PASCAL on EMAS
--------------

CONTENTS

1.  INTRODUCTION

2.  COMPILATION
2.1 Compilation Options
2.2 Compilation Listings

3.  EXECUTION
3.1 Use of Input and Output Media
3.2 The Standard PASCAL Files INPUT and OUTPUT
3.3 Separate Compilation of PASCAL Procedures and Functions
3.4 Mixed Language Programming

4.  BASIC ERROR ACTION
4.1 Error Reporting
4.2 Compile-time Control
4.3 Restrictions and Limitations

5.  EMAS 2900 PASCAL
5.1 Lexical Conventions
5.2 Language Restrictions
5.3 Language Extensins
5.4 The Program Heading and External Files
5.5 Hardware Representation
5.6 The Standard Procedure WRITE

Appendix 1 Compilation error messages 
Appendix 2 Execution Error Messages


1.  INTRODUCTION

    The 2900 PASCAL Compiler is based on the "PASCAL User Manual and
    Report" (2nd Edition) by Kathleen Jensen and Nicklaus Wirth.  The
    language implemented conforms to the standard defined there, subject
    to the restrictions, extensions and further  specifications given
    in this documentation.

    The EMAS 2900 PASCAl compiler was developed by Southampton
    University and is distributed by ICL. PASCAL version 20 is
    currently in use.



2.  COMPILATION

    The command used to compile a PASCAL source program is

         PASCAL(<source>,<object>/.NULL,<clist>/.NULL)

    The three parameters are:

    1.  The name of the source file containing the program or routines
        to be compiled. Several files can be concatenated with the "+" 
        operator; for example A+B+C.

    2.  The name of the object file to be generated by the compiler.
        If a file of the given name exists it will be overwritten, if
        not a file will be created. The keyword .NULL is a valid
        alternative if no object file is required. This is a useful
        facility if the program is known to contain faults, since it
        reduces the compilation time. Note that this parameter cannot be
        a member of a partitioned file. An object file can, however, be
        copied into a partitioned file and used from there.

    3.  The name of a listing file or device. If this parameter is
        omitted a file created by the Subsystem with the name T#LIST
        will be used. Valid alternatives include output devices. Again,
        .NULL can be used if no listing is required. Note that T#LIST is
        destroyed when the user logs off,  and is overwritten if used
        in another compilation in the current session.

    eg. PASCAL(PTEST,PTESTY,.OUT)

2.1 Compilation Options

    The user can select various options for the PASCAL compilation
    process, for which default settings are provided.

    Global options are effective throughout the compilation process. 
    They are specified by the EMAS PARM command (see 'EMAS 2900 Users'
    Guide', 11-2) or by global PASCAL directives, which override any
    set by the PARM command.

    Local options can be changed during the compilation process and are
    specified by local PASCAL directives. Some local options can be
    switched  on/off al global level, using global directives. If a 
    local option is switched off globally, it is permanently off,
    i.e. all relevant local directives are ignored.

    The following table summarises the options available, with defaults
    specified first. Each option is detailed in 2.1.2 - 2.1.4.

    GLOBAL OPTIONS
        %CHARCODE=ISO/EBCDIC    or    PARM(ISO/EBCDIC)
        %DIAGLINEMAP=YES/NO     or    default/PARM(NODIAG)
        %DIAGNAMETABLE=YES/NO   or    default/PARM(NODIAG)
        %SOURCE=YES/NO          or    default/PARM(NOLIST)
        %RANGECHECKS=YES/NO     or    default/PARM(NOCHECK)
           PARM with no parameters resets the defaults.

    LOCAL OPTIONS
        %SOURCE ON/OFF
        %RANGECHECKS ON/OFF
        %KEYEDENTRY ON/OFF  (see also 3.3)
        %RHMARGIN <n>  where 1<=n<=96  (96 default)


2.1.1 Syntax of PASCAL Directives

    Each PASCAL directive occupies a single line of the source
    program and begins with the character % in the first position.
    A PASCAL directive line is not considered to be part of the source
    program  and syntactically incorrect or misplaced directives are
    ignored. In a directive line, any text to the right of the directive
    proper and clearly separated from it, for example by blanks, will
    be ignored.

    The syntax of a directive depends on its status as global or
    local.

    A global directive takes the form:
        %<option identifier>=<value>
    and must occur before the program heading.

    A local directive takes the form:
        %<option identifier> <value>
    and may appear anywhere in the text.

2.1.2 Global Options

   1. The character code option (CHARCODE)

      The character code option specifies the internal code to be 
      used to represent type CHAR in the object program. The possible
      values are EBCDIC/ISO. The default is %CHARCODE = ISO.

      This option is controlled by the EBCDIC/ISO parameter
      to PARM and by the CHARCODE directive which has
      the possible values EBCDIC/ISO.

   2. The diagnostics line map and name table options
      (DIAGLINEMAP and DIAGNAMETABLE)

      These options specify whether or not source line numbers and name
      tables respectively are to be recorded in the object file for use
      in diagnostics. By default, both options are on and only in
      unusual or specialised circumstances will the user require to
      switch them off.

      They are controlled by the DIAGLINEMAP and DIAGNAMETABLE 
      directives, which both have the possible values YES(on) and
      NO(off), and by the PARM command. PARM(NODIAG) specifies a
      setting of off. The default, on, can be set by calling PARM with
      no parameter.

2.1.3 Local Options with Global Switches

   1. The Source Listing Option (SOURCE)

      The source listing option specifies whether or not source lines 
      are to be listed by the compiler (see 2.2 for details of
      listings). The default setting is on.

      This option is controlled globally by the SOURCE global directive,
      which has the possible values YES(on) and NO(off), and by the
      PARM command. PARM(NOLIST) specifies a setting of off. The
      default, on, can be set by calling PARM with no paramter.

      The option is controlled locally by the SOURCE local directive,
      whose possible values are ON(on) and OFF(off).

   2. The Range Checks Option (RANGECHECKS)

      The range checks option specifies whether or not the compiler
      is to generate extra code where necessary to check that
      expressions of ordinal type lie within the required bounds.
      The default setting is on.

      The option is controlled globally by the RANGECHECKS global
      directive, which has the possible values YES(on) and NO(off), and
      by the PARM command. PARM(NOCHECK) specifies a setting of off.
      The default, on, can be set by calling PARM with no parameter.

      The option is controlled locally by the RANGECHECKS local
      directive, whose possible values are ON(on) and OFF(off).

    N.B.If a local option is switched off globally, it is permanently
        off, i.e. all relevant local directives are ignored.

2.1.4 Local Options

   1. The Keyed Entry Points Option (KEYEDENTRY)

      The keyed entry points option contols the entry status of global
      procedures/functions in the object file. If it is on for a
      particular procedure/function, the name  becomes an entry point
      in the object file and as such can be called from another program
      (see section 3.3.). The default setting is off.

      This option is controlled by the KEYEDENTRY directive whose
      possible values are ON(on) and OFF(off).

      Note:
      a) the setting in force when the procedure/function heading
         is encountered determines the entry status for that  procedure/
         function.
      b) The procedure/function must be global.

   2. The Right-hand Margin Option (RHMARGIN)

      The right-hand margin option specifies which character position 
      in each source line is to be treated as the last significant 
      character in that line. Characters beyond this position will 
      be listed but otherwise ignored by the compiler. The possible
      settings for this option are integers in the range 1 to 96.
      The initial setting is 96.

      This option is controlled by the RHMARGIN directive which
      takes integer values in the range 1 to 96.

2.2 Compilation Listings

    During compilation a printer listing is generated which lists the
    source program together with messages describing any errors found.
    The amount of information printed is under the control of the user
    via parameters to the PARM command and certain PASCAL directives
    as described in 2.1. The contents of the standard listing are:

    1. Headings

    2. Start of compilation listing

    3. Source lines

    4. Diagnostic messages

    5. End of compilation listing

2.2.1 Headings

    A heading is always output to the listing file, regardless of
    the listing options selected.  The heading provides such information
    as the source and object file names, the date and time of the 
    compilation, the PARM options selected and the version of the
    compiler.

2.2.2 Start of Compilation Listing

    This is a list of the compilation options (including defaults if
    used) together with the values to which they have been set for the
    compilation. The list cannot be suppressed.

2.2.3 Source Listing

    The source listing consists of each line of the source program 
    preceded by the following information:

    1. The line number

    2. None or more of the letters U,T,R,P and E, indicating the
       current status of certain compiler options. The option
       settings corresponding to the presence of these letters
       are:
       Letter    Option
         U       RANGECHECKS=OFF
         T       not used at this release
         R       not used at this release
         P       not used at this release
         E       KEYEDENTRY=ON

    3. The character ! which indicates the beginning of the source 
       line.

    PARM(NOLIST) can be used to specify the global suppression of source
    listing. Source lines in which errors are detected are listed even
    when source listing is suppressed.

    The listing of lines containing PASCAL directives is controlled by
    the same means as for lines containing standard PASCAL source.

    The presence of any non-blank characters beyond the right-hand
    margin is flagged in the source listing but is not regarded as an
    error.

2.2.4 Diagnostic Messages

    Any line of source that contains a compiler detected error is
    followed by a line containing the characters -
    ****** ERROR ******
    and one or more numbers, indicating the position and error number(s) 
    of the error(s) in this line. The line containing the error is
    produced even when the source listing option is off.

    A list of compilation error numbers and their meanings is
    given in Appendix 1.

2.2.5 End of Compilation Summary

    This summary provides the following information:

    1. A COMPILATION COMPLETE message, which also gives the  number
       of errors reported.

    2. If errors were reported,an ERROR SUMMARY,which lists the
       relevant error numbers with explanatory messages.

    3. If no errors were reported and an object program was generated,
       the virtual store requirements for the compiled program and the
       global option settings that affect the compiled program (for
       example,checks and character code).

    4. The number of lines in the source program, the OCP time for the
       compilation and, if object file was produced, its name.


3.  EXECUTION

    The command used to load and execute a compiled PASCAL program is 
    RUN. It takes one parameter, the name of the object file which
    contains the compiled program.

         EG. RUN(PTESTY)

    Before running a program, any external files in the PASCAL
    program heading must be associated with actual EMAS files using the
    DEFINE or PDEFINE command (see 3.1).

    In addition, it is possible for PASCAL programs to make use of
    separately compiled PASCAL procedures and functions, details
    of which are given in 3.3.

    Diagnostic facilites are described in section 4.

3.1 Use of Input and Output Media

    Before an object program can be executed, all the external files
    in the program heading must be associated with EMAS files.
    This is achieved by:-

    1. use of  the DEFINE command (see 'EMAS 2900 Users' Guide',
       chpt. 7), to associate an EMAS file with a channel number. PASCAL
       determines the channel number for a file from the position of the
       file in the program heading, so that the nth named file is
       connectec via channel n.

    2. use of the command PDEFINE(<PASCAL file>,<EMAS file>) which
       associates the PASCAL file directly with the EMAS file.

    For example, if the program heading is
                PROGRAM TRANSFER(INPUT,OUTPUT,FILEA);
    the command
                DEFINE(3,NEWRESULTS)
    connects FILEA to the EMAS file NEWRESULTS,

    or the command
                PDEFINE(FILEA,FRED)
    connects FILEA to the EMAS file FRED.

N.B. All file assosiations must be done before running a PASCAL
     program, i.e. PDEFINE or DEFINE cannot be called from within
     the PASCAL program.

    PCLEAR clears any file associations that have been PDEFINEd.
    (N.B. PCLEAR calls the command CLEAR, so any files that have been
    DEFINEd will also be cleared.)

3.2 The Standard PASCAL Files INPUT and OUTPUT

    The method of file association described in 3.1 may be used
    for the standard PASCAL files INPUT and OUTPUT. However, the
    following default connections are made for them if no other file
    association has been DEFINEd:

    1. The standard file parameter INPUT is assigned to .IN.

    2. The standard file parameter OUTPUT is assigned to .OUT.

N.B.If INPUT is specified in the program heading, the PASCAL file INPUT 
    is automaticallly initialised and thus input is expected from
    this file, even if no further reference is made to it within the
    program.

3.3 Separate Compilation of PASCAL Procedures and Functions

    If required, the user may create separately compiled  'library'
    files of PASCAL procedures and functions. Such procedures and
    functions are referred to as external routines.

    The source for a PASCAL library file takes the form of a
    standard PASCAL source program with the following special features:

    1. The program heading should be parameterless and no attempt
       should ever be made to execute the file.

    2. The external routines, which must be declared globally, must 
       have their status as entry points specified by appropriate
       control of the keyed entry option using the KEYEDENTRY
       directive (see 2.1.4).

    3. External routines cannot have procedures/functions as formal 
       parameters.

    4. If any input/output is to be done in an external routine, then
       the i/o files must be declared as formal var type parameters to
       the routine.

    5. The statement part of the program should consist of an empty
       component staement, i.e. the source should end with BEGIB END.

    A PASCAL program may use such external routines by declaring them
    globally to be external, using the special symbol EXTERN.

    For example, to create a library file with entry routines FNINT and
    FNRL:-
            PROGRAM NEWFUNS;
            %KEYEDENTRY ON
            FUNCTION FNINT(A:INTEGER):INTEGER;
            BEGIN
              |
              |
              |
            END;
            FUNCTION FNRL(A:REAL):REAL;
            BEGIN
              |
              |
              |
            END;
            BEGIN
            END.

    To use FNINT it must be declared globally, i.e. before the first
    BEGIN, within a program thus:-
            FUNCTION FNINT(A:INTEGER):INTEGER;EXTERN;

    Note that PASCAL library files may themselves have external
    procedure and function declarations, thus referring to other
    PASCAL library files.

3.4 Mixed Language Programming

    It is possible to call IMP routines, ALGOL procedures and
    FORTRAN subroutines from PASCAL by specifying them to be external
    procedures as above. Similarly, certain types of IMP functions, 
    ALGOL function procedures and FORTRAN functions can be called from 
    PASCAL by specifying them as external functions.

    The following table indicates the correspondence between 
    the PASCAL formal parameters and the ALGOL/IMP/FORTRAN formal
    parameters which are valid between the languages.

         PASCAL            ALGOL             IMP            FORTRAN

    INTEGER value     integer by value  %INTEGER         no equivalence
                      ------- -- -----
    INTEGER variable  integer           %INTEGERNAME     INTEGER*4
                      -------
    REAL value        real by value     %LONGREAL        no equivalence
                      ---- -- -----
    REAL variable     real              %LONGREALNAME    REAL*8
                      ----
    BOOLEAN value     boolean by value  %INTEGER         no equivalence
                      ------- -- -----
    BOOLEAN variable  boolean           %INTEGERNAME     LOGICAL*4
                      -------
    anything else     no equivalence    no equivalence   no equivalence

    In the case of functions, it is also necessary to ensure that the
    function result corresponds to a suitable type. The following table
    shows the correspondence between valid function types.

         PASCAL            ALGOL             IMP            FORTRAN

         INTEGER          integer         %INTEGER         INTEGER*4
                          -------
         REAL             real            %LONGREAL        REAL*8
                          ----
         BOOLEAN          boolean         %INTEGER         LOGICAL*4
                          -------

    NOTE
    1. PASCAL booleans appear to IMP as integers taking the value 0
       for FALSE and 1 for TRUE.

    2. Procedures can only be passed to another language if all the
       parameters of the procedure can be represented in the other
       language.

    3. Expressions must be assigned to a local variable before being 
       passed to FORTRAN - there is no call by value.

3.4.1 Calling EMAS Commands from PASCAL

    Calling a system command from within PASCAL requires specifying it as
    an external procedure, as  described in 3.3, but with a parameter
    corresponding to an IMP string (see 'EMAS 2900 User's Guide', 15-1).
    As string parameters are not compatible between PASCAL and IMP, a 
    mechanism for doing this has been worked out. The following example
    shows a call of the LIST command.

          PROGRAM DEMO;
          CONST MAGIC=402653440; (* =x'18000100' *)
          TYPE STR7=PACKED ARRAY[1..7] OF CHAR;
               POINT7=^STR7;
          VAR P7:POINT7;
          PROCEDURE LIST(K:INTEGER;AD:POINT7);EXTERN;
          BEGIN
          NEW(P7);
          P7^:=' filec ';
          P7^[1]:=CHR(5);
          LIST(MAGIC,P7);
          END.

    Note
    1. The first character of the string being passed must always be 
       set to the length of the string in IMP terms.

    2. The PASCAL array is at risk of alteration by the IMP routine.

    3. The commands DEFINE and PDEFINE (see 3.1) can be called in this
       way to associate the PASCAL files in the program heading with
       their EMAS files. A call for a particular file must be made
       before any further reference is made to that file. Tthe default
       files INPUT and OUTPUT  cannot be assigned in this way.

4.  BASIC ERROR ACTION

    A post-mortem diagnostic report system, consistent with that
    used by other compilers on EMAS 2900 has been implemented for use
    with the PASCAL compiler.

4.1 Error Reporting

    When an error occurs during program execution, the following
    information is output to the listing file:-

    1. An error message of the general form:

       PASCAL ERROR: <error number>
       DESCRIPTION: <error message>
       LOCATION: LINE <line number>
                 IN <block-type> <block-name>
                 OF MODULE <module name> COMPILED ON <date> AT <time>

       <error number> and <error message> identify and describe the
       error.
       <line number> identifies the line in the PASCAL source
       program at which the error occurred.
       <block-type> is MAIN PROGRAM BLOCK, PROCEDURE or FUNCTION
       <block-name> is the name of the program, procedure or function
       in which the error occurred. This value is not given  if the
       error occurred in the main program block.
       <module name> is the name of the object program file and is
       different from the program name.

    2. Contents of variables for each stack frame.

    There are three classes of errors relevant to PASCAL:

    1. Compiler detected errors arise:
       a) as a result of range checking (see 2.1.3)
       b) from use of the standard procedures NEW and DISPOSE.
       c) from file-accessing failures

      The error numbers, and their corresponding error messages, are
      listed in Appendix 2. All compiler detected errors are terminal
      errors; the current run of the object program is forced to stop
      when the error is detected.

    2. Hardware interrupts or contingencies are caused by an error in a 
       PASCAL program, such as integer overflow, or by a software error.

       Certain hardware-detected errors, such as division by zero,
       are handled in the same way as compiler-detected errors and
       the error number given is the system defined contingency number.

    3. Mathematical library errors give the system defined library
       error number.

4.2 Compile-time Control

    Production of data to support error reporting in source program
    terms may be controlled by the global compilation options,
    DIAGLINEMAP and DIAGNAMETABLE (see 2.1.2). By default these options
    are on and the user will only require to change them in unusual or
    specialised circumstances. Both options must be on to give the full
    information described in section 4.1.

4.3 Restrictions and Limitations

    1. There is no unassigned variable checking. The value of an
       unassigned variable is undefined.

    2. Line numbers given for failure may be out by one or two lines
       because of parallel execution of instructions.

    3. PASCAL tracebacks for mixed language failures may be unreliable.

    4. Infinite recursion will not be trapped, and will result in the 
       user process failing and being automatically logged off.

    5. Due to the pipelining of instructions, the diagnostics issued for
       failures just before the end of a program may give byte
       displacements instead of line numbers in the intial error
       message report.


5.  EMAS 2900 PASCAL

    This section defines the way in which EMAS 2900 PASCAL differs from
    the language defined by Jensen and Wirth in 'PASCAL User Manual and
    Report'.
 
5.1 Lexical Conventions

    The list below defines the lexical conventions used for EMAS
    2900 PASCAL source programs:

    1. The number of characters in each source line that are treated
       as significant by the compiler is specified by the RHMARGIN
       compiler option (see 2.1.4). The default value is 96 characters.
       No source line should be larger than 96 characters.

    2. Special PASCAL word symbols such as MOD and NIL are represented
       by the appropriate spelling in accordance with the conventions
       for identifiers in items 3 and 4. The user should not attempt to
       declare an identifier with an equivalent spelling to a word
       symbol.

    3. Identifiers and word symbols may include both upper and lower-
       case letters. For example, the word symbol BEGIN can be spelt
       as Begin, begin or bEgiN; THISID, thisid and ThisId are all
       equivalent spellings for the same identifier.

    4. Break characters can be inserted in any positions in an
       identifier or word symbol after the initial letter. These
       characters are included for the purposes of legibility;
       they have no significance but are retained in listings.

    5. Identifiers may be of any length within a single source line, but
       only the first 32 characters are significant to the compiler.

    6. The character pairs (* and *) may be used instead of the
       standard delimiters { and } for the purpose of delimiting
       comments. These delimiters are interchangeable; a comment
       starting with { can be ended by *) and a comment starting 
       with (* can be ended by }.

   7.  The character used in 'PASCAL User Manual and Report' to
       denote buffer and pointer variables is represented
       by either of the characters ^ or @.

    8. A source line beginning with the % character is treated as a
       PASCAL directive (see 2.1.1).

5.2 Language Restrictions

    1.  Local file declarations are not permitted, i.e. all files 
        used by a program must be external files specified in the 
        program heading.

    2.  Components of FILE type may not appear as components  in
        RECORD-TYPES,  ARRAY-TYPES or (other) FILE types.

    3.  Calls to the standard procedures READ and WRITE may not be made
        with a file of non-textfile type as the first parameter, so GET
        and PUT must be used for records of non-textfiles.

    4.  If x and y are the first and last ordinal values in the base
        type of a SET-type, then ord(x) must be greater than or equal
        to 0 and ord(y) must not be greater than 63.
        In particular, this means that SETs of type CHAR cannot be
        defined.

    5.  Contrary to Jensen and Wirth, a textfile is not the same as a
        FILE OF CHAR, but has been implemented as a PACKED FILE OF CHAR.

5.3 Language Extensions

    The language implemented contains the following facilities not 
    defined in Chapters 1 to 12 of the 'PASCAL User Manual and Report':

    1.  There is an additional predefined type ALFA, with definition
        TYPE ALFA=PACKED ARRAY [1..32] OF CHAR;

    2.  There is an additional predefined type ALFA8, with definition
        TYPE ALFA8=PACKED ARRAY [1..8] OF CHAR;

    3.  A program may access external procedures and functions, that is,
        separately compiled procedures and functions that exist outside
        the program and are declared but not defined within the program.
        External procedures and functions must be declared globally.  
        The declaration for such a procedure or function consists of a 
        standard procedure or function-declaration followed by the word
        EXTERN. For example

        FUNCTION SINH(X:REAL):REAL;EXTERN;

        External procedures and functions are described more fully in
        section  3.3.

    4.  The following additional predefined procedures and functions are 
        available:

        HALT (S)           Causes executions of the program to be
                           terminated.  The string S is written to the 
                           job journal.

        DISPOSE(P) or      Used to indicate that the storage space 
        DISPOSE(P,T1,..Tn) occupied by the variable P (with tag fields
                           T1,..Tn) is no longer required. The 
                           parameters to the call of DISPOSE should 
                           correspond to those for the call to NEW that
                           was used to allocate the space.

        CLOCK              The value of this INTEGER function is the 
                           amount of OCP time in milliseconds used by 
                           the program since the start of execution.

        DATEANDTIME(D,T)   Assigns character representations of the
                           current date and time. The current date is 
                           assigned in the form YY/MM/DD to the VAR
                           paramter D; the current time is assigned in
                           the form HH:MM:SS to the VAR parameter t.
                           Both these parameters are of type
                           PACKED ARRAY[1..8] OF CHAR.

    5. The break character may be used in identifiers and word symbols
       as described in 5.1.

    6. The character pairs (* and *) may be used as comment
       delimiters as described in 5.1.

5.4 The Program Heading and External Files

    The formal parameters in the program heading must be formal FILE 
    parameters; there must be one such parameter for each external file 
    used by the program, i.e. one for each file that exists outside
    the program before or after execution.  These files, except INPUT 
    and OUTPUT, must be declared as variables of FILE-type in the 
    program block like any ordinary local variables.  If used, the files
    INPUT and OUTPUT must be specified as formal parameters in the 
    program heading; however they must not be declared in the program 
    block as they are already assumed to be of type TEXT.

    Before an external file is read from or written to, reading or
    writing must be initiated by a call to the appropriate procedure
    RESET or REWRITE. Note, however, these initialising statements  
    must never be specified for INPUT or OUTPUT.

    The use of EMAS files to support the external files specified in
    the program heading is described in 3.1.

5.5 Hardware Representation

    Values of type INTEGER are represented in ICL 2900 Series 32-bit 
    fixed-point format (two's complement), which permits values in the 
    range -2,147,483,648 to  2,147,483,647 (that is, -2@31 to 2@31-1). 
    The constant MAXINT is predefined:

    CONST MAXINT = 2147483647

    Integer constants in a PASCAL program must lie in the range 
    -MAXINT to MAXINT.

    Values of type REAL are represented in ICL 2900 Series 64-bit 
    floating-point format which permits values in the approximate range
    -7.2*10@75 to 7.2*10@75. Values in the approximate range 
    -5.4*10@-79 to 5.4*10@-79 are represented as 0.  All values are 
    represented to a precision of 16 decimal digits.

    Note that the results of all arithmetic operations on real values 
    are truncated not rounded.

    Values of type CHAR are held as 8-bit bytes.  These values are 
    represented internally according to one of two character codes.  
    The required code is selected by the code directive as described in
    section 2.2. By default, ISO code is used.  The user need not 
    normally be concerned about the internal representation of values 
    of type CHAR, except with data structures such as:

    1.  FILE OF RECORD...END where the record contains fields of type 
        CHAR, or FILE OF ARRAY[...] OF CHAR

    2.  SET OF CHAR (in view of the restrictions on the base-type of a
        SET-type, specified in section 5.2)

    External files are represented by EMAS filestore files; 
    textfiles may also be represented by the default input and output
    streams .IN and .OUT respectively.

    Textfiles consist of a sequence of records of variable length.
    Character code conversion is performed automatically for textfiles
    as described above. Records for use with textfiles should not be
    longer than 160 characters; on input, characters beyond the 160th
    position will be ignored. Output of more than 160 successive
    characters, without an intervening WRITELN operation, will cause
    an automatic WRITELN operation to be performed after every 160
    characters.

    Non-textfiles consist of a sequence of records, each record
    representing a single component element and thus, of the same 
    length. The length is determined by the internal PASCAL
    representation for the component type. The EMAS file description
    for a file used as a non-textfile should specify a maximum record
    size that can accommodate items of the component type. No
    representational conversions are performed on components of non- 
    textfiles on transfer to or from the EMAS file.

N.B. The default value for the format parameter to the DEFINE
    command is large enough to accommodate most applications, thus it
    is rarely necessary to specify the format parameter.

5.5.1 Use of Packed

    In general, the use of PACKED in PASCAL type specifications has no 
    effect on the representation of data of the specified type. There
    is one exception, a type specification of the form:

    PACKED ARRAY[<index type>] OF <component type>
    where
    <component type> is an ordinal type
    ORD(<min>)>=0, ORD(<max>)<=225
    <min> and <max> are the min. and max. values of <component type>,

    This causes successive components of the packed array type to
    be allocated storage in successive bytes.

5.6 The Standard Procedure WRITE

    If no minimum field width parameter is specified in a call to WRITE,
    then the following default values are assumed

    Type of value     Default width

    INTEGER           12

    BOOLEAN           5

    CHAR              1

    STRING            Length of string

    REAL              25



Appendix 1

Compilation Error Messages

1     ERROR IN SIMPLE TYPE
2     SYMBOL EXPECTED WAS IDENTIFIER
4     SYMBOL EXPECTED WAS ')'
5     SYMBOL EXPECTED WAS ':'
6     UNEXPECTED SYMBOL
7     ERROR IN PARAMETER LIST
8     SYMBOL EXPECTED WAS 'OF'
9     SYMBOL EXPECTED WAS '('
10    ERROR IN TYPE
11    SYMBOL EXPECTED WAS '['
12    SYMBOL EXPECTED WAS ']'
13    SYMBOL EXPECTED WAS 'END'
14    SYMBOL EXPECTED WAS ';'
15    SYMBOL EXPECTED WAS INTEGER CONSTANT
16    SYMBOL EXPECTED WAS '='
17    SYMBOL EXPECTED WAS 'BEGIN'
18    ERROR IN DECLARATION PART
19    ERROR IN FIELD-LIST
20    SYMBOL EXPECTED WAS '/'
21    MISORDERED DECLARATIONS?
50    ERROR IN CONSTANT
51    SYMBOL EXPECTED WAS ':='
52    SYMBOL EXPECTED WAS 'THEN'
53    SYMBOL EXPECTED WAS 'UNTIL'
54    SYMBOL EXPECTED WAS 'DO'
55    SYMBOL EXPECTED WAS 'TO' OR 'DOWNTO'
56    SYMBOL EXPECTED WAS 'IF'
58    ERROR IN FACTOR
59    ERROR IN VARIABLE
90    PREMATURE END OF SOURCE TEXT
101   IDENTIFIER DECLARED TWICE
102   LOW BOUND EXCEEDS HIGH BOUND
103   IDENTIFIER IS NOT OF APPROPRIATE CLASS
104   IDENTIFIER NOT DECLARED
105   SIGN NOT ALLOWED
105   NUMBER EXPECTED
107   INCOMPATIBLE SUBRANGE TYPES
108   FILE NOT ALLOWED HERE
109   TYPE MUST NOT BE REAL
110   TAGFIELD TYPE MUST BE ORDINAL OR SUBRANGE
111   INCOMPATIBLE WITH TAGFIELD TYPE
112   INDEX TYPE MUST NOT BE REAL
113   INDEX TYPE MUST BE ORDINAL OR SUBRANGE
114   BASE TYPE MUST NOT BE REAL
115   BASE TYPE MUST BE ORDINAL OR SUBRANGE
116   ERROR IN TYPE OF STANDARD PROCEDURE PARAMETER
117   UNSATISFIED FORWARD REFERENCE
119   FORWARD DECLARED: REPETITION OF PARAMETER LIST NOT ALLOWED
120   FUNCTION RESULT TYPE MUST BE ORDINAL, SUBRANGE OR POINTER
121   FILE VALUE PARAMETER NOT ALLOWED
122   FORWARD DECLARED FUNCTION; REPETITION OF RESULT TYPE NOT ALLOWED
123   MISSING RESULT TYPE IN FUNCTION DECLARATION
124   F - FORMAT FOR REAL ONLY
125   ERROR IN TYPE OF STANDARD FUNCTION PARAMETER
126   NUMBER OF PARAMETERS DOES NOT AGREE WITH DECLARATION
128   RESULT TYPE OF PARAMETER FUNCTION DOES NOT AGREE WITH DECLARATION
129   TYPE CONFLICT OF OPERANDS
130   EXPRESSION IS NOT OF SET TYPE
131   TESTS ON EQUALITY ALLOWED ONLY
132   STRICT INCLUSION NOT ALLOWED
133   FILE COMPARISON NOT ALLOWED
134   ILLEGAL TYPE OF OPERAND(S)
135   TYPE OF OPERAND MUST BE BOOLEAN
136   SET ELEMENT TYPE MUST BE ORDINAL OR SUBRANGE
137   SET ELEMENT TYPES NOT COMPATIBLE
138   TYPE OF VARIABLE IS NOT ARRAY
139   INDEX TYPE IS NOT COMPATIBLE WITH DECLARATION
140   TYPE OF VARIABLE IS NOT RECORD
141   TYPE OF VARIABLE MUST BE FILE OR POINTER
142   ILLEGAL PARAMETER SUBSTITUTION
143   ILLEGAL TYPE OF LOOP CONTROL VARIABLE
144   ILLEGAL TYPE OF EXPRESSION
145   TYPE CONFLICT
146   ASSIGNMENT OF FILES NOT ALLOWED
147   LABEL TYPE INCOMPATIBLE WITH SELECTING EXPRESSION
148   SUBRANGE BOUNDS MUST BE ORDINAL
149   INDEX TYPE MUST NOT BE INTEGER
150   ASSIGNMENT TO STANDARD FUNCTION IS NOT ALLOWED
151   ASSIGNMENT TO FORMAL FUNCTION IS NOT ALLOWED
152   NO SUCH FIELD IN THIS RECORD
154   ACTUAL PARAMETER MUST BE A VARIABLE
155   CONTROL VARIABLE MUST NOT BE FORMAL
156   MULTIDEFINED CASE LABEL
158   MISSING CORRESPONDING VARIANT DECLARATION
159   REAL OR STRING TAGFIELDS NOT ALLOWED
151   AGAIN FORWARD DECLARED
162   PARAMETER SIZE MUST BE CONSTANT
164   SUBSTITUTION OF STANDARD PROC/FUNC NOT ALLOWED
165   MULTIDEFINED LABEL
166   MULTIDECLARED LABEL
167   UNDECLARED LABEL
168   UNDEFINED LABEL IN BLOCK JUST ENDED
169   ERROR IN BASE SET
170   PROCEDURE OR FUNCTION WITH ONLY VALUE PARAMETERS EXPECTED
172   UNDECLARED EXTERNAL FILE
173   PACKED COMPONENT CANNOT BE PASSED AS VARIABLE PARAMETER
174   FORWARD PROCEDURE OR FUNCTION NOT FOUND IN BLOCK JUST ENDED
175   SUBRANGE BOUNDS CANNOT BE REAL
176   LABEL NOT DEFINED AT CORRECT LEVEL
177   PERMANENT FILE UNDECLARED
178   NO INPUT OR OUTPUT FILE SPECIFIED
201   ERROR IN REAL CONSTANT: DIGIT EXPECTED
202   STRING CONSTANT MUST NOT EXCEED SOURCE LINE
203   INTEGER CONSTANT EXCEEDS RANGE
205   STRING OF LENGTH ZERO NOT ALLOWED
250   TOO MANY NESTED SCOPES OF IDENTIFIERS
251   ERROR IN INITIALISATION OF GLOBAL VARIABLES
253   PROCEDURE TOO LONG
255   TOO MANY ERRORS IN THIS SOURCE LINE
294   TOO MUCH OBJECT CODE IN THIS LINE
295   R-H MARGIN SETTING TOO LARGE
296   PROGRAM SOURCE LINE TOO LONG
297   UNRECOGNISED DIRECTIVE IDENTIFIER
298   MISPLACED DIRECTIVE
299   INVALID DIRECTIVE
300   DIVISION BY ZERO
302   VALUE OUTSIDE EXPECTED RANGE
303   OVERFLOW OR DIVISION BY ZERO IN INTEGER ARITHMETIC
398   IMPLEMENTATION RESTRICTION


APPENDIX 2

Execution Error Messages

1100   VALUE OUT OF RANGE OF ORDINAL TYPE
1200   NO LABEL FOR VALUE OF "CASE" STATEMENT SELECTOR EXPRESSION
1300   "NEW" - INVALID OR CORRUPT CALL
1301   "NEW" - NOT ENOUGH SPACE ON HEAP
1302   "DISPOSE" - INVALID POINTER VALUE
1303   "DISPOSE" - INCORRECT POINTER AND/OR VARIANT VALUES
1304   "DISPOSE" - INCORRECT POINTER VALUE OR HEAP CORRUPT
1400   FILE NOT DEFINED OR DEFINED INCORRECTLY
1401   "RESET"/"REWRITE" NOT ALLOWED ON STANDARD FILES "INPUT"/"OUTPUT"
1402   COMPONENT ELEMENTS OF NON-TEXTFILE ARE LARGER THAN FILE
       DESCRIPTION ALLOWS
1403   "RESET"/"GET"(NON-TEXT) - ELEMENT OF INCORRECT SIZE
1404   "PUT"(NON-TEXT) - FILE IS FULL
1405   ATTEMPT TO INPUT WHEN FILE NOT OPEN FOR INPUT
1406   ATTEMPT TO INPUT WHEN "EOF" IS TRUE
1407   ATTEMPT TO OUTPUT WHEN FILE NOT OPEN FOR OUTPUT
1408   FIELD WIDTH NON-POSITIVE FOR CHARACTER "WRITE"
1409   FIELD WIDTH NON-POSITIVE FOR STRING "WRITE"
1410   FIELD WIDTH NON-POSITIVE FOR INTEGER "WRITE"
1411   "EOF" ENCOUNTERED DURING INTEGER "READ"
1412   INVALID CHARACTER FOR FIRST DIGIT FOR INTEGER "READ"
1413   "READ(LN)" - INTEGER VALUE OUT OF RANGE
1414   INVALID CHARACTER ENCOUNTERED DURING REAL "READ"
1415   "READ(LN)" - REAL VALUE OUT OF RANGE
1416   "WRITE(LN)"/"PUT"(TEXT) - FILE IS FULL
1500   PASCAL SYSTEM FAILURE - INVALID SUPPORT CALL
10000  PASCAL SUPPORT SOFTWARE FAILURE


R.Soutar
01.05.81