Gpc

gpc@gnu.de

5400 discussions

Pointer and object checks
by Frank Heckenbach 20 Mar '05

20 Mar '05

I'm considering adding runtime checks for nil pointer access. The benefits are not as big as one might suppose, since most systems catch a nil pointer access anyway and give a segmentation fault automatically. Whereas explicit runtime checks might add considerable runtime overhead. Additionally, some valid (well, let's at least say, working) BP and similar programs might fail then, e.g. if they pass dereferenced pointers (which may be nil) as var parameters, and the called routine either doesn't access the parameter or explicitly checks if it's address is nil. On the positive side, such a check might provide a somewhat nicer error message (proper runtime error instead of segfault), and it can be caught like other runtime errors (e.g. via `AtExit') rather than with a signal handler as for segfaults. Also, strict standard conformance makes it an error, even in a situation as described above where the var parameter is never used. Therefore, perhaps, the check should be off by default, and on only in standard modes, and of course, triggered by its own new option (`--[no-]pointer-checking'?). Another runtime check (which BTW should also be rather easy to implement now), is for objects, whether they point to a valid VMT, using the `NegatedSize' VMT field, just like BP does. AFAIR, BP does it on virtual method calls, where the VMT is used. In principle, the check could be done on every object operation/access, but that might be overkill. Since a simple check per virtual method call is not much overhead, I propose the check to be on by default, in all dialects. Of course, it will also get its own option (`--[no-]object-checking'?). BP couples it to range-checking (I can only guess why -- perhaps because they only have short, i.e. one-letter, compiler directives, and they wanted to save a letter). For strict compatibility, we could add a combined option (`--[no-]range-and-object-checking'), and make `{$R[+-]}' equivalent to that. Frank -- Frank Heckenbach, frank(a)g-n-u.de, http://fjf.gnu.de/, 7977168E GPC To-Do list, latest features, fixed bugs: http://www.gnu-pascal.de/todo.html GPC download signing key: ACB3 79B2 7EB2 B7A7 EFDE D101 CD02 4C9D 0FE0 E5E8

3 13

assert -> gcc_assert
by James A. Morrison 20 Mar '05

20 Mar '05

Hi, GCC 4.0.0 and up use gcc_assert and gcc_unreachable for assertions and unreachable code. This patch updates gpc's assert to be gcc_assert and gcc_unreachable and have the same sematics as the upstream assertion functions. This along with the other patch was built and tested against gcc 3.3 cvs on ia64-linux. -- Thanks, Jim http://phython.blogspot.com

2 2

Re: bug: ncurses + slackware 10.0, 10.1
by Frank Heckenbach 18 Mar '05

18 Mar '05

Thomas Dickey wrote: > On Thu, 17 Mar 2005, Russell Whitaker wrote: > > > Program below illustrates a bug: lower case a, lower case f thru z, > > and number 0 are printed with graphic characters. This bug did not > > appear in earlier versions of slackware. > > I don't have Slackware 10.0 or 10.1; a quick try at testing this with > Debian isn't working out, since the Debian package apparently doesn't > build any of the units - just the compiler. Please use the option `--automake' when calling gpc, then it will build the units together with the program. Also, you might have to remove the `FrameWin' unit (it's not a standard unit, and the call to it is already commented out). > However, if it's an ncurses change (nothing offhand occurs to me, reading > the example), one could use ltrace to get a reasonable idea of the calls > that are being made in ncurses (not 100%, since some of the interface may > modify the WINDOW struct directly, but probably good enough to see this > sort of problem). I don't have Slackware 10.x here either, and the program works correctly for me. Unless you can reproduce a bug with the hints above, I guess I'll have to check it with the respective ncurses version to find out if it's a GPC or CRT bug (which would be my responsibility ...) Russ, which ncurses version is it that you have? Look in /usr/include/ncurses.h for: #define NCURSES_VERSION_MAJOR #define NCURSES_VERSION_MINOR #define NCURSES_VERSION_PATCH Frank -- Frank Heckenbach, frank(a)g-n-u.de, http://fjf.gnu.de/, 7977168E GPC To-Do list, latest features, fixed bugs: http://www.gnu-pascal.de/todo.html GPC download signing key: ACB3 79B2 7EB2 B7A7 EFDE D101 CD02 4C9D 0FE0 E5E8

2 1

ada libplot bindings uploaded
by Rick Engebretson 18 Mar '05

18 Mar '05

I've uploaded the ada-libplot bindings; http://www.pinenet.com/~eng/plotutils/ Not all links will work because other source was not uploaded. Also, my earlier email did not seem to go through. But to repeat; the _p_initialize problem is probably limited to the postscript plotter code and can probably be dealt with.

1 0

Re: bug: ncurses + slackware 10.0, 10.1
by Frank Heckenbach 18 Mar '05

18 Mar '05

Thomas Dickey wrote: > ok (I'll try this when I get home). My ncurses shared libraries are > current, but I do have static libraries for different versions, against > which I may be able to test (though I'm not sure how to tell gpc to do > that). GPC basically accepts all GCC options, so you can pass library file name on the command line. In addition you might have to edit the CRT unit (`gpc --print-file-name=units/crt.pas' to get its file name) and change this line (~790): {$L crtc.c, panel, ncurses} to just: {$L crtc.c} (The deleted part yields the effects of `-lpanel -lncurses'. Since explicitly names files take precedence on the linker command line AFAIK, this might not be necessary, but I'm not sure.) Frank -- Frank Heckenbach, frank(a)g-n-u.de, http://fjf.gnu.de/, 7977168E GPC To-Do list, latest features, fixed bugs: http://www.gnu-pascal.de/todo.html GPC download signing key: ACB3 79B2 7EB2 B7A7 EFDE D101 CD02 4C9D 0FE0 E5E8

1 0

Re: NCurses
by Prof A Olowofoyeku (The African Chief) 16 Mar '05

16 Mar '05

Hi My GPC port of the FPC ncurses unit: http://gnu-pascal.de/contrib/chief/ncurses.zip Note that there were linker problems with the "acs_map" variable. You can avoid the linker problems by treating acs_map as an internal (not external) variable. But this probably means missing some functionality. You might have to rebuild your ncurses library, configuring with "--enable-broken_linker" to solve it. I don't know about these things. Best regards, The Chief -------- Prof. Abimbola A. Olowofoyeku (The African Chief) web: http://www.greatchief.plus.com/

3 3

How do I create records that point to a record of the same type?
by loop lopy 16 Mar '05

16 Mar '05

I have changed from gpc into turbo pascal. The code that compiled for me in gpc doesnt compile now in turbo pascal. The code: type NumNode = record Value: integer; pNext: ^ NumNode; end; The compile error I get: Undefined type in pointer definition (NUMNODE) __________________________________ Do you Yahoo!? Yahoo! Small Business - Try our new resources site! http://smallbusiness.yahoo.com/resources/

5 4

Re: PowerPC and ncurses
by Frank Heckenbach 16 Mar '05

16 Mar '05

Forwarded from marcov(a)stack.nl: > In answer to Rick's remark: > > fpimage has nothing to do with GNU Plotutils, and to my knowledge there is > no libplot binding in FPC. > > fpimage is totally standalone pascal code, delphi dialect. So not much > chance of running it on GPC short term. (while it now uses paszlib and > pasjpeg, it is switchable to use the C versions of those libraries when > building a FPC release) [Rather than porting this, it might be easier to translate the Plotutils interface ... -- Frank] Frank -- Frank Heckenbach, frank(a)g-n-u.de http://fjf.gnu.de/ GnuPG and PGP keys: http://fjf.gnu.de/plan (7977168E)

2 2

problems with assignment of whole records on PowerPC
by Igor Marnat 15 Mar '05

15 Mar '05

Hello, List! I've successfully built cross-compiler for PowerPC GPC configuration attached below. Anyway, I cannot execute simple program built with my new compiler. Program looks like: program test; type MyRecord = record x: word; y: word; end; var t1, t2: MyRecord; begin writeln ('SizeOf (word)', Sizeof (word)); writeln ('SizeOf (longword)', Sizeof (longword)); writeln ('SizeOf (CInteger)', Sizeof (CInteger)); t1.x := 1; t1.y := 1; writeln ('before assignment'); t2 := t1; writeln ('after assignment'); end. Program output: bash-2.05b# ./test SizeOf (word)4 SizeOf (longword)8 SizeOf (CInteger)4 before assignment Illegal instruction When I change type of fields fo MyRecord to longword, it works fine. Is it anything wrong with alignment or packed records? But my record isn't packed ... I could avoid usage of whole records assignments and assign the values field by field but there are two reason why I don't want to do that: 1. it shows that something is wrong, isn't it? 2. assignment of the whole records is used in crt.pas unit, for example, in function ScreenSize of crt.pas unit which return the value of TPoint type which is record (this assignment doesn't work as well). I really don't want change anything in crt.pas unit:). Please, any ideas or hints ... Thank you, regards, Igor Marnat GPC configuration: (Reading specs from /home/igor/denx_tools/usr/lib/gcc-lib/ppc-linux/3.3.3/specs Configured with: ../configure --prefix=/home/igor/denx_tools/usr --mandir=/usr/share/man --infodir=/usr/share/info --enable-shared --enable-threads=posix --disable-checking --with-system-zlib --enable-__cxa_atexit --with-newlib --enable-languages=c,c++,pascal --disable-libgcj --host=i386-redhat-linux --target=ppc-linux Thread model: posix gpc version 20050217, based on gcc-3.3.3 (DENX ELDK custom version 3.3.3-8) ). ~

2 2

Fwd: PowerPC and ncurses.
by Frank Heckenbach 15 Mar '05

15 Mar '05

Forwarded from marcov(a)stack.nl: > Some small comments > > Powerpc: The 603 has floating point alignment requirements iirc. If he has > a 603, check what the default ppc processor is, and if it is a ppc603 (quite > old, think 100-200 MHz range) try to specify codegeneration for a ppc603 > > ncurses: > > The addition of library to external symbols is standard delphi syntax. I > think it has to do with the way Windows loads DLLs (OS X also seems to > require it sometimes, Mach O can have multiple namespaces IIRC). > > A nice sideeffect could be that it allows the linker to better pinpoint the cause > of a missing symbol (not symbol x is missing, but symbol x is missing from > lib y), but LD doesn't do that yet. Frank -- Frank Heckenbach, frank(a)g-n-u.de http://fjf.gnu.de/ GnuPG and PGP keys: http://fjf.gnu.de/plan (7977168E)

2 1

Jump to page:

2025

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