LOCAL_NAME,
            [, [External =>] EXTERNAL_SYMBOL,]
            [, [Form     =>] Ada | VMS]
            [, [Code     =>] static_integer_EXPRESSION]);
          
          EXTERNAL_SYMBOL ::=
            IDENTIFIER
          | static_string_EXPRESSION

     This pragma is implemented only in the OpenVMS implementation of
     GNAT.  It allows OpenVMS conditions (for example, from OpenVMS
     system services or other OpenVMS languages) to be propagated to
     Ada programs as Ada exceptions.  The pragma specifies that the
     exception associated with an exception declaration in an Ada
     program be defined externally (in non-Ada code).  For further
     details on this pragma, see the DEC Ada Language Reference Manual,
     section 13.9a.3.1.

`Import_Function ...'
     Syntax:

          pragma Import_Function (
               [Internal                 =>] LOCAL_NAME,
            [, [External                 =>] EXTERNAL_SYMBOL]
            [, [Parameter_Types          =>] PARAMETER_TYPES]
            [, [Result_Type              =>] SUBTYPE_MARK]
            [, [Mechanism                =>] MECHANISM]
            [, [Result_Mechanism         =>] MECHANISM_NAME]
            [, [First_Optional_Parameter =>] IDENTIFIER]);
          
          EXTERNAL_SYMBOL ::=
            IDENTIFIER
          | static_string_EXPRESSION
          
          PARAMETER_TYPES ::=
            null
          | SUBTYPE_MARK {, SUBTYPE_MARK}
          
          MECHANISM ::=
            MECHANISM_NAME
          | (MECHANISM_ASSOCIATION {, MECHANISM_ASSOCIATION})
          
          MECHANISM_ASSOCIATION ::=
            [formal_parameter_NAME =>] MECHANISM_NAME
          
          MECHANISM_NAME ::=
            Value
          | Reference
          | Descriptor [([Class =>] CLASS_NAME)]
          
          CLASS_NAME ::= ubs | ubsb | uba | s | sb | a | nca

     This pragma is used in conjunction with a pragma `Import' to
     specify additional information for an imported function. The pragma
     `Import' (or equivalent pragma `Interface') must precede the
     `Import_Function' pragma and both must appear in the same
     declarative part as the function specification.

     The INTERNAL_NAME argument must uniquely designate the function to
     which the pragma applies. If more than one function name exists of
     this name in the declarative part you must use the
     `Parameter_Types' and RESULT_TYPE parameters to achieve the
     required unique designation. Subtype marks in these parameters
     must exactly match the subtypes in the corresponding function
     specification, using positional notation to match parameters with
     subtype marks.

     You may optionally use the MECHANISM and RESULT_MECHANISM
     parameters to specify passing mechanisms for the parameters and
     result. If you specify a single mechanism name, it applies to all
     parameters.  Otherwise you may specify a mechanism on a parameter
     by parameter basis using either positional or named notation. If
     the mechanism is not specified, the default mechanism is used.

     Passing by descriptor is supported only on the to OpenVMS ports of
     GNAT

     `First_Optional_Parameter' applies only to OpenVMS ports of GNAT.
     It specifies that the designated parameter and all following
     parameters are optional, meaning that they are not passed at the
     generated code level (this is distinct from the notion of optional
     parameters in Ada where the parameters are passed anyway with the
     designated optional parameters). All optional parameters must be
     of mode `IN' and have default parameter values that are either
     known at compile time expressions, or uses of the
     `'Null_Parameter' attribute.

`pragma Import_Object'
     Syntax:

          pragma Import_Object
               [Internal =>] LOCAL_NAME,
            [, [External =>] EXTERNAL_SYMBOL],
            [, [Size     =>] EXTERNAL_SYMBOL])
          
          EXTERNAL_SYMBOL ::=
            IDENTIFIER
          | static_string_EXPRESSION

     This pragma designates an object as imported, and apart from the
     extended rules for external symbols, is identical in effect to the
     use of the normal `Import' pragma applied to an object.  Unlike the
     subprogram case, you need not use a separate `Import' pragma,
     although you may do so (and probably should do so from a
     portability point of view). SIZE is syntax checked, but otherwise
     ignored by GNAT.

`pragma Import_Procedure'
     Syntax:

          pragma Import_Procedure (
               [Internal                 =>] LOCAL_NAME,
            [, [External                 =>] EXTERNAL_SYMBOL]
            [, [Parameter_Types          =>] PARAMETER_TYPES]
            [, [Mechanism                =>] MECHANISM]
            [, [First_Optional_Parameter =>] IDENTIFIER]);
          
          EXTERNAL_SYMBOL ::=
            IDENTIFIER
          | static_string_EXPRESSION
          
          PARAMETER_TYPES ::=
            null
          | SUBTYPE_MARK {, SUBTYPE_MARK}
          
          MECHANISM ::=
            MECHANISM_NAME
          | (MECHANISM_ASSOCIATION {, MECHANISM_ASSOCIATION})
          
          MECHANISM_ASSOCIATION ::=
            [formal_parameter_NAME =>] MECHANISM_NAME
          
          MECHANISM_NAME ::=
            Value
          | Reference
          | Descriptor [([Class =>] CLASS_NAME)]
          
          CLASS_NAME ::= ubs | ubsb | uba | s | sb | a | nca

     This pragma is identical to `Import_Function' except that it
     applies to a procedure rather than a function and the parameters
     `Result_Type' and `Result_Mechanism' are not permitted.

`pragma Import_Valued_Procedure ...'
     Syntax:

          pragma Import_Valued_Procedure (
               [Internal                 =>] LOCAL_NAME,
            [, [External                 =>] EXTERNAL_SYMBOL]
            [, [Parameter_Types          =>] PARAMETER_TYPES]
            [, [Mechanism                =>] MECHANISM]
            [, [First_Optional_Parameter =>] IDENTIFIER]);
          
          EXTERNAL_SYMBOL ::=
            IDENTIFIER
          | static_string_EXPRESSION
          
          PARAMETER_TYPES ::=
            null
          | SUBTYPE_MARK {, SUBTYPE_MARK}
          
          MECHANISM ::=
            MECHANISM_NAME
          | (MECHANISM_ASSOCIATION {, MECHANISM_ASSOCIATION})
          
          MECHANISM_ASSOCIATION ::=
            [formal_parameter_NAME =>] MECHANISM_NAME
          
          MECHANISM_NAME ::=
            Value
          | Reference
          | Descriptor [([Class =>] CLASS_NAME)]
          
          CLASS_NAME ::= ubs | ubsb | uba | s | sb | a | nca

     This pragma is identical to `Import_Procedure' except that the
     first parameter of LOCAL_NAME, which must be present, must be of
     mode `OUT', and externally the subprogram is treated as a function
     with this parameter as the result of the function. The purpose of
     this capability is to allow the use of `OUT' and `IN OUT'
     parameters in interfacing to external functions (which are not
     permitted in Ada functions).  You may optionally use the
     `Mechanism' parameters to specify passing mechanisms for the
     parameters.  If you specify a single mechanism name, it applies to
     all parameters.  Otherwise you may specify a mechanism on a
     parameter by parameter basis using either positional or named
     notation. If the mechanism is not specified, the default mechanism
     is used.

`pragma Inline_Always'
     Syntax:

          pragma Inline_Always (NAME [, NAME]);

     Similar to pragma  `Inline' except that inlining is not subject to
     the use of option `-gnatn' for inter-unit inlining.

`pragma Inline_Generic'
     Syntax:

          pragma Inline_Generic (generic_package_NAME)

     This is implemented for compatibility with DEC Ada 83 and is
     recognized, but otherwise ignored, by GNAT. All generic
     instantiations are inlined by default when using GNAT.

`pragma Interface_Name'
     Syntax:

          pragma Interface_Name (
               [Entity        =>] LOCAL_NAME
            [, [External_Name =>] static_string_EXPRESSION]
            [, [Link_Name     =>] static_string_EXPRESSION]);

     This pragma provides an alternative way of specifying the
     interface name for an interfaced subprogram, and is provided for
     compatibility with Ada 83 compilers that use the pragma for this
     purpose. You must provide at least one of EXTERNAL_NAME or
     LINK_NAME.

`pragma Linker_Alias'
     Syntax:

          pragma Linker_Alias (
            [Entity =>] LOCAL_NAME
            [Alias  =>] static_string_EXPRESSION);

     This pragma establishes a linker alias for the given named entity.
     For further details on the exact effect, consult the GCC manual.

`pragma Linker_Section'
     Syntax:

          pragma Linker_Section (
            [Entity  =>] LOCAL_NAME
            [Section =>] static_string_EXPRESSION);

     This pragma specifies the name of the linker section for the given
     entity.  For further details on the exact effect, consult the GCC
     manual.

`pragma No_Runtime'
     Syntax:

          pragma No_Runtime;

     This is a configuration pragma that makes sure the user code does
     not use nor need anything from the GNAT run time. This is mostly
     useful in context where code certification is required. Please
     consult the GNORT product documentation for additional information.

`pragma Normalize_Scalars'
     Syntax:

          pragma Normalize_Scalars;

     This is a language defined pragma which is fully implemented in
     GNAT. The effect is to cause all scalar objects that are not
     otherwise initialized to be initialized. The initial values are
     implementation dependent and are as follows:

    `Standard.Character'
          Objects whose root type is Standard.Character are initialized
          to Character'Last. This will be out of range of the subtype
          only if the subtype range excludes this value.

    `Standard.Wide_Character'
          Objects whose root type is Standard.Wide_Character are
          initialized to Wide_Character'Last. This will be out of range
          of the subtype only if the subtype range excludes this value.

    `Integer types'
          Objects of an integer type are initialized to
          base_type'First, where base_type is the base type of the
          object type. This will be out of range of the subtype only if
          the subtype range excludes this value. For example, if you
          declare the subtype:

               subtype Ityp is integer range 1 .. 10;

          then objects of type x will be initialized to Integer'First,
          a negative number that is certainly outside the range of
          subtype `Ityp'.

    `Real types'
          Objects of all real types (fixed and floating) are
          initialized to base_type'First, where base_Type is the base
          type of the object type.  This will be out of range of the
          subtype only if the subtype range excludes this value.

    `Modular types'
          Objects of a modular type are initialized to typ'Last. This
          will be out of range of the subtype only if the subtype
          excludes this value.

    `Enumeration types'
          Objects of an enumeration type are initialized to all
          one-bits, i.e. to the value 2 ** typ'Size - 1. This will be
          out of range of the enumeration subtype in all cases except
          where the subtype contains exactly 2**8, 2**16, or 2**32.

`pragma Long_Float'
     Syntax:

          pragma Long_Float (FLOAT_FORMAT);
          
          FLOAT_FORMAT ::= D_Float | G_Float

     This pragma is implemented only in the OpenVMS implementation of
     GNAT.  It allows control over the internal representation chosen
     for the predefined type `Long_Float' and for floating point type
     representations with `digits' specified in the range 7 .. 15.  For
     further details on this pragma, see the DEC Ada Language Reference
     Manual, section 3.5.7b. Note that to use this pragma, the standard
     runtime libraries must be recompiled. See the description of the
     `GNAT LIBRARY' command in the OpenVMS version of the GNAT Users
     Guide for details on the use of this command.

`pragma Machine_Attribute ...'
     Syntax:

          pragma Machine_Attribute (
            [Attribute_Name =>] string_EXPRESSION,
            [Entity         =>] LOCAL_NAME);

     Machine dependent attributes can be specified for types and/or
     declarations. Currently only subprogram entities are supported.
     This pragma is semantically equivalent to `__attribute__((
     STRING_EXPRESSION))' in GNU C, where `string_expression'> is
     recognized by the GNU C macros `VALID_MACHINE_TYPE_ATTRIBUTE' and
     `VALID_MACHINE_DECL_ATTRIBUTE' which are defined in the
     configuration header file `tm.h' for each machine.  See the GCC
     manual for further information.

`pragma Main_Storage'
     Syntax:

          pragma Main_Storage
            (MAIN_STORAGE_OPTION [, MAIN_STORAGE_OPTION]);
          
          MAIN_STORAGE_OPTION ::=
            [WORKING_STORAGE =>] static_SIMPLE_EXPRESSION
          | [TOP_GUARD       =>] static_SIMPLE_EXPRESSION

     This pragma is provided for compatibility with OpenVMS Vax
     Systems. It has no effect in GNAT, other than being syntax
     checked. Note that the pragma also has no effect in DEC Ada 83 for
     OpenVMS Alpha Systems.

`pragma No_Return'
     Syntax:

          pragma No_Return (procedure_LOCAL_NAME);

     PROCEDURE_LOCAL_NAME must refer to one or more procedure
     declarations in the current declarative part. A procedure to which
     this pragma is applied may not contain any explicit `return'
     statements, and also may not contain any implicit return
     statements from falling off the end of a statement sequence. One
     use of this pragma is to identify procedures whose only purpose is
     to raise an exception.

     Another use of this pragma is to suppress incorrect warnings about
     missing returns in functions, where the last statement of a
     function statement sequence is a call to such a procedure.

`pragma Passive'
     Syntax:

          pragma Passive ([Semaphore | No]);

     Syntax checked, but otherwise ignored by GNAT. This is recognized
     for compatibility with DEC Ada 83 implementations, where it is
     used within a task definition to request that a task be made
     passive.  If the argument `Semaphore' is present, or no argument
     is omitted, then DEC Ada 83 treats the pragma as an assertion that
     the containing task is passive and that optimization of context
     switch with this task is permitted and desired.  If the argument
     `No' is present, the task must not be optimized. GNAT does not
     attempt to optimize any tasks in this manner (since protected
     objects are available in place of passive tasks).

`pragma Polling'
     Syntax:

          pragma Polling (ON | OFF);

     This pragma controls the generation of polling code. This is
     normally off.  If `pragma Polling (ON)' is used then periodic
     calls are generated to the routine Ada.Exceptions.Poll. This
     routine is a separate unit in the runtime library, and can be
     found in file a-excpol.adb.

     Pragma polling can appear as a configuration pragma (for example
     it can be placed in the gnat.adc file) to enable polling globally,
     or it can be used in the statement or declaration sequence to
     control polling more locally.

     A call to the polling routine is generated at the start of every
     loop and at the start of every subprogram call. This guarantees
     that the Poll routine is called frequently, and places an upper
     bound (determined by the complexity of the code) on the period
     between two Poll calls.

     The primary purpose of the polling interface is to enable
     asynchronous aborts on targets that cannot otherwise support it
     (for example Windows NT), but it may be used for any other purpose
     requiring periodic polling.  The standard version is null, and can
     be replaced by a user program. This will require re-compilation of
     the Ada.Exceptions package that can be found in files
     a-except.ads/adb.

     A standard alternative unit (called 4wexcpol.adb in the standard
     GNAT distribution) is used to enable the asynchronous abort
     capability on targets that do not normally support the capability.
     The version of Poll in this file makes a call to the appropriate
     runtime routine to test for an abort condition.

     Note that polling can also be enabled by use of the -gnatP switch.
     See the GNAT User's Guide for details.

`pragma Propagate_Exceptions'
     Syntax:

          pragma Propagate_Exceptions (subprogram_LOCAL_NAME);

     This pragma indicates that the given entity, which is the name of
     an imported foreign-langauge subprogram may receive an Ada
     exception, and that the exception should be propagated. It is
     relevant only if zero cost exception handling is in use, and is
     thus never needed if the alternative longjmp/setjmp implementation
     of exceptions is used (although it is harmless to use it in such
     cases).

     The implementation of fast exceptions always properly propagates
     exceptions through Ada code, as described in the Ada Reference
     Manual.  However, this manual is silent about the propagation of
     exceptions through foreign code. For example, consider the
     situation where `P1' calls `P2', and `P2' calls `P3', where `P1'
     and `P3' are in Ada, but `P2' is in C.  `P3' raises an Ada
     exception. The question is whether or not it will be propagated
     through `P2' and can be handled in `P1'.

     For the longjmp/setjmp implementation of exceptions, the answer is
     always yes. For some targets on which zero cost exception handling
     is implemented, the answer is also always yes. However, there are
     some targets, notably in the current version all x86 architecture
     targets, in which the answer is that such propagation does not
     happen automatically. If such propagation is required on these
     targets, it is mandatory to use `Propagate_Exceptions' to name all
     foreign language routines through which Ada exceptions may be
     propagated.

`pragma Psect_Object'
     Syntax:

          pragma Psect_Object
               [Internal =>] LOCAL_NAME,
            [, [External =>] EXTERNAL_SYMBOL]
            [, [Size     =>] EXTERNAL_SYMBOL]
          
          EXTERNAL_SYMBOL ::=
            IDENTIFIER
          | static_string_EXPRESSION

     This pragma is identical in effect to pragma `Common_Object'.

`Pure_Function'
     Syntax:

          pragma Pure_Function ([Entity =>] function_LOCAL_NAME);

     This pragma appears in the same declarative part as a function
     declaration (or a set of function declarations if more than one
     overloaded declaration exists, in which case the pragma applies to
     all entities). If specifies that the function `Entity' is to be
     considered pure for the purposes of code generation. This means
     that the compiler can assume that there are no side effects, and
     in particular that two calls with identical arguments produce the
     same result. It also means that the function can be used in an
     address clause.

     Note that, quite deliberately, there are no static checks to try
     to ensure that this promise is met, so PURE_FUNCTION can be used
     with functions that are conceptually pure, even if they do modify
     global variables. For example, a square root function that is
     instrumented to count the number of times it is called is still
     conceptually pure, and can still be optimized, even though it
     modifies a global variable (the count). Memo functions are another
     example (where a table of previous calls is kept and consulted to
     avoid re-computation).

     Note: All functions in a `Pure' package are automatically pure, and
     there is no need to use pragma `Pure_Function' in this case.

     Note: If pragma `Pure_Function' is applied to a renamed function,
     it applies to the underlying renamed function. This can be used to
     disambiguate cases of overloading where some but not all functions
     in a set of overloaded functions are to be designated as pure.

`pragma Ravenscar'
     Syntax:

          pragma Ravenscar

     A configuration pragma that establishes the following set of
     restrictions:

    `No_Abort_Statements'
          [RM D.7] There are no abort_statements, and there are no
          calls to Task_Identification.Abort_Task.

    `No_Select_Statements'
          There are no select_statements.

    `No_Task_Hierarchy'
          [RM D.7] All (non-environment) tasks depend directly on the
          environment task of the partition.

    `No_Task_Allocators'
          [RM D.7] There are no allocators for task types or types
          containing task subcomponents.

    `No_Dynamic_Priorities'
          [RM D.7] There are no semantic dependencies on the package
          Dynamic_Priorities.

    `No_Terminate_Alternatives'
          [RM D.7] There are no selective_accepts with
          terminate_alternatives

    `No_Nested_Finalization'
          [RM D.7] Objects with controlled parts access types that
          designate such objects shall be declared only at library
          level.

    `No_IO'
          [RM H.4] Semantic dependence on any of the library units
          Sequential_IO, Direct_IO, Text_IO, Wide_Text_IO, or Stream_IO
          is not allowed.

    `No_Streams'
          There are no semantic dependencies on Ada.Streams.

    `No_Exception_Handlers'
          There are no exception handlers. Raise statements are allowed.

    `No_Dynamic_Interrupts'
          There are no semantic dependencies on Ada.Interrupts.

    `No_Protected_Type_Allocators'
          There are no allocators for protected types or types
          containing protected subcomponents.

    `No_Local_Protected_Objects'
          Protected objects and access types that designate such
          objects shall be declared only at library level.

    `No_Requeue'
          Requeue statements are not allowed.

    `No_Calendar'
          There are no semantic dependencies on the package
          Ada.Calendar.

    `No_Relative_Delay'
          There are no delay_relative_statements.

    `No_Task_Attributes'
          There are no semantic dependencies on the Ada.Task_Attributes
          package and there are no references to the attributes
          Callable and Terminated [RM 9.9].

    `No_Enumeration_Maps'
          Similar to Discard_Names in C.5, but this is a restriction
          instead.  For enumeration types with representation clauses
          the following attributes are not allowed (for subtype S with
          an enumeration clause):

             * S'Pos

             * S'Val

             * S'Succ

             * S'Pred

             * S'Valid

          Also not allowed are FOR loops where the index is of type S,
          nor arrays which are indexed by type S.  Also not allowed are
          generic instantiations where the actual parameter is type S
          and the formal parameter is a discrete type.

    `Static_Storage_Size'
          The expression for pragma Storage_Size is static.

    `Boolean_Entry_Barriers'
          Entry barrier condition expressions shall be boolean objects
          which are declared in the protected type which contains the
          entry.

    `Max_Asynchronous_Select_Nesting =  0'
          [RM D.7] Specifies the maximum dynamic nesting level of
          asynchronous_selects.  A value of zero prevents the use of
          any asynchronous_select.

    `Max_Task_Entries =  0'
          [RM D.7] Specifies the maximum number of entries per task.
          The bounds of every entry family of a task unit shall be
          static, or shall be defined by a discriminant of a subtype
          whose corresponding bound is static.  A value of zero
          indicates that no rendezvous are possible.  For the Ravenscar
          pragma, the value of Max_Task_Entries is always 0 (zero).

    `Max_Protected_Entries = 1'
          [RM D.7] Specifies the maximum number of entries per
          protected type.  The bounds of every entry family of a
          protected unit shall be static, or shall be defined by a
          discriminant of a subtype whose corresponding bound is
          static.  For the Ravenscar pragma the value of
          Max_Protected_Entries is always 1.

    `Max_Select_Alternatives = 0'
          [RM D.7] Specifies the maximum number of alternatives in a
          selective_accept.  For the Ravenscar pragma the value if
          always 0.

    `No_Task_Termination'
          Tasks which terminate are erroneous.

    `No_Entry_Queue'
          No task can be queued on a protected entry. Note that this
          restrictions is checked at run time. The violation of this
          restriction generates a Program_Error exception.

     This set of restrictions corresponds to the definition of the
     "Ravenscar Profile" for limited tasking, devised and published by
     the International Workshop On Real Time Ada", 1997.

     The above set is a superset of the restrictions provided by pragma
     `Restricted_Run_Time', it includes eight additional restrictions
     (`Boolean_Entry_Barriers', `No_Select_Statements',
     `No_Exception_Handlers', `No_Calendar', `Static_Storage_Size',
     `No_Relative_Delay', `No_Enumeration_Maps', and
     `No_Task_Termination'). This means that pragma Ravenscar, like the
     pragma Restricted_Run_Time, automatically causes the use of a
     simplified, more efficient version of the tasking run-time system.

`pragma Restricted_Run_Time'
     Syntax:

          pragma Restricted_Run_Time

     A configuration pragma that establishes the following set of
     restrictions:

        * No_Abort_Statements

        * No_Entry_Queue

        * No_Task_Hierarchy

        * No_Task_Allocators

        * No_Dynamic_Priorities

        * No_Terminate_Alternatives

        * No_Nested_Finalization

        * No_IO

        * No_Streams

        * No_Dynamic_Interrupts

        * No_Protected_Type_Allocators

        * No_Local_Protected_Objects

        * No_Requeue

        * No_Task_Attributes

        * Max_Asynchronous_Select_Nesting =  0

        * Max_Task_Entries =  0

        * Max_Protected_Entries = 1

        * Max_Select_Alternatives = 0

     This set of restrictions causes the automatic selection of a
     simplified version of the run time that provides improved
     performance for the limited set of tasking functionality permitted
     by this set of restrictions.

`pragma Share_Generic'
     Syntax:

          pragma Share_Generic (NAME {, NAME});

     This pragma is recognized for compatibility with other Ada
     compilers but is ignored by GNAT. GNAT does not provide the
     capability for sharing of generic code. All generic instantiations
     result in making an inlined copy of the template with appropriate
     substitutions.

`pragma Source_File_Name'
     Syntax:

          pragma Source_File_Name (
            [Unit_Name   =>] unit_NAME,
            [FNAME_DESIG =>] static_string_EXPRESSION);
          
          FNAME_DESIG => Body_File_Name | Spec_File_Name

     Use this to override the normal naming convention. It is a
     configuration pragma, and so has the usual applicability of
     configuration pragmas (i.e. it applies to either an entire
     partition, or to all units in a compilation, or to a single unit,
     depending on how it is used.  UNIT_NAME is mapped to
     FILE_NAME_LITERAL. The identifier for the second argument is
     required, and indicates whether this is the file name for the spec
     or for the body.

`pragma Source_Reference'
     Syntax:

          pragma Source_Reference (INTEGER_LITERAL, STRING_LITERAL);

     This pragma typically appears as the first line of a source file.
     INTEGER_LITERAL is the logical line number of the line following
     the pragma line (for use in error messages and debugging
     information). STRING_LITERAL is a static string constant that
     specifies the file name to be used in error messages and debugging
     information. This is most notably used for the output of `gnatchop'
     with the `-r' switch, to make sure that the original unchopped
     source file is the one referred to.

     The second argument must be a string literal, it cannot be a static
     string expression other than a string literal. This is because its
     value is needed for error messages issued by all phases of the
     compiler.

`pragma Stream_Convert'
     Syntax:

          pragma Stream_Convert (
            [Entity =>] type_LOCAL_NAME,
            [Read   =>] function_NAME,
            [Write  =>] function NAME);

     This pragma provides an efficient way of providing stream
     functions for types defined in packages. Not only is it simpler to
     use than declaring the necessary functions with attribute
     representation clauses, but more significantly, it allows the
     declaration to made in such a way that the stream packages are not
     loaded unless they are needed. The use of the Stream_Convert
     pragma adds no overhead at all, unless the stream attributes are
     actually used on the designated type.

     The first argument specifies the type for which stream functions
     are provided. The second parameter provides a function used to
     read values of this type. It must name a function whose argument
     type may be any subtype, and whose returned type must be the type
     given as the first argument to the pragma.

     The meaning of the READ parameter is that if a stream attribute
     directly or indirectly specifies reading of the type given as the
     first parameter, then a value of the type given as the argument to
     the Read function is read from the stream, and then the Read
     function is used to convert this to the required target type.

     Similarly the WRITE parameter specifies how to treat write
     attributes that directly or indirectly apply to the type given as
     the first parameter.  It must have an input parameter of the type
     specified by the first parameter, and the return type must be the
     same as the input type of the Read function.  The effect is to
     first call the Write function to convert to the given stream type,
     and then write the result type to the stream.

     The Read and Write functions must not be overloaded subprograms.
     If necessary renamings can be supplied to meet this requirement.
     The usage of this attribute is best illustrated by a simple
     example, taken from the GNAT implementation of package
     Ada.Strings.Unbounded:

          function To_Unbounded (S : String) return Unbounded_String
            renames To_Unbounded_String;
          
          pragma Stream_Convert
            (Unbounded_String, To_Unbounded, To_String);

     The specifications of the referenced functions, as given in the
     Ada 95 Reference Manual are:

          function To_Unbounded_String (Source : String)
            return Unbounded_String;
          
          function To_String (Source : Unbounded_String)
            return String;

     The effect is that if the value of an unbounded string is written
     to a stream, then the representation of the item in the stream is
     in the same format used for `Standard.String', and this same
     representation is expected when a value of this type is read from
     the stream.

`pragma Subtitle'
     Syntax:

          pragma Subtitle ([Subtitle =>] STRING_LITERAL);

     This pragma is recognized for compatibility with other Ada
     compilers but is ignored by GNAT.

`pragma Suppress_All'
     Syntax:

          pragma Suppress_All;

     This pragma can only appear immediately following a compilation
     unit. The effect is to apply `Suppress (All_Checks)' to the unit
     which it follows. This pragma is implemented for compatibility
     with DEC Ada 83 usage. The use of pragma `Suppress (All_Checks)'
     as a normal configuration pragma is the preferred usage in GNAT.

`pragma Suppress_Initialization'
     Syntax:

          pragma Suppress_Initialization ([Entity =>] type_Name);

     This pragma suppresses any implicit or explicit initialization
     associated with the given type name for all variables of this type.

`pragma Task_Info'
     Syntax

          pragma Task_Info (EXPRESSION);

     This pragma appears within a task definition (like pragma
     `Priority') and applies to the task in which it appears. The
     argument must be of type `System.Task_Info.Task_Info_Type'.  The
     `Task_Info' pragma provides system dependent control over aspect
     of tasking implementation, for example, the ability to map tasks
     to specific processors. For details on the facilities available
     for the version of GNAT that you are using, see the documentation
     in the specification of package System.Task_Info in the runtime
     library.

`pragma Task_Storage'
     Syntax:

          pragma Task_Storage
            [Task_Type =>] LOCAL_NAME,
            [Top_Guard =>] static_integer_EXPRESSION);

     This pragma specifies the length of the guard area for tasks.  The
     guard area is an additional storage area allocated to a task.  A
     value of zero means that either no guard area is created or a
     minimal guard area is created, depending on the target. This
     pragma can appear anywhere a `Storage_Size' attribute definition
     clause is allowed for a task type.

`pragma Time_Slice'
     Syntax:

          pragma Time_Slice (static_duration_EXPRESSION);

     For implementations of GNAT on operating systems where it is
     possible to supply a time slice value, this pragma may be used for
     this purpose.  It is ignored if it is used in a system that does
     not allow this control, or if it appears in other than the main
     program unit.  Note that the effect of this pragma is identical to
     the effect of the DEC Ada 83 pragma of the same name when
     operating under OpenVMS systems.

`pragma Title'
     Syntax:

          pragma Title (TITLING_OPTION [, TITLING OPTION]);
          
          TITLING_OPTION ::=
            [Title    =>] STRING_LITERAL,
          | [Subtitle =>] STRING_LITERAL

     Syntax checked but otherwise ignored by GNAT. This is a listing
     control pragma used in DEC Ada 83 implementations to provide a
     title and/or subtitle for the program listing. The program listing
     generated by GNAT does not have titles or subtitles.

     Unlike other pragmas, the full flexibility of named notation is
     allowed for this pragma, i.e. the parameters may be given in any
     order if named notation is used, and named and positional notation
     can be mixed following the normal rules for procedure calls in Ada.

`pragma Unchecked_Union'
     Syntax:

          pragma Unchecked_Union (first_subtype_LOCAL_NAME)

     This pragma is used to declare that the specified type should be
     represented in a manner equivalent to a C union type, and is
     intended only for use in interfacing with C code that uses union
     types. In Ada terms, the named type must obey the following rules:

        * It is a non-tagged non-limited record type.

        * It has a single discrete discriminant with a default value.

        * The component list consists of a single variant part.

        * Each variant has a component list with a single component.

        * No nested variants are allowed.

        * No component has an explicit default value.

        * No component has a non-static constraint.

     In addition, given a type that meets the above requirements, the
     following restrictions apply to its use throughout the program:

        * The discriminant name can be mentioned only in an aggregate.

        * No subtypes may be created of this type.

        * The type may not be constrained by giving a discriminant
          value.

        * The type cannot be passed as the actual for a generic formal
          with a discriminant.

     Equality and inequality operations on `unchecked_unions' are not
     available, since there is no discriminant to compare and the
     compiler does not even know how many bits to compare. It is
     implementation dependent whether this is detected at compile time
     as an illegality or whether it is undetected and considered to be
     an erroneous construct. In GNAT, a direct comparison is illegal,
     but GNAT does not attempt to catch the composite case (where two
     composites are compared that contain an unchecked union
     component), so such comparisons are simply considered erroneous.

     The layout of the resulting type corresponds exactly to a C union,
     where each branch of the union corresponds to a single variant in
     the Ada record. The semantics of the Ada program is not changed in
     any way by the pragma, i.e. provided the above restrictions are
     followed, and no erroneous incorrect references to fields or
     erroneous comparisons occur, the semantics is exactly as described
     by the Ada reference manual.  Pragma `Suppress
     (Discriminant_Check)' applies implicitly to the type and the
     default convention is C

`pragma Unimplemented_Unit'
     Syntax:

          pragma Unimplemented_Unit;

     If this pragma occurs in a unit that is processed by the compiler,
     GNAT aborts with the message `XXX not implemented', where XXX is
     the name of the current compilation unit.  This pragma is intended
     to allow the compiler to handle unimplemented library units in a
     clean manner.

     The abort only happens if code is being generated.  Thus you can
     use specs of unimplemented packages in syntax or semantic checking
     mode.

`pragma Unreserve_All_Interrupts'
     Syntax:

          pragma Unreserve_All_Interrupts;

     Normally certain interrupts are reserved to the implementation.
     Any attempt to attach an interrupt causes Program_Error to be
     raised, as described in RM C.3.2(22). A typical example is the
     `SIGINT' interrupt used in many systems for an `Ctrl-C' interrupt.
     Normally this interrupt is reserved to the implementation, so that
     `Ctrl-C' can be used to interrupt execution.

     If the pragma Unreserve_All_Interrupts appears anywhere in any
     unit in a program, then all such interrupts are unreserved. This
     allows the program to handle these interrupts, but disables their
     standard functions. For example, if this pragma is used, then
     pressing `Ctrl-C' will not automatically interrupt execution.
     However, a program can then handle the `SIGINT' interrupt as it
     chooses.

     For a full list of the interrupts handled in a specific
     implementation, see the source code for the specification of
     Ada.Interrupts.Names in file s-intnam.ads. This is a target
     dependent file that contains the list of interrupts recognized for
     a given target. The documentation in this file also specifies what
     interrupts are affected by the use of the Unreserve_All_Interrupts
     pragma.

`pragma Unsuppress'
     Syntax:

          pragma Unsuppress (IDENTIFIER [, [On =>] NAME]);

     This pragma undoes the effect of a previous pragma `Suppress'. If
     there is no corresponding pragma `Suppress' in effect, it has no
     effect. The range of the effect is the same as for pragma
     `Suppress'. The meaning of the arguments is identical to that used
     in pragma `Suppress'.

     One important application is to ensure that checks are on in cases
     where code depends on the checks for its correct functioning, so
     that the code will compile correctly even if the compiler switches
     are set to suppress checks.

`pragma Use_VADS_Size'
     Syntax:

          pragma Use_VADS_Size;

     This is a configuration pragma. In a unit to which it applies, any
     use of the 'Size attribute is automatically interpreted as a use
     of the 'VADS_Size attribute. Note that this may result in
     incorrect semantic processing of valid Ada 95 programs. This is
     intended to aid in the handling of legacy code which depends on
     the interpretation of Size as implemented in the VADS compiler.
     See description of the VADS_Size attribute for further details.

`pragma Volatile'
     Syntax:

          pragma Volatile (local_NAME)

     This pragma is defined by the Ada 95 Reference Manual, and the GNAT
     implementation is fully conformant with this definition. The
     reason it is mentioned in this section is that a pragma of the
     same name was supplied in some Ada 83 compilers, including DEC Ada
     83. The Ada 95 implementation of pragma Volatile is upwards
     compatible with the implementation in Dec Ada 83.

`pragma Warnings'
     Syntax:

          pragma Warnings (On | Off [, LOCAL_NAME]);

     Normally warnings are enabled, with the output being controlled by
     the command line switch. Warnings (`Off') turns off generation of
     warnings until a Warnings (`On') is encountered or the end of the
     current unit. If generation of warnings is turned off using this
     pragma, then no warning messages are output, regardless of the
     setting of the command line switches.

     The form with a single argument is a configuration pragma.

     If the LOCAL_NAME parameter is present, warnings are suppressed for
     the specified entity. This suppression is effective from the point
     where it occurs till the end of the extended scope of the variable
     (similar to the scope of `Suppress').

`pragma Weak_External'
     Syntax:

          pragma Weak_External ([Entity =>] LOCAL_NAME);

     This pragma specifies that the given entity should be marked as a
     weak external (one that does not have to be resolved) for the
     linker. For further details, consult the GCC manual.

Implementation Defined Attributes
*********************************

   Ada 95 defines (throughout the Ada 95 reference manual, summarized
in annex K), a set of attributes that provide useful additional
functionality in all areas of the language. These language defined
attributes are implemented in GNAT and work as described in the Ada 95
Reference Manual.

   In addition, Ada 95 allows implementations to define additional
attributes whose meaning is defined by the implementation. GNAT provides
a number of these implementation-dependent attributes which can be used
to extend and enhance the functionality of the compiler. This section of
the GNAT reference manual describes these additional attributes.

   Note that any program using these attributes may not be portable to
other compilers (although GNAT implements this set of attributes on all
platforms). Therefore if portability to other compilers is an important
consideration, you should minimize the use of these attributes.

`Abort_Signal'
     `Standard'Abort_Signal' (`Standard' is the only allowed prefix)
     provides the entity for the special exception used to signal task
     abort or asynchronous transfer of control. Normally this attribute
     should only be used in the tasking runtime (it is highly peculiar,
     and completely outside the normal semantics of Ada, for a user
     program to intercept the abort exception).

`Address_Size'
     `Standard'Address_Size' (`Standard' is the only allowed prefix) is
     a static constant giving the number of bits in an `Address'. It is
     used primarily for constructing the definition of `Memory_Size' in
     package `Standard', but may be freely used in user programs.

`Asm_Input'
     The `Asm_Input' attribute denotes a function that takes two
     parameters.  The first is a string, the second is an expression of
     the type designated by the prefix. The first (string) argument is
     required to be a static expression, and is the constraint for the
     parameter, (e.g. what kind of register is required). The second
     argument is the value to be used as the input argument. The
     possible values for the constant are the same as those used in the
     RTL, and are dependent on the configuration file used to built the
     GCC back end.  *Note Machine Code Insertions::

`Asm_Output'
     The `Asm_Output' attribute denotes a function that takes two
     parameters.  The first is a string, the second is the name of a
     variable of the type designated by the attribute prefix. The first
     (string) argument is required to be a static expression and
     designates the constraint for the parameter (e.g. what kind of
     register is required). The second argument is the variable to be
     updated with the result. The possible values for constraint are
     the same as those used in the RTL, and are dependent on the
     configuration file used to build the GCC back end.  If there are
     no output operands, then this argument may either be omitted, or
     explicitly given as `No_Output_Operands'.  *Note Machine Code
     Insertions::

`AST_Entry'
     This attribute is implemented only in OpenVMS versions of GNAT.
     Applied to the name of an entry, it yields a value of the
     predefined type AST_Handler (declared in the predefined package
     System, as extended by the use of pragma Extend_System (Aux_DEC)).
     This value enables the given entry to be called when an AST
     occurs. For further details, refer to the DEC Ada Language
     Reference Manual, section 9.12a.

`Bit'
     `OBJ'Bit', where OBJ is any object, yields the bit offset within
     the storage unit (byte) that contains the first bit of storage
     allocated for the object. The value of this attribute is of the
     type `Universal_Integer', and is always a non-negative number not
     exceeding the value of `System.Storage_Unit'.

     For an object that is a variable or a constant allocated in a
     register, the value is zero. (The use of this attribute does not
     force the allocation of a variable to memory).

     For an object that is a formal parameter, this attribute applies
     to either the matching actual parameter or to a copy of the
     matching actual parameter.

     For an access object the value is zero. Note that `OBJ.all'Bit' is
     subject to an `Access_Check' for the designated object. Similarly
     for a record component `X.C'Bit' is subject to a discriminant
     check and `X(I).Bit' and `X(I1..I2)'Bit' are subject to index
     checks.

     This attribute is designed to be compatible with the DEC Ada 83
     definition and implementation of the `Bit' attribute.

`Bit_Position'
     `R.C'Bit', where R is a record object and C is one of the fields
     of the record type, yields the bit offset within the record
     contains the first bit of storage allocated for the object. The
     value of this attribute is of the type `Universal_Integer'. The
     value depends only on the field C and is independent of the
     alignment of the containing record R.

`Code_Address'
     The `'Address' attribute may be applied to subprograms in Ada 95,
     but the intended effect from the Ada 95 reference manual seems to
     be to provide an address value which can be used to call the
     subprogram by means of an address clause as in the following
     example:

          procedure K is ...
          
          procedure L;
          for L'Address use K'Address;
          pragma Import (Ada, L);

     A call to L is then expected to result in a call to K. In Ada 83,
     where there were no access-to-subprogram values, this was a common
     work around for getting the effect of an indirect call.  GNAT
     implements the above use of Address and the technique illustrated
     by the example code works correctly.

     However, for some purposes, it is useful to have the address of
     the start of the generated code for the subprogram. On some
     architectures, this is not necessarily the same as the Address
     value described above. For example, the Address value may
     reference a subprogram descriptor rather than the subprogram
     itself.

     The `'Code_Address' attribute, which can only be applied to
     subprogram entities, always returns the address of the start of the
     generated code of the specified subprogram, which may or may not be
     the same value as is returned by the corresponding `'Address'
     attribute.

`Default_Bit_Order'
     `Standard'Default_Bit_Order' (`Standard' is the only permissible
     prefix), provides the value `System.Default_Bit_Order' as a `Pos'
     value (0 for `High_Order_First', 1 for `Low_Order_First'). This is
     used to construct the definition of `Default_Bit_Order' in package
     `System'.

`Elaborated'
     The prefix of the `'Elaborated' attribute must be a unit name. The
     value is a Boolean which indicates whether or not the given unit
     has been elaborated. This attribute is primarily intended for
     internal use by the generated code for dynamic elaboration
     checking, but it can also be used in user programs. The value will
     always be True once elaboration of all units has been completed.

`Elab_Body'
     This attribute can only be applied to a program unit name. It
     returns the entity for the corresponding elaboration procedure for
     elaborating the body of the referenced unit. This is used in the
     main generated elaboration procedure by the binder and is not
     normally used in any other context.  However, there may be
     specialized situations in which it is useful to be able to call
     this elaboration procedure from Ada code, e.g. if it is necessary
     to do selective re-elaboration to fix some error.

`Elab_Spec'
     This attribute can only be applied to a program unit name. It
     returns the entity for the corresponding elaboration procedure for
     elaborating the specification of the referenced unit. This is used
     in the main generated elaboration procedure by the binder and is
     not normally used in any other context.  However, there may be
     specialized situations in which it is useful to be able to call
     this elaboration procedure from Ada code, e.g. if it is necessary
     to do selective re-elaboration to fix some error.

`Emax'
     The `Emax' attribute is provided for compatibility with Ada 83. See
     the Ada 83 reference manual for an exact description of the
     semantics of this attribute.

`Enum_Rep'
     For every enumeration subtype S, `S'Enum_Rep' denotes a function
     with the following specification:

          function S'Enum_Rep (Arg : S'Base) return Universal_Integer;

     It is also allowable to apply Enum_Rep directly to an object of an
     enumeration type or to a non-overloaded enumeration literal. In
     this case `S'Enum_Rep' is equivalent to `TYP'Enum_Rep(S)' where
     TYP is the type of the enumeration literal or object.

     The function returns the representation value for the given
     enumeration value. This will be equal to value of the `Pos'
     attribute in the absence of an enumeration representation clause.
     This is a static attribute (i.e. the result is static if the
     argument is static).

     S'Enum_Rep can also be used with integer types and objects, in
     which case it simply returns the integer value. The reason for
     this is to allow it to be used for (<>) discrete formal arguments
     in a generic unit that can be instantiated with either enumeration
     types or integer types.

`Epsilon'
     The `Epsilon' attribute is provided for compatibility with Ada 83.
     See the Ada 83 reference manual for an exact description of the
     semantics of this attribute.

`Fixed_Value'
     For every fixed-point type S, `S'Fixed_Value' denotes a function
     with the following specification:

          function S'Fixed_Value (Arg : Universal_Integer) return S;

     The value returned is the fixed-point value V such that

          V = Arg * S'Small

     The effect is thus equivalent to first converting the argument to
     the integer type used to represent S, and then doing an unchecked
     conversion to the fixed-point type. This attribute is primarily
     intended for use in implementation of the input-output functions
     for fixed-point values.

`Has_Discriminants'
     The prefix of the `Has_Discriminants' attribute is a type. The
     result is a Boolean value which is True if the type has
     discriminants, and False otherwise. The intended use of this
     attribute is in conjunction with generic definitions. If the
     attribute is applied to a generic private type, it indicates
     whether or not the corresponding actual type has discriminants.

`Img'
     The `Img' attribute differs from `Image' in that it may be applied
     to objects as well as types, in which case it gives the `Image'
     for the subtype of the object. This is convenient for debugging:

          Put_Line ("X = " & X'Img);

     has the same meaning as the more verbose:

          Put_Line ("X = " & TYPE'Image (X));

     where TYPE is the subtype of the object X.

`Integer_Value'
     For every integer type S, `S'Integer_Value' denotes a function
     with the following specification:

          function S'Integer_Value (Arg : Universal_Fixed) return S;

     The value returned is the integer value V, such that

          Arg = V * TYPE'Small

     The effect is thus equivalent to first doing an unchecked convert
     from the fixed-point type to its corresponding implementation
     type, and then converting the result to the target integer type.
     This attribute is primarily intended for use in implementation of
     the standard input-output functions for fixed-point values.

`Large'
     The `Large' attribute is provided for compatibility with Ada 83.
     See the Ada 83 reference manual for an exact description of the
     semantics of this attribute.

`Machine_Size'
     This attribute is identical to the `Object_Size' attribute. It is
     provided for compatibility with the DEC Ada 83 attribute of this
     name.

`Mantissa'
     The `Mantissa' attribute is provided for compatibility with Ada
     83. See the Ada 83 reference manual for an exact description of
     the semantics of this attribute.

`Max_Interrupt_Priority'
     `Standard'Max_Interrupt_Priority' (`Standard' is the only
     permissible prefix), provides the value
     `System.Max_Interrupt_Priority' and is intended primarily for
     constructing this definition in package `System'.

`Max_Priority'
     `Standard'Max_Priority' (`Standard' is the only permissible
     prefix) provides the value `System.Max_Priority' and is intended
     primarily for constructing this definition in package `System'.

`Maximum_Alignment'
     `Standard'Maximum_Alignment' (`Standard' is the only permissible
     prefix) provides the maximum useful alignment value for the
     target. This is a static value that can be used to specify the
     alignment for an object, guaranteeing that it is properly aligned
     in all cases. This is useful when an external object is imported
     and its alignment requirements are unknown.

`Mechanism_Code'
     `FUNCTION'Mechanism_Code' yields an integer code for the mechanism
     used for the result of function, and `SUBPROGRAM'Mechanism_Code
     (N)' yields the mechanism used for formal parameter number N (a
     static integer value with 1 meaning the first parameter) of
     SUBPROGRAM. The code returned is:

    1
          by copy (value)

    2
          by reference

    3
          by descriptor (default descriptor class)

    4
          by descriptor (UBS: unaligned bit string)

    5
          by descriptor (UBSB: aligned bit string with arbitrary bounds)

    6
          by descriptor (UBA: unaligned bit array)

    7
          by descriptor (S: string, also scalar access type parameter)

    8
          by descriptor (SB: string with arbitrary bounds)

    9
          by descriptor (A: contiguous array)

    10
          by descriptor (NCA: non-contiguous array)

     Values from 3-10 are only relevant to Digital OpenVMS
     implementations.

`Null_Parameter'
     A reference `T'Null_Parameter' denotes an imaginary object of type
     or subtype T allocated at machine address zero. The attribute is
     allowed only as the default expression of a formal parameter, or as
     an actual expression of a subprogram call. In either case, the
     subprogram must be imported.

     The identity of the object is represented by the address zero in
     the argument list, independent of the passing mechanism (explicit
     or default).

     This capability is needed to specify that a zero address should be
     passed for a record or other composite object passed by reference.
     There is no way of indicating this without the `Null_Parameter'
     attribute.

`Object_Size'
     The size of an object is not necessarily the same as the size of
     the type of an object. This is because by default object sizes are
     increased to be a multiple of the alignment of the object. For
     example, `Natural'Size' is 31, but by default objects of type
     `Natural' will have a size of 32 bits.  Similarly, a record
     containing an integer and a character:

          type Rec is record
             I : Integer;
             C : Character;
          end record;

     will have a size of 40 (that is `Rec'Size' will be 40. The
     alignment will be 4, because of the integer field, and so the
     default size of record objects for this type will be 64 (8 bytes).

     The `TYPE'Object_Size' attribute has been added to GNAT to allow
     the default object size of a type to be easily determined. For
     example, `Natural'Object_Size' is 32, and `Rec'Object_Size' (for
     the record type in the above example) will be 64. Note also that,
     unlike the situation with the `Size' attribute as defined in the
     Ada RM, the `Object_Size' attribute can be specified individually
     for different subtypes. For example:

          type R is new Integer;
          subtype R1 is R range 1 .. 10;
          subtype R2 is R range 1 .. 10;
          for R2'Object_Size use 8;

     In this example, `R'Object_Size' and `R1'Object_Size' are both 32
     since the default object size for a subtype is the same as the
     object size for the the parent subtype. This means that objects of
     type `R' or `R1' will by default be 32 bits (four bytes). But
     objects of type `R2' will be only 8 bits (one byte), since
     `R2'Object_Size' has been set to 8.

`Passed_By_Reference'
     `TYPE'Passed_By_Reference' for any subtype TYPE returns a value of
     type `Boolean' value that is `True' if the type is normally passed
     by reference and `False' if the type is normally passed by copy in
     calls. For scalar types, the result is always `False' and is
     static. For non-scalar types, the result is non-static.

`Range_Length'
     `TYPE'Range_Length' for any discrete type TYPE yields the number
     of values represented by the subtype (zero for a null range). The
     result is static for static subtypes. `Range_Length' applied to
     the index subtype of a one dimensional array always gives the same
     result as `Range' applied to the array itself.

`Safe_Emax'
     The `Safe_Emax' attribute is provided for compatibility with Ada
     83. See the Ada 83 reference manual for an exact description of
     the semantics of this attribute.

`Safe_Large'
     The `Safe_Large' attribute is provided for compatibility with Ada
     83. See the Ada 83 reference manual for an exact description of
     the semantics of this attribute.

`Safe_Large'
     The `Safe_Large' attribute is provided for compatibility with Ada
     83. See the Ada 83 reference manual for an exact description of
     the semantics of this attribute.

`Small'
     The `Small' attribute is defined in Ada 95 only for fixed-point
     types.  GNAT also allows this attribute to be applied to
     floating-point types for compatibility with Ada 83. See the Ada 83
     reference manual for an exact description of the semantics of this
     attribute when applied to floating-point types.

`Storage_Unit'
     `Standard'Storage_Unit' (`Standard' is the only permissible
     prefix) provides the value `System.Storage_Unit' and is intended
     primarily for constructing this definition in package `System'.

`Tick'
     `Standard'Tick' (`Standard' is the only permissible prefix)
     provides the value of `System.Tick' and is intended primarily for
     constructing this definition in package `System'.

`To_Address'
     The `System'To_Address' (`System' is the only permissible prefix)
     denotes a function identical to
     `System.Storage_Elements.To_Address' except that it is a static
     attribute. This means that if its argument is a static expression,
     then the result of the attribute is a static expression. The
     result is that such an expression can be used in contexts (e.g.
     preelaborable packages) which require a static expression and
     where the function call could not be used (since the function call
     is always non-static, even if its argument is static).

`Type_Class'
     `TYPE'Type_Class' for any type or subtype TYPE yields the value of
     the type class for the full type of TYPE. If TYPE is a generic
     formal type, the value is the value for the corresponding actual
     subtype.  The value of this attribute is of type
     `System.Aux_DEC.Type_Class', which has the following definition:

            type Type_Class is
              (Type_Class_Enumeration,
               Type_Class_Integer,
               Type_Class_Fixed_Point,
               Type_Class_Floating_Point,
               Type_Class_Array,
               Type_Class_Record,
               Type_Class_Access,
               Type_Class_Task,
               Type_Class_Address);

     Protected types yield the value `Type_Class_Task', which thus
     applies to all concurrent types. This attribute is designed to be
     compatible with the DEC Ada 83 attribute of the same name.

`UET_Address'
     The `UET_Address' attribute can only be used for a prefix which
     denotes a library package. It yields the address of the unit
     exception table when zero cost exception handling is used. This
     attribute is intended only for use within the GNAT implementation.
     See the unit `Ada.Exceptions' in files `a-except.ads,a-except.adb'
     for details on how this attribute is used in the implementation.

`Universal_Literal_String'
     The prefix of `Universal_Literal_String' must be a named number.
     The static result is the string consisting of the characters of
     the number as defined in the original source. This allows the user
     program to access the actual text of named numbers without
     intermediate conversions and without the need to enclose the
     strings in quotes (which would preclude their use as numbers).
     This is used internally for the construction of values of the
     floating-point attributes from the file `ttypef.ads', but may also
     be used by user programs.

`Unrestricted_Access'
     The `Unrestricted_Access' attribute is similar to `Access' except
     that all accessibility and aliased view checks are omitted. This
     is a user-beware attribute.  It is similar to `Address', for which
     it is a desirable replacement where the value desired is an access
     type. In other words, its effect is identical to first applying
     the `Address' attribute and then doing an unchecked conversion to
     a desired access type. In GNAT, but not necessarily in other
     implementations, the use of static chains for inner level
     subprograms means that `Unrestricted_Access' applied to a
     subprogram yields a value that can be called as long as the
     subprogram is in scope (normal Ada 95 accessibility rules restrict
     this usage).

`VADS_Size'
     The `'VADS_Size' attribute is intended to make it easier to port
     legacy code which relies on the semantics of `'Size' as implemented
     by the VADS Ada 83 compiler. GNAT makes a best effort at
     duplicating the same semantic interpretation. In particular,
     `'VADS_Size' applied to a predefined or other primitive type with
     no Size clause yields the Object_Size (for example, `Natural'Size'
     is 32 rather than 31 on typical machines). In addition
     `'VADS_Size' applied to an object gives the result that would be
     obtained by applying the attribute to the corresponding type.

`Value_Size'
     `TYPE'Value_Size' is the number of bits required to represent a
     value of the given subtype. It is the same as `TYPE'Size', but,
     unlike `Size', may be set for non-first subtypes.

`Word_Size'
     `Standard'Word_Size' (`Standard' is the only permissible prefix)
     provides the value `System.Word_Size' and is intended primarily
     for constructing this definition in package `System'.

Implementation Advice
*********************

   The main text of the Ada 95 Reference Manual describes the required
behavior of all Ada 95 compilers, and the GNAT compiler conforms to
these requirements.

   In addition, there are sections throughout the Ada 95 reference
manual headed by the phrase "implementation advice". These sections are
not normative, i.e. they do not specify requirements that all compilers
must follow. Rather they provide advice on generally desirable
behavior. You may wonder why they are not requirements. The most
typical answer is that they describe behavior that seems generally
desirable, but cannot be provided on all systems, or which may be
undesirable on some systems.

   As far as practical, GNAT follows the implementation advice sections
in the Ada 95 Reference Manual. This chapter contains a table giving the
reference manual section number, paragraph number and several keywords
for each advice.  Each entry consists of the text of the advice followed
by the GNAT interpretation of this advice. Most often, this simply says
"followed", which means that GNAT follows the advice. However, in a
number of cases, GNAT deliberately deviates from this advice, in which
case the text describes what GNAT does and why.

*1.1.3(20): Error Detection*
     If an implementation detects the use of an unsupported Specialized
     Needs Annex feature at run time, it should raise `Program_Error' if
     feasible.
     Not relevant. All specialized needs annex features are either
     supported, or diagnosed at compile time.

*1.1.3(31): Child Units*
     If an implementation wishes to provide implementation-defined
     extensions to the functionality of a language-defined library
     unit, it should normally do so by adding children to the library
     unit.
     Followed.

*1.1.5(12): Bounded Errors*
     If an implementation detects a bounded error or erroneous
     execution, it should raise `Program_Error'.
     Followed in all cases in which the implementation detects a bounded
     error or erroneous execution. Not all such situations are detected
     at runtime.

*2.8(16): Pragmas*
     Normally, implementation-defined pragmas should have no semantic
     effect for error-free programs; that is, if the
     implementation-defined pragmas are removed from a working program,
     the program should still be legal, and should still have the same
     semantics.
     The following implementation defined pragmas are exceptions to this
     rule:

    `Abort_Defer'
          Affects semantics

    `Ada_83'
          Affects legality

    `Assert'
          Affects semantics

    `CPP_Class'
          Affects semantics

    `CPP_Constructor'
          Affects semantics

    `CPP_Destructor'
          Affects semantics

    `CPP_Virtual'
          Affects semantics

    `CPP_Vtable'
          Affects semantics

    `Debug'
          Affects semantics

    `Interface_Name'
          Affects semantics

    `Machine_Attribute'
          Affects semantics

    `Unimplemented_Unit'
          Affects legality

    `Unchecked_Union'
          Affects semantics

     In each of the above cases, it is essential to the purpose of the
     pragma that this advice not be followed. For details see the
     separate section on implementation defined pragmas.

*2.8(17-19): Pragmas*
     Normally, an implementation should not define pragmas that can
     make an illegal program legal, except as follows:

     A pragma used to complete a declaration, such as a pragma `Import';

     A pragma used to configure the environment by adding, removing, or
     replacing `library_items'.
     See response to paragraph 16 of this same section.

*3.5.2(5): Alternative Character Sets*
     If an implementation supports a mode with alternative
     interpretations for `Character' and `Wide_Character', the set of
     graphic characters of `Character' should nevertheless remain a
     proper subset of the set of graphic characters of
     `Wide_Character'. Any character set "localizations" should be
     reflected in the results of the subprograms defined in the
     language-defined package `Characters.Handling' (see A.3) available
     in such a mode. In a mode with an alternative interpretation of
     `Character', the implementation should also support a
     corresponding change in what is a legal `identifier_letter'.
     Not all wide character modes follow this advice, in particular the
     JIS and IEC modes reflect standard usage in Japan, and in these
     encoding, the upper half of the Latin-1 set is not part of the
     wide-character subset, since the most significant bit is used for
     wide character encoding. However, this only applies to the
     external forms. Internally there is no such restriction.

*3.5.4(28): Integer Types*
     An implementation should support `Long_Integer' in addition to
     `Integer' if the target machine supports 32-bit (or longer)
     arithmetic. No other named integer subtypes are recommended for
     package `Standard'. Instead, appropriate named integer subtypes
     should be provided in the library package `Interfaces' (see B.2).
     `Long_Integer' is supported. Other standard integer types are
     supported so this advice is not fully followed. These types are
     supported for convenient interface to C, and so that all hardware
     types of the machine are easily available.

*3.5.4(29): Integer Types*
     An implementation for a two's complement machine should support
     modular types with a binary modulus up to `System.Max_Int*2+2'. An
     implementation should support a non-binary modules up to
     `Integer'Last'.
     Followed.

*3.5.5(8): Enumeration Values*
     For the evaluation of a call on `S'Pos' for an enumeration
     subtype, if the value of the operand does not correspond to the
     internal code for any enumeration literal of its type (perhaps due
     to an un-initialized variable), then the implementation should
     raise `Program_Error'. This is particularly important for
     enumeration types with noncontiguous internal codes specified by an
     enumeration_representation_clause.
     Followed.

*3.5.7(17): Float Types*
     An implementation should support `Long_Float' in addition to
     `Float' if the target machine supports 11 or more digits of
     precision. No other named floating point subtypes are recommended
     for package `Standard'. Instead, appropriate named floating point
     subtypes should be provided in the library package `Interfaces'
     (see B.2).
     `Short_Float' and `Long_Long_Float' are also provided. The former
     provides improved compatibility with other implementations
     supporting this type. The latter corresponds to the highest
     precision floating-point type supported by the hardware. On most
     machines, this will be the same as `Long_Float', but on some
     machines, it will correspond to the IEEE extended form. On the
     Silicon Graphics processors, which do not support IEEE extended
     form, `Long_Long_Float' is the same as `Long_Float'.

*3.6.2(11): Multidimensional Arrays*
     An implementation should normally represent multidimensional
     arrays in row-major order, consistent with the notation used for
     multidimensional array aggregates (see 4.3.3). However, if a
     pragma `Convention' (`Fortran', ...) applies to a multidimensional
     array type, then column-major order should be used instead (see
     B.5, "Interfacing with Fortran").
     Followed.

*9.6(30-31): Duration'Small*
     Whenever possible in an implementation, the value of
     `Duration'Small' should be no greater than 100 microseconds.
     Followed. (`Duration'Small' = 10**(-9)).

     The time base for `delay_relative_statements' should be monotonic;
     it need not be the same time base as used for `Calendar.Clock'.
     Followed.

*10.2.1(12): Consistent Representation*
     In an implementation, a type declared in a pre-elaborated package
     should have the same representation in every elaboration of a
     given version of the package, whether the elaborations occur in
     distinct executions of the same program, or in executions of
     distinct programs or partitions that include the given version.
     Followed, except in the case of tagged types. Tagged types involve
     implicit pointers to a local copy of a dispatch table, and these
     pointers have representations which thus depend on a particular
     elaboration of the package. It is not easy to see how it would be
     possible to follow this advice without severely impacting
     efficiency of execution.

*11.4.1(19): Exception Information*
     `Exception_Message' by default and `Exception_Information' should
     produce information useful for debugging. `Exception_Message'
     should be short, about one line. `Exception_Information' can be
     long. `Exception_Message' should not include the `Exception_Name'.
     `Exception_Information' should include both the `Exception_Name'
     and the `Exception_Message'.
     Followed. For each exception that doesn't have a specified
     `Exception_Message', the compiler generates one containing the
     location of the raise statement. This location has the form
     "file:line", where file is the short file name (without path
     information) and line is the line number in the file. Note that in
     the case of the Zero Cost Exception mechanism, these messages
     become redundant with the Exception_Information that contains a
     full backtrace of the calling sequence, so they are disabled.  To
     disable explicitly the generation of the source location message,
     use the Pragma `Discard_Names'.

*11.5(28): Suppression of Checks*
     The implementation should minimize the code executed for checks
     that have been suppressed.
     Followed.

*13.1 (21-24): Representation Clauses*
     The recommended level of support for all representation items is
     qualified as follows:

     An implementation need not support representation items containing
     non-static expressions, except that an implementation should
     support a representation item for a given entity if each
     non-static expression in the representation item is a name that
     statically denotes a constant declared before the entity.
     Followed. GNAT does not support non-static expressions in
     representation clauses unless they are constants declared before
     the entity. For example:

               X : typ;
               for X'Address use To_address (16#2000#);

     will be rejected, since the To_Address expression is non-static.
     Instead write:

               X_Address : constant Address : =
               To_Address    ((16#2000#);
               X : typ;
               for X'Address use X_Address;

     An implementation need not support a specification for the `Size'
     for a given composite subtype, nor the size or storage place for an
     object (including a component) of a given composite subtype,
     unless the constraints on the subtype and its composite
     subcomponents (if any) are all static constraints.
     Followed. Size Clauses are not permitted on non-static components,
     as described above.

     An aliased component, or a component whose type is by-reference,
     should always be allocated at an addressable location.
     Followed.

*13.2(6-8): Packed Types*
     If a type is packed, then the implementation should try to minimize
     storage allocated to objects of the type, possibly at the expense
     of speed of accessing components, subject to reasonable complexity
     in addressing calculations.

     The recommended level of support pragma `Pack' is:

     For a packed record type, the components should be packed as
     tightly as possible subject to the Sizes of the component
     subtypes, and subject to any `record_representation_clause' that
     applies to the type; the implementation may, but need not, reorder
     components or cross aligned word boundaries to improve the
     packing. A component whose `Size' is greater than the word size
     may be allocated an integral number of words.
     Followed. Tight packing of arrays is supported for all component
     sizes up to 32-bits, which is the word size on typical
     implementations of GNAT.

     An implementation should support Address clauses for imported
     subprograms.
     Followed.

*13.3(14-19): Address Clauses*
     For an array X, `X'Address' should point at the first component of
     the array, and not at the array bounds.
     Followed.

     The recommended level of support for the `Address' attribute is:

     `X'Address' should produce a useful result if X is an object that
     is aliased or of a by-reference type, or is an entity whose
     `Address' has been specified.
     Followed.  A valid address will be produced even if none of those
     conditions have been met.  If necessary, the object is forced into
     memory to ensure the address is valid.

     An implementation should support `Address' clauses for imported
     subprograms.
     Followed.

     Objects (including subcomponents) that are aliased or of a
     by-reference type should be allocated on storage element
     boundaries.
     Followed.

     If the `Address' of an object is specified, or it is imported or
     exported, then the implementation should not perform optimizations
     based on assumptions of no aliases.
     Followed.

*13.3(29-35): Alignment Clauses*
     The recommended level of support for the `Alignment' attribute for
     subtypes is:

     An implementation should support specified Alignments that are
     factors and multiples of the number of storage elements per word,
     subject to the following:
     Followed.

     An implementation need not support specified `Alignment's for
     combinations of `Size's and `Alignment's that cannot be easily
     loaded and stored by available machine instructions.
     Followed.

     An implementation need not support specified `Alignment's that are
     greater than the maximum `Alignment' the implementation ever
     returns by default.
     Followed.

     The recommended level of support for the `Alignment' attribute for
     objects is:

     Same as above, for subtypes, but in addition:
     Followed.

     For stand-alone library-level objects of statically constrained
     subtypes, the implementation should support all `Alignment's
     supported by the target linker. For example, page alignment is
     likely to be supported for such objects, but not for subtypes.
     Followed.

*13.3(42-43): Size Clauses*
     The recommended level of support for the `Size' attribute of
     objects is:

     A `Size' clause should be supported for an object if the specified
     `Size' is at least as large as its subtype's `Size', and
     corresponds to a size in storage elements that is a multiple of the
     object's `Alignment' (if the `Alignment' is nonzero).
     Followed.

*13.3(50-56): Size Clauses*
     If the `Size' of a subtype is specified, and allows for efficient
     independent addressability (see 9.10) on the target architecture,
     then the `Size' of the following objects of the subtype should
     equal the `Size' of the subtype:

     Aliased objects (including components).
     Followed.

     `Size' clause on a composite subtype should not affect the
     internal layout of components.
     Followed.

     The recommended level of support for the `Size' attribute of
     subtypes is:

     The `Size' (if not specified) of a static discrete or fixed point
     subtype should be the number of bits needed to represent each value
     belonging to the subtype using an unbiased representation, leaving
     space for a sign bit only if the subtype contains negative values.
     If such a subtype is a first subtype, then an implementation
     should support a specified `Size' for it that reflects this
     representation.
     Followed.

     For a subtype implemented with levels of indirection, the `Size'
     should include the size of the pointers, but not the size of what
     they point at.
     Followed.

*13.3(71-73): Component Size Clauses*
     The recommended level of support for the `Component_Size'
     attribute is:

     An implementation need not support specified `Component_Sizes'
     that are less than the `Size' of the component subtype.
     Followed.

     An implementation should support specified `Component_Size's that
     are factors and multiples of the word size. For such
     `Component_Size's, the array should contain no gaps between
     components. For other `Component_Size's (if supported), the array
     should contain no gaps between components when packing is also
     specified; the implementation should forbid this combination in
     cases where it cannot support a no-gaps representation.
     Followed.

*13.4(9-10): Enumeration Representation Clauses*
     The recommended level of support for enumeration representation
     clauses is:

     An implementation need not support enumeration representation
     clauses for boolean types, but should at minimum support the
     internal codes in the range `System.Min_Int.System.Max_Int'.
     Followed.

*13.5.1(17-22): Record Representation Clauses*
     The recommended level of support for
     `record_representation_clauses' is:

     An implementation should support storage places that can be
     extracted with a load, mask, shift sequence of machine code, and
     set with a load, shift, mask, store sequence, given the available
     machine instructions and run-time model.
     Followed.

     A storage place should be supported if its size is equal to the
     `Size' of the component subtype, and it starts and ends on a
     boundary that obeys the `Alignment' of the component subtype.
     Followed.

     If the default bit ordering applies to the declaration of a given
     type, then for a component whose subtype's `Size' is less than the
     word size, any storage place that does not cross an aligned word
     boundary should be supported.
     Followed.

     An implementation may reserve a storage place for the tag field of
     a tagged type, and disallow other components from overlapping that
     place.
     Followed. The storage place for the tag field is the beginning of
     the tagged record, and its size is Address'Size. GNAT will reject
     an explicit component clause for the tag field.

     An implementation need not support a `component_clause' for a
     component of an extension part if the storage place is not after
     the storage places of all components of the parent type, whether
     or not those storage places had been specified.
     Followed. The above advice on record representation clauses is
     followed, and all mentioned features are implemented.

*13.5.2(5): Storage Place Attributes*
     If a component is represented using some form of pointer (such as
     an offset) to the actual data of the component, and this data is
     contiguous with the rest of the object, then the storage place
     attributes should reflect the place of the actual data, not the
     pointer. If a component is allocated discontinuously from the rest
     of the object, then a warning should be generated upon reference
     to one of its storage place attributes.
     Followed. There are no such components in GNAT.

*13.5.3(7-8): Bit Ordering*
     The recommended level of support for the non-default bit ordering
     is:

     If `Word_Size' = `Storage_Unit', then the implementation should
     support the non-default bit ordering in addition to the default
     bit ordering.
     Followed. Word size does not equal storage size in this
     implementation.  Thus non-default bit ordering is not supported.

*13.7(37): Address as Private*
     `Address' should be of a private type.
     Followed.

*13.7.1(16): Address Operations*
     Operations in `System' and its children should reflect the target
     environment semantics as closely as is reasonable. For example, on
     most machines, it makes sense for address arithmetic to "wrap
     around."  Operations that do not make sense should raise
     `Program_Error'.
     Followed. Address arithmetic is modular arithmetic that wraps
     around. No operation raises `Program_Error', since all operations
     make sense.

*13.9(14-17): Unchecked Conversion*
     The `Size' of an array object should not include its bounds; hence,
     the bounds should not be part of the converted data.
     Followed.

     The implementation should not generate unnecessary run-time checks
     to ensure that the representation of S is a representation of the
     target type. It should take advantage of the permission to return
     by reference when possible. Restrictions on unchecked conversions
     should be avoided unless required by the target environment.
     Followed. There are no restrictions on unchecked conversion. A
     warning is generated if the source and target types do not have
     the same size since the semantics in this case may be target
     dependent.

     The recommended level of support for unchecked conversions is:

     Unchecked conversions should be supported and should be reversible
     in the cases where this clause defines the result. To enable
     meaningful use of unchecked conversion, a contiguous
     representation should be used for elementary subtypes, for
     statically constrained array subtypes whose component subtype is
     one of the subtypes described in this paragraph, and for record
     subtypes without discriminants whose component subtypes are
     described in this paragraph.
     Followed.

*13.11(23-25): Implicit Heap Usage*
     An implementation should document any cases in which it dynamically
     allocates heap storage for a purpose other than the evaluation of
     an allocator.
     Followed, the only other points at which heap storage is
     dynamically allocated are as follows:

        * At initial elaboration time, to allocate dynamically sized
          global objects.

        * To allocate space for a task when a task is created.

        * To extend the secondary stack dynamically when needed. The
          secondary stack is used for returning variable length results.

     A default (implementation-provided) storage pool for an access-to-
     constant type should not have overhead to support de-allocation of
     individual objects.
     Followed.

     A storage pool for an anonymous access type should be created at
     the point of an allocator for the type, and be reclaimed when the
     designated object becomes inaccessible.
     Followed.

*13.11.2(17): Unchecked De-allocation*
     For a standard storage pool, `Free' should actually reclaim the
     storage.
     Followed.

*13.13.2(17): Stream Oriented Attributes*
     If a stream element is the same size as a storage element, then the
     normal in-memory representation should be used by `Read' and
     `Write' for scalar objects. Otherwise, `Read' and `Write' should
     use the smallest number of stream elements needed to represent all
     values in the base range of the scalar type.
     Followed.

*A.1(52): Implementation Advice*
     If an implementation provides additional named predefined integer
     types, then the names should end with `Integer' as in
     `Long_Integer'. If an implementation provides additional named
     predefined floating point types, then the names should end with
     `Float' as in `Long_Float'.
     Followed.

*A.3.2(49): `Ada.Characters.Handling'*
     If an implementation provides a localized definition of `Character'
     or `Wide_Character', then the effects of the subprograms in
     `Characters.Handling' should reflect the localizations. See also
     3.5.2.
     Followed. GNAT provides no such localized definitions.

*A.4.4(106): Bounded-Length String Handling*
     Bounded string objects should not be implemented by implicit
     pointers and dynamic allocation.
     Followed. No implicit pointers or dynamic allocation are used.

*A.5.2(46-47): Random Number Generation*
     Any storage associated with an object of type `Generator' should be
     reclaimed on exit from the scope of the object.
     Followed.

     If the generator period is sufficiently long in relation to the
     number of distinct initiator values, then each possible value of
     `Initiator' passed to `Reset' should initiate a sequence of random
     numbers that does not, in a practical sense, overlap the sequence
     initiated by any other value. If this is not possible, then the
     mapping between initiator values and generator states should be a
     rapidly varying function of the initiator value.
     Followed. The generator period is sufficiently long for the first
     condition here to hold true.

*A.10.7(23): `Get_Immediate'*
     The `Get_Immediate' procedures should be implemented with
     unbuffered input. For a device such as a keyboard, input should be
     "available" if a key has already been typed, whereas for a disk
     file, input should always be available except at end of file. For
     a file associated with a keyboard-like device, any line-editing
     features of the underlying operating system should be disabled
     during the execution of `Get_Immediate'.
     Followed.

*B.1(39-41): Pragma `Export'*
     If an implementation supports pragma `Export' to a given language,
     then it should also allow the main subprogram to be written in that
     language. It should support some mechanism for invoking the
     elaboration of the Ada library units included in the system, and
     for invoking the finalization of the environment task. On typical
     systems, the recommended mechanism is to provide two subprograms
     whose link names are `adainit' and `adafinal'. `adainit' should
     contain the elaboration code for library units. `adafinal' should
     contain the finalization code. These subprograms should have no
     effect the second and subsequent time they are called.
     Followed.

     Automatic elaboration of pre-elaborated packages should be
     provided when pragma Export is supported.
     Followed when the main program is in Ada. If the main program is
     in a foreign language, then `adainit' must be called to elaborate
     pre-elaborated packages.

     For each supported convention L other than `Intrinsic', an
     implementation should support `Import' and `Export' pragmas for
     objects of L-compatible types and for subprograms, and pragma
     `Convention' for L-eligible types and for subprograms, presuming
     the other language has corresponding features. Pragma `Convention'
     need not be supported for scalar types.
     Followed.

*B.2(12-13): Package `Interfaces'*
     For each implementation-defined convention identifier, there
     should be a child package of package Interfaces with the
     corresponding name. This package should contain any declarations
     that would be useful for interfacing to the language
     (implementation) represented by the convention. Any declarations
     useful for interfacing to any language on the given hardware
     architecture should be provided directly in `Interfaces'.
     Followed. An additional package not defined in the Ada 95
     Reference Manual is `Interfaces.CPP', used for interfacing to C++.

     An implementation supporting an interface to C, COBOL, or Fortran
     should provide the corresponding package or packages described in
     the following clauses.
     Followed. GNAT provides all the packages described in this section.

*B.3(63-71): Interfacing with C*
     An implementation should support the following interface
     correspondences between Ada and C.
     Followed.

     An Ada procedure corresponds to a void-returning C function.
     Followed.

     An Ada function corresponds to a non-void C function.
     Followed.

     An Ada `in' scalar parameter is passed as a scalar argument to a C
     function.
     Followed.

     An Ada `in' parameter of an access-to-object type with designated
     type T is passed as a `T*' argument to a C function, where T is
     the C type corresponding to the Ada type T.
     Followed.

     An Ada access T parameter, or an Ada `out' or `in out' parameter
     of an elementary type T, is passed as a `T*' argument to a C
     function, where T is the C type corresponding to the Ada type T.
     In the case of an elementary `out' or `in out' parameter, a
     pointer to a temporary copy is used to preserve by-copy semantics.
     Followed.

     An Ada parameter of a record type T, of any mode, is passed as a
     `T*' argument to a C function, where T is the C structure
     corresponding to the Ada type T.
     Followed. This convention may be overridden by the use of the
     C_Pass_By_Copy pragma, or Convention, or by explicitly specifying
     the mechanism for a given call using an extended import or export
     pragma.

     An Ada parameter of an array type with component type T, of any
     mode, is passed as a `T*' argument to a C function, where T is the
     C type corresponding to the Ada type T.
     Followed.

     An Ada parameter of an access-to-subprogram type is passed as a
     pointer to a C function whose prototype corresponds to the
     designated subprogram's specification.
     Followed.

*B.4(95-98): Interfacing with COBOL*
     An Ada implementation should support the following interface
     correspondences between Ada and COBOL.
     Followed.

     An Ada access T parameter is passed as a "BY REFERENCE" data item
     of the COBOL type corresponding to T.
     Followed.

     An Ada in scalar parameter is passed as a "BY CONTENT" data item of
     the corresponding COBOL type.
     Followed.

     Any other Ada parameter is passed as a "BY REFERENCE" data item of
     the COBOL type corresponding to the Ada parameter type; for
     scalars, a local copy is used if necessary to ensure by-copy
     semantics.
     Followed.

*B.5(22-26): Interfacing with Fortran*
     An Ada implementation should support the following interface
     correspondences between Ada and Fortran:
     Followed.

     An Ada procedure corresponds to a Fortran subroutine.
     Followed.

     An Ada function corresponds to a Fortran function.
     Followed.

     An Ada parameter of an elementary, array, or record type T is
     passed as a T argument to a Fortran procedure, where T is the
     Fortran type corresponding to the Ada type T, and where the INTENT
     attribute of the corresponding dummy argument matches the Ada
     formal parameter mode; the Fortran implementation's parameter
     passing conventions are used. For elementary types, a local copy
     is used if necessary to ensure by-copy semantics.
     Followed.

     An Ada parameter of an access-to-subprogram type is passed as a
     reference to a Fortran procedure whose interface corresponds to the
     designated subprogram's specification.
     Followed.

*C.1(3-5): Access to Machine Operations*
     The machine code or intrinsic support should allow access to all
     operations normally available to assembly language programmers for
     the target environment, including privileged instructions, if any.
     Followed.

     The interfacing pragmas (see Annex B) should support interface to
     assembler; the default assembler should be associated with the
     convention identifier `Assembler'.
     Followed.

     If an entity is exported to assembly language, then the
     implementation should allocate it at an addressable location, and
     should ensure that it is retained by the linking process, even if
     not otherwise referenced from the Ada code. The implementation
     should assume that any call to a machine code or assembler
     subprogram is allowed to read or update every object that is
     specified as exported.
     Followed.

*C.1(10-16): Access to Machine Operations*
     The implementation should ensure that little or no overhead is
     associated with calling intrinsic and machine-code subprograms.
     Followed for both intrinsics and machine-code subprograms.

     It is recommended that intrinsic subprograms be provided for
     convenient access to any machine operations that provide special
     capabilities or efficiency and that are not otherwise available
     through the language constructs.
     Followed. A full set of machine operation intrinsic subprograms is
     provided.

     Atomic read-modify-write operations - e.g., test and set, compare
     and swap, decrement and test, enqueue/dequeue.
     Followed on any target supporting such operations.

     Standard numeric functions - e.g., sin, log.
     Followed on any target supporting such operations.

     String manipulation operations - e.g., translate and test.
     Followed on any target supporting such operations.

     Vector operations - e.g., compare vector against thresholds.
     Followed on any target supporting such operations.

     Direct operations on I/O ports.
     Followed on any target supporting such operations.

*C.3(28): Interrupt Support*
     If the `Ceiling_Locking' policy is not in effect, the
     implementation should provide means for the application to specify
     which interrupts are to be blocked during protected actions, if
     the underlying system allows for a finer-grain control of
     interrupt blocking.
     Followed. The underlying system does not allow for finer-grain
     control of interrupt blocking.

*C.3.1(20-21): Protected Procedure Handlers*
     Whenever possible, the implementation should allow interrupt
     handlers to be called directly by the hardware.
     Followed on any target where the underlying operating system
     permits such direct calls.

     Whenever practical, violations of any implementation-defined
     restrictions should be detected before run time.
     Followed. Compile time warnings are given when possible.

*C.3.2(25): Package `Interrupts'*
     If implementation-defined forms of interrupt handler procedures are
     supported, such as protected procedures with parameters, then for
     each such form of a handler, a type analogous to
     `Parameterless_Handler' should be specified in a child package of
     `Interrupts', with the same operations as in the predefined
     package Interrupts.
     Followed.

*C.4(14): Pre-elaboration Requirements*
     It is recommended that pre-elaborated packages be implemented in
     such a way that there should be little or no code executed at run
     time for the elaboration of entities not already covered by the
     Implementation Requirements.
     Followed. Executable code is generated in some cases, e.g. loops
     to initialize large arrays.

*C.5(8): Pragma `Discard_Names'*
     If the pragma applies to an entity, then the implementation should
     reduce the amount of storage used for storing names associated
     with that entity.
     Followed.

*C.7.2(30): The Package Task_Attributes*
     Some implementations are targeted to domains in which memory use
     at run time must be completely deterministic. For such
     implementations, it is recommended that the storage for task
     attributes will be pre-allocated statically and not from the heap.
     This can be accomplished by either placing restrictions on the
     number and the size of the task's attributes, or by using the
     pre-allocated storage for the first N attribute objects, and the
     heap for the others. In the latter case, N should be documented.
     Not followed. This implementation is not targeted to such a domain.

*D.3(17): Locking Policies*
     The implementation should use names that end with `_Locking' for
     locking policies defined by the implementation.
     Followed. No such implementation-defined locking policies exist.

*D.4(16): Entry Queuing Policies*
     Names that end with `_Queuing' should be used for all
     implementation-defined queuing policies.
     Followed. No such implementation-defined queueing policies exist.

*D.6(9-10): Preemptive Abort*
     Even though the `abort_statement' is included in the list of
     potentially blocking operations (see 9.5.1), it is recommended
     that this statement be implemented in a way that never requires
     the task executing the `abort_statement' to block.
     Followed.

     On a multi-processor, the delay associated with aborting a task on
     another processor should be bounded; the implementation should use
     periodic polling, if necessary, to achieve this.
     Followed.

*D.7(21): Tasking Restrictions*
     When feasible, the implementation should take advantage of the
     specified restrictions to produce a more efficient implementation.
     GNAT currently takes advantage of these restrictions by providing
     an optimized run time when the Ravenscar profile and the GNAT
     restricted run time set of restrictions are specified. See pragma
     `Ravenscar' and pragma `Restricted_Run_Time' for more details.

*D.8(47-49): Monotonic Time*
     When appropriate, implementations should provide configuration
     mechanisms to change the value of `Tick'.
     Such configuration mechanisms are not appropriate to this
     implementation and are thus not supported.

     It is recommended that `Calendar.Clock' and `Real_Time.Clock' be
     implemented as transformations of the same time base.
     Followed.

     It is recommended that the "best" time base which exists in the
     underlying system be available to the application through `Clock'.
     "Best" may mean highest accuracy or largest range.
     Followed.

*E.5(28-29): Partition Communication Subsystem*
     Whenever possible, the PCS on the called partition should allow for
     multiple tasks to call the RPC-receiver with different messages and
     should allow them to block until the corresponding subprogram body
     returns.
     Followed by GLADE, a separately supplied PCS that can be used with
     GNAT. For information on GLADE, contact Ada Core Technologies.

     The `Write' operation on a stream of type `Params_Stream_Type'
     should raise `Storage_Error' if it runs out of space trying to
     write the `Item' into the stream.
     Followed by GLADE, a separately supplied PCS that can be used with
     GNAT. For information on GLADE, contact Ada Core Technologies.

*F(7): COBOL Support*
     If COBOL (respectively, C) is widely supported in the target
     environment, implementations supporting the Information Systems
     Annex should provide the child package `Interfaces.COBOL'
     (respectively, `Interfaces.C') specified in Annex B and should
     support a `convention_identifier' of COBOL (respectively, C) in
     the interfacing pragmas (see Annex B), thus allowing Ada programs
     to interface with programs written in that language.
     Followed.

*F.1(2): Decimal Radix Support*
     Packed decimal should be used as the internal representation for
     objects of subtype S when S'Machine_Radix = 10.
     Not followed. GNAT ignores S'Machine_Radix and always uses binary
     representations.

*G: Numerics*
     If Fortran (respectively, C) is widely supported in the target
     environment, implementations supporting the Numerics Annex should
     provide the child package `Interfaces.Fortran' (respectively,
     `Interfaces.C') specified in Annex B and should support a
     `convention_identifier' of Fortran (respectively, C) in the
     interfacing pragmas (see Annex B), thus allowing Ada programs to
     interface with programs written in that language.
     Followed.

*G.1.1(56-58): Complex Types*
     Because the usual mathematical meaning of multiplication of a
     complex operand and a real operand is that of the scaling of both
     components of the former by the latter, an implementation should
     not perform this operation by first promoting the real operand to
     complex type and then performing a full complex multiplication. In
     systems that, in the future, support an Ada binding to IEC
     559:1989, the latter technique will not generate the required
     result when one of the components of the complex operand is
     infinite. (Explicit multiplication of the infinite component by
     the zero component obtained during promotion yields a NaN that
     propagates into the final result.) Analogous advice applies in the
     case of multiplication of a complex operand and a pure-imaginary
     operand, and in the case of division of a complex operand by a
     real or pure-imaginary operand.
     Not followed.

     Similarly, because the usual mathematical meaning of addition of a
     complex operand and a real operand is that the imaginary operand
     remains unchanged, an implementation should not perform this
     operation by first promoting the real operand to complex type and
     then performing a full complex addition. In implementations in
     which the `Signed_Zeros' attribute of the component type is `True'
     (and which therefore conform to IEC 559:1989 in regard to the
     handling of the sign of zero in predefined arithmetic operations),
     the latter technique will not generate the required result when
     the imaginary component of the complex operand is a negatively
     signed zero. (Explicit addition of the negative zero to the zero
     obtained during promotion yields a positive zero.)  Analogous
     advice applies in the case of addition of a complex operand and a
     pure-imaginary operand, and in the case of subtraction of a
     complex operand and a real or pure-imaginary operand.
     Not followed.

     Implementations in which `Real'Signed_Zeros' is `True' should
     attempt to provide a rational treatment of the signs of zero
     results and result components. As one example, the result of the
     `Argument' function should have the sign of the imaginary
     component of the parameter `X' when the point represented by that
     parameter lies on the positive real axis; as another, the sign of
     the imaginary component of the `Compose_From_Polar' function
     should be the same as (respectively, the opposite of) that of the
     `Argument' parameter when that parameter has a value of zero and
     the `Modulus' parameter has a nonnegative (respectively, negative)
     value.
     Followed.

*G.1.2(49): Complex Elementary Functions*
     Implementations in which `Complex_Types.Real'Signed_Zeros' is
     `True' should attempt to provide a rational treatment of the signs
     of zero results and result components. For example, many of the
     complex elementary functions have components that are odd
     functions of one of the parameter components; in these cases, the
     result component should have the sign of the parameter component
     at the origin. Other complex elementary functions have zero
     components whose sign is opposite that of a parameter component at
     the origin, or is always positive or always negative.
     Followed.

*G.2.4(19): Accuracy Requirements*
     The versions of the forward trigonometric functions without a
     `Cycle' parameter should not be implemented by calling the
     corresponding version with a `Cycle' parameter of
     `2.0*Numerics.Pi', since this will not provide the required
     accuracy in some portions of the domain. For the same reason, the
     version of `Log' without a `Base' parameter should not be
     implemented by calling the corresponding version with a `Base'
     parameter of `Numerics.e'.
     Followed.

*G.2.6(15): Complex Arithmetic Accuracy*
     The version of the `Compose_From_Polar' function without a `Cycle'
     parameter should not be implemented by calling the corresponding
     version with a `Cycle' parameter of `2.0*Numerics.Pi', since this
     will not provide the required accuracy in some portions of the
     domain.
     Followed.

Implementation Defined Characteristics
**************************************

   In addition to the implementation dependent pragmas and attributes,
and the implementation advice, there are a number of other features of
Ada 95 that are potentially implementation dependent. These are
mentioned throughout the Ada 95 Reference Manual, and are summarized in
annex M.

   A requirement for conforming Ada compilers is that they provide
documentation describing how the implementation deals with each of these
issues. In this chapter, you will find each point in annex M listed
followed by a description in italic font of how GNAT handles the
implementation dependence.

   You can use this chapter as a guide to minimizing implementation
dependent features in your programs if portability to other compilers
and other operating systems is an important consideration.  The numbers
in each section below correspond to the paragraph number in the Ada 95
Reference Manual.

*2*.  Whether or not each recommendation given in Implementation Advice
is followed. See 1.1.2(37).

*Note Implementation Advice::.

*3*.  Capacity limitations of the implementation. See 1.1.3(3).

The complexity of programs that can be processed is limited only by the
total amount of available virtual memory, and disk space for the
generated object files.

*4*.  Variations from the standard that are impractical to avoid given
the implementation's execution environment. See 1.1.3(6).

There are no variations from the standard.

*5*.  Which `code_statement's cause external interactions. See
1.1.3(10).

Any `code_statement' can potentially cause external interactions.

*6*.  The coded representation for the text of an Ada program. See
2.1(4).

See separate section on source representation.

*7*.  The control functions allowed in comments. See 2.1(14).

See separate section on source representation.

*8*.  The representation for an end of line. See 2.2(2).

See separate section on source representation.

*9*.  Maximum supported line length and lexical element length. See
2.2(15).

The maximum line length is 255 characters an the maximum length of a
lexical element is also 255 characters.

*10*.  Implementation defined pragmas. See 2.8(14).

*Note Implementation Defined Pragmas::.

*11*.  Effect of pragma `Optimize'. See 2.8(27).

Pragma `Optimize', if given with a `Time' or `Space' parameter, checks
that the optimization flag is set, and aborts if it is not.

*12*.  The sequence of characters of the value returned by `S'Image'
when some of the graphic characters of `S'Wide_Image' are not defined
in `Character'. See 3.5(37).

The sequence of characters is as defined by the wide character encoding
method used for the source. See section on source representation for
further details.

*13*.  The predefined integer types declared in `Standard'. See
3.5.4(25).

`Short_Short_Integer'
     8 bit signed

`Short_Integer'
     (Short) 16 bit signed

`Integer'
     32 bit signed

`Long_Integer'
     32 bit signed

`Long_Long_Integer'
     64 bit signed

*14*.  Any nonstandard integer types and the operators defined for
them. See 3.5.4(26).

There are no nonstandard integer types.

*15*.  Any nonstandard real types and the operators defined for them.
See 3.5.6(8).

There are no nonstandard real types.

*16*.  What combinations of requested decimal precision and range are
supported for floating point types. See 3.5.7(7).

The precision and range is as defined by the IEEE standard.

*17*.  The predefined floating point types declared in `Standard'. See
3.5.7(16).

`Short_Float'
     32 bit IEEE short

`Float'
     (Short) 32 bit IEEE short

`Long_Float'
     64 bit IEEE long

`Long_Long_Float'
     64 bit IEEE long

*18*.  The small of an ordinary fixed point type. See 3.5.9(8).

`Fine_Delta' is 2**(-63)

*19*.  What combinations of small, range, and digits are supported for
fixed point types. See 3.5.9(10).

Any combinations are permitted that do not result in a small less than
`Fine_Delta' and do not result in a mantissa larger than 63 bits.

*20*.  The result of `Tags.Expanded_Name' for types declared within an
unnamed `block_statement'. See 3.9(10).

Block numbers of the form `BNNN', where NNN is a decimal integer are
allocated.

*21*.  Implementation-defined attributes. See 4.1.4(12).

*Note Implementation Defined Attributes::.

*22*.  Any implementation-defined time types. See 9.6(6).

There are no implementation-defined time types.

*23*.  The time base associated with relative delays.

See 9.6(20). The time base used is that provided by the C library
function `gettimeofday'.

*24*.  The time base of the type `Calendar.Time'. See 9.6(23).

The time base used is that provided by the C library function
`gettimeofday'.

*25*.  The time zone used for package `Calendar' operations. See
9.6(24).

The time zone used by package `Calendar' is the current system time zone
setting for local time, as accessed by the C library function
`localtime'.

*26*.  Any limit on `delay_until_statements' of `select_statements'.
See 9.6(29).

There are no such limits.

*27*.  Whether or not two non overlapping parts of a composite object
are independently addressable, in the case where packing, record
layout, or `Component_Size' is specified for the object. See 9.10(1).

Separate components are independently addressable if they do not share
overlapping storage units.

*28*.  The representation for a compilation. See 10.1(2).

A compilation is represented by a sequence of files presented to the
compiler in a single invocation of the `gcc' command.

*29*.  Any restrictions on compilations that contain multiple
compilation_units. See 10.1(4).

No single file can contain more than one compilation unit, but any
sequence of files can be presented to the compiler as a single
compilation.

*30*.  The mechanisms for creating an environment and for adding and
replacing compilation units. See 10.1.4(3).

See separate section on compilation model.

*31*.  The manner of explicitly assigning library units to a partition.
See 10.2(2).

See separate section on binding and linking programs.

*32*.  The implementation-defined means, if any, of specifying which
compilation units are needed by a given compilation unit. See 10.2(2).

See separate section on compilation unit.

*33*.  The manner of designating the main subprogram of a partition.
See 10.2(7).

The main program is designated by providing the name of the
corresponding ali file as the input parameter to the binder.

*34*.  The order of elaboration of `library_items'. See 10.2(18).

The first constraint on ordering is that it meets the requirements of
chapter 10 of the Ada 95 Reference Manual. This still leaves some
implementation dependent choices, which are resolved by first
elaborating bodies as early as possible (i.e. in preference to specs
where there is a choice), and second by evaluating the immediate with
clauses of a unit to determine the probably best choice, and third by
elaborating in alphabetical order of unit names where a choice still
remains.

*35*.  Parameter passing and function return for the main subprogram.
See 10.2(21).

The main program has no parameters. It may be a procedure, or a function
returning an integer type. In the latter case, the returned integer
value is the return code of the program.

*36*.  The mechanisms for building and running partitions. See 10.2(24).

GNAT itself supports programs with only a single partition. The GNATDIST
tool provided with the GLADE package (which also includes an
implementation of the PCS) provides a completely flexible method for
building and running programs consisting of multiple partitions. See
the separate GLADE manual for details.

*37*.  The details of program execution, including program termination.
See 10.2(25).

See separate section on compilation model.

*38*.  The semantics of any non-active partitions supported by the
implementation. See 10.2(28).

Passive partitions are supported on targets where shared memory is
provided by the operating system. See the GLADE reference manual for
further details.

*39*.  The information returned by `Exception_Message'. See 11.4.1(10).

Exception message returns the null string unless a specific message has
been passed by the program.

*40*.  The result of `Exceptions.Exception_Name' for types declared
within an unnamed `block_statement'. See 11.4.1(12).

Blocks have implementation defined names of the form `BNNN' where NNN
is an integer.

*41*.  The information returned by `Exception_Information'. See
11.4.1(13).

`Exception_Information' contains the expanded name of the exception in
upper case, and no other information.

*42*.  Implementation-defined check names. See 11.5(27).

No implementation-defined check names are supported.

*43*.  The interpretation of each aspect of representation. See
13.1(20).

See separate section on data representations.

*44*.  Any restrictions placed upon representation items. See 13.1(20).

See separate section on data representations.

*45*.  The meaning of `Size' for indefinite subtypes. See 13.3(48).

Size for an indefinite subtype is the maximum possible size, except that
for the case of a subprogram parameter, the size of the parameter object
is the actual size.

*46*.  The default external representation for a type tag. See 13.3(75).

The default external representation for a type tag is the fully expanded
name of the type in upper case letters.

*47*.  What determines whether a compilation unit is the same in two
different partitions. See 13.3(76).

A compilation unit is the same in two different partitions if and only
if it derives from the same source file.

*48*.  Implementation-defined components. See 13.5.1(15).

The only implementation defined component is the tag for a tagged type,
which contains a pointer to the dispatching table.

*49*.  If `Word_Size' = `Storage_Unit', the default bit ordering. See
13.5.3(5).

`Word_Size' (32) is not the same as `Storage_Unit' (8) for this
implementation, so no non-default bit ordering is supported. The default
bit ordering corresponds to the natural endianness of the target
architecture.

*50*.  The contents of the visible part of package `System' and its
language-defined children. See 13.7(2).

See the definition of these packages in files `system.ads' and
`s-stoele.ads'.

*51*.  The contents of the visible part of package
`System.Machine_Code', and the meaning of `code_statements'. See
13.8(7).

See the definition and documentation in file `s-maccod.ads'.

*52*.  The effect of unchecked conversion. See 13.9(11).

Unchecked conversion between types of the same size and results in an
uninterpreted transmission of the bits from one type to the other. If
the types are of unequal sizes, then in the case of discrete types, a
shorter source is first zero or sign extended as necessary, and a
shorter target is simply truncated on the left.  For all non-discrete
types, the source is first copied if necessary to ensure that the
alignment requirements of the target are met, then a pointer is
constructed to the source value, and the result is obtained by
dereferencing this pointer after converting it to be a pointer to the
target type.

*53*.  The manner of choosing a storage pool for an access type when
`Storage_Pool' is not specified for the type. See 13.11(17).

There are 3 different standard pools used by the compiler when
`Storage_Pool' is not specified depending whether the type is local to
a subprogram or defined at the library level and whether
`Storage_Size'is specified or not. See documentation in the runtime
library units `System.Pool_Global', `System.Pool_Size' and
`System.Pool_Local' in files `s-poosiz.ads', `s-pooglo.ads' and
`s-pooloc.ads' for full details on the default pools used.

*54*.  Whether or not the implementation provides user-accessible names
for the standard pool type(s). See 13.11(17).

See documentation in the sources of the run time mentioned in paragraph
*53* . All these pools are accessible by means of `with''ing these
units.

*55*.  The meaning of `Storage_Size'. See 13.11(18).

`Storage_Size' is measured in storage units, and refers to the total
space available for an access type collection, or to the primary stack
space for a task.

*56*.  Implementation-defined aspects of storage pools. See 13.11(22).

See documentation in the sources of the run time mentioned in paragraph
*53* for details on GNAT-defined aspects of storage pools.

*57*.  The set of restrictions allowed in a pragma `Restrictions'. See
13.12(7).

All RM defined Restriction identifiers are implemented. The following
additional restriction identifiers are provided:

`No_Implementation_Attributes'
     This restriction checks at compile time that no GNAT-defined
     attributes are present. With this restriction, the only attributes
     that can be used are those defined in the Ada 95 Reference Manual.

`No_Implementation_Pragmas'
     This restriction checks at compile time that no GNAT-defined
     pragmas are present. With this restriction, the only pragmas that
     can be used are those defined in the Ada 95 Reference Manual.

`No_Elaboration_Code'
     This restriction ensures at compile time that no elaboration code
     is generated. Note that this is not the same condition as is
     enforced by pragma Preelaborate. There are cases in which pragma
     Preelaborate still permits code to be generated (e.g. code to
     initialize a large array to all zeroes), and there are cases of
     units which do not meet the requirements for pragma Preelaborate,
     but for which no elaboration code is generated. Generally, it is
     the case that preelaborable units will meet the restrictions, with
     the exception of large aggregates initialized by others.

`No_Complex_Barriers'
     This restriction ensures at compile time that barriers in entry
     declarations for protected types are restricted to references to
     simple boolean variables defined in the private part of the
     protected type. No other form of entry barriers is permitted. This
     is one of the restrictions of the Ravenscar profile for limited
     tasking (see also pragma Ravenscar).

`No_Select_Statements'
     This restriction ensures at compile time no select statements of
     any kind are permitted, that is the keyword `select' may not
     appear.  This is one of the restrictions of the Ravenscar profile
     for limited tasking (see also pragma Ravenscar).

*58*.  The consequences of violating limitations on `Restrictions'
pragmas. See 13.12(9).

Restrictions that can be checked at compile time result in illegalities
if violated. Currently there are no other consequences of violating
restrictions.

*59*.  The representation used by the `Read' and `Write' attributes of
elementary types in terms of stream elements. See 13.13.2(9).

The representation is the in-memory representation of the base type of
the type, using the number of bits corresponding to the `TYPE'Size'
value, and the natural ordering of the machine.

*60*.  The names and characteristics of the numeric subtypes declared
in the visible part of package `Standard'. See A.1(3).

See items describing the integer and floating-point types supported.

*61*.  The accuracy actually achieved by the elementary functions. See
A.5.1(1).

The elementary functions correspond to the functions available in the C
library. Only fast math mode is implemented.

*62*.  The sign of a zero result from some of the operators or
functions in `Numerics.Generic_Elementary_Functions', when
`Float_Type'Signed_Zeros' is `True'. See A.5.1(46).

The sign of zeroes follows the requirements of the IEEE 754 standard on
floating-point.

*63*.  The value of `Numerics.Float_Random.Max_Image_Width'. See
A.5.2(27).

Maximum image width is 649, see library file `a-numran.ads'.

*64*.  The value of `Numerics.Discrete_Random.Max_Image_Width'. See
A.5.2(27).

Maximum image width is 80, see library file `a-nudira.ads'.

*65*.  The algorithms for random number generation. See A.5.2(32).

The algorithm is documented in the source files `a-numran.ads' and
`a-numran.adb'.

*66*.  The string representation of a random number generator's state.
See A.5.2(38).

See the documentation contained in the file `a-numran.adb'.

*67*.  The minimum time interval between calls to the time-dependent
Reset procedure that are guaranteed to initiate different random number
sequences. See A.5.2(45).

The minimum period between reset calls to guarantee distinct series of
random numbers is one microsecond.

*68*.  The values of the `Model_Mantissa', `Model_Emin',
`Model_Epsilon', `Model', `Safe_First', and `Safe_Last' attributes, if
the Numerics Annex is not supported. See A.5.3(72).

See the source file `ttypef.ads' for the values of all numeric
attributes.

*69*.  Any implementation-defined characteristics of the input-output
packages. See A.7(14).

There are no special implementation defined characteristics for these
packages.

*70*.  The value of `Buffer_Size' in `Storage_IO'. See A.9(10).

All type representations are contiguous, and the `Buffer_Size' is the
value of `TYPE'Size' rounded up to the next storage unit boundary.

*71*.  External files for standard input, standard output, and standard
error See A.10(5).

These files are mapped onto the files provided by the C streams
libraries. See source file `i-cstrea.ads' for further details.

*72*.  The accuracy of the value produced by `Put'. See A.10.9(36).

If more digits are requested in the output than are represented by the
precision of the value, zeroes are output in the corresponding least
significant digit positions.

*73*.  The meaning of `Argument_Count', `Argument', and `Command_Name'.
See A.15(1).

These are mapped onto the `argv' and `argc' parameters of the main
program in the natural manner.

*74*.  Implementation-defined convention names. See B.1(11).

The following convention names are supported

`Ada'
     Ada

`Asm'
     Assembly language

`Assembler'
     Assembly language

`C'
     C

`C_Pass_By_Copy'
     Treated like C, except for record types

`COBOL'
     COBOL

`CPP'
     C++

`Default'
     Treated the same as C

`DLL'
     DLL (used for Windows implementations only) is handled like the
     Stdcall convention. This convention is used to access variables
     and functions (with Stdcall convention) in a DLL.

`External'
     Treated the same as C

`Fortran'
     Fortran

`Intrinsic'
     Intrinsic

`Stdcall'
     Stdcall (used for Windows implementations only). This convention
     correspond to the WINAPI (previously called Pascal convention)
     C/C++ convention under Windows. A function with this convention
     clean the stack before exit.

In addition, all otherwise unrecognized convention names are also
treated as being synonymous with convention C. In all implementations
except for VMS, use of such other names results in a warning. In VMS
implementations, these names are accepted silently.

*75*.  The meaning of link names. See B.1(36).

Link names are the actual names used by the linker.

*76*.  The manner of choosing link names when neither the link name nor
the address of an imported or exported entity is specified. See B.1(36).

The default linker name is that which would be assigned by the relevant
external language, interpreting the Ada name as being in all lower case
letters.

*77*.  The effect of pragma `Linker_Options'. See B.1(37).

The string passed to `Linker_Options' is presented uninterpreted as an
argument to the link command.

*78*.  The contents of the visible part of package `Interfaces' and its
language-defined descendants. See B.2(1).

See files with prefix `i-' in the distributed library.

*79*.  Implementation-defined children of package `Interfaces'. The
contents of the visible part of package `Interfaces'. See B.2(11).

See files with prefix `i-' in the distributed library.

*80*.  The types `Floating', `Long_Floating', `Binary', `Long_Binary',
`Decimal_ Element', and `COBOL_Character'; and the initialization of
the variables `Ada_To_COBOL' and `COBOL_To_Ada', in `Interfaces.COBOL'.
See B.4(50).

`Floating'
     Float

`Long_Floating'
     (Floating) Long_Float

`Binary'
     Integer

`Long_Binary'
     Long_Long_Integer

`Decimal_Element'
     Character

`COBOL_Character'
     Character

For initialization, see the file `i-cobol.ads' in the distributed
library.

*81*.  Support for access to machine instructions. See C.1(1).

See documentation in file `s-maccod.ads' in the distributed library.

*82*.  Implementation-defined aspects of access to machine operations.
See C.1(9).

See documentation in file `s-maccod.ads' in the distributed library.

*83*.  Implementation-defined aspects of interrupts. See C.3(2).

Interrupts are mapped to signals or conditions as appropriate. See
definition of unit `Ada.Interrupt_Names' in source file `a-intnam.ads'
for details on the interrupts supported on a particular target.

*84*.  Implementation-defined aspects of pre-elaboration. See C.4(13).

GNAT does not permit a partition to be restarted without reloading,
except under control of the debugger.

*85*.  The semantics of pragma `Discard_Names'. See C.5(7).

Pragma `Discard_Names' causes names of enumeration literals to be
suppressed. In the presence of this pragma, the Image attribute
provides the image of the Pos of the literal, and Value accepts Pos
values.

*86*.  The result of the `Task_Identification.Image' attribute. See
C.7.1(7).

The result of this attribute is an 8-digit hexadecimal string
representing the virtual address of the task control block.

*87*.  The value of `Current_Task' when in a protected entry or
interrupt handler. See C.7.1(17).

Protected entries or interrupt handlers can be executed by any
convenient thread, so the value of `Current_Task' is undefined.

*88*.  The effect of calling `Current_Task' from an entry body or
interrupt handler. See C.7.1(19).

The effect of calling `Current_Task' from an entry body or interrupt
handler is to return the identification of the task currently executing
the code.

*89*.  Implementation-defined aspects of `Task_Attributes'. See
C.7.2(19).

There are no implementation-defined aspects of `Task_Attributes'.

*90*.  Values of all `Metrics'. See D(2).

Information on metrics is not yet available.

*91*.  The declarations of `Any_Priority' and `Priority'. See D.1(11).

See declarations in file `system.ads'.

*92*.  Implementation-defined execution resources. See D.1(15).

There are no implementation-defined execution resources.

*93*.  Whether, on a multiprocessor, a task that is waiting for access
to a protected object keeps its processor busy. See D.2.1(3).

On a multi-processor, a task that is waiting for access to a protected
object does not keep its processor busy.

*94*.  The affect of implementation defined execution resources on task
dispatching. See D.2.1(9).

Tasks map to threads in the threads package used by GNAT. Where possible
and appropriate, these threads correspond to native threads of the
underlying operating system.

*95*.  Implementation-defined `policy_identifiers' allowed in a pragma
`Task_Dispatching_Policy'. See D.2.2(3).

There are no implementation-defined policy-identifiers allowed in this
pragma.

*96*.  Implementation-defined aspects of priority inversion. See
D.2.2(16).

Execution of a task cannot be preempted by the implementation processing
of delay expirations for lower priority tasks.

*97*.  Implementation defined task dispatching. See D.2.2(18).

The policy is the same as that of the underlying threads implementation.

*98*.  Implementation-defined `policy_identifiers' allowed in a pragma
`Locking_Policy'. See D.3(4).

There are no implementation defined policy identifiers allowed in this
pragma.

*99*.  Default ceiling priorities. See D.3(10).

The ceiling priority of protected objects of the type
`System.Interrupt_Priority'Last' as described in the Ada 95 Reference
Manual D.3(10),

*100*.  The ceiling of any protected object used internally by the
implementation. See D.3(16).

The ceiling priority of internal protected objects is
`System.Priority'Last'.

*101*.  Implementation-defined queuing policies. See D.4(1).

There are no implementation-defined queueing policies.

*102*.  On a multiprocessor, any conditions that cause the completion
of an aborted construct to be delayed later than what is specified for
a single processor. See D.6(3).

The semantics for abort on a multi-processor is the same as on a single
processor, there are no further delays.

*103*.  Any operations that implicitly require heap storage allocation.
See D.7(8).

The only operation that implicitly requires heap storage allocation is
task creation.

*104*.  Implementation-defined aspects of pragma `Restrictions'. See
D.7(20).

There are no such implementation-defined aspects.

*105*.  Implementation-defined aspects of package `Real_Time'. See
D.8(17).

There are no implementation defined aspects of package `Real_Time'.

*106*.  Implementation-defined aspects of `delay_statements'. See
D.9(8).

Any difference greater than one microsecond will cause the task to be
delayed (see D.9(7)).

*107*.  The upper bound on the duration of interrupt blocking caused by
the implementation. See D.12(5).

The upper bound is determined by the underlying operating system. In no
cases is it more than 10 milliseconds.

*108*.  The means for creating and executing distributed programs. See
E(5).

The GLADE package provides a utility GNATDIST for creating and executing
distributed programs. See the GLADE reference manual for further
details.

*109*.  Any events that can result in a partition becoming
inaccessible. See E.1(7).

See the GLADE reference manual for full details on such events.

*110*.  The scheduling policies, treatment of priorities, and
management of shared resources between partitions in certain cases. See
E.1(11).

See the GLADE reference manual for full details on these aspects of
multi-partition execution.

*111*.  Events that cause the version of a compilation unit to change.
See E.3(5).

Editing the source file of a compilation unit, or the source files of
any units on which it is dependent in a significant way cause the
version to change. No other actions cause the version number to change.
All changes are significant except those which affect only layout,
capitalization or comments.

*112*.  Whether the execution of the remote subprogram is immediately
aborted as a result of cancellation. See E.4(13).

See the GLADE reference manual for details on the effect of abort in a
distributed application.

*113*.  Implementation-defined aspects of the PCS. See E.5(25).

See the GLADE reference manual for a full description of all
implementation defined aspects of the PCS.

*114*.  Implementation-defined interfaces in the PCS. See E.5(26).

See the GLADE reference manual for a full description of all
implementation defined interfaces.

*115*.  The values of named numbers in the package `Decimal'. See
F.2(7).

`Max_Scale'
     +18

`Min_Scale'
     -18

`Min_Delta'
     1.0E-18

`Max_Delta'
     1.0E+18

`Max_Decimal_Digits'
     18

*116*.  The value of `Max_Picture_Length' in the package
`Text_IO.Editing'. See F.3.3(16).

64

*117*.  The value of `Max_Picture_Length' in the package
`Wide_Text_IO.Editing'. See F.3.4(5).

64

*118*.  The accuracy actually achieved by the complex elementary
functions and by other complex arithmetic operations. See G.1(1).

Standard library functions are used for the complex arithmetic
operations. Only fast math mode is currently supported.

*119*.  The sign of a zero result (or a component thereof) from any
operator or function in `Numerics.Generic_Complex_Types', when
`Real'Signed_Zeros' is True. See G.1.1(53).

The signs of zero values are as recommended by the relevant
implementation advice.

*120*.  The sign of a zero result (or a component thereof) from any
operator or function in
`Numerics.Generic_Complex_Elementary_Functions', when
`Real'Signed_Zeros' is `True'. See G.1.2(45).

The signs of zero values are as recommended by the relevant
implementation advice.

*121*.  Whether the strict mode or the relaxed mode is the default. See
G.2(2).

The strict mode is the default. There is no separate relaxed mode. GNAT
provides a highly efficient implementation of strict mode.

*122*.  The result interval in certain cases of fixed-to-float
conversion. See G.2.1(10).

For cases where the result interval is implementation dependent, the
accuracy is that provided by performing all operations in 64-bit IEEE
floating-point format.

*123*.  The result of a floating point arithmetic operation in overflow
situations, when the `Machine_Overflows' attribute of the result type
is `False'. See G.2.1(13).

Infinite and Nan values are produced as dictated by the IEEE
floating-point standard.

*124*.  The result interval for division (or exponentiation by a
negative exponent), when the floating point hardware implements division
as multiplication by a reciprocal. See G.2.1(16).

Not relevant, division is IEEE exact.

*125*.  The definition of close result set, which determines the
accuracy of certain fixed point multiplications and divisions. See
G.2.3(5).

Operations in the close result set are performed using IEEE long format
floating-point arithmetic. The input operands are converted to
floating-point, the operation is done in floating-point, and the result
is converted to the target type.

*126*.  Conditions on a `universal_real' operand of a fixed point
multiplication or division for which the result shall be in the perfect
result set. See G.2.3(22).

The result is only defined to be in the perfect result set if the result
can be computed by a single scaling operation involving a scale factor
representable in 64-bits.

*127*.  The result of a fixed point arithmetic operation in overflow
situations, when the `Machine_Overflows' attribute of the result type
is `False'. See G.2.3(27).

Not relevant, `Machine_Overflows' is `True' for fixed-point types.

*128*.  The result of an elementary function reference in overflow
situations, when the `Machine_Overflows' attribute of the result type
is `False'. See G.2.4(4).

IEEE infinite and Nan values are produced as appropriate.

*129*.  The value of the angle threshold, within which certain
elementary functions, complex arithmetic operations, and complex
elementary functions yield results conforming to a maximum relative
error bound. See G.2.4(10).

Information on this subject is not yet available.

*130*.  The accuracy of certain elementary functions for parameters
beyond the angle threshold. See G.2.4(10).

Information on this subject is not yet available.

*131*.  The result of a complex arithmetic operation or complex
elementary function reference in overflow situations, when the
`Machine_Overflows' attribute of the corresponding real type is
`False'. See G.2.6(5).

IEEE infinite and Nan values are produced as appropriate.

*132*.  The accuracy of certain complex arithmetic operations and
certain complex elementary functions for parameters (or components
thereof) beyond the angle threshold. See G.2.6(8).

Information on those subjects is not yet available.

*133*.  Information regarding bounded errors and erroneous execution.
See H.2(1).

Information on this subject is not yet available.

*134*.  Implementation-defined aspects of pragma `Inspection_Point'.
See H.3.2(8).

Pragma `Inspection_Point' ensures that the variable is live and can be
examined by the debugger at the inspection point.

*135*.  Implementation-defined aspects of pragma `Restrictions'. See
H.4(25).

There are no implementation-defined aspects of pragma `Restrictions'.
The use of pragma `Restrictions [No_Exceptions]' has no effect on the
generated code. Checks must suppressed by use of pragma `Suppress'.

*136*.  Any restrictions on pragma `Restrictions'. See H.4(27).

There are no restrictions on pragma `Restrictions'.

Standard Library Routines
*************************

The Ada 95 Reference Manual contains in Annex A a full description of an
extensive set of standard library routines that can be used in any Ada
program, and which must be provided by all Ada compilers. They are
analogous to the standard C library used by C programs.

   GNAT implements all of the facilities described in annex A, and for
most purposes the description in the Ada 95 reference manual, or
appropriate Ada text book, will be sufficient for making use of these
facilities.

   In the case of the input-output facilities, *Note The Implementation
of Standard I/O::, gives details on exactly how GNAT interfaces to the
file system. For the remaining packages, the Ada 95 reference manual
should be sufficient. The following is a list of the packages included,
together with a brief description of the functionality that is provided.

   For completeness, references are included to other predefined library
routines defined in other sections of the Ada 95 reference manual
(these are cross-indexed from annex A).

`Ada (A.2)'
     This is a parent package for all the standard library packages. It
     is usually included implicitly in your program, and itself
     contains no useful data or routines.

`Ada.Calendar (9.6)'
     `Calendar' provides time of day access, and routines for
     manipulating times and durations.

`Ada.Characters (A.3.1)'
     This is a dummy parent package that contains no useful entities

`Ada.Characters.Handling (A.3.2)'
     This package provides some basic character handling capabilities,
     including classification functions for classes of characters (e.g.
     test for letters, or digits).

`Ada.Characters.Latin_1 (A.3.3)'
     This package includes a complete set of definitions of the
     characters that appear in type CHARACTER. It is useful for writing
     programs that will run in international environments. For example,
     if you want an upper case E with an acute accent in a string, it
     is often better to use the definition of `UC_E_Acute' in this
     package. Then your program will print in an understandable manner
     even if your environment does not support these extended
     characters.

`Ada.Command_Line (A.15)'
     This package provides access to the command line parameters and
     the name of the current program (analogous to the use of argc and
     argv in C), and also allows the exit status for the program to be
     set in a system-independent manner.

`Ada.Decimal (F.2)'
     This package provides constants describing the range of decimal
     numbers implemented, and also a decimal divide routine (analogous
     to the COBOL verb DIVIDE .. GIVING .. REMAINDER ..)

`Ada.Direct_IO (A.8.4)'
     This package provides input-output using a model of a set of
     records of fixed-length, containing an arbitrary definite Ada
     type, indexed by an integer record number.

`Ada.Dynamic_Priorities (D.5)'
     This package allows the priorities of a task to be adjusted
     dynamically as the task is running.

`Ada.Exceptions (11.4.1)'
     This package provides additional information on exceptions, and
     also contains facilities for treating exceptions as data objects,
     and raising exceptions with associated messages.

`Ada.Finalization (7.6)'
     This package contains the declarations and subprograms to support
     the use of controlled types, providing for automatic
     initialization and finalization (analogous to the constructors and
     destructors of C++)

`Ada.Interrupts (C.3.2)'
     This package provides facilities for interfacing to interrupts,
     which includes the set of signals or conditions that can be raised
     and recognized as interrupts.

`Ada.Interrupts.Names (C.3.2)'
     This package provides the set of interrupt names (actually signal
     or condition names) that can be handled by GNAT.

`Ada.IO_Exceptions (A.13)'
     This package defines the set of exceptions that can be raised by
     use of the standard IO packages.

`Ada.Numerics'
     This package contains some standard constants and exceptions used
     throughout the numerics packages. Note that the constants pi and e
     are defined here, and it is better to use these definitions than
     rolling your own.

`Ada.Numerics.Complex_Elementary_Functions'
     Provides the implementation of standard elementary functions (such
     as log and trigonometric functions) operating on complex numbers
     using the standard `Float' and the `Complex' and `Imaginary' types
     created by the package `Numerics.Complex_Types'.

`Ada.Numerics.Complex_Types'
     This is a predefined instantiation of
     `Numerics.Generic_Complex_Types' using `Standard.Float' to build
     the type `Complex' and `Imaginary'.

`Ada.Numerics.Discrete_Random'
     This package provides a random number generator suitable for
     generating random integer values from a specified range.

`Ada.Numerics.Float_Random'
     This package provides a random number generator suitable for
     generating uniformly distributed floating point values.

`Ada.Numerics.Generic_Complex_Elementary_Functions'
     This is a generic version of the package that provides the
     implementation of standard elementary functions (such as log an
     trigonometric functions) for an arbitrary complex type.

     The following predefined instantiations of this package exist

    `Short_Float'
          `Ada.Numerics.Short_Complex_Elementary_Functions'

    `Float'
          `Ada.Numerics.Complex_Elementary_Functions'

    `Long_Float'
          `Ada.Numerics.Long_Complex_Elementary_Functions'

`Ada.Numerics.Generic_Complex_Types'
     This is a generic package that allows the creation of complex
     types, with associated complex arithmetic operations.

     The following predefined instantiations of this package exist
    `Short_Float'
          `Ada.Numerics.Short_Complex_Complex_Types'

    `Float'
          `Ada.Numerics.Complex_Complex_Types'

    `Long_Float'
          `Ada.Numerics.Long_Complex_Complex_Types'

`Ada.Numerics.Generic_Elementary_Functions'
     This is a generic package that provides the implementation of
     standard elementary functions (such as log an trigonometric
     functions) for an arbitrary float type.

     The following predefined instantiations of this package exist

    `Short_Float'
          `Ada.Numerics.Short_Elementary_Functions'

    `Float'
          `Ada.Numerics.Elementary_Functions'

    `Long_Float'
          `Ada.Numerics.Long_Elementary_Functions'

`Ada.Real_Time (D.8)'
     This package provides facilities similar to those of `Calendar',
     but operating with a finer clock suitable for real time control.

`Ada.Sequential_IO (A.8.1)'
     This package provides input-output facilities for sequential files,
     which can contain a sequence of values of a single type, which can
     be any Ada type, including indefinite (unconstrained) types.

`Ada.Storage_IO (A.9)'
     This package provides a facility for mapping arbitrary Ada types
     to and from a storage buffer. It is primarily intended for the
     creation of new IO packages.

`Ada.Streams (13.13.1)'
     This is a generic package that provides the basic support for the
     concept of streams as used by the stream attributes (`Input',
     `Output', `Read' and `Write').

`Ada.Streams.Stream_IO (A.12.1)'
     This package is a specialization of the type `Streams' defined in
     package `Streams' together with a set of operations providing
     Stream_IO capability. The Stream_IO model permits both random and
     sequential access to a file which can contain an arbitrary set of
     values of one or more Ada types.

`Ada.Strings (A.4.1)'
     This package provides some basic constants used by the string
     handling packages.

`Ada.Strings.Bounded (A.4.4)'
     This package provides facilities for handling variable length
     strings. The bounded model requires a maximum length. It is thus
     somewhat more limited than the unbounded model, but avoids the use
     of dynamic allocation or finalization.

`Ada.Strings.Fixed (A.4.3)'
     This package provides facilities for handling fixed length strings.

`Ada.Strings.Maps (A.4.2)'
     This package provides facilities for handling character mappings
     and arbitrarily defined subsets of characters. For instance it is
     useful in defining specialized translation tables.

`Ada.Strings.Maps.Constants (A.4.6)'
     This package provides a standard set of predefined mappings and
     predefined character sets. For example, the standard upper to
     lower case conversion table is found in this package. Note that
     upper to lower case conversion is non-trivial if you want to take
     the entire set of characters, including extended characters like E
     with an acute accent, into account. You should use the mappings in
     this package (rather than adding 32 yourself) to do case mappings.

`Ada.Strings.Unbounded (A.4.5)'
     This package provides facilities for handling variable length
     strings. The unbounded model allows arbitrary length strings, but
     requires the use of dynamic allocation and finalization.

`Ada.Strings.Wide_Bounded (A.4.7)'
`Ada.Strings.Wide_Fixed (A.4.7)'
`Ada.Strings.Wide_Maps (A.4.7)'
`Ada.Strings.Wide_Maps.Constants (A.4.7)'
`Ada.Strings.Wide_Unbounded (A.4.7)'
     These package provide analogous capabilities to the corresponding
     packages without `Wide_' in the name, but operate with the types
     `Wide_String' and `Wide_Character' instead of `String' and
     `Character'.

`Ada.Synchronous_Task_Control (D.10)'
     This package provides some standard facilities for controlling task
     communication in a synchronous manner.

`Ada.Tags'
     This package contains definitions for manipulation of the tags of
     tagged values.

`Ada.Task_Attributes'
     This package provides the capability of associating arbitrary
     task-specific data with separate tasks.

`Ada.Text_IO'
     This package provides basic text input-output capabilities for
     character, string and numeric data. The subpackages of this
     package are listed next.

`Ada.Text_IO.Decimal_IO'
     Provides input-output facilities for decimal fixed-point types

`Ada.Text_IO.Enumeration_IO'
     Provides input-output facilities for enumeration types.

`Ada.Text_IO.Fixed_IO'
     Provides input-output facilities for ordinary fixed-point types.

`Ada.Text_IO.Float_IO'
     Provides input-output facilities for float types. The following
     predefined instantiations of this generic package are available:

    `Short_Float'
          `Short_Float_Text_IO'

    `Float'
          `Float_Text_IO'

    `Long_Float'
          `Long_Float_Text_IO'

`Ada.Text_IO.Integer_IO'
     Provides input-output facilities for integer types. The following
     predefined instantiations of this generic package are available:

    `Short_Short_Integer'
          `Ada.Short_Short_Integer_Text_IO'

    `Short_Integer'
          `Ada.Short_Integer_Text_IO'

    `Integer'
          `Ada.Integer_Text_IO'

    `Long_Integer'
          `Ada.Long_Integer_Text_IO'

    `Long_Long_Integer'
          `Ada.Long_Long_Integer_Text_IO'

`Ada.Text_IO.Modular_IO'
     Provides input-output facilities for modular (unsigned) types

`Ada.Text_IO.Complex_IO (G.1.3)'
     This package provides basic text input-output capabilities for
     complex data.

`Ada.Text_IO.Editing (F.3.3)'
     This package contains routines for edited output, analogous to the
     use of pictures in COBOL. The picture formats used by this package
     are a close copy of the facility in COBOL.

`Ada.Text_IO.Text_Streams (A.12.2)'
     This package provides a facility that allows Text_IO files to be
     treated as streams, so that the stream attributes can be used for
     writing arbitrary data, including binary data, to Text_IO files.

`Ada.Unchecked_Conversion (13.9)'
     This generic package allows arbitrary conversion from one type to
     another of the same size, providing for breaking the type safety in
     special circumstances.

`Ada.Unchecked_Deallocation (13.11.2)'
     This generic package allows explicit freeing of storage previously
     allocated by use of an allocator.

`Ada.Wide_Text_IO (A.11)'
     This package is similar to `Ada.Text_IO', except that the external
     file supports wide character representations, and the internal
     types are `Wide_Character' and `Wide_String' instead of `Character'
     and `String'. It contains generic subpackages listed next.

`Ada.Wide_Text_IO.Decimal_IO'
     Provides input-output facilities for decimal fixed-point types

`Ada.Wide_Text_IO.Enumeration_IO'
     Provides input-output facilities for enumeration types.

`Ada.Wide_Text_IO.Fixed_IO'
     Provides input-output facilities for ordinary fixed-point types.

`Ada.Wide_Text_IO.Float_IO'
     Provides input-output facilities for float types. The following
     predefined instantiations of this generic package are available:

    `Short_Float'
          `Short_Float_Wide_Text_IO'

    `Float'
          `Float_Wide_Text_IO'

    `Long_Float'
          `Long_Float_Wide_Text_IO'

`Ada.Wide_Text_IO.Integer_IO'
     Provides input-output facilities for integer types. The following
     predefined instantiations of this generic package are available:

    `Short_Short_Integer'
          `Ada.Short_Short_Integer_Wide_Text_IO'

    `Short_Integer'
          `Ada.Short_Integer_Wide_Text_IO'

    `Integer'
          `Ada.Integer_Wide_Text_IO'

    `Long_Integer'
          `Ada.Long_Integer_Wide_Text_IO'

    `Long_Long_Integer'
          `Ada.Long_Long_Integer_Wide_Text_IO'

`Ada.Wide_Text_IO.Modular_IO'
     Provides input-output facilities for modular (unsigned) types

`Ada.Wide_Text_IO.Complex_IO (G.1.3)'
     This package is similar to `Ada.Text_IO.Complex_IO', except that
     the external file supports wide character representations.

`Ada.Wide_Text_IO.Editing (F.3.4)'
     This package is similar to `Ada.Text_IO.Editing', except that the
     types are `Wide_Character' and `Wide_String' instead of
     `Character' and `String'.

`Ada.Wide_Text_IO.Streams (A.12.3)'
     This package is similar to `Ada.Text_IO.Streams', except that the
     types are `Wide_Character' and `Wide_String' instead of
     `Character' and `String'.

The Implementation of Standard I/O
**********************************

GNAT implements all the required input-output facilities described in
A.6 through A.14. These sections of the Ada 95 reference manual
describe the required behavior of these packages from the Ada point of
view, and if you are writing a portable Ada program that does not need
to know the exact manner in which Ada maps to the outside world when it
comes to reading or writing external files, then you do not need to
read this chapter. As long as your files are all regular files (not
pipes or devices), and as long as you write and read the files only
from Ada, the description in the Ada 95 reference manual is sufficient.

   However, if you want to do input-output to pipes or other devices,
such as the keyboard or screen, or if the files you are dealing with are
either generated by some other language, or to be read by some other
language, then you need to know more about the details of how the GNAT
implementation of these input-output facilities behaves.

   In this chapter we give a detailed description of exactly how GNAT
interfaces to the file system. As always, the sources of the system are
available to you for answering questions at an even more detailed level,
but for most purposes the information in this chapter will suffice.

   Another reason that you may need to know more about how input-output
is implemented arises when you have a program written in mixed languages
where, for example, files are shared between the C and Ada sections of
the same program. GNAT provides some additional facilities, in the form
of additional child library packages, that facilitate this sharing, and
these additional facilities are also described in this chapter.

Standard I/O Packages
=====================

The Standard I/O packages described in Annex A for

   * Ada.Text_IO

   * Ada.Text_IO.Complex_IO

   * Ada.Text_IO.Text_Streams,

   * Ada.Wide_Text_IO

   * Ada.Wide_Text_IO.Complex_IO,

   * Ada.Wide_Text_IO.Text_Streams

   * Ada.Stream_IO

   * Ada.Sequential_IO

   * Ada.Direct_IO

are implemented using the C library streams facility; where

   * All files are opened using `fopen'.

   * All input/output operations use `fread'/`fwrite'.

   There is no internal buffering of any kind at the Ada library level.
The only buffering is that provided at the system level in the
implementation of the C library routines that support streams. This
facilitates shared use of these streams by mixed language programs.

FORM Strings
============

The format of a FORM string in GNAT is:

     "keyword=value,keyword=value,...,keyword=value"

where letters may be in upper or lower case, and there are no spaces
between values. The order of the entries is not important. Currently
there are two keywords defined.

     SHARED=[YES|NO]
     WCEM=[n|h|u|s\e]

   The use of these parameters is described later in this section.

Direct_IO
=========

Direct_IO can only be instantiated for definite types. This is a
restriction of the Ada language, which means that the records are fixed
length (the length being determined by `TYPE'Size', rounded up to the
next storage unit boundary if necessary).

   The records of a Direct_IO file are simply written to the file in
index sequence, with the first record starting at offset zero, and
subsequent records following. There is no control information of any
kind. For example, if 32-bit integers are being written, each record
takes 4-bytes, so the record at index K starts at offset (K - 1)*4.

   There is no limit on the size of Direct_IO files, they are expanded
as necessary to accommodate whatever records are written to the file.

Sequential_IO
=============

Sequential_IO may be instantiated with either a definite (constrained)
or indefinite (unconstrained) type.

   For the definite type case, the elements written to the file are
simply the memory images of the data values with no control information
of any kind. The resulting file should be read using the same type, no
validity checking is performed on input.

   For the indefinite type case, the elements written consist of two
parts. First is the size of the data item, written as the memory image
of a `Interfaces.C.size_t' value, followed by the memory image of the
data value. The resulting file can only be read using the same
(unconstrained) type. Normal assignment checks are performed on these
read operations, and if these checks fail, `Data_Error' is raised. In
particular, in the array case, the lengths must match, and in the
variant record case, if the variable for a particular read operation is
constrained, the discriminants must match.

   Note that it is not possible to use Sequential_IO to write variable
length array items, and then read the data back into different length
arrays. For example, the following will raise `Data_Error':

      package IO is new Sequential_IO (String);
      F : IO.File_Type;
      S : String (1..4);
      ...
      IO.Create (F)
      IO.Write (F, "hello!")
      IO.Reset (F, Mode=>In_File);
      IO.Read (F, S);
      Put_Line (S);

   On some Ada implementations, this will print `hell', but the program
is clearly incorrect, since there is only one element in the file, and
that element is the string `hello!'.

   In Ada 95, this kind of behavior can be legitimately achieved using
Stream_IO, and this is the preferred mechanism. In particular, the above
program fragment rewritten to use Stream_IO will work correctly.

Text_IO
=======

Text_IO files consist of a stream of characters containing the following
special control characters:

     LF (line feed, 16#0A#) Line Mark
     FF (form feed, 16#0C#) Page Mark

   A canonical Text_IO file is defined as one in which the following
conditions are met:

   * The character `LF' is used only as a line mark, i.e. to mark the
     end of the line.

   * The character `FF' is used only as a page mark, i.e. to mark the
     end of a page and consequently can appear only immediately
     following a `LF' (line mark) character.

   * The file ends with either `LF' (line mark) or `LF'-`FF' (line
     mark, page mark). In the former case, the page mark is implicitly
     assumed to be present.

   A file written using Text_IO will be in canonical form provided that
no explicit `LF' or `FF' characters are written using `Put' or
`Put_Line'. There will be no `FF' character at the end of the file
unless an explicit `New_Page' operation was performed before closing
the file.

   A canonical Text_IO file that is a regular file, i.e. not a device
or a pipe, can be read using any of the routines in Text_IO. The
semantics in this case will be exactly as defined in the Ada 95
reference manual and all the routines in Text_IO are fully implemented.

   A text file that does not meet the requirements for a canonical
Text_IO file has one of the following:

   * The file contains `FF' characters not immediately following a `LF'
     character.

   * The file contains `LF' or `FF' characters written by `Put' or
     `Put_Line', which are not logically considered to be line marks or
     page marks.

   * The file ends in a character other than `LF' or `FF', i.e. there
     is no explicit line mark or page mark at the end of the file.

   Text_IO can be used to read such non-standard text files but
subprograms to do with line or page numbers do not have defined
meanings. In particular, a `FF' character that does not follow a `LF'
character may or may not be treated as a page mark from the point of
view of page and line numbering. Every `LF' character is considered to
end a line, and there is an implied `LF' character at the end of the
file.

Stream Pointer Positioning
--------------------------

`Ada.Text_IO' has a definition of current position for a file that is
being read. No internal buffering occurs in Text_IO, and usually the
physical position in the stream used to implement the file corresponds
to this logical position defined by Text_IO. There are two exceptions:

   * After a call to `End_Of_Page' that returns `True', the stream is
     positioned past the `LF' (line mark) that precedes the page mark.
     Text_IO maintains an internal flag so that subsequent read
     operations properly handle the logical position which is unchanged
     by the `End_Of_Page' call.

   * After a call to `End_Of_File' that returns `True', if the Text_IO
     file was positioned before the line mark at the end of file before
     the call, then the logical position is unchanged, but the stream
     is physically positioned right at the end of file (past the line
     mark, and past a possible page mark following the line mark. Again
     Text_IO maintains internal flags so that subsequent read
     operations properly handle the logical position.

   These discrepancies have no effect on the observable behavior of
Text_IO, but if a single Ada stream is shared between a C program and
Ada program, or shared (using `shared=yes' in the form string) between
two Ada files, then the difference may be observable in some situations.

Reading and Writing Non-Regular Files
-------------------------------------

A non-regular file is a device (such as a keyboard), or a pipe. Text_IO
can be used for reading and writing. Writing is not affected and the
sequence of characters output is identical to the normal file case, but
for reading, the behavior of Text_IO is modified to avoid undesirable
look-ahead as follows:

   An input file that is not a regular file is considered to have no
page marks. Any `Ascii.FF' characters (the character normally used for a
page mark) appearing in the file are considered to be data characters.
In particular:

   * `Get_Line' and `Skip_Line' do not test for a page mark following a
     line mark. If a page mark appears, it will be treated as a data
     character.

   * This avoids the need to wait for an extra character to be typed or
     entered from the pipe to complete one of these operations.

   * `End_Of_Page' always returns `False'

   * `End_Of_File' will return `False' if there is a page mark at the
     end of the file.

   Output to non-regular files is the same as for regular files. Page
marks may be written to non-regular files using `New_Page', but as noted
above they will not be treated as page marks on input if the output is
piped to another Ada program.

   Another important discrepancy when reading non-regular files is that
the end of file indication is not "sticky". If an end of file is
entered, e.g. by pressing the `EOT' key, then end of file is signalled
once (i.e. the test `End_Of_File' will yield `True', or a read will
raise `End_Error'), but then reading can resume to read data past that
end of file indication, until another end of file indication is entered.

Get_Immediate
-------------

Get_Immediate returns the next character (including control characters)
from the input file. In particular, Get_Immediate will return LF or FF
characters used as line marks or page marks. Such operations leave the
file positioned past the control character, and it is thus not treated
as having its normal function. This means that page, line and column
counts after this kind of Get_Immediate call are set as though the mark
did not occur. In the case where a Get_Immediate leaves the file
positioned between the line mark and page mark (which is not normally
possible), it is undefined whether the FF character will be treated as a
page mark.

Treating Text_IO Files as Streams
---------------------------------

The package `Text_IO.Streams' allows a Text_IO file to be treated as a
stream. Data written to a Text_IO file in this stream mode is binary
data. If this binary data contains bytes 16#0A# (`LF') or 16#0C#
(`FF'), the resulting file may have non-standard format. Similarly if
read operations are used to read from a Text_IO file treated as a
stream, then `LF' and `FF' characters may be skipped and the effect is
similar to that described above for `Get_Immediate'.

Wide_Text_IO
============

`Wide_Text_IO' is similar in most respects to Text_IO, except that both
input and output files may contain special sequences that represent
wide character values. The encoding scheme for a given file may be
specified using a FORM parameter:

     WCEM=X

as part of the FORM string (WCEM = wide character encoding method),
where X is one of the following characters

`h'
     Hex ESC encoding

`u'
     Upper half encoding

`s'
     Shift-JIS encoding

`e'
     EUC Encoding

`8'
     UTF-8 encoding

`b'
     Brackets encoding

   The encoding methods match those that can be used in a source
program, but there is no requirement that the encoding method used for
the source program be the same as the encoding method used for files,
and different files may use different encoding methods.

   The default encoding method for the standard files, and for opened
files for which no WCEM parameter is given in the FORM string matches
the wide character encoding specified for the main program (the default
being brackets encoding if no coding method was specified with -gnatW).

Hex Coding
     In this encoding, a wide character is represented by a five
     character sequence:

          ESC a b c d

     where A, B, C, D are the four hexadecimal characters (using upper
     case letters) of the wide character code. For example, ESC A345 is
     used to represent the wide character with code 16#A345#. This
     scheme is compatible with use of the full `Wide_Character' set.

Upper Half Coding
     The wide character with encoding 16#abcd#, where the upper bit is
     on (i.e. a is in the range 8-F) is represented as two bytes 16#ab#
     and 16#cd#. The second byte may never be a format control
     character, but is not required to be in the upper half. This
     method can be also used for shift-JIS or EUC where the internal
     coding matches the external coding.

Shift JIS Coding
     A wide character is represented by a two character sequence 16#ab#
     and 16#cd#, with the restrictions described for upper half
     encoding as described above. The internal character code is the
     corresponding JIS character according to the standard algorithm
     for Shift-JIS conversion. Only characters defined in the JIS code
     set table can be used with this encoding method.

EUC Coding
     A wide character is represented by a two character sequence 16#ab#
     and 16#cd#, with both characters being in the upper half. The
     internal character code is the corresponding JIS character
     according to the EUC encoding algorithm. Only characters defined
     in the JIS code set table can be used with this encoding method.

UTF-8 Coding
     A wide character is represented using UCS Transformation Format 8
     (UTF-8) as defined in Annex R of ISO 10646-1/Am.2.  Depending on
     the character value, the representation is a one, two, or three
     byte sequence:

          16#0000#-16#007f#: 2#0xxxxxxx#
          16#0080#-16#07ff#: 2#110xxxxx# 2#10xxxxxx#
          16#0800#-16#ffff#: 2#1110xxxx# 2#10xxxxxx# 2#10xxxxxx#

     where the xxx bits correspond to the left-padded bits of the the
     16-bit character value. Note that all lower half ASCII characters
     are represented as ASCII bytes and all upper half characters and
     other wide characters are represented as sequences of upper-half
     (The full UTF-8 scheme allows for encoding 31-bit characters as
     6-byte sequences, but in this implementation, all UTF-8 sequences
     of four or more bytes length will raise a Constraint_Error, as
     will all illegal UTF-8 sequences.)

Brackets Coding
     In this encoding, a wide character is represented by the following
     eight character sequence:

          [ " a b c d " ]

     Where `a', `b', `c', `d' are the four hexadecimal characters
     (using uppercase letters) of the wide character code. For example,
     `["A345"]' is used to represent the wide character with code
     `16#A345#'.  This scheme is compatible with use of the full
     Wide_Character set.  On input, brackets coding can also be used
     for upper half characters, e.g. `["C1"]' for lower case a.
     However, on output, brackets notation is only used for wide
     characters with a code greater than `16#FF#'.

   For the coding schemes other than Hex and Brackets encoding, not all
wide character values can be represented. An attempt to output a
character that cannot be represented using the encoding scheme for the
file causes Constraint_Error to be raised. An invalid wide character
sequence on input also causes Constraint_Error to be raised.

Stream Pointer Positioning
--------------------------

`Ada.Wide_Text_IO' is similar to `Ada.Text_IO' in its handling of
stream pointer positioning (*note Text_IO::.). There is one additional
case:

   If `Ada.Wide_Text_IO.Look_Ahead' reads a character outside the
normal lower ASCII set (i.e. a character in the range:

     Wide_Character'Val (16#0080#) .. Wide_Character'Val (16#FFFF#)

then although the logical position of the file pointer is unchanged by
the `Look_Ahead' call, the stream is physically positioned past the
wide character sequence. Again this is to avoid the need for buffering
or backup, and all `Wide_Text_IO' routines check the internal
indication that this situation has occurred so that this is not visible
to a normal program using `Wide_Text_IO'. However, this discrepancy can
be observed if the wide text file shares a stream with another file.

Reading and Writing Non-Regular Files
-------------------------------------

As in the case of Text_IO, when a non-regular file is read, it is
assumed that the file contains no page marks (any form characters are
treated as data characters), and `End_Of_Page' always returns `False'.
Similarly, the end of file indication is not sticky, so it is possible
to read beyond an end of file.

Stream_IO
=========

A stream file is a sequence of bytes, where individual elements are
written to the file as described in the Ada 95 reference manual. The
type `Stream_Element' is simply a byte. There are two ways to read or
write a stream file.

   * The operations `Read' and `Write' directly read or write a
     sequence of stream elements with no control information.

   * The stream attributes applied to a stream file transfer data in the
     manner described for stream attributes.

Shared Files
============

Section A.14 of the Ada 95 Reference Manual allows implementations to
provide a wide variety of behavior if an attempt is made to access the
same external file with two or more internal files.

   To provide a full range of functionality, while at the same time
minimizing the problems of portability caused by this implementation
dependence, GNAT handles file sharing as follows:

   * In the absence of a `shared=XXX' form parameter, an attempt to
     open two or more files with the same full name is considered an
     error and is not supported. The exception `Use_Error' will be
     raised. Note that a file that is not explicitly closed by the
     program remains open until the program terminates.

   * If the form parameter `shared=no' appears in the form string, the
     file can be opened or created with its own separate stream
     identifier, regardless of whether other files sharing the same
     external file are opened. The exact effect depends on how the C
     stream routines handle multiple accesses to the same external
     files using separate streams.

   * If the form parameter `shared=yes' appears in the form string for
     each of two or more files opened using the same full name, the same
     stream is shared between these files, and the semantics are as
     described in Ada 95 Reference Manual, Section A.14.

   When a program that opens multiple files with the same name is ported
from another Ada compiler to GNAT, the effect will be that `Use_Error'
is raised.

   The documentation of the original compiler and the documentation of
the program should then be examined to determine if file sharing was
expected, and `shared=XXX' parameters added to `Open' and `Create'
calls as required.

   When a program is ported from GNAT to some other Ada compiler, no
special attention is required unless the `shared=XXX' form parameter is
used in the program. In this case, you must examine the documentation
of the new compiler to see if it supports the required file sharing
semantics, and form strings modified appropriately. Of course it may be
the case that the program cannot be ported if the target compiler does
not support the required functionality. The best approach in writing
portable code is to avoid file sharing (and hence the use of the
`shared=XXX' parameter in the form string) completely.

   One common use of file sharing in Ada 83 is the use of
instantiations of Sequential_IO on the same file with different types,
to achieve heterogenous input-output. Although this approach will work
in GNAT if `shared=yes' is specified, it is preferable in Ada 95 to use
Stream_IO for this purpose (using the stream attributes)

Open Modes
==========

`Open' and `Create' calls result in a call to `fopen' using the mode
shown in Table 6.1

               Table 6-1 `Open' and `Create' Call Modes

                                    OPEN            CREATE
     Append_File                    "r+"             "w+"
     In_File                        "r"              "w+"
     Out_File (Direct_IO)           "r+"             "w"
     Out_File (all other cases)     "w"              "w"
     Inout_File                     "r+"             "w+"

   If text file translation is required, then either `b' or `t' is
added to the mode, depending on the setting of Text. Text file
translation refers to the mapping of CR/LF sequences in an external file
to LF characters internally. This mapping only occurs in DOS and
DOS-like systems, and is not relevant to other systems.

   A special case occurs with Stream_IO. As shown in the above table,
the file is initially opened in `r' or `w' mode for the `In_File' and
`Out_File' cases. If a `Set_Mode' operation subsequently requires
switching from reading to writing or vice-versa, then the file is
reopened in `r+' mode to permit the required operation.

Operations on C Streams
=======================

   The package `Interfaces.C_Streams' provides an Ada program with
direct access to the C library functions for operations on C streams:

     package Interfaces.C_Streams is
       -- Note: the reason we do not use the types that are in
       -- Interfaces.C is that we want to avoid dragging in the
       -- code in this unit if possible.
       subtype chars is System.Address;
       -- Pointer to null-terminated array of characters
       subtype FILEs is System.Address;
       -- Corresponds to the C type FILE*
       subtype voids is System.Address;
       -- Corresponds to the C type void*
       subtype int is Integer;
       subtype long is Long_Integer;
       -- Note: the above types are subtypes deliberately, and it
       -- is part of this spec that the above correspondences are
       -- guaranteed. This means that it is legitimate to, for
       -- example, use Integer instead of int. We provide these
       -- synonyms for clarity, but in some cases it may be
       -- convenient to use the underlying types (for example to
       -- avoid an unnecessary dependency of a spec on the spec
       -- of this unit).
       type size_t is mod 2 ** Standard'Address_Size;
       NULL_Stream : constant FILEs;
       -- Value returned (NULL in C) to indicate an
       -- fdopen/fopen/tmpfile error
       ----------------------------------
       -- Constants Defined in stdio.h --
       ----------------------------------
       EOF : constant int;
       -- Used by a number of routines to indicate error or
       -- end of file
       IOFBF : constant int;
       IOLBF : constant int;
       IONBF : constant int;
       -- Used to indicate buffering mode for setvbuf call
       SEEK_CUR : constant int;
       SEEK_END : constant int;
       SEEK_SET : constant int;
       -- Used to indicate origin for fseek call
       function stdin return FILEs;
       function stdout return FILEs;
       function stderr return FILEs;
       -- Streams associated with standard files
       --------------------------
       -- Standard C functions --
       --------------------------
       -- The functions selected below are ones that are
       -- available in DOS, OS/2, UNIX and Xenix (but not
       -- necessarily in ANSI C). These are very thin interfaces
       -- which copy exactly the C headers. For more
       -- documentation on these functions, see the Microsoft C
       -- "Run-Time Library Reference" (Microsoft Press, 1990,
       -- ISBN 1-55615-225-6), which includes useful information
       -- on system compatibility.
       procedure clearerr (stream : FILEs);
       function fclose (stream : FILEs) return int;
       function fdopen (handle : int; mode : chars) return FILEs;
       function feof (stream : FILEs) return int;
       function ferror (stream : FILEs) return int;
       function fflush (stream : FILEs) return int;
       function fgetc (stream : FILEs) return int;
       function fgets (strng : chars; n : int; stream : FILEs)
           return chars;
       function fileno (stream : FILEs) return int;
       function fopen (filename : chars; Mode : chars)
           return FILEs;
       -- Note: to maintain target independence, use
       -- text_translation_required, a boolean variable defined in
       -- a-sysdep.c to deal with the target dependent text
       -- translation requirement. If this variable is set,
       -- then  b/t should be appended to the standard mode
       -- argument to set the text translation mode off or on
       -- as required.
       function fputc (C : int; stream : FILEs) return int;
       function fputs (Strng : chars; Stream : FILEs) return int;
       function fread
          (buffer : voids;
           size : size_t;
           count : size_t;
           stream : FILEs)
           return size_t;
       function freopen
          (filename : chars;
           mode : chars;
           stream : FILEs)
           return FILEs;
       function fseek
          (stream : FILEs;
           offset : long;
           origin : int)
           return int;
       function ftell (stream : FILEs) return long;
       function fwrite
          (buffer : voids;
           size : size_t;
           count : size_t;
           stream : FILEs)
           return size_t;
       function isatty (handle : int) return int;
       procedure mktemp (template : chars);
       -- The return value (which is just a pointer to template)
       -- is discarded
       procedure rewind (stream : FILEs);
       function rmtmp return int;
       function setvbuf
          (stream : FILEs;
           buffer : chars;
           mode : int;
           size : size_t)
           return int;
     
       function tmpfile return FILEs;
       function ungetc (c : int; stream : FILEs) return int;
       function unlink (filename : chars) return int;
       ---------------------
       -- Extra functions --
       ---------------------
       -- These functions supply slightly thicker bindings than
       -- those above. They are derived from functions in the
       -- C Run-Time Library, but may do a bit more work than
       -- just directly calling one of the Library functions.
       function is_regular_file (handle : int) return int;
       -- Tests if given handle is for a regular file (result 1)
       -- or for a non-regular file (pipe or device, result 0).
       ---------------------------------
       -- Control of Text/Binary Mode --
       ---------------------------------
       -- If text_translation_required is true, then the following
       -- functions may be used to dynamically switch a file from
       -- binary to text mode or vice versa. These functions have
       -- no effect if text_translation_required is false (i.e. in
       -- normal UNIX mode). Use fileno to get a stream handle.
       procedure set_binary_mode (handle : int);
       procedure set_text_mode (handle : int);
       ----------------------------
       -- Full Path Name support --
       ----------------------------
       procedure full_name (nam : chars; buffer : chars);
       -- Given a NUL terminated string representing a file
       -- name, returns in buffer a NUL terminated string
       -- representing the full path name for the file name.
       -- On systems where it is relevant the   drive is also
       -- part of the full path name. It is the responsibility
       -- of the caller to pass an actual parameter for buffer
       -- that is big enough for any full path name. Use
       -- max_path_len given below as the size of buffer.
       max_path_len : integer;
       -- Maximum length of an allowable full path name on the
       -- system, including a terminating NUL character.
     end Interfaces.C_Streams;

Interfacing to C Streams
========================

The packages in this section permit interfacing Ada files to C Stream
operations.

      with Interfaces.C_Streams;
      package Ada.Sequential_IO.C_Streams is
         function C_Stream (F : File_Type)
            return Interfaces.C_Streams.FILEs;
         procedure Open
           (File : in out File_Type;
            Mode : in File_Mode;
            C_Stream : in Interfaces.C_Streams.FILEs;
            Form : in String := "");
      end Ada.Sequential_IO.C_Streams;
     
       with Interfaces.C_Streams;
       package Ada.Direct_IO.C_Streams is
          function C_Stream (F : File_Type)
             return Interfaces.C_Streams.FILEs;
          procedure Open
            (File : in out File_Type;
             Mode : in File_Mode;
             C_Stream : in Interfaces.C_Streams.FILEs;
             Form : in String := "");
       end Ada.Direct_IO.C_Streams;
     
       with Interfaces.C_Streams;
       package Ada.Text_IO.C_Streams is
          function C_Stream (F : File_Type)
             return Interfaces.C_Streams.FILEs;
          procedure Open
            (File : in out File_Type;
             Mode : in File_Mode;
             C_Stream : in Interfaces.C_Streams.FILEs;
             Form : in String := "");
       end Ada.Text_IO.C_Streams;
     
     
       with Interfaces.C_Streams;
       package Ada.Wide_Text_IO.C_Streams is
          function C_Stream (F : File_Type)
             return Interfaces.C_Streams.FILEs;
          procedure Open
            (File : in out File_Type;
             Mode : in File_Mode;
             C_Stream : in Interfaces.C_Streams.FILEs;
             Form : in String := "");
      end Ada.Wide_Text_IO.C_Streams;
     
      with Interfaces.C_Streams;
      package Ada.Stream_IO.C_Streams is
         function C_Stream (F : File_Type)
            return Interfaces.C_Streams.FILEs;
         procedure Open
           (File : in out File_Type;
            Mode : in File_Mode;
            C_Stream : in Interfaces.C_Streams.FILEs;
            Form : in String := "");
      end Ada.Stream_IO.C_Streams;

   In each of these five packages, the `C_Stream' function obtains the
`FILE' pointer from a currently opened Ada file. It is then possible to
use the `Interfaces.C_Streams' package to operate on this stream, or
the stream can be passed to a C program which can operate on it
directly. Of course the program is responsible for ensuring that only
appropriate sequences of operations are executed.

   One particular use of relevance to an Ada program is that the
`setvbuf' function can be used to control the buffering of the stream
used by an Ada file. In the absence of such a call the standard default
buffering is used.

   The `Open' procedures in these packages open a file giving an
existing C Stream instead of a file name. Typically this stream is
imported from a C program, allowing an Ada file to operate on an
existing C file.

Interfacing to Other Languages
******************************

The facilities in annex B of the Ada 95 Reference Manual are fully
implemented in GNAT, and in addition, a full interface to C++ is
provided.

Interfacing to C
================

Interfacing to C with GNAT can use one of two approaches:

  1. The types in the package `Interfaces.C' may be used.

  2. Standard Ada types may be used directly. This may be less portable
     to other compilers, but will work on all GNAT compilers, which
     guarantee correspondence between the C and Ada types.

Pragma `Convention C' maybe applied to Ada types, but mostly has no
effect, since this is the default. The following table shows the
correspondence between Ada scalar types and the corresponding C types.

`Integer'
     `int'

`Short_Integer'
     `short'

`Short_Short_Integer'
     `signed char'

`Long_Integer'
     `long'

`Long_Long_Integer'
     `long long'

`Short_Float'
     `float'

`Float'
     `float'

`Long_Float'
     `double'

`Long_Long_Float'
     This is the longest floating-point type supported by the hardware.
     Sometimes, this is the same as `Long_Float', i.e. as the C type
     `double'.  Otherwise, it is a wider type which is also available
     as `long double' in GNU C.

   * Ada enumeration types map to C enumeration types directly if pragma
     `Convention C' is specified, which causes them to have int length.
     Without pragma `Convention C', Ada enumeration types map to 8, 16,
     or 32 bits (i.e. C types signed char, short, int respectively)
     depending on the number of values passed. This is the only case in
     which pragma `Convention C' affects the representation of an Ada
     type.

   * Ada access types map to C pointers, except for the case of
     pointers to unconstrained types in Ada, which have no direct C
     equivalent.

   * Ada arrays map directly to C arrays.

   * Ada records map directly to C structures.

   * Packed Ada records map to C structures where all members are bit
     fields of the length corresponding to the `TYPE'Size' value in Ada.

Interfacing to C++
==================

The interface to C++ makes use of the following pragmas, which are
primarily intended to be constructed automatically using a binding
generator tool, although it is possible to construct them by hand. Ada
Core Technologies does not currently supply a suitable binding
generator tool.

   Using these pragmas it is possible to achieve complete
inter-operability between Ada tagged types and C class definitions.
*Note Implementation Defined Pragmas:: for more details.

`pragma CPP_Class ([Entity =>] LOCAL_NAME)'
     The argument denotes an entity in the current declarative region
     that is declared as a tagged or untagged record type. It indicates
     that the type corresponds to an externally declared C++ class
     type, and is to be laid out the same way that C++ would lay out
     the type.

`pragma CPP_Constructor ([Entity =>] LOCAL_NAME)'
     This pragma identifies an imported function (imported in the usual
     way with pragma `Import') as corresponding to a C++ constructor.

`pragma CPP_Vtable ...'
     One `CPP_Vtable' pragma can be present for each component of type
     `CPP.Interfaces.Vtable_Ptr' in a record to which pragma `CPP_Class'
     applies.

Interfacing to COBOL
====================

Interfacing to COBOL is achieved as described in section B.4 of the Ada
95 reference manual.

Interfacing to Fortran
======================

Interfacing to Fortran is achieved as described in section B.5 of the
reference manual. The pragma `Convention Fortran', applied to a multi-
dimensional array causes the array to be stored in column-major order
as required for convenient interface to Fortran.

Interfacing to non-GNAT Ada code
================================

   It is possible to specify the convention Ada in a pragma Import or
pragma Export. However this refers to the calling conventions used by
GNAT, which may or may not be similar enough to those used by some
other Ada 83 or Ada 95 compiler to allow interoperation.

   If arguments types are kept simple, and if the foreign compiler
generally follows system calling conventions, then it may be possible
to integrate files compiled by other Ada compilers, provided that the
elaboration issues are adequately addressed (for example by eliminating
the need for any load time elaboration).

   In particular, GNAT running on VMS is designed to be highly
compatible with the DEC Ada 83 compiler, so this is one case in which
it is possible to import foreign units of this type, provided that the
data items passed are restricted to simple scalar values or simple
record types without variants, or simple array types with fixed bounds.

Machine Code Insertions
***********************

Package `Machine_Code' provides machine code support as described in
the Ada 95 Reference Manual in two separate forms:
   * Machine code statements, consisting of qualified expressions that
     fit the requirements of RM section 13.8.

   * An intrinsic callable procedure, providing an alternative
     mechanism of including machine instructions in a subprogram.

   The two features are similar, and both closely related to the
mechanism provided by the asm instruction in the GNU C compiler. Full
understanding and use of the facilities in this package requires
understanding the asm instruction as described in `Using and Porting
GNU CC' by Richard Stallman.  Calls to the function `Asm' and the
procedure `Asm' have identical semantic restrictions and effects as
described below.  Both are provided so that the procedure call can be
used as a statement, and the function call can be used to form a
code_statement.

   The first example given in the GNU CC documentation is the C `asm'
instruction:
        asm ("fsinx %1 %0" : "=f" (result) : "f" (angle));

The equivalent can be written for GNAT as:

        Asm ("fsinx %1 %0",
             My_Float'Asm_Output ("=f", result),
             My_Float'Asm_Input  ("f",  angle));

   The first argument to `Asm' is the assembler template, and is
identical to what is used in GNU CC. This string must be a static
expression.  The second argument is the output operand list. It is
either a single `Asm_Output' attribute reference, or a list of such
references enclosed in parentheses (technically an array aggregate of
such references).

   The `Asm_Output' attribute denotes a function that takes two
parameters.  The first is a string, the second is the name of a variable
of the type designated by the attribute prefix. The first (string)
argument is required to be a static expression and designates the
constraint for the parameter (e.g. what kind of register is required).
The second argument is the variable to be updated with the result. The
possible values for constraint are the same as those used in the RTL,
and are dependent on the configuration file used to build the GCC back
end.  If there are no output operands, then this argument may either be
omitted, or explicitly given as `No_Output_Operands'.

   The second argument of `MY_FLOAT'Asm_Output' functions as though it
were an `out' parameter, which is a little curious, but all names have
the form of expressions, so there is no syntactic irregularity, even
though normally functions would not be permitted `out' parameters.  The
third argument is the list of input operands. It is either a single
`Asm_Input' attribute reference, or a list of such references enclosed
in parentheses (technically an array aggregate of such references).

   The `Asm_Input' attribute denotes a function that takes two
parameters.  The first is a string, the second is an expression of the
type designated by the prefix. The first (string) argument is required
to be a static expression, and is the constraint for the parameter,
(e.g. what kind of register is required). The second argument is the
value to be used as the input argument. The possible values for the
constant are the same as those used in the RTL, and are dependent on
the configuration file used to built the GCC back end.

   If there are no input operands, this argument may either be omitted,
or explicitly given as `No_Input_Operands'.  The fourth argument, not
present in the above example, is a list of register names, called the
"clobber" argument. This argument, if given, must be a static string
expression, and is a space or comma separated list of names of registers
that must be considered destroyed as a result of the `Asm' call. If
this argument is the null string (the default value), then the code
generator assumes that no additional registers are destroyed.

   The fifth argument, not present in the above example, called the
"volatile" argument, is by default `False'. It can be set to the
literal value `True' to indicate to the code generator that all
optimizations with respect to the instruction specified should be
suppressed, and that in particular, for an instruction that has outputs,
the instruction will still be generated, even if none of the outputs are
used. See the full description in the GCC manual for further details.

   The `Asm' subprograms may be used in two ways. First the procedure
forms can be used anywhere a procedure call would be valid, and
correspond to what the RM calls "intrinsic" routines. Such calls can be
used to intersperse machine instructions with other Ada statements.
Second, the function forms, which return a dummy value of the limited
private type `Asm_Insn', can be used in code statements, and indeed
this is the only context where such calls are allowed.  Code statements
appear as aggregates of the form:

        Asm_Insn'(Asm (...));
        Asm_Insn'(Asm_Volatile (...));

   In accordance with RM rules, such code statements are allowed only
within subprograms whose entire body consists of such statements.  It is
not permissible to intermix such statements with other Ada statements.

   Typically the form using intrinsic procedure calls is more convenient
and more flexible. The code statement form is provided to meet the RM
suggestion that such a facility should be made available.  The following
is the exact syntax of the call to `Asm' (of course if named notation is
used, the arguments may be given in arbitrary order, following the
normal rules for use of positional and named arguments)

       ASM_CALL ::= Asm (
                        [Template =>] static_string_EXPRESSION
                      [,[Outputs  =>] OUTPUT_OPERAND_LIST      ]
                      [,[Inputs   =>] INPUT_OPERAND_LIST       ]
                      [,[Clobber  =>] static_string_EXPRESSION ]
                      [,[Volatile =>] static_boolean_EXPRESSION] )
       OUTPUT_OPERAND_LIST ::=
         No_Output_Operands
       | OUTPUT_OPERAND_ATTRIBUTE
       | (OUTPUT_OPERAND_ATTRIBUTE {,OUTPUT_OPERAND_ATTRIBUTE})
       OUTPUT_OPERAND_ATTRIBUTE ::=
         SUBTYPE_MARK'Asm_Output (static_string_EXPRESSION, NAME)
       INPUT_OPERAND_LIST ::=
         No_Input_Operands
       | INPUT_OPERAND_ATTRIBUTE
       | (INPUT_OPERAND_ATTRIBUTE {,INPUT_OPERAND_ATTRIBUTE})
       INPUT_OPERAND_ATTRIBUTE ::=
         SUBTYPE_MARK'Asm_Input (static_string_EXPRESSION, EXPRESSION)

GNAT Implementation of Tasking
******************************

Mapping Ada Tasks onto the Underlying Kernel Threads
====================================================

   GNAT run-time system comprises two layers:

   * GNARL (GNAT Run-time  Layer)

   * GNULL (GNAT Low-level Library)

   In GNAT, Ada's tasking services rely on a platform and OS independent
layer known as GNARL. This code is responsible for implementing the
correct semantics of Ada's task creation, rendez-vous, protected
operations etc.

   GNARL decomposes Ada's tasking semantics into simpler lower level
operations such as create a thread, set the priority of a thread,
yield, create a lock, lock/unlock, etc. The spec for these low-level
operations constitutes GNULLI, the GNULL Interface. This interface is
directly inspired from the POSIX real-time API.

   If the underlying executive or OS implements the POSIX standard
faithfully, the GNULL Interface maps as is to the services offered by
the underlying kernel. Otherwise, some target dependent glue code maps
the services offered by the underlying kernel to the semantics expected
by GNARL.

   Whatever the underlying OS (VxWorks, UNIX, OS/2, Windows NT, etc.)
the key point is that each Ada task is mapped on a thread in the
underlying kernel. For example, in the case of VxWorks

   1 Ada task = 1 VxWorks task

   In addition Ada task priorities map onto the underlying thread
priorities.  Mapping Ada tasks onto the underlying kernel threads has
several advantages:

  1. The underlying scheduler is used to schedule the Ada tasks. This
     makes Ada tasks as efficient as kernel threads from a scheduling
     standpoint.

  2. Interaction with code written in C containing threads is eased
     since at the lowest level Ada tasks and C threads map onto the same
     underlying kernel concept.

  3. When an Ada task is blocked during I/O the remaining Ada tasks are
     able to proceed.

  4. On multi-processor systems Ada Tasks can execute in parallel.

Ensuring Compliance with the Real-Time Annex
============================================

   The reader will be quick to notice that while mapping Ada tasks onto
the underlying threads has significant advantages, it does create some
complications when it comes to respecting the scheduling semantics
specified in the real-time annex (Annex D).

   For instance Annex D requires that for the FIFO_Within_Priorities
scheduling policy we have:

   9  When the active priority of a ready task that is not running
 changes, or the setting of its base priority takes effect, the
task is removed from the ready queue for its old active priority
and is added at the tail of the ready queue for its new active
priority, except in the case where the active priority is lowered
due to the loss of inherited priority, in which case the task is
added at the head of the ready queue for its new active priority.

   While most kernels do put tasks at the end of the priority queue when
a task changes its priority, (which respects the main
FIFO_Within_Priorities requirement), almost none keep a thread at the
beginning of its priority queue when its priority drops from the loss
of inherited priority.

   As a result most vendors have provided incomplete Annex D
implementations.

   The GNAT run-time, has a nice cooperative solution to this problem
which ensures that accurate FIFO_Within_Priorities semantics are
respected.

   The principle is as follows. When an Ada task T is about to start
running, it checks whether some other Ada task R with the same priority
as T has been suspended due to the loss of priority inheritance. If
this is the case, T yields and is placed at the end of its priority
queue. When R arrives at the front of the queue it executes.

   Note that this simple scheme preserves the relative order of the
tasks that were ready to execute in the priority queue where R has been
placed at the end.

Specialized Needs Annexes
*************************

Ada 95 defines a number of specialized needs annexes, which are not
required in all implementations. However, as described in this chapter,
GNAT implements all of these special needs annexes:

Systems Programming (Annex C)
     The systems programming annex is fully implemented.

Real-Time Systems (Annex D)
     The real-time systems annex is fully implemented.

Distributed Systems (Annex E)
     Stub generation is fully implemented, but no PCS is provided yet,
     so distributed systems cannot yet be constructed with this version
     of GNAT.

Information Systems (Annex F)
     The information systems annex is fully implemented.

Numerics (Annex G)
     The numerics annex is fully implemented.

Safety and Security (Annex H)
     The safety and security annex is fully implemented.

Obsolescent Features (Annex I)
     The obsolescent features annex is fully implemented.

Language Defined Attributes (Annex J)
     The language defined attributes annex is fully implemented.

Language Defined Pragmas (Annex K)
     The language defined pragmas annex is fully implemented.

Compatibility Guide
*******************

This chapter contains sections that describe compatibility issues
between GNAT and other Ada 83 and Ada 95 compilation systems, to aid in
porting applications developed in other Ada environments.

Compatibility with Ada 83
=========================

Ada 95 is designed to be highly upwards compatible with Ada 83. In
particular, the design intention is that the difficulties associated
with moving from Ada 83 to Ada 95 should be no greater than those that
occur when moving from one Ada 83 system to another.

   However, there are a number of points at which there are minor
incompatibilities. The Ada 95 Annotated Reference Manual contains full
details of these issues, and should be consulted for a complete
treatment.  In practice the following are the most likely issues to be
encountered.

Character range
     The range of Standard.Character is now the full 256 characters of
     Latin-1, whereas in most Ada 83 implementations it was restricted
     to 128 characters.  This may show up as compile time or runtime
     errors. The desirable fix is to adapt the program to accommodate
     the full character set, but in some cases it may be convenient to
     define a subtype or derived type of Character that covers only the
     restricted range.

New reserved words
     The identifiers `abstract', `aliased', `protected', `requeue',
     `tagged', and `until' are reserved in Ada 95.  Existing Ada 83
     code using any of these identifiers must be edited to use some
     alternative name.

Freezing rules
     The rules in Ada 95 are slightly different with regard to the
     point at which entities are frozen, and representation pragmas and
     clauses are not permitted past the freeze point. This shows up
     most typically in the form of an error message complaining that a
     representation item appears too late, and the appropriate
     corrective action is to move the item nearer to the declaration of
     the entity to which it refers.

     A particular case is that representation pragmas (including the
     extended DEC Ada 83 compatibility pragmas such as
     Export_Procedure), cannot be applied to a subprogram body. If
     necessary, a separate subprogram declaration must be introduced to
     which the pragma can be applied.

Optional bodies for library packages
     In Ada 83, a package that did not require a package body was
     nevertheless allowed to have one. This lead to certain surprises
     in compiling large systems (situations in which the body could be
     unexpectedly ignored). In Ada 95, if a package does not require a
     body then it is not permitted to have a body. To fix this problem,
     simply remove a redundant body if it is empty, or, if it is
     non-empty, introduce a dummy declaration into the spec that makes
     the body required. One approach is to add a private part to the
     package declaration (if necessary), and define a parameterless
     procedure called Requires_Body, which must then be given a dummy
     procedure body in the package body, which then becomes required.

Numeric_Error is now the same as Constraint_Error
     In Ada 95, the exception Numeric_Error is a renaming of
     Constraint_Error.  This means that it is illegal to have separate
     exception handlers for the two exceptions. The fix is simply to
     remove the handler for the Numeric_Error case (since even in Ada
     83, a compiler was free to raise Constraint_Error in place of
     Numeric_Error in all cases).

Indefinite subtypes in generics
     In Ada 83, it was permissible to pass an indefinite type (e.g.
     String) as the actual for a generic formal private type, but then
     the instantiation would be illegal if there were any instances of
     declarations of variables of this type in the generic body. In Ada
     95, to avoid this clear violation of the contract model, the
     generic declaration clearly indicates whether or not such
     instantiations are permitted. If a generic formal parameter has
     explicit unknown discriminants, indicated by using (<>) after the
     type name, then it can be instantiated with indefinite types, but
     no variables can be declared of this type. Any attempt to declare
     a variable will result in an illegality at the time the generic is
     declared. If the (<>) notation is not used, then it is illegal to
     instantiate the generic with an indefinite type. This will show up
     as a compile time error, and the fix is usually simply to add the
     (<>) to the generic declaration.

   All implementations of GNAT provide a switch that causes GNAT to
operate in Ada 83 mode. In this mode, some but not all compatibility
problems of the type described above are handled automatically. For
example, the new Ada 95 protected keywords are not recognized in this
mode. However, in practice, it is usually advisable to make the
necessary modifications to the program to remove the need for using
this switch.

Compatibility with Other Ada 95 Systems
=======================================

Providing that programs avoid the use of implementation dependent and
implementation defined features of Ada 95, as documented in the Ada 95
reference manual, there should be a high degree of portability between
GNAT and other Ada 95 systems. The following are specific items which
have proved troublesome in moving GNAT programs to other Ada 95
compilers, but do not affect porting code to GNAT.

Ada 83 Pragmas and Attributes
     Ada 95 compilers are allowed, but not required, to implement the
     missing Ada 83 pragmas and attributes that are no longer defined
     in Ada 95.  GNAT implements all such pragmas and attributes,
     eliminating this as a compatibility concern, but some other Ada 95
     compilers reject these pragmas and attributes.

Special-needs Annexes
     GNAT implements the full set of special needs annexes. At the
     current time, it is the only Ada 95 compiler to do so. This means
     that programs making use of these features may not be portable to
     other Ada 95 compilation systems.

Representation Clauses
     Some other Ada 95 compilers implement only the minimal set of
     representation clauses required by the Ada 95 reference manual.
     GNAT goes far beyond this minimal set, as described in the next
     section.

Representation Clauses
======================

The Ada 83 reference manual was quite vague in describing both the
minimal required implementation of representation clauses, and also
their precise effects. The Ada 95 reference manual is much more
explicit, but the minimal set of capabilities required in Ada 95 is
quite limited.

   GNAT implements the full required set of capabilities described in
the Ada 95 reference manual, but also goes much beyond this, and in
particular an effort has been made to be compatible with existing Ada
83 usage to the greatest extent possible.

   A few cases exist in which Ada 83 compiler behavior is incompatible
with requirements in the Ada 95 reference manual. These are instances of
intentional or accidental dependence on specific implementation
dependent characteristics of these Ada 83 compilers. The following is a
list of the cases most likely to arise in existing legacy Ada 83 code.

Implicit Packing
     Some Ada 83 compilers allowed a Size specification to cause
     implicit packing of an array or record. This is specifically
     disallowed by implementation advice in the Ada 83 reference manual
     (for good reason, this usage can cause expensive implicit
     conversions to occur in the code). The problem will show up as an
     error message rejecting the size clause. The fix is simply to
     provide the explicit pragma Pack.

Meaning of Size Attribute
     The Size attribute in Ada 95 for discrete types is defined as
     being the minimal number of bits required to hold values of the
     type. For example, on a 32-bit machine, the size of Natural will
     typically be 31 and not 32 (since no sign bit is required). Some
     Ada 83 compilers gave 31, and some 32 in this situation. This
     problem will usually show up as a compile time error, but not
     always. It is a good idea to check all uses of the 'Size attribute
     when porting Ada 83 code. The GNAT specific attribute Object_Size
     can provide a useful way of duplicating the behavior of some Ada
     83 compiler systems.

Size of Access Types
     A common assumption in Ada 83 code is that an access type is in
     fact a pointer, and that therefore it will be the same size as a
     System.Address value. This assumption is true for GNAT in most
     cases with one exception. For the case of a pointer to an
     unconstrained array type (where the bounds may vary from one value
     of the access type to another), the default is to use a "fat
     pointer", which is represented as two separate pointers, one to
     the bounds, and one to the array. This representation has a number
     of advantages, including improved efficiency. However, it may
     cause some difficulties in porting existing Ada 83 code which
     makes the assumption that, for example, pointers fit in 32 bits on
     a machine with 32-bit addressing.

     To get around this problem, GNAT also permits the use of "thin
     pointers" for access types in this case (where the designated type
     is an unconstrained array type). These thin pointers are indeed
     the same size as a System.Address value.  To specify a thin
     pointer, use a size clause for the type, for example:

          type X is access all String;
          for X'Size use System.Address'Size;

     which will cause the type X to be represented using a single
     pointer. When using this representation, the bounds are right
     behind the array. This representation is slightly less efficient,
     and does not allow quite such flexibility in the use of foreign
     pointers or in using the Unrestricted_Access attribute to create
     pointers to non-aliased objects. But for any standard portable use
     of the access type it will work in a functionally correct manner
     and allow porting of existing code. Note that another way of
     forcing a thin pointer representation is to use a component size
     clause for the element size in an array, or a record
     representation clause for an access field in a record.

Compatibility with DEC Ada 83
=============================

The VMS version of GNAT fully implements all the pragmas and attributes
provided by DEC Ada 83, as well as providing the standard DEC Ada 83
libraries, including Starlet. In addition, data layouts and parameter
passing conventions are highly compatible. This means that porting
existing DEC Ada 83 code to GNAT in VMS systems should be easier than
most other porting efforts. The following are some of the most
significant differences between GNAT and DEC Ada 83.

Default floating-point representation
     In GNAT, the default floating-point format is IEEE, whereas in DEC
     Ada 83, it is VMS format. GNAT does implement the necessary pragmas
     (Long_Float, Float_Representation) for changing this default.

System
     The package System in GNAT exactly corresponds to the definition
     in the Ada 95 reference manual, which means that it excludes many
     of the DEC Ada 83 extensions. However, a separate package Aux_DEC
     is provided that contains the additional definitions, and a
     special pragma, Extend_System allows this package to be treated
     transparently as an extension of package System.

To_Address
     The definitions provided by Aux_DEC are exactly compatible with
     those in the DEC Ada 83 version of System, with one exception. DEC
     Ada provides the following declarations:

             TO_ADDRESS(INTEGER)
             TO_ADDRESS(UNSIGNED_LONGWORD)
             TO_ADDRESS(universal_integer)

     The version of TO_ADDRESS taking a universal integer argument is
     in fact an extension to Ada 83 not strictly compatible with the
     reference manual.  In GNAT, we are constrained to be exactly
     compatible with the standard, and this means we cannot provide
     this capability. In DEC Ada 83, the point of this definition is to
     deal with a call like:

             TO_ADDRESS (16#12777#);

     Normally, according to the Ada 83 standard, one would expect this
     to be ambiguous, since it matches both the INTEGER and
     UNSIGNED_LONGWORD forms of TO_ADDRESS. However, in DEC Ada 83,
     there is no ambiguity, since the definition using
     universal_integer takes precedence.

     In GNAT, since the version with universal_integer cannot be
     supplied, it is not possible to be 100% compatible. Since there
     are many programs using numeric constants for the argument to
     TO_ADDRESS, the decision in GNAT was to change the name of the
     function in the UNSIGNED_LONGWORD case, so the declarations
     provided in the GNAT version of AUX_Dec are:

             function To_Address (X : Integer) return Address;
             pragma Pure_Function (To_Address);
          
             function To_Address_Long (X : Unsigned_Longword) return Address;
             pragma Pure_Function (To_Address_Long);

     This means that programs using TO_ADDRESS for UNSIGNED_LONGWORD
     must change the name to TO_ADDRESS_LONG.

Task_Id values
     The Task_Id values assigned will be different in the two systems,
     and GNAT does not provide a specified value for the Task_Id of the
     environment task, which in GNAT is treated like any other declared
     task.

   For full details on these and other less significant compatibility
issues, see appendix E of the Digital publication entitled "DEC Ada,
Technical Overview and Comparison on DIGITAL Platforms".

   For GNAT running on other than VMS systems, all the DEC Ada 83
pragmas and attributes are recognized, although only a subset of them
can sensibly be implemented. The description of pragmas in this
reference manual indicates whether or not they are applicable to
non-VMS systems.

Index
*****

* Menu:

* Abort_Defer:                           Implementation Defined Pragmas.
* Abort_Signal:                          Implementation Defined Attributes.
* Access, unrestricted:                  Implementation Defined Attributes.
* Accuracy requirements:                 Implementation Advice.
* Accuracy, complex arithmetic:          Implementation Advice.
* Ada 83 attributes:                     Implementation Defined Attributes.
* Ada 95 ISO/ANSI Standard:              What This Reference Manual Contains.
* Ada.Characters.Handling:               Implementation Advice.
* Ada_83:                                Implementation Defined Pragmas.
* Ada_95:                                Implementation Defined Pragmas.
* Address clauses:                       Implementation Advice.
* Address of subprogram code:            Implementation Defined Attributes.
* Address, as private type:              Implementation Advice.
* Address, operations of:                Implementation Advice.
* Address_Size:                          Implementation Defined Attributes.
* Alignment clauses:                     Implementation Advice.
* Alignment, maximum:                    Implementation Defined Attributes.
* Alignments of components:              Implementation Defined Pragmas.
* Alternative Character Sets:            Implementation Advice.
* Annotate:                              Implementation Defined Pragmas.
* Argument passing mechanisms:           Implementation Defined Pragmas.
* Arrays, multidimensional:              Implementation Advice.
* Asm_Input:                             Implementation Defined Attributes.
* Asm_Output:                            Implementation Defined Attributes.
* Assert:                                Implementation Defined Pragmas.
* AST_Entry:                             Implementation Defined Attributes.
* Ast_Entry:                             Implementation Defined Pragmas.
* Big endian:                            Implementation Defined Attributes.
* Bit:                                   Implementation Defined Attributes.
* Bit ordering:                          Implementation Advice.
* Bit_Position:                          Implementation Defined Attributes.
* Bounded errors:                        Implementation Advice.
* Bounded-length strings:                Implementation Advice.
* C, interfacing with:                   Implementation Advice.
* C_Pass_By_Copy:                        Implementation Defined Pragmas.
* Character Sets:                        Implementation Advice.
* Checks, suppression of:                Implementation Advice.
* Child Units:                           Implementation Advice.
* COBOL support:                         Implementation Advice.
* COBOL, interfacing with:               Implementation Advice.
* Code_Address:                          Implementation Defined Attributes.
* Comment:                               Implementation Defined Pragmas.
* Common_Object:                         Implementation Defined Pragmas.
* Complex arithmetic accuracy:           Implementation Advice.
* Complex elementary functions:          Implementation Advice.
* Complex types:                         Implementation Advice.
* Complex_Representation:                Implementation Defined Pragmas.
* Component_Alignment:                   Implementation Defined Pragmas.
* Component_Size:                        Implementation Defined Pragmas.
* Component_Size clauses:                Implementation Advice.
* Component_Size_4:                      Implementation Defined Pragmas.
* Conventions:                           Conventions.
* CPP_Class:                             Implementation Defined Pragmas.
* CPP_Constructor:                       Implementation Defined Pragmas.
* CPP_Destructor:                        Implementation Defined Pragmas.
* CPP_Virtual:                           Implementation Defined Pragmas.
* CPP_Vtable:                            Implementation Defined Pragmas.
* Debug:                                 Implementation Defined Pragmas.
* Dec Ada 83:                            Implementation Defined Pragmas.
* Decimal radix support:                 Implementation Advice.
* Default_Bit_Order:                     Implementation Defined Attributes.
* Deferring aborts:                      Implementation Defined Pragmas.
* Discriminants, testing for:            Implementation Defined Attributes.
* Duration'Small:                        Implementation Advice.
* Elab_Body:                             Implementation Defined Attributes.
* Elab_Spec:                             Implementation Defined Attributes.
* Elaborated:                            Implementation Defined Attributes.
* Eliminate:                             Implementation Defined Pragmas.
* Elimination of unused subprograms:     Implementation Defined Pragmas.
* Emax:                                  Implementation Defined Attributes.
* Entry queuing policies:                Implementation Advice.
* Enum_Rep:                              Implementation Defined Attributes.
* Enumeration representation clauses:    Implementation Advice.
* Enumeration values:                    Implementation Advice.
* Epsilon:                               Implementation Defined Attributes.
* Error detection:                       Implementation Advice.
* Exception information:                 Implementation Advice.
* Export:                                Implementation Advice.
* Export_Exception:                      Implementation Defined Pragmas.
* Export_Function:                       Implementation Defined Pragmas.
* Export_Object:                         Implementation Defined Pragmas.
* Export_Procedure:                      Implementation Defined Pragmas.
* Export_Valued_Procedure:               Implementation Defined Pragmas.
* Extend_System:                         Implementation Defined Pragmas.
* Finalize_Storage_Only:                 Implementation Defined Pragmas.
* Fixed_Value:                           Implementation Defined Attributes.
* Float types:                           Implementation Advice.
* Float_Representation:                  Implementation Defined Pragmas.
* Fortran, interfacing with:             Implementation Advice.
* Get_Immediate:                         Implementation Advice.
* Has_Discriminants:                     Implementation Defined Attributes.
* Heap usage, implicit:                  Implementation Advice.
* Ident:                                 Implementation Defined Pragmas.
* Img:                                   Implementation Defined Attributes.
* Implementation-dependent features:     About This Guide.
* Import_Exception:                      Implementation Defined Pragmas.
* Import_Function:                       Implementation Defined Pragmas.
* Import_Object:                         Implementation Defined Pragmas.
* Import_Procedure:                      Implementation Defined Pragmas.
* Import_Valued_Procedure:               Implementation Defined Pragmas.
* Initialization, suppression of:        Implementation Defined Pragmas.
* Inline_Always:                         Implementation Defined Pragmas.
* Inline_Generic:                        Implementation Defined Pragmas.
* Integer types:                         Implementation Advice.
* Integer_Value:                         Implementation Defined Attributes.
* Interface_Name:                        Implementation Defined Pragmas.
* Interfaces:                            Implementation Advice.
* Interfacing to C++:                    Implementation Defined Pragmas.
* Interfacing with C++:                  Implementation Defined Pragmas.
* Interrupt priority, maximum:           Implementation Defined Attributes.
* Interrupt support:                     Implementation Advice.
* Interrupts:                            Implementation Advice.
* Large:                                 Implementation Defined Attributes.
* Linker_Alias:                          Implementation Defined Pragmas.
* Linker_Section:                        Implementation Defined Pragmas.
* Little endian:                         Implementation Defined Attributes.
* Locking Policies:                      Implementation Advice.
* Long_Float:                            Implementation Defined Pragmas.
* Machine operations:                    Implementation Advice.
* Machine_Attribute:                     Implementation Defined Pragmas.
* Machine_Size:                          Implementation Defined Attributes.
* Main_Storage:                          Implementation Defined Pragmas.
* Mantissa:                              Implementation Defined Attributes.
* Max_Interrupt_Priority:                Implementation Defined Attributes.
* Max_Priority:                          Implementation Defined Attributes.
* Maximum_Alignment:                     Implementation Defined Attributes.
* Mechanism_Code:                        Implementation Defined Attributes.
* Multidimensional arrays:               Implementation Advice.
* Named numbers, representation of:      Implementation Defined Attributes.
* No_Complex_Barriers:                   Implementation Defined Characteristics.
* No_Elaboration_Code:                   Implementation Defined Characteristics.
* No_Implementation_Attributes:          Implementation Defined Characteristics.
* No_Implementation_Pragmas:             Implementation Defined Characteristics.
* No_Return:                             Implementation Defined Pragmas.
* No_Runtime:                            Implementation Defined Pragmas.
* No_Select_Statements:                  Implementation Defined Characteristics.
* Normalize_Scalars:                     Implementation Defined Pragmas.
* Null_Parameter:                        Implementation Defined Attributes.
* Object_Size:                           Implementation Defined Attributes.
* OpenVMS <1>:                           Implementation Defined Attributes.
* OpenVMS:                               Implementation Defined Pragmas.
* Operations, on Address:                Implementation Advice.
* Package Interfaces:                    Implementation Advice.
* Package Interrupts:                    Implementation Advice.
* Package Task_Attributes:               Implementation Advice.
* Packed types:                          Implementation Advice.
* Parameters, passing mechanism:         Implementation Defined Attributes.
* Parameters, when passed by reference:  Implementation Defined Attributes.
* Partition communication subsystem:     Implementation Advice.
* Passed_By_Reference:                   Implementation Defined Attributes.
* Passing by copy:                       Implementation Defined Pragmas.
* Passing by descriptor:                 Implementation Defined Pragmas.
* Passive:                               Implementation Defined Pragmas.
* PCS:                                   Implementation Advice.
* Polling:                               Implementation Defined Pragmas.
* Portability:                           About This Guide.
* Pragmas:                               Implementation Advice.
* Pre-elaboration requirements:          Implementation Advice.
* Preemptive abort:                      Implementation Advice.
* Priority, maximum:                     Implementation Defined Attributes.
* Propagate_Exceptions:                  Implementation Defined Pragmas.
* Protected procedure handlers:          Implementation Advice.
* Psect_Object:                          Implementation Defined Pragmas.
* Pure:                                  Implementation Defined Pragmas.
* Pure_Function:                         Implementation Defined Pragmas.
* Random number generation:              Implementation Advice.
* Range_Length:                          Implementation Defined Attributes.
* Ravenscar:                             Implementation Defined Pragmas.
* Record representation clauses:         Implementation Advice.
* Representation clauses:                Implementation Advice.
* Representation clauses, enumeration:   Implementation Advice.
* Representation clauses, records:       Implementation Advice.
* Representation of enums:               Implementation Defined Attributes.
* Restricted_Run_Time:                   Implementation Defined Pragmas.
* Return values, passing mechanism:      Implementation Defined Attributes.
* Safe_Emax:                             Implementation Defined Attributes.
* Safe_Large:                            Implementation Defined Attributes.
* Share_Generic:                         Implementation Defined Pragmas.
* Size clauses:                          Implementation Advice.
* Size of Address:                       Implementation Defined Attributes.
* Size, setting for not-first subtype:   Implementation Defined Attributes.
* Size, used for objects:                Implementation Defined Attributes.
* Size, VADS compatibility <1>:          Implementation Defined Attributes.
* Size, VADS compatibility:              Implementation Defined Pragmas.
* Small:                                 Implementation Defined Attributes.
* Source_File_Name:                      Implementation Defined Pragmas.
* Source_Reference:                      Implementation Defined Pragmas.
* Storage place attributes:              Implementation Advice.
* Storage_Unit <1>:                      Implementation Defined Attributes.
* Storage_Unit:                          Implementation Defined Pragmas.
* Stream oriented attributes:            Implementation Advice.
* Stream_Convert:                        Implementation Defined Pragmas.
* Subprogram address:                    Implementation Defined Attributes.
* Subtitle:                              Implementation Defined Pragmas.
* Suppress_All:                          Implementation Defined Pragmas.
* Suppress_Initialization:               Implementation Defined Pragmas.
* Suppressing initialization:            Implementation Defined Pragmas.
* Suppression of checks:                 Implementation Advice.
* system, extending:                     Implementation Defined Pragmas.
* Task_Attributes:                       Implementation Advice.
* Task_Info:                             Implementation Defined Pragmas.
* Task_Storage:                          Implementation Defined Pragmas.
* Tasking restrictions:                  Implementation Advice.
* Tick:                                  Implementation Defined Attributes.
* Time, monotonic:                       Implementation Advice.
* Time_Slice:                            Implementation Defined Pragmas.
* Title:                                 Implementation Defined Pragmas.
* To_Address:                            Implementation Defined Attributes.
* Type_Class:                            Implementation Defined Attributes.
* Typographical conventions:             Conventions.
* UET_Address:                           Implementation Defined Attributes.
* Unchecked conversion:                  Implementation Advice.
* Unchecked deallocation:                Implementation Advice.
* Unchecked_Union:                       Implementation Defined Pragmas.
* Unimplemented_Unit:                    Implementation Defined Pragmas.
* Unions in C:                           Implementation Defined Pragmas.
* Universal_Literal_String:              Implementation Defined Attributes.
* Unreserve_All_Interrupts:              Implementation Defined Pragmas.
* Unrestricted_Access:                   Implementation Defined Attributes.
* Unsuppress:                            Implementation Defined Pragmas.
* Use_VADS_Size:                         Implementation Defined Pragmas.
* VADS_Size:                             Implementation Defined Attributes.
* Value_Size:                            Implementation Defined Attributes.
* Volatile:                              Implementation Defined Pragmas.
* Warnings:                              Implementation Defined Pragmas.
* Weak_External:                         Implementation Defined Pragmas.
* Word_Size:                             Implementation Defined Attributes.
* Zero address, passing:                 Implementation Defined Attributes.
* Zero Cost Exceptions:                  Implementation Defined Pragmas.

