Home

User Guide - Intelligent Interfaces

1. INCLUDE Statement An INCLUDE statement allows a file containing source text to be read in by the compiler at the point where the INCLUDE statement occurs The syntax of the statement is INCLUDE filename Line numbers in the INCLUDE file are not recorded in the object file and will therefore not appear in a backtrace The correct line numbers are shown in the program listing and in the error messages Type Names COMP LEX respectively COMPLEX 16 REAL 8 may be used as an alternative to DOUBLE PRECISION The type names LOGICAL 4 INTEGER 4 REAL 4 and COMPLEX 8 are alternatives to LOGICAL INTEGER REAL and A COMPLEX 16 value consists of a pair of DOUBLE PRECISION numbers representing the real and imaginary parts of a complex number The rules for the use of COMPLEX 16 are the 7 same for COMPLEX with a few exceptions Combining a COMPLEX 16 with a REAL or COMPLEX gives a COMPLEX 16 result Combining a COMPLEX with a DOUBLE PRECISION gives a COMPLEX 16 result A complex constant containing a double precision value is a COMPLEX 16 The intrinsic function DIMAG is used to extract the imaginary part of a COMPLEX 16 DCMPLX is used to convert to COMPLEX 16 it may have one or two arguments The rules for memory layout and EQUIVALENCE of COMPLEX 16 are the same as for COMPLEX except that the individual parts are
2. duplicate use of filename The same filename has been used more than once in an OPEN statement invalid unit for operation BACKSPACE REWIND ENDF ILE attempted on unit connected for direct access error detected previously An I O error has been detected previously on this unit and trapped with ERR direct access without OPEN A direct access READ or WRITE has been used without an OPEN statement for the unit invalid use of unit Inconsistent use of unit formatted mixed with unformatted sequential mixed with direct access or ENDF ILE done previously input and output mixed Input and output mixed on a sequential unit without intervening REWIND or OPEN direct access not open for input The direct access file could not be opened for input for example file is write only direct access not open for output The direct access file could not be opened for output for example file is read only end of file on output An attempt has been made to write beyond the end of a sequential file In practice this will only occur with internal files invalid logical in input 1021 1022 1023 1024 1025 1026 Formatted input file contains bad logical value invalid number in input Bad number range or syntax in formatted I D E F or G input Bad complex data Bad COMPLEX constant in list directed input LD repeat not integer Repeat count in list directed input is not valid
3. LD input data not REAL Syntax or range error in REAL list directed input value LD input data not INTEGER Syntax or range error in INTEGER list directed input value LD input data not DP Syntax or range error in DOUBLE PRECISION list directed input value 25 1027 1028 1029 1030 2000 2001 2002 LD input data not LOGICAL Syntax error in LOGICAL list directed input value LD input data not COMPLEX Syntax or range error in COMPLEX list directed input value LD input data not CHARACTER Syntax error in CHARACTER list directed input value LD repeat split CHARACTER Attempt to split a repeated character constant across a record boundary This is strictly legal but almost impossible to implement correctly not available BACKSPACE operation is not available bad unformatted record message A record in an unformatted file does not have the required structure invalid access to terminal file message 2003 2004 2005 2006 2007 2009 2010 Attempt to use terminal or other output device as an unformatted or direct access file More detail is given sequential open failed message The actual reason for the failure for example Bad name is given in the brackets direct access open failed message The actual reason for the failure for example Bad name is given in the brackets direct access IO failed message For ex
4. IARRAY 37 Parameters ICODE integer RO ICODE IARRAY integer array R1 pointer to IARRAY one dimensional integer array the OS_Word parameter block Results Placed in IARRAY IFOSCLI Purpose To make OS_CLI calls Example OGICAL FUNCTION IFOSCLI STRING OGICAL STATUS STATUS IFOSCLI STRING Parameters STRING character command line terminated by a carriage return Results STATUS logical if the command is executed without error STATUS TRUE or if an error occurs STATUS FALSE IFOSGETERROR Purpose To return the error number and error message immediately after OSCLI has returned with STATUS FALSE E Example CALL IFOSG T TERROR IERRNO ERRSTR Parameters None Results ERRNO integer the error number RRSTR character the error IFFILEEXISTS Purpose To check that a file exists H zal Example OGICAL FUNCTION IFFILEEXISTS STRING OGICAL EXISTS EXISTS IFFILEEXISTS FILENAME Parameters FILENAME character Results EXISTS logical if the file exists EXISTS TRUE or if the file does not exist EXISTS FALS CJ 38 Appendix J The IFLib Utility Library This is a small library which includes routines to return the addresses of variables make SWI calls and read and write memory Examples illustrating the use of the routines a
5. etc The appearance of line numbers in the backtrace is controlled by the compiler L option level 1 is the default If a hardware trap occurs in a program compiled with a line number option level 1 it may not be possible to determine the exact line number ERROR 3000 hardware trap Routine data area line ABC amp 00005514 5 16 F77_MAIN amp 000054EC 3 19 Here the main program called ABc failed with a hardware trap between the lines 5 and 16 inclusive If the program is recompiled with line number option level 2 the exact line number will be displayed Code 1000 errors There are a number of simple run time errors producing error messages which all have the same error number of 1000 These are listed in Appendix B Array and Substring Errors There are two errors which may be generated by a program unit which has been compiled with the bound checking option ERROR 1050 array bound error An illegal array subscript has been used ERROR 1051 substring bound error An illegal substring has been used Input Output Errors I O errors are those which may be trapped by the use of END and ERR specifiers in FORTRAN 77 statements If these specifiers are not used an error message and code are generated as described below Otherwise execution continues with the error code available through the use of the TOSTAT specifier All the messages have the general form ERROR N PREFIX UNIT
6. reason N is the error code PREF 1X describes the I O operation being attempted which may be OPEN CLOSE ENDFILE REWIND Of READ WRITE and UNIT is the unit number with given for one of the asterisk units and internal for an internal file The rest of the message gives more information about the error End of file on input may be trapped with the END specifier The IOSTAT value in this case is 1 If END is not used then the message end of file is generated with code 1000 Other errors may be trapped with the ERR specifier The IOSTAT value is the corresponding error code as listed in Appendix B Tracing To specify that calls to special trace routines are to be included in the code select the T option when compiling These routines will cause trace information to be output when entering the program unit leaving the program unit a labelled statement is about to be executed the THEN clause of an IF THEN Or ELSEIF THEN construct is about to be executed the ELSE clause of an IF THEN Or ELSEIF THEN construct is about to be executed a DO statement is about to be executed 20 another subprogram unit is about to be executed The trace routines will output a message which starts with T and indicates the type of trace point encountered For some of these it will also indicate a count modulo 32768 of the number of times this trace point
7. Commands Include the following line to ensure that the older version of the linker is used by the 77 771k linkf77 d77 d 771k and dlinkf77 commands If lt F77 Running gt Then Set Run Path lt Run Path gt lt Fortran77S Dir gt Execlib OldLink Include the following line to ensure that the newer version of the linker is used by the 77 771k linkf77 d77 d 771k and dlinkf77 commands If lt F77 Running gt Then Set Run Path lt Run Path gt lt Fortran77S Dir gt Execlib NewLink Command format oldlink output filename options objectfilel objectfile2 oldlink output filename options via viafile General options Capitals are used to denote the abbreviated form of the keyword Output filename Specifies the name of the output file as filename The 771ink command can be used to check for unresolved references in object files by specifying the file name as the device nu11 The linked output will be discarded Dbug An output file is produced which can be used with the Acorn Symbolic Debugger asd Verbose Gives information as files are linked VIA filename The object and library files listed in the text file filename are linked Note that this option cannot be used on computers fitted with a StrongARM processor Special options Case Ignore case when symbol matching Base n Specify base of image prefix n with amp for hex postfix with k for 21 m for 220
8. DOUBLE PRECISION rather than REAL There are new specific names for intrinsic functions with COMPLEX 16 arguments These are Generic Specific ABS CDABS CONJG DCONJG SORT CDSORT EXP CDEXP LOG CDLOG SIN CDSIN COS CDCOS Bit Manipulation Functions There are eight intrinsic functions concerned with bit manipulation on INTEGER arguments These are IAND I J logical and of and J IOR I J logical or of and J IEOR I J logical exclusive or of and J NOT I logical complement of ISHFT I J return shifted left J places if J is positive or shifted right J places if J is negative The result is undefined if J is notin the range 32 to 32 Bits shifted out at the end are lost zeros are introduced at the other end IBSET I J return with bit J set to one Bit zero is the least significant bit The result is undefined if J is not in the range 0 31 IBCLR I d return with bit J set to zero BTEST I J test bit J of and return a LOGICAL result TRUE if the bit is set and FALSE if it is clear lt Note that BTEST returns a LOGICAL result whilst the other functions return an INTEGER result Example IF BTEST IX 0 tests to see if IX is odd I IAND I IFF clears all but the least significant byte of I I ISHFT J 24 extracts the most significant byte of J Relaxed Rules for List Directed Input When readin
9. Relocatable Relocatable AIF Note that Dbug and Relocatable are mutually exclusive options 33 Appendix G The Newer Linker link Command ARM Linker Version 5 06 Acorn Computers Ltd Jan 11 1995 To ensure that the newer version of the linker newlink is used the lines in the FORTRAN Fortran77 Run obey file should read as follows Location of Command Line Commands Include the following line to ensure that the older version of the linker is used by the 77 771k linkf77 d77 d 771k and linkf77 commands If lt F77 Running gt Then Set Run Path lt Run Path gt lt Fortran77S Dir gt Execlib OldLink Include the following line to ensure that the newer version of the linker is used by the 77 771k linkf77 d77 d 771k and linkf77 commands If lt F77 Running gt Then Set Run Path lt Run Path gt lt Fortran77S Dir gt Execlib NewLink Command format newlink output filename options objectfilel objectfile2 newlink output filename options via viafile General options Capitals are used to denote the abbreviated form of the keyword Output filename Specifies the name of the output file as filename The newlink command can be used to check for unresolved references in object files by specifying the file name as the device nu11 The linked output will be discarded Debug An output file is produced which can be used with the Desktop Debugger DDT see Appendix L ERRORS
10. The map gives the name type and location of local and COMMON variables in each program unit The location is relative to the start of the static area for a local variable and is the offset in the block for a COMMON variable The offset of each statement number from the start of the code is also given debug level The debug keyword specifies the 1eve1 of symbolic debugging information to be included in the aof format output file The Zevel must be one of the following none No information This is the default min Subroutine and function names only vars Subroutine and function names and variable name information lines Subroutine and function names and line number information all Subroutine function variable and line information max is an alternative synonym for all Normally none is used for a working program and a11 for programs under development The use of all increases the size of the program considerably and so should be avoided when not debugging The intermediate levels can be used to provide some debugging information without increasing the size of the program to the same extent source filenam The source keyword specifies the filename of the original FORTRAN 77 source text file for inclusion in debugging information when the debug keyword is used with a debugging level other than the default of none help The help keyword gives a summary of the keywords and arguments available Examples f77cg tmp fcode to a
11. are Image ROS Base Address of the start of the read only program area Image SROS Limit Address of the byte beyond the end of program area Image ZI Base Address of the start of run time zero initialised area Image ZI Limit Address of the byte beyond the zero initialised area Image RW Base Address of the start of the read write data area ImageSSRWSSLimit Address of the byte beyond the end of the data area Although it will often be the case it cannot be guaranteed that the end of the read only area corresponds to the start of the read write area These symbols can be imported as relocatable addresses by assembly language routines that 31 might need them Note that programs can reside in read write areas as they sometimes contain local writable data eg self modifying code and it is possible to have read only data eg floating point constants and string literals The linker joins all areas from all input files with the same name and attributes together to form a single area It then creates the two symbols name Base and name Limit to mark the start and end of the area It is an error for two areas to have the same name but different attributes 32 Appendix F The Older Linker The oldlink Command Acorn Object Format Linker ARM Arthur AIF 595 M To ensure that the older version of the linker oldlink is used the lines in the FORTRAN Fortran77 Run obey file should read as follows Location of Command Line
12. arising from the use of any information or particulars in or an error or omission in this User Guide CONTENTS Introduction Conventions Used Installation Checking the Installation Directories Configuring the Fortran77 Application Using the Fortran77 Application Extensions to the Standard Hexadecimal Constants Naming Loops Random Number Generators INCLUDE Statement TYPE Names COMPLEX 16 Bit Manipulation Functions Relaxed Rules for List Directed Input RISC OS Interface Routines Input Output Unit Numbers and Files Sequential Files Direct Access Files OPEN and CLOSE INQUIRE BACKSPACE ENDFILE REWIND Format Decoding Graphics Errors and Debugging Front End Error Messages Warning Messages Code Generator Error Messages Code Generator Limits Run time Errors Array and Substring Errors Input Output Errors Tracing Appendix A Code Generator Error Messages Appendix B Run time Error Messages Input Output Errors Appendix C The Front End The f77fe Command Onn hy A OO WA ONNNDDOD Q0 22 22 24 24 25 27 27 Appendix D The Code Generator The f77cg Command Appendix E The Linkers Appendix F The Older Linker The oldlink Command Appendix G The Newer Linker The newlink Command Appendix H The 77 f771k and linkf77 Commands The df77 df771k and dlinkf77 Commands Appendix The IFExt Utility Library Appendix J The IFLib Utility Library Appendix K Call
13. behaves in an unexpected way Level 1 warns for example of statements that will not be executed because they follow a GOTO statement and are unlabelled Level 2 warns of the use of extensions to standard FORTRAN 77 These extensions may give problems if the program it be re compiled on another make of computer eg an IBM PC Levels 3 and 4 warn of source text that conforms to the standard syntax but is of unusual style and therefore could possibly be a mistake The strict FORTRAN 77 option 7 is used to control warnings about language extensions If unset warnings are not generated Otherwise messages are generated if the warning level Wn is 2 the default or greater Option 7 is unset by default so that the extensions may be used without generating messages whatever the warning level Code Generator Error Messages These are not always as explicit as front end error messages and are listed in Appendix A with a brief explanation of the most likely cause As was the case with front end error messages 18 errors do necessarily occur at the line where the message is given Code Generator Limits The code generator has certain internal limits on the complexity of each program unit These are code size 2 Mbytes number of labels 4096 number of local variables 8192 number of constants 8192 number of COMMON blocks 2048 number of external symbols 2048 These limits should never be exceeded Normally the code generator will run
14. case when it reads the source text so all statements identifiers etc may be in lower case Names may be up to 255 characters long However there is no limit on the length of CHARACTER values Loops WHILE ENDWHILE This loop construct has the syntax WHILE logical expr DO ENDWHILE WHILE and ENDWHILE must be nested correctly and neither statement may be used as the terminal statement of a DO loop or in a logical IF The loop is equivalent to 11 IF NOT logical expr GOTO 12 GOTO 11 12 DO WHILE This loop construct has the syntax DO n WHILE logical expr The rules regarding nesti Block DO ng and the terminal statement are exactly as for normal DO loops The syntax of DO and DO WHILE loops has been extended so that the terminal statement number may be omitted DO v vl v2 v3 END DO The loop is then terminated by an END Do statement or DO WHILE logical expr END DO END DO may not be used as the terminal statement in a labelled DO loop Random Number Generators The compiler has two routines for random number generation REAL FUNCTION RNDO1 returns a pseudo random number in the range 0 0 lt r lt 1 0 SUBROUTINE SETRND I selects a new random sequence If is zero the sequence is non repeatable The generator is initialised with a call to SETRND 0 so that successive runs will produce different sequences
15. everywhere cases are distinguished only in quoted strings and nH descriptors and in the D E and G edit descriptors see below Extraneous repeat counts Unexpected repeat counts are ignored ie before T S and B edit descriptors before the sign of a P edit descriptor or before a comma or closing parenthesis Edit descriptor separators A comma may be omitted except where the omission would cause ambiguity or a change in meaning It cannot be omitted between a repeatable edit descriptor such as 15 and an nH edit descriptor such as 11Habcdefghi jk Numeric edit descriptors As well as the standard forms Iw Iw m Fw d Ew d Ew dEe Dw d Gw d and Gw dEe additional forms are Fw Dw dDe Gw dDe Dw dEe Ew dDe Zw and Z When the exponent field width is specified the letter used to introduce it is used in the same case in the output form If no exponent field width is specified then except for G edit descriptors the initial character of the descriptor is used in the same case in the output form If an exponent field width is given as zero a field width of 2 is assumed If on output the given exponent field width is just too small for the exponent the character introducing the exponent field is suppressed The z edit descriptor provides input and output of numeric data in hexadecimal form A field width of zero implies the correct width for the data type being transferred z by itself is an
16. file when the first RI EN statement which refers to a non existent file will not EAD or WRITE is attempted and can be trapped by using EAD or EN and ERR The routine copies a named file to SUBROUTINE COPY TEXTFILE CHARACTER TEXTFILE LINE 72 OPEN 10 FILE TEXTFILE ERR 100 1 READ 10 A END 100 ERR 100 LINE PRINT A LINE GOTO 1 100 CLOSE 10 END A sequential file may be used without an explicit OPEN statement The file is opened when the first READ or WRIT Formatted I O Formatted and list directed READS and WRIT A formatted R1 input records are assumed to be extended in may refer to more characters than are actuall uses normal line editing conventions includin end of file code which may be trapped by specifying END in the RI E statement which refers to its associated unit number is executed ES are permitted on all files EAD statement causes one or more records to be read from a file or terminal All definitely with spaces Therefore an input format y present in the record Input from a terminal g cursor copying lt CTRL D gt 04 is treated as the EAD statement For file input the carriage return OD or line feed OA codes are recognised as record terminators Form feed 0C codes are ignor characters then the rest are ignored The com ed If the record contains more than 512 data bination carriage retur
17. format specification 10 in these examples must not be less than the number of VDU codes in the list The format descriptor must be used to suppress the final newline The format can be given as a character constant as in the examples above or in a separate statement PRINT 100 CHAR codel CHAR code2 100 FORMAT S 10A For example to change to mode 12 PRINT 2A CHAR 22 CHAR 12 or to change the palette so that colour 1 refers to colour 6 PRINT 6A CHAR 19 CHAR 1 CHAR 6 CHAR 0 CHAR 0 CHAR 0 Most move and draw operations require a pair of 16 bit coordinates These should be output 16 as a pair of bytes For example the following subroutine provides a general PLOT command VDU code 25 SUBROUTINE PLOT TYPE X Y INTEGER TYPE X Y PRINT 6A CHAR 25 CHAR TYPE CHAR IAND X 255 CHAR ISHFT X 8 CHAR IAND Y 255 CHAR ISHFT Y 8 END Move is a TYP GI 4 and draw is a TYP GI 5 call to the subroutine PLOT 17 Errors and Debugging Errors can be detected both by the compiler and by the run time library In addition to generating error messages the compiler may also generate warning messages which indicate that the program may not behave as anticipated An example of this is using a variable that has not been declared An example of a fault which is not detected by the compiler but by the run time library is attem
18. gt Invalid character lt ch gt in FORMAT FORMAT bad numeric descriptor Bad syntax for numeric FORMAT descriptor FORMAT cannot use when reading Quoted string used in input FORMAT FORMAT unexpected format end End of FORMAT inside quoted string FORMAT cannot use H when reading nH used in input FORMAT FORMAT bad scale factor Bad nP or nP construct FORMAT too many opening parentheses More than 20 nested opening parentheses including the first FORMAT trouble with reversion No value has been or written by the repeated part of the format this would cause an infinite loop if not trapped The following program fragment illustrates the trouble with reversion format error WRITE 1 10 i j 10 FORMAT i5 1x FORMAT width missing or zero Bad width in numeric edit descriptor Unformatted output too long Unformatted record length exceeds maximum permitted This can occur with direct access output only Unformatted input record too short Input record does not contain sufficient data mismatched use of ACCESS RECL in OPEN ACCESS DIRECT has been quoted in an OPEN which does not contain a RECL specifier or 24 vice versa Input Output Errors 1001 1002 1003 1004 1005 1006 1007 1008 1009 1010 1011 1020 invalid unit number Unit number not in range 1 60 invalid attribute Invalid attribute used in OPEN statement
19. has been met A special routine called TRACE can be called with a single LOGICAL argument to turn this tracing information on and off Note that even if the trace output is off the counting will still be done so the values produced will be correct if tracing is turned on again If the main program is compiled with tracing on the user will be asked if trace output is to be produced or suppressed If the main program is compiled without tracing then trace output is initially enabled In addition to the TRACE routine two further subroutines are available The first of these HISTOR short for History causes information to be output about the last few traced subprogram calls Each line of history information consists of a name which may be preceded by gt or by lt A right arrow indicates a traced call of a subprogram a left arrow indicates a traced exit from a program unit and a line with neither type of arrow indicates a traced entry to a program unit Note that the name given when tracing entry and exit from a program unit is the name of the program unit itself rather than the name of the entry called by the user The second routine provided is BACKTR short for Backtrace which outputs information on the current nesting of program unit calls The routine should be given a single logical argument If this is TRUE then the HISTOR subroutine is called after the backtrace information has been generated Under RISC OS all tracing output i
20. in DATA statement There are more constants than variables in the DATA statement too many program units in compilation type mismatch in DATA statement The type of the constant is illegal for the corresponding variable variable initialised more than once in DATA A variable has been initialised more than once by DATA statements in this program unit wrong number of hex bytes for constant of TYPE type A hex constant has been given with the wrong number of digits zero increment in DO loop A DO loop with a constant zero increment value has been used inconsistent use of NAME The external subroutine or function NAME has been used with inconsistent argument types This error message would occur with the following program CALL ABC 1 0 CALL ABC 2 END 23 Appendix B Run time Error Messages Code 1000 errors lt ch gt edit descriptor cannot handle logical list item Format descriptor used with a LOGICAL list item is not L lt ch gt is the actual descriptor used lt ch gt edit descriptor cannot handle character list item Format descriptor used with a CHARACTER list item is not A lt ch gt is the actual descriptor used lt ch gt edit descriptor cannot handle numeric list item Invalid descriptor for numeric value lt ch gt is the actual descriptor used Z field width unsuitable Wrong number of digits in hex z input field for given type FORMAT unexpected character lt ch
21. may be used to construct bytes for output as VDU control codes The following will switch the screen to mode 3 WRITE 3 CHAR 22 CHAR 3 3 FORMAT 2A Note that the format descriptor has been used to suppress the final new line During formatted input of numeric values blanks are either ignored or treated as zeros depending on the use of the BZ and BN format specifiers and the BLANK status of the unit All pre assigned units those opened without explicit use of OPEN have BLANK ZERO as the default status any unit connected by an OPEN statement has BLANK NULL as the default The difference in the defaults was introduced for compatibility with FORTRAN 66 and the FORTRAN 77 subset language in FORTRAN 66 blanks are always treated as zeros Unformatted I O Unformatted READS and WRITEs are permitted on disc files only Unformatted and formatted operations may not be mixed on any unit unless the unit is CLOSEd and reoPENed Each unformatted WRITE statement writes a single record to the file The record may be read back later by any READ which quotes the same number of or fewer variables as illustrated below WRITE 1 1 WRITE 1 6 7 8 REWIND 1 R R EAD 1 I EAD 1 J l is read as 1 and J is read as 6 The first record contains 5x4 20 bytes of data and the second 3x4 12 bytes of data Records of the same length could be achieved by padding all unformatted records but this would le
22. newlink Edit the FORTRAN Fortran77 Run obey file as shown in Appendix G and then use the df77 df77Ik and dlinkf77 commands 49
23. out of memory before this happens Run time Errors A program may compile and link but when it is run error messages are generated These error messages are generated by the run time library and have the following form ERROR N text followed by a backtrace N is an error number and text is a sentence describing the error A backtrace is a re tracing of the steps which the run time library has taken in attempting to run the program Each line of the backtrace output gives the name of a program unit the address of the corresponding static data area and the line number The data area address may be used in conjunction with the storage map produced by the code generator to examine the values of local variables The address of the data area is given in hexadecimal Note that a name in a backtrace refers to the main entry point of the program unit and so may not be the actual name used in a call ERROR 1025 LD input data not INTEGER Routine data area line F77_INIT amp 000100D8 F77_1067 amp 00010000 ERR2 amp 0000FF04 16 ERR1 amp 0000F9B4 10 F77_MAIN amp 0000F9BO 6 In this example the main program with default name F77_MATN has called ERR1 which in turn has called ERR2 which has attempted to read an integer using list directed input F77_1067 and F77_INIT are internal routines in the run time library The call to ERR1 in the main program was on line 6 the call to ERR2 in ERR1 was on line 10
24. Desktop Fortran 77 for Acorn RISC OS based Computer Systems User Guide Intelligent Interfaces Ltd October 1999 The Fortran77 Application Copyright D J amp K M Crennell 1993 This software must not be copied or altered in any way without the written approval of the authors In no circumstances shall the authors be liable for any damage loss of profits or any indirect or consequential loss arising out of the use of this software or inability to use this software even if they have been advised of the possibility of such a loss User Guide Copyright Intelligent Interfaces Lid 1999 Neither the whole or any part of the information contained in this User Guide may be adapted or reproduced in any material form except with the written approval of Intelligent Interfaces Ltd All information is given by Intelligent Interfaces in good faith However it is acknowledged that there may be errors or omissions in this User Guide Intelligent Interfaces welcomes comments and suggestions relating to this User Guide All correspondence should be addressed to Technical Queries Intelligent Interfaces Ltd P O Box 80 Eastleigh Hampshire S053 2YX Tel 023 8026 1514 Fax 087 0052 1281 E mail support intint demon co uk URL http www intint demon co uk This User Guide is intended only to assist the reader in the use of the software and therefore Intelligent Interfaces shall not be liable for any loss or damage whatsoever
25. FHWRDB Purpose To return the byte stored at the given hardware address and return it as an integer Note that the function does not check that the address is in the valid hardware address range Example inum IFHWRDB iaddress Parameters iaddress integer Results inun integer byte stored at iaddress IFHWWRB Purpose To store a byte at the given address Note that the function does not check that the address is in the valid hardware address range 42 Example CALL IFHWWRB inum iaddress Parameters inun integer byte to be stored at iaddress iaddress integer IFHWRD16 Purpose To return the 16 bits stored at the given hardware address and return them as an integer Note that the address must be on a word boundary and be in the valid address range Example inum IFHWRD16 iaddress Parameters iaddress integer must be on a word boundary ie be a multiple of 4 Results inumn integer 16 bits stored at iaddress IFHWWR16 Purpose To store 16 bits at the given address Note that the address must be on a word boundary and be in the valid address range Example CALL IFHWWR16 inum iaddress Parameters inumn integer 16 bits to be stored at iaddress iaddress integer must be on a word boundary ie be a multiple of 4 IFHWRDW Purpose To return the word 32 bits stored at the given hardware address and return it as an integer Note that the address must be on a word boundary and be in t
26. FORTRAN 77 Example lerror IFSWIX numswi ilregsin iregsout iflags Parameters numswi integer SWI number the function sets bit 17 amp 00020000 to return on error iregsin integer array one dimensional array with subscripts in the range 0 to 9 iregsin 0 RO0 iregsin 1 R1 etc Results ierror integer if no error occurs ierror 0 If an error occurs the function returns and ierror address ofa standard error block iregsout integer array one dimensional array with subscripts in the range 0 to 9 iregsout 0 R0 iregsout 1 R1 etc iflags integer processor flag bits bit0 V flag bit C flag bit2 Z flag bit3 N flag 44 Appendix K Calling Functions and Subroutines Written in Assembler from FORTRAN The FORTRAN compiler does not conform to the ARM Procedure Call Standard APCS R Only one argument is passed on calling register variables vl v6 and 4 f7 are not preserved and the register binding of the APCS R is not used Register Conventions The register binding used is ro Pointer to a list of the addresses of the arguments given on calling ri r9 Scratch registers fp r10 Frame pointer used as a pointer to a list of the addresses of the arguments given on calling within the assembler routine sp r12 Stack pointer sb r13 Static base used to refer to local data within the assembler routine Argument Lists Every call in FORTRAN passes one argument in ro This is a pointer to
27. IFADR intl 1 iadrint1 IFADR int1 PRINT Integer 1 is int1 and its address is iadrintl STOP END Assuming that the source text of the assembler routine is in the file s IFADR and the source text of the FORTRAN program is in the file 77 Example they should be compiled and linked as follows f77 Example objasm s IFADR o IFADR APCS NONE link output aif Example o Example o IFADR lt F77 Lib gt f77 Run the program by typing aif Example The following should be displayed Integer 1 is 1 and its address is 106204 STOP 48 Appendix L Notes on Using a Debugger The version of the linker used by the df77 df77Ik and dlinkf77 commands can be selected by editing the FORTRAN Fortran77 Run obey file as shown in Appendices F and G The Acorn Symbolic Debugger Run from the command line by typing x asd aif filename This can only be used with programs linked by the older linker oldlink Edit the FORTRAN Fortran77 Run obey file as shown in Appendix F and then use the df77 df77Ik and dlinkf77 commands Before using the Acorn Symbolic Debugger on a computer fitted with a StrongARM processor it is essential to turn the cache off From the command line type cache off To turn the cache on from the command line type cache on The Desktop Debugger DDT Run from the command line by typing debugaif aif filename This can only be used with programs linked by the newer linker
28. R sb DATAPTR MOV fp r0 list code LDMFD sp r1 fp sb pc Standard END An Example FORTRAN Callable Assembler Routine s IFADR i contains IFADR 07 June 1993 Version 0 00 A Purpose 46 r T r RE E To return the address of an integer variable Example iadr IFADR inum Parameters inum integer Results iadr integer address of variable inum REGISTERS Use the RN directive to defin AoA A A A AA e A OMANI HDHUOBWNE OC Z ee p A AA Z K ws Z aor FBPWWNN EF OC OO Z O O o b p oO p h h h E m DOE D oO m E h Z K on Use the FN directive to defin ARM register names floating point register names A A A A ee aa A yy Z YO OF W O O zZ DATA D x inal D SULT o A CODE ARI za D NAME DCB IFADR m F77 Code COD F77 SData DATA result location T EADONLY 47 DATAPTR DCD F77 SData eal x PORT IFADR IFADR ADR rl NAME 12 standard entry sequence STMFD sp r1 fp sb l1r LDR sb DATAPTR MOV fp r0 fp points to param block LDR ri Eps load address of integer sb points to data area STR rl sb store address of integer LDMFD sp xrl1 fp sb pc standard exit sequenc eal ND An Example FORTRAN Program Which Calls the Assembler Routine PROGRAM Example C Demonstrates the use of
29. a list of the addresses of the arguments given in the call every argument in FORTRAN is passed by reference Thus the address of the first argument is at r0 0 the second at r0 4 etc Normally the address of the list is copied to fp r10 which is preserved by calls If an assembler routine does not call any other routines the address of the list can be left in ro For a CHARACTER argument the address in the argument list does not point directly to the data It points at a character descriptor which is a two word block containing the address of the character string in its first word and its length in the second For example if the third argument in a call is a CHARACTER value the following loads its address into r1 and its length into r2 LDR v1 r0 8 Descriptor address LDMIA EL el r24 Address and length Function Results For non CHARACTER functions the address of the result is returned in r0 A CHARACTER function is implemented as a subroutine with the address of the result a character descriptor passed as an additional argument inserted before the other arguments thus the first argument in the call appears as the second argument etc A subroutine with alternate returns s in the argument list is implemented as an INTEGER function The result should be zero for the main return one for the first alternate return two for the second etc The alternate return specifiers do not appear i
30. abbreviation for Z0 A editing 15 The A edit descriptor can also handle numeric list items the effects are as recommended in Appendix C Hollerith of the FORTRAN 77 standard If the field width is zero the system will automatically use the right value for the data type being transferred 4 or 8 It must be emphasised that this use of A editing was introduced solely to aid in the transfer of FORTRAN 66 programs It should not be used otherwise Abbreviations symbol abbreviation OP P X X Tl T L1 TL R1 TR AO A Transfer of numeric items The I edit descriptor can be used to transfer real and double precision values F E D and G can be used to output an integer value Note that the external form of a value that is to be transferred to an integer variable must not have a fractional part or a negative exponent and descriptors A or descriptor in a format specification suppresses the final newline when writing to a non printer file Graphics FORTRAN programs can write control codes to the RISC OS VDU drivers to produce graphics The CHAR function is used to convert an integer code to a character for output The basic form of WRITE statement to generate graphics is WRITE 10A CHAR codel CHAR code 2 or PRINT 10A CHAR codel CHAR code2 The WRITE statement uses the standard asterisk output unit Any non printer unit 1 49 could be used instead The repeat count in the
31. ad to wasted file space in many cases The system includes a record length before every unformatted record when it is output and always reads the right amount when the record is read again The internal file format of the record is the characters UF a four byte count giving the number of data bytes followed by the data bytes The UF characters are used as a check that the file contains valid unformatted records The two records written in the example above would contain the following bytes 55 46 14 00 00 00 UF no of data bytes 20 01 00 00 00 02 00 00 00 03 00 00 00 04 00 00 00 05 00 00 00 data 55 46 OC 00 00 00 U F no of data bytes 12 13 06 00 00 00 07 00 00 00 08 00 00 00 data Direct Access Files A direct access file consists of a number of records all of the same length which may be read and written in any order The records are either all formatted or all unformatted An OPEN statement specifying the record length RECL must be used for a direct access file The record length is measured in bytes and formatted records are padded with spaces to this length The internal file format of a direct access file is the characters DA followed by a four byte count giving the record length It is permissible to OPEN a direct access file specifying a smaller record length than was given when the file was created The maximum permitted record length for a formatted direct access OPEN is 512 bytes there is no limit for unformatted fi
32. age level n 0 suppresses all warnings ton 4 print all warnings See the chapter Errors and Debugging Xn This option specifies the cross reference listing width 18 or more for legibility n 0 suppresses cross referencing The maximum value of n depends on where the listing is being sent for example the printer Cross reference information is given immediately after the END statement of a program unit For each name the type is given together with the lines on which it is referenced For each statement label the type executable or non executable and the line number of the statement is given as well as the lines on which the label is referenced 7 This option is used to control warnings about the use of FORTRAN 77 language extensions If it is not enabled warnings are not generated If it is enabled warnings are generated when the warning level wn is 2 the default or greater Warnings are not enabled by default so the extensions may be used without warnings being generated whatever the warning level ver filename The ver keyword specifies the filename of the output file for compiler and error messages generated If the keyword is not used the messages are output to the screen help The help keyword gives a summary of the keywords and arguments available Examples f77fe f77 prog to tmp fcode Compiles the source text in the file 77 prog to fcode format in the file tmp fcode f 77fe 77 prog ver x Compiles th
33. ample attempt to read beyond the end of the file record length too large The record length specified in a formatted direct access OPEN exceeds the permitted maximum 512 bytes bad direct access file message A file used for direct access has invalid initial data or an insufficient record length bad command line syntax sequential write failed message I O error on sequential output for example cannot extend 26 Appendix C The Front End The f77fe Command This reads FORTRAN 77 source text and converts it to a special intermediate form known as fcode The front end has the options x W T 6 and 7 The default settings are xow2 T67 The front end has the following command format f 77fe from filename to filename list filename opt options ver filename from filename The from keyword specifies the filename of the FORTRAN 77 source text input file to filename The to keyword specifies the filename of the fcode format output file If this keyword is not used then no output is produced list filename The list keyword specifies the filename of the list output file for a line numbered listing of the source text together with any error messages generated If this keyword is not used no listing is produced and error messages are output to the screen opt options The opt keyword specifies the options The options T 6 and 7 are enabled or disabled by preceeding them with or Th
34. ange 0 to 9 iregs 0 RO0 iregs 1 R1 etc Results ierror integer if no error occurs ierror 0 If an error occurs the function returns and ierror address of a standard error block iregs integer array one dimensional array with subscripts in the range 0 to 9 iregs 0 RO0 iregs 1 R1 etc IFRDB Purpose To return the byte stored at the given address and return it as an integer Example inum IFRDB iaddress Parameters iaddress integer Results inum integer byte stored at iaddress 41 IFWRB Purpose To store a byte at the given address Example CALL IFWRB inum iaddress Parameters inum integer byte to be stored at iaddress iaddress integer IFRDI IFRDW Purpose To return the integer word stored at the given address Note that the address must be on a word boundary and is in the valid address range Example inum IFRDI iaddress or inum IFRDW iaddress Parameters iaddress integer must be on a word boundary ie be a multiple of 4 Results inun integer integer word stored at iaddress IFWRI IFWRW Purpose To store an integer word at the given address Note that the address must be on a word boundary and is in the valid address range Example CALL IFWRI inum iaddress or CALL IFWRW inum iaddress Parameters inun integer integer word to be stored at iaddress iaddress integer must be on a word boundary ie be a multiple of 4 I
35. constant too large A REAL constant exceeds the permitted range double constant too large A DOUBLE PRECISION constant exceeds the permitted range inconsistent equivalencing involving name The given variable is involved in inconsistent EQUIVALENCE statements ncrement in DATA implied DO loop is zero A DATA statement implied Do loop has a zero increment pis nsufficient store for code generation The code generator has run out of memory pn nsufficient values in DATA constant list There are more variables than constants in a DATA statement fars integer invalid for length or size A value which is not positive has been used for a CHARACTER length or array size 22 lower bound exceeds upper bound in substring In a substring a constant lower bound exceeds the constant upper bound lower bound of substring is less than one A constant substring lower bound is less than one upper bound exceeds length in substring A constant substring upper bound exceeds the length of the character variable stack overflow program must be simplified The internal expression stack has overflowed The offending statement must be simplified subscript below lower bound in dimension N A constant array subscript is less than the lower bound in the given dimension subscript exceeds upper bound in the dimension N A constant array subscript exceeds the upper bound in the given dimension too many constants
36. e The 1ink 77 command links the object file filename with the IFExt IFLib and 77 libraries to produce the program aif filename The df77 df77Ik and dlinkf77 Commands The df77 df77 k and dlinkf77 commands are similar to the 77 f77Ik and linkf77 commands but include the debug all rather than the default options 36 Appendix The IFExt Utility Library This is a small library which includes routines to make OS_Byte OS_Word and OS_CLI calls Examples illustrating the use of the routines are included in the FORTRAN Examples IFExt directory IFOSBYTE Purpose To make OS _Byte calls which do not return any values Example CALL IFOSBYTE IFUNC IARG1 IARG2 Parameters IFUNC integer RO IFUNC IARG1 integer R1 IARG1 IARG2 integer R2 IARG2 IFOSBYTE1 Purpose To make OS _Byte calls that return one value Example CALL IFOSBYTE1 IFUNC IARG1 IARG2 IRES1 Parameters IFUNC integer RO IFUNC IARG1 integer Rl IARG1 IARG2 integer R2 IARG2 Results IRES1 integer IRES1 R1 IFOSBYTE2 Purpose To make OS _Byte calls that return two values Example CALL IFOSBYTE2 IFUNC IARG1 IARG2 IRES1 IRES2 Parameters IFUNC integer RO IFUNC IARG1 integer R1 IARG1 IARG2 integer R2 IARG2 Results IRES1 integer IRES1 R1 IRES2 integer IRES2 R2 IFOSWORD Purpose To make OS_Word calls Example CALL IFOSWORD ICODE
37. e otherwise they are in a directory tmp These are deleted when they are no longer required Extensions to the Standard The FORTRAN 77 compiler offers several extensions to the standard Further extensions concerning input output are described in the next chapter Hexadecimal Constants The compiler allows hexadecimal constants to be used and has the following form lt type gt lt digits gt type is a letter specifying the type of the constant It must be one of I R D C L H or Q for INTEGER REAL DOUBLE PRECISION COMPLEX LOGICAL CHARACTER and COMPLEX 16 respectively The type letter is followed by hexadecimal digits 0 9 A F There must always be an even number of digits that is an exact number of bytes The bytes in a CHARACTER hexadecimal constant are given in the order in which they are to appear in memory With other constants the most significant byte is given first If the type of the constant is REAL DOUBLE PRECISION COMPLEX or COMPLEX 16 the number of bytes must match the size of the item in memory 4 8 or 16 for INTEGER and LOGICAL constants there may be fewer bytes Example CHARACTER WINDOW PARAMETER WINDOW H1CO5141E0C J 11234 WINDOW consists of the bytes 1c 05 14 1E Oc and Jis set to the decimal value 4660 Naming The compiler converts all lower case letters apart from FORMATS and CHARACTER constants to upper
38. e options W and X must be followed by a number The options have the following meanings 6 This option allows FORTRAN 66 source text to be compiled Constructs which have a different meaning in FORTRAN 77 are interpreted according to the FORTRAN 66 definition In particular DO loops will always execute at least once Hollerith nH constants are allowed in DATA and CALL statements and quoted constants in calls are not of CHARACTER type Non CHARACTER array names are allowed as format specifiers When the FORTRAN 66 option is used Hollerith and quoted constants are treated in the same way when used as arguments in CALLs they are not of CHARACTER type The option is provided for use with FORTRAN 66 programs which store character information in numeric data types For example the following calls will have identical effects at run time if the FORTRAN 66 option is used CALL jim abcd CALL jim 4habcd If the FORTRAN 66 option is used run time FORMATS specifiers may also be non CHARACTER array names For example DOUBLE PRECISION d 3 num DATA d 1 d 3 8h 1X D20 5h 15 DATA num 2h10 27 This option was introduced to allow FORTRAN 66 programs to be compiled It isstrongly recommended that new programs conform to the FORTRAN 77 standard T This option causes special trace routines to be include in the code See the chapter Errors and Debugging wn This option specifies the warning mess
39. e source text in the file 77 prog producing no fcode output but with messages output to the file x f77fe 77 prog to tmp fcode list list prog Compiles the source text in the file 77 prog to fcode format in the file tmp fcode and also outputs a source listing to the file List prog 77fe 77 prog to tmp fcode opt T Compiles the source text in the file 77 prog to fcode format in the file tmp fcode with tracing calls included 28 Appendix D The Code Generator The f77cg Command This reads fcode format and generates aof format and or assembler source text format The code generator has the options L B and H The option 6 may be used instead of H The default settings are L1 BH The front end options T 7 w and x are ignored by the code generator whilst the front end ignores B and L so that the same option string may be given to both programs if required The code generator has the following command format f77cg fcode filename to filename asm filename ver filename map filename opt options debug level source name fcode filename The fcode keyword specifies the filename of the fcode format input file to filename The to keyword specifies the filename of the aof format output file generated If the keyword is not used then no aof format output file is generated asm filename The asm keyword specifies the filename of the assembler source text format output file equ
40. ent linking when the newer version of the linker is used see Appendix G Configuring the Fortran77 Application The Fortran77 Run obey file sets the following operating system variables to configure Fortran77 F77cl1 Dir points to the directory containing the compiler front end f77fe code generator f77cg and linker link as supplied lt Fortran77 Dir gt Library F771libsSDir points to the directory containing the Fortran libraries as supplied Fortran77 lib MaxF77S Libs sets the maximum number of libraries as supplied 20 MaxF77 Files sets the maximum number of source or object files as supplied default 20 The Fortran77 Run obey file also sets the following operating system variables to configure command line operation options F77 Tmp points to the directory used for temporary scratch files created during compilation as supplied Fortran77 tmp F77 Lib points to the directory containing the Fortran libraries as supplied Fortran77 1lib Run Path points to the directory containing the f77 f77Ik linkf77 d77 df77 k and dlinkf77 commands as supplied Fortran77 Execlib NewLink Using the Fortran77 Application Before carrying out the selected operations the Fortran77 application sets the current directory to the directory containing the 77 directory and creates any aof and aif directories which it needs and which do not already exist If any INCLUDE files are to be used they should be stored i
41. es a computer fitted with 4 Mbyte of RAM a hard disc RISC OS version 3 1 to 3 71 or RISC OS version 4 02 or greater and is StrongARM compatible Conventions Used Text entered by the user and as it appears on the screen is shown as follows This is text as it appears on the screen Arguments to commands and options are shown as follows debug arguments The chosen value must be entered for arguments Optional arguments are shown in square brackets map file Installation 1 Open a directory viewer on a Suitable directory for the FORTRAN directory on the destination filing system If this is anything other than the root directory Fortran77 must be reconfigured as described in the next section 2 Open a directory viewer on Distribution Disc 1 3 Drag the FORTRAN directory from Distribution Disc 1 to the suitable directory on the destination filing system 4 Remove Distribution Disc 1 and keep it in a safe place Open a directory viewer on Distribution Disc 2 Drag the FORTRAN directory from Distribution Disc 2 to the suitable directory on the destination filing system 7 Remove Distribution Disc 2 and keep it in a safe place 8 Open a directory viewer a suitable directory for the Library Directory on the destination filing system If this is anything other than the root directory Fortran77 must be reconfigured as described in the next section 9 Open a directory viewer on Distribution Disc 3 10 Drag the Library directory from D
42. filename Diagnostic information is output to the file filename LIST filename Map and Xref information is output to the file filename not stdout MAP Prints an area map to the standard output Symbols filename Symbol definitions are output to the file filename Verbose Gives information as files are linked VIA filename The object and library files listed in the text file filename are linked Xref Prints an area cross reference list to the standard output Output options AIF Absolute AIF the default 34 AIF Relocatable AIF R Workspace nnn AOF BIN BIN AIF IHF SPLIT SHL filename SHL filename REENTrant RMF OVerlay filename Special options RO base n Base n RW base n DATA n Entry n Entry n obj area Case MATCH n FIRST obj area LAST obj area NOUNUSEDareas Unresolved sym Ct Relocatable AIF Self moving AIF Partially linked AOF Plain binary Plain binary described by a prepended AIF header Intellec Hex Formay readable text Output RO and RW sections to separate files BIN IHF Shared library stub as described in filename Shared library reentrant stub RISC OS Module Overlaid image as described in filename Specify base of image Specify separate base for image s data Specify entry address Specify entry as offset within area prefix n with amp or Ox for hex postfix with K for 21 M for 22 Ignore case when s
43. g a complex value using list directed free format input an integer or real constant can be given the imaginary part of the value is set to zero When reading a character value if the constant does not start with a quote is contained on a single record does not contain an embedded space comma or character does not start with digits followed by a then the delimiting quotes may be omitted and embedded quotes are not doubled RISC OS Interface Routines The small utility library IFExt see Appendix contains alternative routines to those listed in this section The small utility library IFLib see Appendix J includes routines to return the addresses of variables make SWI calls and read and write memory The FORTRAN run time library contains the following routines to interface to the operating system Examples illustrating the use of these routines are included in the FORTRAN Examples General directory OSBYTE Purpose To make OS _Byte calls which do not return any values Example CALL OSBYTE IFUNC IARG1 IARG2 Parameters IFUNC integer RO IFUNC IARG1 integer Rl IARG1 IARG2 integer R2 IARG2 OSBYTE1 Purpose To make OS _Byte calls that return one value Example CALL OSBYTE1 IFUNC IARG1 IARG2 IRES1 Parameters IFUNC integer RO IFUNC IARG1 integer R1 IARG1 IARG2 integer R2 IARG2 Results IRES1 integer IRES1 R1 OSBYTE2 Purpose To make OS _By
44. he screen can be redirected to output to a file using the standard RISC OS syntax gt filename and the input stream from the keyboard can be redirected to input from a file using the standard RISC OS syntax lt filename Any OPEN files are closed when a program terminates When writing to a sequential formatted file a distinction is made between files which are to be printed and those which are not When writing to files which are to be printed the first character of each record is a carriage control code and does not form part of the data in the record All units in the range 50 60 assume printer output format by default On other units specifying FORM PRINTER in the first OPEN statement for the unit causes printer output format to be assumed for that unit This is an extension to the standard Note that the printer output format does not imply output to any physical printer The carriage control codes which are recognised and their representation in files are described in the section Formatted I O 11 Sequential Files OPEN and CLOSE The OP WRITE statement is executed An OP fail The error will occur when a RI ERR in the RI EAD or WRITE statement The following subroutine shows the use of OP the terminal using unit 10 EN statement for a sequential file does not specify whether the file is to be read from or written to Therefore the operating system is called to open the
45. he valid address range Example inum IFHWRDW iaddress Parameters iaddress integer must be on a word boundary ie be a multiple of 4 Results inun integer word stored at iaddress IFHWWRW Purpose To store a word 32 bits at the given address Note that the address must be on a word boundary and be in the valid address range Example CALL IFHWWRW inum iaddress Parameters inum integer word to be stored at iaddress iaddress integer must be on a word boundary ie be a multiple of 4 43 IFSWI Purpose To enable SWI s to be called from FORTRAN 77 Example lerror IFSWI numswi iregsin iregsout iflags Parameters numswi integer SWI number Clear bit 17 800020000 to abort on error Set bit 17 amp 00020000 to return on error iregsin integer array one dimensional array with subscripts in the range 0 to 9 iregsin 0 RO0 iregsin 1 R1 etc Results ierror integer if no error occurs ierror 0 If bit 17 of numswi amp 00020000 is clear and an error occurs the function does not return If bit 17 of numswi amp 00020000 is set and an error occurs the function returns and ierror address of a standard error block iregsout integer array one dimensional array with subscripts in the range 0 to 9 iregsout 0 R0 iregsout 1 R1 etc iflags integer processor flag bits bitO V flag bit C flag bit2 Z flag bit3 N flag IFSWIX Purpose To enable SWI s to be called from
46. if the unit is in the valid range Itis not possible to return accurate responses for SEQUENTIAL DIRECT FORMATTED and UNFORMATTED YES is returned if the unit is currently being used for the relevant access type otherwise UNKNOWN is returned Note that NAMED can only be used if FILE was specified in the OPEN statement for the unit Command line file assignments are not available to INQUIRE INQUIRE by file 14 If FILE was specified in an OPEN statement for a unit and not CLOSEd information deduced from that association is returned for example DIRECT is returned as YES if the file is open for direct access and the file is assumed to exist Otherwise if the file exists EXIST returns TRUE and SEQUENTIAL DIRECT FORMATTED and UNFORMATTED return UNKNOWN BACKSPACE BACKSPACE is not implemented ENDFILE ENDFILE sets the end of file status and prevents further file access REWIND REWIND is implemented as a CLOSE followed by an OPEN After executing a REWIND the file is in a similar state to that arising after an OPEN statement the operating system is called to open the file when the first READ or WRITE statement is executed Format Decoding Format specifications are decoded in a more liberal manner than as defined by the FORTRAN 77 standard Lower case Lower case can be used instead of upper case
47. ing Functions and Subroutines Written in Assembler from FORTRAN Appendix L Notes on Using a Debugger 29 29 31 31 33 33 34 34 36 36 36 37 37 39 39 45 45 49 49 Introduction For the seriously scientific user the Desktop Fortran 77 package enables an Acorn RISC OS based computer to be used as a cost effective workstation for developing large Fortran programs The Fortran77 application enables Fortran programs to be compiled linked and run in the RISC OS Desktop environment When used with the editor supplied SrcEdit it can throwback errors by highlighting the line containing the error in the source text The compiler fully conforms to the ANSI FORTRAN X3 9 1978 standard and in addition provides a number of optional extensions The package contains the Fortran77 application the compiler front end f77fe the compiler code generator f77cg a choice of linkers oldlink and newlink the source editor SrcEdit the symbolic debugger asd the IFExt and IFLib utility libraries which include routines to return the addresses of variables make SWI calls and read and write memory and the DrawF Graphics SpriteOp Utils and Wimp public domain libraries A text file helpF77 is supplied to enable SrcEdit to provide on line help The User Guide describes the installation and use of the compiler on Acorn RISC OS based computers but is not a tutorial on Fortran programming The package requir
48. istribution Disc 3 to the root directory of the destination filing system 11 Oncomputers running RISC OS 3 60 or earlier update the System application by dragging the System application from Distribution Disc 3 to the System application of the computer 12 Remove Distribution Disc 3 and keep it in a safe place 13 Re set the computer Checking the Installation 1 Open a directory viewer on the FORTRAN Examples General directory 2 Double click on the Test obey file 3 The following should be displayed Topexpress FORTRAN 77 front end version 1 19 Program WORLD Compiled otal workspace used 6016 ARM FORTRAN 77 code generator version 1 62 Main program WORLD code 104 data 20 Total code size 104 data size 20 ARM Linker Warning Attribute conflict within AREA F77 Data conflict first found with rts F77 Data ARM Linker attribute difference 0 INIT ARM Linker Warning Symbol ImageS DataLimit referenced ImageSSRWSSLimit used A m RM Linker finished 1 informational 2 warning and 0 error essages Hello Fortran world STOP Press SPACE or click mouse to continue Directories A directory for FORTRAN programs must in turn contain the following directories 77 contains FORTRAN source text files aof contains Acorn Object Format files for subsequent linking aif contains executable Acorn Image Format files 2 o contains Acorn Object Format files for subsequ
49. ivalent to the aof format generated If the keyword is not used then no assembler source text file is produced ver filenam The ver keyword specifies the filename of the output file for compiler and error messages generated If the keyword is not used the messages are output to the screen opt options The opt keyword specifies the options The options B and H are enabled or disabled by preceeding them with or The option L must be followed by a number The options have the following meanings B When enabled bound checking code is included Array or substring subscripts out of range will cause run time errors H When enabled Hollerith constants can be used in DATA statements to initialisenon character variables for example INTEGER Ln The number following this option indicates the level of line numbering included in the code for backtrace purposes see the chapter Errors and Debugging The levels available are 0 no line numbering 1 numbers lines containing subprogram calls 2 numbers statements which can cause a run time exception gt 2 numbers every line Higher levels cause more code to be generated If a hardware exception occurs ina program unit compiled with level 1 the backtrace system will not be able to determine the exact line number A range of numbers will be given for example 100 106 and the error will be between them 29 map filename The map keyword specifies the filename of map output file
50. les If the file has been OPENed for updating or input the first six bytes of the file are read and checked The OPEN will fail if these bytes are invalid or the specified record length is greater than the value used when the file was created As it is possible both to read from and write to a direct access file the operating system is called to open the file when the OPEN statement is executed rather than being delayed until the first READ or WRITE which occurs with an OPEN statement for a sequential file Therefore any errors which occur may be trapped by specifying ERR in the OPEN statement The following program uses direct access to write to and read from a file OPEN 42 ACCESS DIRECT FILE EGDATA RECL 16 ERR 100 IOSTAT IERR DO 1 J 20 1 1 1 WRITE 42 REC J J J 1 J J J 1 DO 2 J 1 10 READ 42 REC J K L M 2 WRITE 3 K L M 3 FORMAT 1X 3I5 STOP 100 PRINT OPEN FAIL IERR END Note that unformatted records are the default for direct access files The file EGDATA used in the above example need not exist but if it does it must be a valid direct access file with a record length greater than or equal to 16 OPEN and CLOSE The OPEN and CLOSE statements have been discussed above Specifying STATUS NEW or STATUS OLD in the OPEN statement has no effect INQUIRE INQUIRE by unit EXIST returns TRUE
51. n line feed or line feed carriage return is treated as a single record terminator When writing a record to a file or terminal the carriage control code or codes are output first followed by the data in the record Trailing spaces in a record are not output The following carriage control codes are recognised space performs a line feed LF 0 performs LF LF extra blank line 1 performs CR FF newpage performs CR overprint no action taken The initial LF space or 0 or CR 1 or is not output before the first record in a file When a file is closed a line feed code is output if the done for all OP final record contained any data characters This is EN files when a program terminates normally When writing to a non printer unit each record is terminated by a new line If a prompt line is required a or character may be included in the format This suppresses the final new line and trailing spaces are not removed from the interactive prompts WRITE ry 6 AS Type an intege 12 final line output This may be used to generate ee The or acts as a normal item like and can occur anywhere in the format except after any unused editing codes since these will be skipped The following program illustrates interaction with a terminal 1 PRINT Sa READ END 3 I WRITE 2 I I I 2 FORMAT 2110 GOTO 1 3 END The CHAR function
52. n the argument list Static Data Static data for an assembler routine should be in a writable area and addressed using sb r13 as this is preserved by calls Section Format The code area of a FORTRAN callable assembler routine should start with the routine name as a twelve character string padded with spaces The address of the first byte following this name must be pushed on to the stack at the beginning of the routine The code area should be named F77SSCode and have the attributes CODE and READONLY The data area if any should be named F77 SData and have the DATA attribute 45 FORTRAN common blocks should be defined as named AREAs with the COMMON and NOINIT attributes An initialised COMMON block equivalent to a BLOCK DATA subprogram should be defined with the COMDEF common definition attribute FORTRAN blank common is given the name F77_BLANK The basic layout of a FORTRAN callable assembler routine is Name padded with spaces to 12 characters Address of data Standard entry sequence Address of data area Copy address of argument xit sequenc TTL name Registers ro RN 0 sag RN T r9 RN 9 fp RN 10 sp RN 12 sb RN 13 LE RN 14 pc RN 15 EO FN 0 77 FN 7 Data AREA F77SS Data DATA data declarations Code AREA F77S Code CODE READONLY NAME DCB ASMROUTINE y F i DATAPTR DCD F77S Data H EXPORT ASMROUTINE ASMROUTINE ADR r1 NAME 12 STMFD sp rl fp sb 1r LD
53. n this directory so that they do not need a directory prefix 1 Install SrcEdit and Fortran77 on the icon bar Click menu over the Fortran77 icon to enable the options to be saved fortran info 4 help 4 save options quit To open the main window either a drag a source text file from an 77 directory onto the Fortran77 icon or b click the select button over the Fortran77 icon and drag the source textfile s from an 77 directory to the Fortran77 window mised Fortran Wm ir Directory Compile Link Squeeze Run 4T aof lib Aif fi v Sal Drag any previously compiled object files to the Fortran77 window The 77 and aof directories must both be sub directories of the same directory Click select over the Compile Link Squeeze or Run icons to select the operations required Caio Fortran i I Directory SCSI SC5IDised FORTRAN Examples Compile Squeeze ETT real kiif Laot Jib aif HelloW fi HelloW v eI gt E 5 Click menu over the Compile icon to display the compiler options Click select to select an option Click adjust to de select an option Click select over Throwback to select throwback of errors and provide on line help Do not select any of the debug options see A
54. nt to the asterisk and may only be used in sequential READS and WRITES Note that the filing system limits the number of files that can be open simultaneously A unit may be associated with an external file either by means of an OPEN statement or by assignments on the command line when the program is run If an OPEN statement with the FILE specifier is used then the unit is associated with the given filename Otherwise the command line arguments are scanned The format of the command line is command filenamel filename2 unitnol filenamel unitno2 filename2 An optional list of filenames is followed by an optional list of assignments of unit numbers to file names The initial list of unassigned filenames are associated with units numbers 1 2 3 etc Each assigned filename is associated with the given unit number All unassigned filenames must precede any assigned filenames Example PROG ABC DEF This associates the file ABC with unit number 1 and DEF with unit number 2 PROG 10 RESULTS This associates the file RESULTS with unit number 10 PROG RESULTS1 32 RESULTS2 3 X This associates RESULTS1 with unit number 1 RESULTS 2 with unit number 32 and x with unit number 3 The two units always refer to the screen and keyboard Any units which are not associated with a file in an OPEN statement or through command line arguments also refer to the screen and keyboard The output stream to t
55. of prog Generates the aof format file aof prog from the fcode format file tmp fcode 77cg tmp fcode asm vdu Generates assembler source text format and outputs it to the screen vdu from the fcode format file tmp fcode f77cg tmp fcode to aof prog map map prog Generates the aof format file aof prog from the fcode format file tmp fcode and sends map output to map prog f77cg tmp fcode to aof prog opt B Generates the aof format file aof prog from the fcode format file tmp fcode with bound checking code included 77cg tmp fcode to aof prog debug all source f 77 prog Generates the aof format file aof prog from the fcode format file tmp fcode with full debugging information and with the name of the FORTRAN source text file specified as 77 prog 30 Appendix E The Linkers Two linkers are supplied the older linker oldlink see Appendix F and the newer linker newlink see Appendix G The linkers combine a number of object files with library files to produce a single executable program Each of the object files must be in Acorn Object Format aof or Acorn Library Format alf A file may contain references to external symbols procedure and variable names which the linker attempts to resolve by searching for definitions in the other files Usually at least one library file will be specified A library is a collection of Acorn Object Format files stored in a single Acorn Library Format file Libra
56. ppendix L F7 Options F66 Code BoundsCheck Hollerith Debug All Lines Vars Min TraceLines OLBIA TraceCode E Warnings omni E 4 F77 Listings Asn K ret 6 Click menu over the Link icon to display the linker options Click select to select an option Click adjust to de select an option Choose the libraries to be included by clicking select over the appropriate library name ys Link Options a Relocatable tap ie Verbose K ref l Libraries inlr lib Binaryio DrawF Graphics IFLib Spritelp Utils Himp E 7 Click menu over the Run icon to enter command line arguments Es Run Options m Hello 8 Click select over the Start icon to carry out the selected operations 9 Click select over an item in a list of files to move it to the top this is the way to change the scanning order of library files Click adjust over an item in a list of files to remove it If there are any compilation or link errors they are written to the file err prog where prog is the name of the source text file The details are displayed by Edit or SrcEdit if RISC OS has seen them similarly asm prog 1is prog map prog are created if the corresponding option has been selected Temporary files used during compiling and linking are stored on a RAM disc if possibl
57. pting to divide by zero Front End Error Messages Errors detected by the compiler front end are of a different type from those detected by the code generator Front end error messages are short obvious statements indicating that the compiler has detected unacceptable syntax These messages are self explanatory There are two classes of error Class 1 errors cause the front end to abandon compilation of the current statement The statement is printed as part of the error message together with the number of the line on which the fault appeared an error number and a description of the error itself Thus if line 211 contained the incorrect FORTRAN statement 100 ERRONEOUS then the message produced would be 211 100 ERRONEOUS L 211 2 Error code 2311 Statement not recognised Class 2 errors may be less obvious in their report of a fault and do not always refer to the line which contains the code which instigated the error For example information about missing labels is given at the end of the program unit rather than where the non existent label was referenced The distinction between these two types of error message has been made in order to show that errors do not necessarily occur at the line where the message is given Warning Messages The w compilation option enables the compiler to generate warnings These warnings are graded in severity from 1 the most serious to 4 and are useful if the program
58. re included in the FORTRAN Examples IFLib directory IFADR Purpose To return the address of an integer variable Example iadr IFADR inum Parameters inum integer Results iadr integer address of variable inum IFADRF Purpose To return the address of a real single precision floating point variable Example REAL fnum iadrf IFADRF fnum Parameters fnum real Results iadrf integer address of variable fnum IFADRD Purpose To return the address of a real double precision floating point variable Example DOUBLE PRECISION dnum iadrd IFADRD dnum Parameters dnum double Results iadrd integer address of variable dnum IFADRC Purpose To return the address of the character descriptor which contains the address of the character string in the first word and its length in the second Example CHARACTER string 3 itadrc IFADRC string Parameters 39 string character Results iadrc integer address of character descriptor IFADRL Purpose To return the address of an logical variable Example LOGICAL status iadr IFADRL status Parameters status logical Results iadr integer address of variable status IFADRCMPLX Purpose To return the address of a complex variable Example COMPLEX voltage Ladr IFADRCMPLX voltage Parameters voltage complex Results iadr integer address of variable voltage IFADRCMPLX16 Purpose To ret
59. ries differ from object files in the way that the linker searches them Object files are searched only once when the linker attempts to resolve external references Libraries are searched as many times as necessary If a required symbol is found in one of the component files of the library then the whole component file is incorporated in the output file Two common errors which occur during linking are caused by unresolved and multiple references In the first case a symbol has been referenced in a file whose name is given in the error but there is no corresponding definition of the symbol This is usually caused by the omission of a required object or library file or the mis spelling of a name in the original source program In the second case a clash of names occurs For example a procedure might have been defined with the same name as a library procedure or as a procedure in another object file The version of the procedure used in any situation is the one local to the reference to it Wildcards can be used in the filenames These will be expanded into the list of files matching the specification For example the name aof bas might be expanded into aof basmain aof basexpr and aof bascmd Predefined Linker Symbols There are several symbols which the linker knows about independently of any of its input files These start with the string Image and along with all other external names containing are reserved by Acorn The symbols
60. s sent to the screen or may be sent to a file using the SPOOL command 21 Appendix A Code Generator Error Messages argument out of range for CHAR The intrinsic function CHAR has been used with a constant argument outside the range 0 255 local data area too large The size of the local storage area for the program unit exceeds memory size array lt name gt has invalid size The size of the given array is negative or exceeds memory size attempt to extend common block name backwards An attempt has been made to extend a COMMON block backwards by means of EQUIVALENCE statements bad length for CHARACTER value A value which is not positive has been used for a CHARACTER length class storage block containing lt name gt is too large class is local or COMMON The storage block containing the named variable exceeds memory size concatenation too long The result of a CHARACTER concatenation may exceed memory size conversion to integer failed A REAL Of DOUBLE PRECISION value is too large for conversion to an xr D to R real conversion failed A DOUBLE PRECISION value is too large for conversion to a RI CI AL DATA statement too complicated The variable list in a DATA statement is too complicated and must be simplified division by zero attempted in constant expression The divisor might be REAL INTEGER DOUBLE PRECISION or COMPLEX real
61. te calls that return two values Example CALL OSBYTE2 IFUNC IARG1 IARG2 IRES1 IRES2 9 Parameters IFUNC integer RO IFUNC IARG1 integer R1 IARG1 IARG2 integer R2 IARG2 Results IRES1 integer IRES1 R1 IRES2 integer IRES2 R2 OSWORD Purpose To make OS_Word calls Example CALL OSWORD ICODE IARRAY Parameters ICODE integer RO ICODE IARRAY integer array R1 pointer to IARRAY one dimensional integer array the OS_Word parameter block Results Placed in IARRAY OSCLI Purpose To make OS_CLI calls Example OGICAL FUNCTION OSCLI STRING OGICAL STATUS STATUS OSCLI STRING Parameters STRING character command line terminated by a carriage return Results STATUS logical if the command is executed without error STATUS TRUE or if an error occurs STATUS FALSE OSGETERROR Purpose To return the error number and error message immediately after OSCLI has returned with STATUS FALSE 7 Example CALL OSGI J TERROR H ERRNO ERRSTR Parameters None Results ERRNO integer the error number RRSTR character the error message H azl 10 Input Output Unit Numbers and Files A FORTRAN unit number is used to refer to a file Unit numbers in the range 1 to 60 may be used as well as the two units for the keyboard and screen Zero is equivale
62. urn the address of a double precision complex variable Example COMPLEX 16 current iadr IFADRCMPLX16 current Parameters current double precision complex Results iadr integer address of variable current IFADRS Purpose To return the address of the character string Example iadrs IFADRS string Parameters string character Results iadrs integer address of character string 40 IFQSWI Purpose To enable SWI s to be called from FORTRAN 77 Example Llerror IFQSWI numswi iregs Parameters numswi integer SWI number Clear bit 17 800020000 to abort on error Set bit 17 amp 00020000 to return on error iregs integer array one dimensional array with subscripts in the range 0 to 9 iregs 0 RO0 iregs 1 R1 etc Results ierror integer if no error occurs ierror 0 If bit 17 of numswi amp 00020000 is clear and an error occurs the function does not return If bit 17 of numswi amp 00020000 is set and an error occurs the function returns and ierror address of a standard error block iregs integer array one dimensional array with subscripts in the range 0 to 9 iregs 0 RO iregs 1 R1 etc IFQSWIX Purpose To enable SWI s to be called from FORTRAN 77 Example Llerror IFQSWIX numswi iregs Parameters numswi integer SWI number the function sets bit 17 00020000 to return on error iregs integer array one dimensional array with subscripts in the r
63. ymbol matching Set last gasp symbol matching option Place area from object obj first in the output image PLace area from object obj last Do not eliminate AREAs unreachable from the AREA containing the entry point AIF images only Make all unresolved references refer to sym Support C external naming conventions 35 Appendix H The f77 f77Ik and linkf77 Commands In order ot use the f77 f771k and linkf77 commands the currently selected directory must be set to the directory which contains the following directories 77 contains FORTRAN source text files aof contains Acorn Object Format files for subsequent linking aif contains executable Acorn Image Format files o contains Acorn Object Format files for subsequent linking when a newer version of the linker is used see Appendix G The version of the linker used either the older p1d1ink or newer newlink can be selected by editing the FORTRAN Fortran77 Boot obey file as shown in Appendices F and G The f77 Command 77 filename The 77 command combines the 77fe and 77cg commands to compile the file 77 filename using the default options The f77Ik Command f771k filename The 771k command combines the 77 e and 77cg commands to compile the file 77 filename using the default options It then links the resulting object file with the IFExt IFLib and 77 libraries to produce the program aif filename The linkf77 Command linkf77 filenam

User Guide - Intelligent Interfaces

Contents

Download Pdf Manuals

Related Search

Related Contents