September 1997 - Gpc

Re: Looking for some random-functions ... and maybe a bug report
by Kevin A. Foss 23 Sep '97

23 Sep '97

On Tue, 23 Sep 1997 15:21:11 +0200, Frank Heckenbach wrote: >But still I don't know how to get the system time in OS/2 or non-DPMI DOS >(I was using the DPMI call only to call a DOS int, since I'm not sure if >calling int 21 directly is OK) If all you need is system time, why do you go through all of the DPMI/int calls/&c, instead of just using extended Pascal's timestamp routines. Or are these not available because you are turning non-Borland features off when compiling these Borland-ish Units? Using gettimestamp() should work on all gpc platforms, right? I guess I must misunderstand what exactly you need for a 'system time.' -Kevin -- Kevin A. Foss --- kfoss(a)mint.net --

1 0

Re: Looking for some random-functions ... and maybe a bug report
by Peter Gerwinski 23 Sep '97

23 Sep '97

According to Kevin A. Foss: > > If all you need is system time, why do you go through all of the > DPMI/int calls/&c, instead of just using extended Pascal's timestamp > routines. These only have a resolution of one second which probably is not "random" enough for this purpose. The 1/100s when a program was started is almost a true random number; if we use the second in the minute it might be too regular. > Or are these not available because you are turning > non-Borland features off when compiling these Borland-ish Units? No. > Using > gettimestamp() should work on all gpc platforms, right? Right. If you find a platform where it doesn't, it's a bug. Greetings, Peter Dipl.-Phys. Peter Gerwinski, Essen, Germany, free physicist and programmer peter.gerwinski(a)uni-essen.de - http://home.pages.de/~peter.gerwinski/ [970201] maintainer GNU Pascal [970714] - http://home.pages.de/~gnu-pascal/ [970125]

1 0

EMX (was: Looking for some random-functions ... and maybe a bug report)
by Peter Gerwinski 23 Sep '97

23 Sep '97

According to Frank Heckenbach: > > According to myself (Peter Gerwinski): > > > You can get this information at run-time - and *only* at run-time since > > one and the same executable can run both on DOS (with and without DPMI) > > and on OS/2. If the variable `__osmode' contains a nonzero value, it > > is OS/2; if it is DOS and the variable `__emx_env' and $0180 is nonzero, > > it is DPMI (RSX). > > But still I don't know how to get the system time in OS/2 or non-DPMI DOS > [...] This was implied: If `__osmode' is zero and `__emx_env' and $0180 is zero as well, then it is DOS without DPMI. Once again the three EMX modes as a tabular: __osmode __emx_env and $0180 system nonzero doesn't matter OS/2 0 nonzero DOS with DPMI 0 0 DOS without DPMI (-; Peter Dipl.-Phys. Peter Gerwinski, Essen, Germany, free physicist and programmer peter.gerwinski(a)uni-essen.de - http://home.pages.de/~peter.gerwinski/ [970201] maintainer GNU Pascal [970714] - http://home.pages.de/~gnu-pascal/ [970125]

1 0

Re: Looking for some random-functions ... and maybe a bug report
by Frank Heckenbach 23 Sep '97

23 Sep '97

Peter Gerwinski wrote: > According to Frank Heckenbach: > > > > Kevin A. Foss wrote: > > > > > Unfortunately, I know of no easy way to test for what operating system > > > EMX is currently using. > > > > Neither do I, that's the problem... > > You can get this information at run-time - and *only* at run-time since > one and the same executable can run both on DOS (with and without DPMI) > and on OS/2. If the variable `__osmode' contains a nonzero value, it > is OS/2; if it is DOS and the variable `__emx_env' and $0180 is nonzero, > it is DPMI (RSX). Thanks for the info! But still I don't know how to get the system time in OS/2 or non-DPMI DOS (I was using the DPMI call only to call a DOS int, since I'm not sure if calling int 21 directly is OK) -- or some other way to init random; AFAICS, the system unit v1.00 doesn't do any random initialization at all... Besides, I couldn't test it with EMX, anyway. So I won't change it now. If someone who has EMX can change it, please do so and post the result here. -- Frank Heckenbach, Erlangen, Germany heckenb(a)mi.uni-erlangen.de http://home.pages.de/~fjf/links.htm

1 0

Re: Looking for some random-functions ... and maybe a bug report
by Frank Heckenbach 23 Sep '97

23 Sep '97

Christian Bockermann wrote: > Hello ! > > After I inserted the random-functions, which I got through this list, I > wanted to use the Rand-Unit by Frank Heckenbach. But I get some error- > messages when compiling this unit : > > [...] > > What exactly does this 'Cardinal'-expression mean ? Cardinal is an unsigned integer. It was introduced after 2.0, sometime around June, IIRC. With 2.0, you might be able to use "__unsigned__" or such a beast, but OTOH the random unit relies on the exact sizes (16, 32 and 64 bits) of some types, and AFAIR there was no way to specify those in 2.0 (you could use long, short and such, but that woudln't be portable). Su I suggest you to upgrade to a newer version (maybe wait for the next beta, or even for 2.1 which hopefully won't take too long now to complete). Sorry that you can't use it now, but these are temporary problems... Frank -- Frank Heckenbach, Erlangen, Germany heckenb(a)mi.uni-erlangen.de http://home.pages.de/~fjf/links.htm

1 0

Re: Looking for some random-functions ... and maybe a bug report
by Frank Heckenbach 23 Sep '97

23 Sep '97

Kevin A. Foss wrote: > >{$IFDEF __DJGPP__}{$DEFINE DOS}{$ENDIF} > >{$IFDEF __EMX__}{$DEFINE DOS}{$ENDIF} > > IMHO, this isn't a good way to do things in a portable unit. Remember > that the EMX version of GCC also works under OS/2. Trying to DOS-ish > things won't go over well if you only check for __EMX__. I assumed such interrupts would be eumlated, but if not... > Unfortunately, I know of no easy way to test for what operating system > EMX is currently using. Neither do I, that's the problem... > >{$W-} > >FUNCTION __DPMI_Simulate_Real_Mode_Interrupt(Vector:Integer;VAR Regs:TDPMIRegs):Integer; C; > >{$W+} > > Does this even work with the DOS version of EMX? Don't know. I've only got Linux and DJGPP. > It looks like some > type of DPMI call but EMX doesn't DPMI without the RSX extender. Oh! I didn't know that. Well anyway, it's not a serious thing at all. This last section of the units only deals with how the random number seed is initialized. If it doesn't work on any system, one can simply do it another way. (My way of doing it exactly the same way as BP under Dos was just a kind of fun, anyway; I don't think anyone would ever care.) If OS/2 has something like GetPID, one can use this or other "pseudo-random" data (such as the system time, received by other means than a Dos interrupt, but I don't know how). > [C:\kevin\pascal]gpc -c rand.pas > gpc: Internal compiler error: program gpc1 got fatal signal 11 > (I also get an OS/2 popup with a generic 3171 error -- this "program... > can not continue.") > > I take it the compiler should reply with some error and not a fatal > signal? Or this one of the things that has been fixed in the later > betas? Probably. I got such errors a lot with betas from Jul/Aug when compiling units. The latest developer version (to be published soon) doesn't seem to cause such problems any more. Oh yes, and there was a bug that caused a serious crash of the DJGPP GPC with the type declarations in the random unit. Perhaps that was in the version you're using, I don't remember exactly. Again, this is fixed in the current GPC. -- Frank Heckenbach, Erlangen, Germany heckenb(a)mi.uni-erlangen.de http://home.pages.de/~fjf/links.htm

2 1

GPC on Solaris
by croy＠osiris.com 23 Sep '97

23 Sep '97

Hello, Boy it has been a long time for me waiting to have a Pascal Compiler on Unix.... C is OK but Pascal just feels better, Thanks to all the people involved Ok That is as much brown nosing I will ever do on this list, promissed... I have build Successfully but with some difficulties GPC-970714 on both Solaris x86 2.5.1 and Solaris Sparc 2.5.0 ... Should I give you access to the build to put it in the binaries distributions or should I wait for a more official 2.1 release.... Again Thanks and if I can be of help I have here quite a few different version of Unix in the Lab.... Good Test bed for the protability... Ciao ______________________________________________________________________________ Claude Roy Osiris Inc. The Network Architects Email : croy(a)osiris.com Http://www.osiris.com/

1 0

Re: Upgrading GPC (was: Looking for some random-functions ...)
by Peter Gerwinski 23 Sep '97

23 Sep '97

According to Orlando Llanes: > I downloaded the same gpc-970714.zip file from the gnu-pascal FTP site, > I even downloaded the gpc-970714.tar.gz, and I look all over through them, > but 0 executables turned up. I thoroughly searched the entire gnu-pascal > directory looking through every directory and subdirectory in it, nothing. > This is getting really frustrating... ????? What about ftp://agnes.dida.physik.uni-essen.de/gnu-pascal/beta/binary/ gpc-970714.i386-djgppv201.zip? Peter Dipl.-Phys. Peter Gerwinski, Essen, Germany, free physicist and programmer peter.gerwinski(a)uni-essen.de - http://home.pages.de/~peter.gerwinski/ [970201] maintainer GNU Pascal [970714] - http://home.pages.de/~gnu-pascal/ [970125]

1 0

Re: Upgrading GPC (was: Looking for some random-functions ...)
by Orlando Llanes 22 Sep '97

22 Sep '97

>What's that - 2.01b? The last official beta version that came with >binaries was gpc-970714; after that there were some (inofficial) I downloaded the same gpc-970714.zip file from the gnu-pascal FTP site, I even downloaded the gpc-970714.tar.gz, and I look all over through them, but 0 executables turned up. I thoroughly searched the entire gnu-pascal directory looking through every directory and subdirectory in it, nothing. This is getting really frustrating... See ya! Orlando Llanes

1 0

Re: GPC assembler
by Peter Gerwinski 22 Sep '97

22 Sep '97

According to cagliostro: > > I want to set the result of the function, directly from the asm > part.Is it possible ? Yes: You can define a return value variable for the function: Function foo ( x, y: Integer ) = bar: Integer begin (* foo *) asm ( ' <do something> ' : '=rm' ( bar ) : (* input *) : (* clobbered *) ); end (* foo *); Even if you don't define one, you can use `Result' for this purpose: Function foo ( x, y: Integer ): Integer begin (* foo *) asm ( ' <do something> ' : '=rm' ( Result ) : (* input *) : (* clobbered *) ); end (* foo *); Hope this helps, Peter Dipl.-Phys. Peter Gerwinski, Essen, Germany, free physicist and programmer peter.gerwinski(a)uni-essen.de - http://home.pages.de/~peter.gerwinski/ [970201] maintainer GNU Pascal [970714] - http://home.pages.de/~gnu-pascal/ [970125]

1 0