Gpc August 1997

gpc@gnu.de

23 participants
88 discussions

Re: gpc questions
by Peter Gerwinski 25 Aug '97

25 Aug '97

Hello! According to Fausto Iannuzzi: > I have read it's possible to call gcc functions from a gpc program or unit > adding ";C" to functions' declaration. Is it also possible to call or use > gpc functions, or generally gpc code, from a gcc program? Yes: To call the Pascal procedure Procedure foo ( Var Bar: Integer ); from a C program, use the header extern void Foo ( int *bar ); Note that the First Character Of The Pascal Procedure Is Capitalized. Compile the Pascal source as a Unit or Module to avoid a `main()' function to be generated from it. > Does gpc support operator overloading? Yes: Type Foo = <whatever>; Operator + ( x, y: Foo ) = z: Foo; begin (* Foo + Foo *) <calculate z out of x and y>; end (* Foo + 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

Re: Can't run gpc_970714 :-(
by Peter Gerwinski 25 Aug '97

25 Aug '97

According to Ken Linder: > > I am not too sure if I am experiencing brain damage or what... Since I > downloaded the binary Linux GPC file gpc-970714_i586-linux.tar and unzipped > it, I have not been able to compile ANY pascal code.:-( > > OS: Linux 2.0.27 > GCC: 2.7.2.1 That sounds rather strange. As you can imagine, this very same compiler worked fine here. > What I get when I try to compile and pascal source... > /tmp/cca00363.i:1: numeric constant with no digits > /tmp/cca00363.i:1: numeric constant with no digits > /tmp/cca00363.i:1: invalid #line > etc... etc... etc... Looks like a problem with the preprocessor. Can you please retry with gpc -v <other options> foo.pas -o foo ^^ and post the result here? > This wouldn't be because I am using 486 hardware would it? I cannot imagine that. GCC and GPC don't have true Pentium optimization (yet;). AFAIK, the code is essentially optimized for the i486 with a few optimizations switched off that would reduce performance on a Pentium (not sure about that). Later, 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: Pointer & GPC question...
by Frank Heckenbach 25 Aug '97

25 Aug '97

Orlando Llanes wrote: > How would I do low level access for memory with a normal user privilege? > Why can't I just define a pointer and write data to it? Ugh, Linux is a > good OS, but it can be a pain in the butox :P Of course you can do it this way for your (i.e. this process's) memory. Sorry if I didn't make this clear. You need root privileges to access all memory -- that's the point of memory protection in a good OS (because a normal program can't accidentally overwrite/crash other processes or the kernel). The other thing is (as far as I've understood it so far) that Linux uses paging, so the same pointer doesn't necessarily mean the same thing in two different processes (because the memory of these processes is paged differently). Therefore, there's the /dev/mem device to allow you (i.e. root) to access memory at absolute addresses. About the video memory, I don't know very much. I think the user who uses a certain console *should* have direct access to its video memory, but I don't know if this is so. (Either this or accessing the ports is the reason why svgalib programs need root privileges.) If this wasn't quite correct/complete, plase someone correct me.

1 0

Re: Variable conformant arrays, and the price of fish...
by Peter Gerwinski 25 Aug '97

25 Aug '97

Hello! According to Michael Meeks: > > PROCEDURE print_str (name: PACKED ARRAY [lo..hi: integer] OF Char) ; {a} > [...] > It compiles to object fine with the {b} labeled code in, but the > {a} labeled code causes a crash : This is a bug in gpc-2.0 or below which has vanished in recent beta versions. To get a GPC without that bug, download gpc-970714.tar.gz from ftp://agnes.dida.physik.uni-essen.de/gnu-pascal/beta/. > Is this perhaps one section of the ISO 10206 spec. that isn't > implemented yet ? or am I doing something extremely daft :-) It's part of level 1 of ISO-7185 Standard Pascal. > TYPE > ulong = 0..16#FFFFFFFF ; Also, gpc-970714 can compils this. In addition, it has an extension type `Cardinal' or `Word' (two names) which has the same size as `Integer' but is unsigned. As another extension, you can write `Cardinal ( 32 )' to be sure to get an unsigned integer type of 32 bits width. (`Integer' may actually have 64 bits or whatever on some machines.) > Thanks in advance for any help, its a great project, keep up the > good work. Thank you as well for your interest. Yes, we'll try to keep it up. :-) If you want to contribute to the project, too: You seem to have some experience with Extended Pascal. When we will face 100% compliance to Extended Pascal in the future we will need some test programs urgently, and useful and/or entertaining example programs and documentation are very welcome, too. Hope this helps, Peter PS: What has all this to do with the price of fish? 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]

2 1

Re: Pointer & GPC question...
by Orlando Llanes 22 Aug '97

22 Aug '97

How would I do low level access for memory with a normal user privilege? Why can't I just define a pointer and write data to it? Ugh, Linux is a good OS, but it can be a pain in the butox :P > From: Frank Heckenbach <heckenb(a)mi.uni-erlangen.de> > It means a bit more. Root (and only root, AFAIK) can directly access all > ... > because today, most Linux systems are single-user machines, AFAIK... See ya! Orlando Llanes

1 0

Re: Pointer & GPC question...
by Frank Heckenbach 22 Aug '97

22 Aug '97

Orlando Llanes wrote: > Cool :) Do I have to write a driver or something? Or will libs suffice? Or > do I even need to worry about the user not being root? suid refers to the > program? Like Ring 1/2/3 in PMode on Intel for example? > > From: Peter Gerwinski <peter(a)agnes.dida.physik.NO-SPAM-PLEASE.de> > > Well ... you can use the `/dev/mem' device to directly access memory, > > ... > > but then the program needs to be "suid root" to be executed. > > (In fact this is how `svgalib' accesses video memory.) It means a bit more. Root (and only root, AFAIK) can directly access all memory and ports, but root also has unlimited access to everything: all files (i.e. can read/modify/delete any private files of any user), all processes (i.e. can kill any user's processes), all devices (i.e. can format harddisks and all these things). So if I were the administrator of a multi-user system, I wouldn't install any unknown setuid root program -- that's the disadvantage. I guess it's not taken too seriously (and therefore not done much about it) by the Linux developers, because today, most Linux systems are single-user machines, AFAIK... (BTW: suid only means that the program runs under the ID and with the privileges of a certain user (usually, a program has the ID of the user who started it). So, a suid guest program would get all the privileges of the guest account, which probably isn't very much. Almost the only interesting thing is suid root...) -- Frank Heckenbach, Erlangen, Germany heckenb(a)mi.uni-erlangen.de http://home.pages.de/~fjf/links.htm

1 0

Re: Pointer & GPC question...
by Orlando Llanes 22 Aug '97

22 Aug '97

Cool :) Do I have to write a driver or something? Or will libs suffice? Or do I even need to worry about the user not being root? suid refers to the program? Like Ring 1/2/3 in PMode on Intel for example? > From: Peter Gerwinski <peter(a)agnes.dida.physik.NO-SPAM-PLEASE.de> > Well ... you can use the `/dev/mem' device to directly access memory, > ... > but then the program needs to be "suid root" to be executed. > (In fact this is how `svgalib' accesses video memory.) Cool, that's the way I want it :) > That's what the LGPL stands for, *and* it allowes to link the library > with non-free software. The ordinary GPL would stand for the same, too, Noted :) > It's a good example how to use existing GCC libraries from GPC. See ya! Orlando Llanes

1 0

Re: Pointer & GPC question...
by Peter Gerwinski 22 Aug '97

22 Aug '97

According to Orlando Llanes: > According to myself (Peter Gerwinski): > > No way. > Bummer :( Well ... you can use the `/dev/mem' device to directly access memory, (* Linux lo-level file system access (here: memory access) *) Function mmap ( addr: Pointer; len: Word; prot, flags, fd, off: Integer ): Word; C; (* In fact it returns a pointer. But we convert it to Word. *) Function open ( filename: CString; Flags: Integer; ... ): Integer; C; Procedure InitTools; Var Dummy: Word; Function Allocate ( address, length: Word ): Pointer; Var segMem: Pointer; result: Word; begin (* Allocate *) segMem:= GetMem ( length + $FFF ); if Word ( segMem ) and $FFF <> 0 then inc ( Word ( segMem ), $1000 - Word ( segMem ) and $FFF ); if segMem = Nil then FatalError ( 'Linux drivers: cannot allocate video memory' ); result:= mmap ( segMem, length, $03, $11, MemoryHandle, address ); if Integer ( result ) <= 0 then FatalError ( 'Linux drivers: cannot mmap video memory' ); Allocate:= Pointer ( result ); end (* Allocate *); ... MemA0000:= Allocate ( $A0000, $10000 ); but then the program needs to be "suid root" to be executed. (In fact this is how `svgalib' accesses video memory.) > I plan to put it in the license where it's Copyrighted Freeware :) > Hopefully it's what the LGPL stands for. I will release the source code, > but I do want to retain the copyright, I won't charge for it's use tho. ^^(corrected) That's what the LGPL stands for, *and* it allowes to link the library with non-free software. The ordinary GPL would stand for the same, too, but would only allow to link the library with other free software. > I saw it, but nothing about memory access, It's a good example how to use existing GCC libraries from GPC. > besides, I don't see why > Borland's BGI should live on. For compatibility's sake. Nothing else. 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

Can't run gpc_970714 :-(
by Ken Linder 22 Aug '97

22 Aug '97

Hi All I am not too sure if I am experiencing brain damage or what... Since I downloaded the binary Linux GPC file gpc-970714_i586-linux.tar and unzipped it, I have not been able to compile ANY pascal code.:-( OS: Linux 2.0.27 GCC: 2.7.2.1 What I get when I try to compile and pascal source... /tmp/cca00363.i:1: numeric constant with no digits /tmp/cca00363.i:1: numeric constant with no digits /tmp/cca00363.i:1: invalid #line etc... etc... etc... This wouldn't be because I am using 486 hardware would it? Thanks for the assist! Ken L.

1 0

Re: Pointer & GPC question...
by Orlando Llanes 21 Aug '97

21 Aug '97

I plan on using them to fill gaps where OS specific stuff is concerned, File I/O, Network I/O, etc. > From: Frank Heckenbach <heckenb(a)mi.uni-erlangen.de> > Use them with care! This way, you can only support some limited set of known > ... > use "sizeof(pointer)"... I know it doens't, but I don't like Borland's API. > (* BTW: BGI2GRX doesn't use Borland's inefficient BGI drivers and such, it > just provides the same (nearly) interface as Borland's Graph unit. *) See ya! Orlando Llanes

1 0

← Newer
1
2
3
4
5
6
7
8
9
Older →

2024

2023

2022

2021

2020

2019

2018

2017

2016

2015

2014

2013

2012

2011

2010

2009

2008

2007

2006

2005

2004

2003

2002

2001

2000

1999

1998

1997

1996

1995

Gpc August 1997