- Gpc - gnu.de

Graphics in Pascal
by Peter Gerwinski 20 Jan '97

20 Jan '97

Brin wrote: > I wrote the following small program to try out some graphics > routines [...] There is already BGI2GRX, a "Graph" compatible Unit, based on GRX20 and BCC2GRX, written by Sven Hilscher. > [...] Here is the program, any idea to > what I am doing wrong! The problem is that you are confusing some sizes of data types: > {****************************************************************} > Program Mode13h; > type > {Define some 'terms'} > byte = __byte__ integer; {8 bit unsigned integer} signed > word = __unsigned__ integer; {16 bit unsigned integer} 32 > Long = __longlong__ integer; {64 bit signed integer} correct, but used in wrong context (see below) If you correct them, your program runs fine. Below are the correct sizes for Intel X86 platforms. (I am thinking about implementing the identifiers below into GPC's standard vocabulary.) Type ByteInt = __byte__ Integer; (* 8 bit signed Integer *) Byte = __unsigned__ ByteInt; (* 8 bit unsigned Integer *) ShortInt = __short__ Integer; (* 16 bit signed Integer *) ShortWord = __unsigned__ ShortInt; (* 16 bit unsigned Integer *) (* Integer = __long__ Integer *) (* 32 bit signed Integer *) Word = __unsigned Integer; (* 32 bit unsigned Integer *) LongInt = __longlong__ Integer; (* 64 bit signed Integer *) LongWord = __unsigned__ LongInt; In addition, "unsigned long" in C has 32 bits, not 64: > {routines taken from include/sys/farptr.h} > {Puts a byte} > procedure _farnspokeb(offset: Long ; value: byte);External;C; > {Gets a byte} > function _farnspeekb(offset: Long ; value: byte):byte;External;C; (* Here's ^^^^^^^^^^^ another error ;*) Procedure FarNoSelectorPokeByte ( Offset: Word; value: Byte ); AsmName '_farnspokeb'; (* ;-) *) Function FarNoSelectorPeekByte ( Offset: Word ): Byte; AsmName '_farnspeekb'; (* (-; *) Hope this helps, Peter e-mail: peter.gerwinski(a)uni-essen.de home address: D\"usseldorfer Str. 35, 45145 Essen, Germany WWW: http://agnes.dida.physik.uni-essen.de/~peter/

1 0

Re: GNU Pascal, go32.h
by Peter Gerwinski 20 Jan '97

20 Jan '97

According to Brin: > I am trying to use the farptr.h library and the go32.h libaray. [...] > To use them I need _dos_ds, which is defined in the C library go32.h > but is not a function or a procedure, but is: #define. I tried > incluing go32.h but got errors (due to it being a C file I think). Yes. GNU Pascal does not parse C source. ;-) > How can I import _dos_ds? Only by re-writing the functionality of the macro (#define), i.e. translating it to Pascal. In this case it is access to a component of a struct (a record in Pascal): Type ByteInt = __byte__ Integer; (* 8 bit signed Integer *) Byte = __unsigned__ ByteInt; (* 8 bit unsigned Integer *) ShortInt = __short__ Integer; (* 16 bit signed Integer *) ShortWord = __unsigned__ ShortInt; (* 16 bit unsigned Integer *) (* Integer = __long__ Integer *) (* 32 bit signed Integer *) Word = __unsigned__ Integer; (* 32 bit unsigned Integer *) Go32InfoBlockRecord = record size_of_this_structure_in_bytes, linear_address_of_primary_screen, linear_address_of_secondary_screen, linear_address_of_transfer_buffer, size_of_transfer_buffer, (* >= 4k *) pid: Word; master_interrupt_controller_base, slave_interrupt_controller_base: Byte; selector_for_linear_memory: ShortWord; linear_address_of_stub_info_structure, linear_address_of_original_psp: Word; run_mode, run_mode_info: ShortWord; end (* Go32InfoBlockRecord *); Var Go32InfoBlock: asmname '_go32_info_block' Go32InfoBlockRecord; Then by typing Go32InfoBlock.selector_for_linear_memory you get the functionality of _dos_ds. Now, of course, you can define it as a macro (*$define _dos_ds Go32InfoBlock.selector_for_linear_memory *) but then the definition is case sensitive. Or: Write a C function which uses the macro, then use that function from Pascal. But that's less effective. Hope this helps, Peter e-mail: peter.gerwinski(a)uni-essen.de home address: D\"usseldorfer Str. 35, 45145 Essen, Germany WWW: http://agnes.dida.physik.uni-essen.de/~peter/

1 0

Re: GPC and G77
by Peter Gerwinski 14 Jan '97

14 Jan '97

According to Nils Bokermann: > If trying to build the gpc-2.0 with the gcc-2.7.2.1 with the GNU Fortran > enabled, will not run. There are some problems with the file toplev.c. Here is > a simple diff, which worked at mine. > [...] > It's from the patch which came with GNU Fortran. Thanks for your work! I am thinking about including the patch into GPC's toplev.c since it doesn't seem to interfere with other parts of GCC, but it might be better to remain consistent with GCC-2.7.2.1 and to put the patch into the documentation. What would you suggest? Greetings, Peter e-mail: peter.gerwinski(a)uni-essen.de home address: D\"usseldorfer Str. 35, 45145 Essen, Germany WWW: http://agnes.dida.physik.uni-essen.de/~peter/

1 0

GPC and G77
by Nils Bokermann 13 Jan '97

13 Jan '97

Hello! If trying to build the gpc-2.0 with the gcc-2.7.2.1 with the GNU Fortran enabled, will not run. There are some problems with the file toplev.c. Here is a simple diff, which worked at mine. --------cut-------------- diff -rcp2N gcc-2.7.2.1/toplev.c gcc-2.7.2.1.f.1/toplev.c *** gpc-2.0/toplev.c Fri Oct 20 17:56:35 1995 --- gpc-2.0.f.1/toplev.c Mon Nov 11 15:02:33 1996 *************** int flag_unroll_loops; *** 388,391 **** --- 388,405 ---- int flag_unroll_all_loops; + /* Nonzero forces all invariant computations in loops to be moved + outside the loop. */ + + int flag_move_all_movables = 0; + + /* Nonzero forces all general induction variables in loops to be + strength reduced. */ + + int flag_reduce_all_givs = 0; + + /* Nonzero gets another run of loop_optimize performed. */ + + int flag_rerun_loop_opt = 0; + /* Nonzero for -fwritable-strings: store string constants in data segment and don't uniquize them. */ *************** struct { char *string; int *variable; in *** 542,545 **** --- 556,562 ---- {"unroll-loops", &flag_unroll_loops, 1}, {"unroll-all-loops", &flag_unroll_all_loops, 1}, + {"move-all-movables", &flag_move_all_movables, 1}, + {"reduce-all-givs", &flag_reduce_all_givs, 1}, + {"rerun-loop-opt", &flag_rerun_loop_opt, 1}, {"writable-strings", &flag_writable_strings, 1}, {"peephole", &flag_no_peephole, 0}, *************** rest_of_compilation (decl) *** 2894,2897 **** --- 2911,2916 ---- { loop_optimize (insns, loop_dump_file); + if (flag_rerun_loop_opt) + loop_optimize (insns, loop_dump_file); }); } --------------cut------------- It's from the patch which came with GNU Fortran. Bye, Nils -- Nils Bokermann Johanneswerkstr. 90 Phone: +49 521 891279 33613 Bielefeld FAX: +49 521 550116 Germany "Wir wollen die Natur nicht erhalten -- Wir wollen nur ihre Dynamik nicht st"oren." _______________________________________________________________________________

1 0

Re: gpc/15: Emulation of Borland Pascal 'Assign' procedure does not work in separate unit
by Peter Gerwinski 12 Jan '97

12 Jan '97

Hello, Axel! > Emulation of the Borland Pascal 'Assign' procedure (as suggested in > the gpc.info) does not work if the 'Assign' procedure is put into a > separate unit. The problem is that String types without specified length do not (yet!;-) survive the transport through a GPI file. Workaround: Use Strings with specified length: > (* begin 'borland.pas' - this goes into a separate file *) > unit borland; > > INTERFACE type mystring = string (255); procedure Assign(var t: text; protected Name: mystring); > IMPLEMENTATION procedure Assign(var t: text; protected Name: mystring); > var > b: BindingType; > begin > unbind(t); > b := binding(t); > b.Name := Name; > bind(t, b); > b := binding(t); > end; > > end. > (* end 'borland.pas' *) However, thanks a lot for the bug report. Without such reports, we would probably never be able to make GPC a stable compiler. Greetings, Peter e-mail: peter.gerwinski(a)uni-essen.de home address: D\"usseldorfer Str. 35, 45145 Essen, Germany WWW: http://agnes.dida.physik.uni-essen.de/~peter/

1 0

gpc/15: Emulation of Borland Pascal 'Assign' procedure does not work in separate unit
by axel＠moore3.cchem.berkeley.edu 12 Jan '97

12 Jan '97

>Number: 15 >Category: gpc >Synopsis: Emulation of Borland Pascal 'Assign' procedure does not work in separate unit >Confidential: no >Severity: serious >Priority: medium >Responsible: peter (Peter Gerwinski) >State: open >Class: sw-bug >Submitter-Id: www >Arrival-Date: Sat Jan 11 17:00:00 1997 >Originator: Axel Mellinger >Organization: UC Berkeley >Release: 2.0(2.7.2.1) >Environment: Linux 2.0.20 >Description: Emulation of the Borland Pascal 'Assign' procedure (as suggested in the gpc.info) does not work if the 'Assign' procedure is put into a separate unit. Execution of the program 'test.pas' (see below) results in the error message %Gpc: Could't open bound file. Doing `unbind(F)' and a file named '\264\316\b@\264\316\b@\020' is created instead of 'test.dat'. >How-To-Repeat: (* begin 'test.pas' *) (* compile with 'gpc -o test --automake test.pas' *) program test; uses borland; var f: text; begin assign(f, 'test.dat'); rewrite(f); writeln(f, 'Success.'); close(f); end. (* end 'test.pas' *) (* begin 'borland.pas' - this goes into a separate file *) unit borland; INTERFACE procedure Assign(var t: text; protected Name: string); IMPLEMENTATION procedure Assign(var t: text; protected Name: string); var b: BindingType; begin unbind(t); b := binding(t); b.Name := Name; bind(t, b); b := binding(t); end; end. (* end 'borland.pas' *) >Fix: Above example works fine when the 'Assign' procedure is put into the main program instead of a separate unit. >Audit-Trail: >Unformatted:

1 0

Re: A couple quick questions: Books, Accessing memory and Ports
by JanJaap van der Heijden 03 Jan '97

03 Jan '97

At 05:30 PM 1/1/97 -0500, you wrote: >Hello everyone, >Happy New Year. Greetings and best wishes from me too. I realize this is a very stupid question and that it should >be painfully obvious, but I cannot seem to find the answer. Basically, I >recently started programming in the DJGPP enviornment and have been able to >do alot the of the basic stuff, I am very familiar with Turbo Pascal. >Unfortunentally, I am not familiar with the other languages that the >comipler is based on (ansi, enhanced etc.) and was wondering if there are >any good books/docs etc on the subject. > >Also, I have been trying to find out about how to write, say a byte, >directly to memory or to read a byte directly from memory. > >ie > >var > P : ^void; > >begin > GetMem(P, The_Space_I_Need); > . > . > . >end. > > > Now how do I read the and write to the memory I have allocated. In Turbo >Pascal, i could use the mem[segment:offset] array, but that is not available >in GNU Pascal... > Ah -- you probably mean physical memory access. This is not easy to accomplish, because of the DPMI trickery required to do such things. That's not GPC's fault, it's just as hard to do in djgpp. Give me some time to hack some code together. >I also wish to be able to read and write to ports, of the material I have >found, I know that the port[] array that was in Turbo Pascal is also no >longer available. What can I do in its place (ie writing to SB registers >and to the Palette registers) A port[] array would meaningless outside the PC world, so it's not part of the ANSI standards and GPC. But djgpp's C library provides simular functionality with the inportb() and outportb() functions. These (and more) are defined in <pc.h> Consider this little program which accesses the PC's realtime clock via port $70 and $71. ============================================================================ program RTC; type byte = __byte__ integer; short = __short__ integer; ushort = __unsigned__ short; const RTCAdrPort = $70; RTCDtaPort = $71; Seconds = $00; Minutes = $02; Hours = $04; Day = $07; Month = $08; Year = $09; HundredYear = $32; StatusA = $0A; StatusB = $0B; StatusC = $0C; StatusD = $0D; Diagnose = $0E; { Declared in djgpp's C library, see <pc.h> } function inportb(port: ushort): byte; external; C; procedure outportb(port: ushort; data: ushort); external; C; { Read a value from an RTC register } function RTCRead(address: ushort): byte; begin outportb(RTCAdrPort, address); { pass RTC address } RTCRead := inportb(RTCDtaPort); { read value } end; { Write a value to an RTC register } procedure RTCWrite(address: ushort; content: byte); begin outportb(RTCAdrPort, address); { pass RTC address } outportb(RTCDtaPort, content); { write new value } end; { Read a BCD date/time memory location from RTC and convert it to binary } function RTCDT(address: ushort): byte; var value: byte; begin if (RTCRead(StatusB) and 2 = 0) { BCD or binary mode ? } then RTCDT := RTCRead(address) else begin value := RTCRead(address); RTCDT := ((value shr 4) and $0F) * 10 + (value and $0F); end; end; begin writeln('Information from the battery operated realtime clock'); writeln('----------------------------------------------------'); if RTCRead(Diagnose) and 128 = 0 then begin writeln('The clock is in ', ((RTCRead(StatusB) and 2)*6+12):2, ' hour mode'); writeln('Current time: ', RTCDT(Hours):2, ':', RTCDT(Minutes):2, ':', RTCDT(Seconds):2); writeln('Today is: ', RTCDT(Day):2, '/', RTCDT(Month):2, '/', RTCDT(HundredYear):2, RTCDT(Year):2); end else writeln(' CMOS battery dead!'); end. ============================================================================ In Turbo Pascal, RTCRead() would have been: function RTCRead(address: ushort): byte; begin Port[RTCAdrPort] := address; RTCRead := Port[RTCDtaPort]; end; > >Any help on any of the above problems would be of a huge help! > If you want to do sound programming, you may want to get the Allegro library for djgpp. Then, rewrite Allegro's C header file for GPC and you're done. That way, you don't have to program the hardware directly. JanJaap -- "Nothing shocks me, I'm a scientist", Indiana Jones

1 0

A couple quick questions: Books, Accessing memory and Ports
by Brin 01 Jan '97

01 Jan '97

Hello everyone, Happy New Year. I realize this is a very stupid question and that it should be painfully obvious, but I cannot seem to find the answer. Basically, I recently started programming in the DJGPP enviornment and have been able to do alot the of the basic stuff, I am very familiar with Turbo Pascal. Unfortunentally, I am not familiar with the other languages that the comipler is based on (ansi, enhanced etc.) and was wondering if there are any good books/docs etc on the subject. Also, I have been trying to find out about how to write, say a byte, directly to memory or to read a byte directly from memory. ie var P : ^void; begin GetMem(P, The_Space_I_Need); . . . end. Now how do I read the and write to the memory I have allocated. In Turbo Pascal, i could use the mem[segment:offset] array, but that is not available in GNU Pascal... I also wish to be able to read and write to ports, of the material I have found, I know that the port[] array that was in Turbo Pascal is also no longer available. What can I do in its place (ie writing to SB registers and to the Palette registers) Any help on any of the above problems would be of a huge help! Thanks in advance -Brin ******************************************************************************** Brin's Humor List SUBSCRIBE!: If you're not on then get signed up! >>>>>Send an EMail to: brin(a)star.net<<<<< leaving the body with lots of funny comments and where you heard about us and Subscribe in the Subject. DISTRIBUTE!: Send this to all your friends, enemies, and to someone very special who you havn't seen in a long time COME AND SEE:The archives http://www.star.net/people/~brin CONTRIBUTE!: Send me funny stuff. DISCLAIMER: Please read the disclaimer at: http://www.star.net/people/~Brin/Humor/disclaim.htm ****************************************************************************

1 0

e-mail address
by Vern McCrae 23 Dec '96

23 Dec '96

Vern McCrae I didn't understand your question or why the e-mail address was not provided when I logged onto your site. Here it is anyway. Vern McCrae VMS Technology Inc. Canada vmst(a)golden.net

1 0

GNU Pascal 2.0 is released
by Peter Gerwinski 19 Dec '96

19 Dec '96

Hello, everybody! GNU Pascal 2.0 is released. The source and binaries are both on kampi.hut.fi and on agnes.dida.physik.uni-essen.de. On agnes we will provide experimental versions in `beta' and `alpha' subdirectories in addition to the "official" version, that's the only difference between the stuff on both sites. The announcement will appear in the NewsGroups, soon. Enjoy, Peter e-mail: peter.gerwinski(a)uni-essen.de home address: D\"usseldorfer Str. 35, 45145 Essen, Germany WWW: http://agnes.dida.physik.uni-essen.de/~peter/

2 1