Re: [Warzone-dev] [Warzone2100-commits] SF.net SVN: warzone2100:[7968] trunk/lib/exceptionhandler/exceptionhandler. c
On Friday, 14 August 2009 at 11:52, the_cybersph...@users.sourceforge.net wrote: Revision: 7968 http://warzone2100.svn.sourceforge.net/warzone2100/?rev=7968view=rev Author: the_cybersphinx Date: 2009-08-14 11:52:46 + (Fri, 14 Aug 2009) Log Message: --- Fix crashdump creation. Was broken in r7850 (#716). I reverted r7850 and then fixed the -Wdeclaration-after-statement warnings instead of poking around trying to find the problem. The question is, why do we need to be C90 compliant in unix-only code at all? r7850 removed quite a few consts that might help prevent problems. I'd prefer both this and r7850 to be reverted completely, but at least now it works. Stiv/Zarel, did you actually test this before committing? ___ Warzone-dev mailing list Warzone-dev@gna.org https://mail.gna.org/listinfo/warzone-dev
Re: [Warzone-dev] [Warzone2100-commits] SF.net SVN: warzone2100:[7968] trunk/lib/exceptionhandler/exceptionhandler. c
Dennis Schridde wrote: No clue, but I support that question. To my knowledge you cannot compile the unix crashdump code with bad compilers anyway. Bad compilers? signature.asc Description: OpenPGP digital signature ___ Warzone-dev mailing list Warzone-dev@gna.org https://mail.gna.org/listinfo/warzone-dev
Re: [Warzone-dev] [Warzone2100-commits] SF.net SVN: warzone2100:[7968] trunk/lib/exceptionhandler/exceptionhandler. c
On Fri, Aug 14, 2009 at 8:02 AM, Christian Ohmchr@gmx.net wrote: The question is, why do we need to be C90 compliant in unix-only code at all? r7850 removed quite a few consts that might help prevent problems. I'd prefer both this and r7850 to be reverted completely, but at least now it works. Stiv/Zarel, did you actually test this before committing? Good question. Stiv came online one day and started talking about how he couldn't compile stuff for some reason. I wasn't sure on the specifics, but I committed his patch under the impression that all it did was move a few variable declarations to the beginnings of blocks... If it doesn't, you'll have to ask Stiv for details. -Zarel ___ Warzone-dev mailing list Warzone-dev@gna.org https://mail.gna.org/listinfo/warzone-dev
Re: [Warzone-dev] [Warzone2100-commits] SF.net SVN: warzone2100:[7968] trunk/lib/exceptionhandler/exceptionhandler. c
On Fri, Aug 14, 2009 at 05:03:18PM +0200, Dennis Schridde wrote: Am Freitag, 14. August 2009 15:02:00 schrieb Christian Ohm: On Friday, 14 August 2009 at 11:52, the_cybersph...@users.sourceforge.net wrote: Revision: 7968 Was broken in r7850 (#716). I reverted r7850 and then fixed the -Wdeclaration-after-statement warnings instead of poking around trying to find the problem. The question is, why do we need to be C90 compliant in unix-only code at all? No clue, but I support that question. To my knowledge you cannot compile the unix crashdump code with bad compilers The problem, as stated in #716 was Running default ./configure gives a bunch of mixed declaration code errors in exceptionhandler.c. gcc 4.3.2 openSUSE linux 11.2 Gcc 4.3.2 hardly fits the 'bad compiler' label, IMHO. (feel free to add your own snarky remarks) It *is* more strict than previous versions. In response to Christian's did you test it?, the answer is apparently, not enough. Zarel is excused because he is a Windozer. Glancing at the code again with fresh eyes and knowlege that something is broken, it looks like the crashdump filename creation got b0rked - a couple of lines and easy to fix. As for The question is, why do we need to be C90 compliant in unix-only code at all? Good question. You tell me. In any case, you should be able to compile WZ on a mainstream linux platform with the default settings. -- Stephen Swaney sswa...@centurytel.net ___ Warzone-dev mailing list Warzone-dev@gna.org https://mail.gna.org/listinfo/warzone-dev
Re: [Warzone-dev] [Warzone2100-commits] SF.net SVN: warzone2100:[7968] trunk/lib/exceptionhandler/exceptionhandler. c
On Friday, 14 August 2009 at 15:12, Stephen Swaney wrote: The problem, as stated in #716 was Running default ./configure gives a bunch of mixed declaration code errors in exceptionhandler.c. gcc 4.3.2 openSUSE linux 11.2 Gcc 4.3.2 hardly fits the 'bad compiler' label, IMHO. (feel free to add your own snarky remarks) It *is* more strict than previous versions. But why does it insist on C90? My gcc 4.3.4 on Debian doesn't do that. Can you post the full gcc command line and error messages? The question is, why do we need to be C90 compliant in unix-only code at all? Good question. You tell me. From my point of view, we don't. Your gcc begs to differ. In any case, you should be able to compile WZ on a mainstream linux platform with the default settings. As long as the default settings aren't stupid, I agree. ___ Warzone-dev mailing list Warzone-dev@gna.org https://mail.gna.org/listinfo/warzone-dev
Re: [Warzone-dev] [Warzone2100-commits] SF.net SVN: warzone2100:[7968] trunk/lib/exceptionhandler/exceptionhandler. c
As long as the default settings aren't stupid, I agree. Ehm, as long as the system isn't being stupid. ___ Warzone-dev mailing list Warzone-dev@gna.org https://mail.gna.org/listinfo/warzone-dev
Re: [Warzone-dev] [Warzone2100-commits] SF.net SVN: warzone2100:[7968] trunk/lib/exceptionhandler/exceptionhandler. c
On Friday, 14 August 2009 at 15:46, Stephen Swaney wrote: gcc -std=gnu99 ... -O0 -g -Wall -Werror -Wno-unused-label -Wno-pointer-to-int-cast -Wmissing-field-initializers -Wcast-align -Wwrite-strings -Wmissing-declarations -Wstrict-prototypes -Wpointer-arith -Wno-format-security -Wdeclaration-after-statement ... The -Wdeclaration-after-statement and -Werror seem to be the significant parts for this discussion. I suspect the They are. The question is, where does -Wdeclaration-after-statement come from? It's not in the repo, says git grep. -Wdeclaration-after-statement is added everywhere because, unlike gcc, the microsoft compilers cannot handle it and no one has the autoconf skills to disable the warning for unix-specific code. ___ Warzone-dev mailing list Warzone-dev@gna.org https://mail.gna.org/listinfo/warzone-dev
Re: [Warzone-dev] [Warzone2100-commits] SF.net SVN: warzone2100:[7968] trunk/lib/exceptionhandler/exceptionhandler. c
On Fri, Aug 14, 2009 at 10:05:41PM +0200, Christian Ohm wrote: On Friday, 14 August 2009 at 15:46, Stephen Swaney wrote: gcc -std=gnu99 ... -O0 -g -Wall -Werror -Wno-unused-label -Wno-pointer-to-int-cast -Wmissing-field-initializers -Wcast-align -Wwrite-strings -Wmissing-declarations -Wstrict-prototypes -Wpointer-arith -Wno-format-security -Wdeclaration-after-statement ... The -Wdeclaration-after-statement and -Werror seem to be the significant parts for this discussion. I suspect the They are. The question is, where does -Wdeclaration-after-statement come from? It's not in the repo, says git grep. Don't know where it comes from. My first thought was it maybe it gets turned on by -Wall, but browsing the gcc docs says no. I had been compiling trunk with --enable-debug=relaxed, but someone on IRC with another linux distro complained about being unable to build with the default ./configure, something that should be possible, so I looked into the problem and here we are! Personally, I like being able to sprinkle declarations around near where they are needed ala C++/C99. If we drop support for MSVC this all goes away! -- Stephen Swaney sswa...@centurytel.net ___ Warzone-dev mailing list Warzone-dev@gna.org https://mail.gna.org/listinfo/warzone-dev