The Ur-Quan Masters Discussion Forum

I'm curreently planning to build uqm under qnx 6.2.1...

what problems would i face?

currently, stuck with libogg/libvorbis/libvorbisfile...

but, going to resolve them...

I've got no experience with QNX, but if it's pretty much POSIX-compliant, the I'd expect few problems. SDL lists QNX as one of the supported platforms (though I doubt it's as well maintained as some of the others); if that works ok, I suspect so will UQM.
Please let me know how it works out.

Yay! I'm really cool.
After 2 hours of lib(SDL*,ogg,vorbis*) fucking, after 3 hours of
unresolved symbols, brain dead build scripts, absent -l and -D, less'ing and grep'ing code, after 50'th attempt of "./build.sh uqm", i have finally got the binary! And it worked FLAWLESS!

i have tried some melee only...
but i will try to test it so in 0.3 alpha there will be full QNX support, i hope...

now, developers, are u here? there will be some minor kicks in the butt for u...

1) build scripts are... they simply does not work

i had to put -lvorbisfile -DSDL_<someshit> and so to build.vars manually. i had to change some @shit@ to #define shit in config.h using my hands.
i will post what problems did i found...

also, they forget to rebuild objects after build.vars changed, which is bad thing, i suggest...

2) i might be wrong
"Damn it, Jim! I'm a doctor, not a programmer!" (c) linux kernel panic.
but somewhere in code there is:
==
static shit wait, signal;
==
which is redefinition; fixed after s/static/extern/
am i wrong or?

oh, i was wrong ;( it's not so stable: got core dump ;(
_GetMusicData(): loading ipanims/orbit5.mod
   decoder: MikMod, rate 44100 format 170202
       'urquan.con' -- 155 bytes
       'lbm/player.fon' -- 1 bytes
       'comm/urquan/urquan.ani' -- 1207 bytes
       'comm/urquan/urquan.fon' -- 1 bytes
       'comm/urquan/urquan.txt' -- 14565 bytes
Found timestamp file: comm/urquan/urquan.ts
       'comm/urquan/urquan.mod' -- 75885 bytes
_GetMusicData(): loading comm/urquan/urquan.mod
   decoder: MikMod, rate 44100 format 170202
SpliceTrack(): loading comm/urquan/urqua027.ogg
       'comandr.con' -- 223 bytes
       'lbm/player.fon' -- 1 bytes
       'comm/comandr/comandr.ani' -- 2731 bytes
       'comm/comandr/comandr.fon' -- 1 bytes
       'comm/comandr/comandr.ct' -- 292 bytes
       'comm/comandr/comandr.mod' -- 67076 bytes
_GetMusicData(): loading comm/comandr/comandr.mod
   decoder: MikMod, rate 44100 format 170202
       'comm/comandr/comandr.txt' -- 15788 bytes
Found timestamp file: comm/comandr/comandr.ts
SpliceTrack(): loading comm/comandr/coman003.ogg
Memory fault (core dumped)

When arriving at starbase... but my gdb and ddd are with me... I'll find clue.
Also, one time i had problems with sound...
(http://www.webcenter.ru/~ilyak/SNAP2.JPG)

The build scripts will be made more intelligent somewhere in the future.
But I'd like to know exactly what you had to change to make this work on QNX.


This stuff is already gone in the cvs version. I suggest you use that one for your own compiles. It might also solve the crash you had. 0.2 had some stability problems.

I might be very stupid, but how can i get everything but 'content'
(or but *.ogg) form cvs? Almost no experience with it...

The standard cvs client can't do it, it seems.

Using command line CVS, you can download one directory at a time either with or without subdirectories. Thus, to get UQM without the content files, use:


Updating this partial checkout is left as an exercise for the student. ;D

It might make a certain amount sense to move the content into a seperate repository, and people can checkout/update the two halves seperately (and for those who like it the way it works now, you can checkout the content volume inside the source volume, and updates will look like they're working identically to the way they work now)

But if the new resource system is almost here, the bulk of the content files are going to get attic'ed anyways, which should drastically reduce update times (and possibly checkout times) since there will be thousands of fewer files to check, so this may all be very moot quite soon.

The thoughts have crossed our minds.

thx, got it...
so, there it goes:
let's continue our... er...
first, let's change:
config_progname:
==
case "$OSNAME" in
     FreeBSD|OpenBSD)
           LIB_vorbisfile_CFLAGS="-I/usr/local/include"
           LIB_vorbisfile_LDFLAGS="-L/usr/local/lib -lvorbisfile -lvorbis"
           ;;
     *)
           LIB_vorbisfile_CFLAGS=""
           LIB_vorbisfile_LDFLAGS="-lvorbisfile -lvorbis"
           ;;
esac
==
or:
==
/x86/usr/lib/libvorbis.so: undefined reference to `sqrt'
/x86/usr/lib/libvorbis.so: undefined reference to `ogg_stream_pagein'
/x86/usr/lib/libvorbis.so: undefined reference to `oggpack_writealign'
/x86/usr/lib/libvorbis.so: undefined reference to `_Sin'
/x86/usr/lib/libvorbis.so: undefined reference to `ogg_sync_wrote'
/x86/usr/lib/libvorbis.so: undefined reference to `ogg_sync_reset'
/x86/usr/lib/libvorbis.so: undefined reference to `ogg_page_granulepos'
/x86/usr/lib/libvorbis.so: undefined reference to `ceil'
/x86/usr/lib/libvorbis.so: undefined reference to `floor'
/x86/usr/lib/libvorbis.so: undefined reference to `ogg_sync_pageseek'
/x86/usr/lib/libvorbis.so: undefined reference to `ogg_page_continued'
/x86/usr/lib/libvorbis.so: undefined reference to `ogg_stream_reset_serialno'
/x86/usr/lib/libvorbis.so: undefined reference to `oggpack_writeinit'
/x86/usr/lib/libvorbis.so: undefined reference to `ldexp'
/x86/usr/lib/libvorbis.so: undefined reference to `ogg_stream_clear'
/x86/usr/lib/libvorbis.so: undefined reference to `oggpack_look'
/x86/usr/lib/libvorbis.so: undefined reference to `rint'
/x86/usr/lib/libvorbis.so: undefined reference to `ogg_stream_packetout'
/x86/usr/lib/libvorbis.so: undefined reference to `ogg_page_eos'
/x86/usr/lib/libvorbis.so: undefined reference to `ogg_sync_init'
/x86/usr/lib/libvorbis.so: undefined reference to `oggpack_bytes'
/x86/usr/lib/libvorbis.so: undefined reference to `oggpack_writeclear'
/x86/usr/lib/libvorbis.so: undefined reference to `pow'
/x86/usr/lib/libvorbis.so: undefined reference to `oggpack_adv'
/x86/usr/lib/libvorbis.so: undefined reference to `_Log'
/x86/usr/lib/libvorbis.so: undefined reference to `ogg_stream_packetpeek'
/x86/usr/lib/libvorbis.so: undefined reference to `exp'
/x86/usr/lib/libvorbis.so: undefined reference to `ogg_stream_init'
/x86/usr/lib/libvorbis.so: undefined reference to `atan'
/x86/usr/lib/libvorbis.so: undefined reference to `oggpack_get_buffer'
/x86/usr/lib/libvorbis.so: undefined reference to `ogg_sync_clear'
/x86/usr/lib/libvorbis.so: undefined reference to `oggpack_readinit'
/x86/usr/lib/libvorbis.so: undefined reference to `ogg_page_serialno'
/x86/usr/lib/libvorbis.so: undefined reference to `oggpack_read'
/x86/usr/lib/libvorbis.so: undefined reference to `ogg_sync_buffer'
/x86/usr/lib/libvorbis.so: undefined reference to `acos'
/x86/usr/lib/libvorbis.so: undefined reference to `oggpack_reset'
/x86/usr/lib/libvorbis.so: undefined reference to `frexp'
/x86/usr/lib/libvorbis.so: undefined reference to `oggpack_write'
collect2: ld returned 1 exit status
==
so:
+
     QNX)
           LIB_vorbisfile_CFLAGS="-I/usr/local/include"
           LIB_vorbisfile_LDFLAGS="-L/usr/local/lib -lvorbisfile -lvorbis -logg -lm"
           ;;
+
or maybe, something else?

we get:
build.vars:
==
COMPILE=gcc
MKDEPEND="gcc -MM"
MAKE="make"
uqm_CFLAGS=" -I/opt/include/SDL -I/opt/include/SDL -g -O0 -W -Wall -DDEBUG_TRACK_SEM -DGFXMODULE_SDL -DSOUNDMODULE_SDL -I/usr/local/include"
uqm_LDFLAGS=" -L/opt/lib -lSDL -L/opt/lib -lSDL -lSDL_image -O0 -L/usr/local/lib -lvorbisfile -lvorbis -logg -lm"
DEBUG="1"
uqm_INSTALL_BINDIR="/usr/local/games/bin/"
uqm_INSTALL_LIBDIR="/usr/local/games/lib/"

# Exported variables are available from all the Makeinfo files
# Non-exported files only where build.vars is explicitely included.
uqm_SOUNDMODULE="mixsdl"
uqm_HAVE_OPENGL="0"
WINDRES=""
export uqm_SOUNDMODULE uqm_HAVE_OPENGL WINDRES
==
became A LOT better compared to 0.2 (when it "forgot" to put in -L/usr/local/lib -lvorbisfile -lvorbis -logg -lm and -DGFXMODULE_SDL -DSOUNDMODULE_SDL

so, we do config?
==
localhost% ./build.sh uqm config
Sed stream editor found.
tr found.
Make found.
GNU C compiler found.
Little-endian machine.
Simple DirectMedia Layer version 1.2.5 found.
SDL_image found.
Vorbisfile found.
Symbol 'strupr' found.
Symbol 'stricmp' found.
Header 'getopt.h' not found.
/usr/bin/ntox86-ld: cannot find -lGL
collect2: ld returned 1 exit status
OpenGL not found.

 -= Main menu =-
 1. Include debugging information        Debugging information
 2. OpenGL graphics support              Don't include OpenGL graphics support
 3. Sound backend                        Use MixSDL for sound
 4. Installation paths                   [...]

Press a number plus <ENTER> if you want to change something,
or just <ENTER> if everything is ok: 1

 -= Include debugging information =-

invalid format string

invalid format string

invalid format string

Select the option you want by typing a number plus <ENTER>
or just <ENTER> if everything is ok:


 -= Main menu =-
 1. Include debugging information        Debugging information
 2. OpenGL graphics support              Don't include OpenGL graphics support
 3. Sound backend                        Use MixSDL for sound
 4. Installation paths                   [...]

Press a number plus <ENTER> if you want to change something,
or just <ENTER> if everything is ok:

Saving choices...
Configuration complete.
==
emm? What strings and what format?

ok; trying to compile?
==
gcc -o obj/debug/./src/options.o -c  -I/opt/include/SDL -I/opt/include/SDL -g -O0 -W -Wall -DDEBUG_TRACK_SEM -DGFXMODULE_SDL -DSOUNDMODULE_SDL -I/usr/local/include -I src -I src/sc2code -I src/sc2code/libs src/options.c
In file included from src/sc2code/starcon.h:44,
                from src/sc2code/libs/graphics/gfx_common.h:25,
                from src/options.c:25:
src/config.h:35: parse error before `@'
make: *** [obj/debug/./src/options.o] Error 1
==
why?
config.h:
==
/* Defined if your system has strupr of its own */
@HAVE_STRUPR@

/* Defined if your system has stricmp of its own */
@HAVE_STRICMP@

/* Defined if your system has getopt.h */
@HAVE_GETOPT_H@

#endif  /* _CONFIG_H */
==
even after ./build.sh uqm config...
ok, changed:


compiling:
==
gcc -o obj/debug/./src/sc2code/velocity.o -c  -I/opt/include/SDL -I/opt/include/SDL -g -O0 -W -Wall -DDEBUG_T
RACK_SEM -DGFXMODULE_SDL -DSOUNDMODULE_SDL -I/usr/local/include -I src -I src/sc2code -I src/sc2code/libs src
/sc2code/velocity.c
gcc -o obj/debug/./src/sc2code/weapon.o -c  -I/opt/include/SDL -I/opt/include/SDL -g -O0 -W -Wall -DDEBUG_TRA
CK_SEM -DGFXMODULE_SDL -DSOUNDMODULE_SDL -I/usr/local/include -I src -I src/sc2code -I src/sc2code/libs src/s
c2code/weapon.c
mkdir: obj/debug/./src/sc2code/comm/arilou/: File exists
make: *** [obj/debug/./src/sc2code/comm/arilou/arilouc.o] Error 1
==
it always says mkdir: obj/debug/./src/sc2code/foo/bar: File exists
and dies when changes dir to existing. should not it check if exist?

ok,
==
==
gcc -o obj/debug/./src/sc2code/comm/yehat/yehatc.o -c  -I/opt/include/SDL -I/opt/include/SDL -g -O0 -W -Wall
-DDEBUG_TRACK_SEM -DGFXMODULE_SDL -DSOUNDMODULE_SDL -I/usr/local/include -I src -I src/sc2code -I src/sc2code
/libs src/sc2code/comm/yehat/yehatc.c
src/sc2code/comm/yehat/yehatc.c:113: warning: missing initializer
src/sc2code/comm/yehat/yehatc.c:113: warning: (near initialization for `yehat_desc.AlienAmbientArray[8].Block
Mask')
src/sc2code/comm/yehat/yehatc.c:120: warning: missing initializer
src/sc2code/comm/yehat/yehatc.c:120: warning: (near initialization for `yehat_desc.AlienAmbientArray[9].Block
Mask')
src/sc2code/comm/yehat/yehatc.c:127: warning: missing initializer
src/sc2code/comm/yehat/yehatc.c:127: warning: (near initialization for `yehat_desc.AlienAmbientArray[10].Bloc
kMask')
src/sc2code/comm/yehat/yehatc.c:134: warning: missing initializer
src/sc2code/comm/yehat/yehatc.c:134: warning: (near initialization for `yehat_desc.AlienAmbientArray[11].Bloc
kMask')
src/sc2code/comm/yehat/yehatc.c:141: warning: missing initializer
src/sc2code/comm/yehat/yehatc.c:141: warning: (near initialization for `yehat_desc.AlienAmbientArray[12].Bloc
kMask')
src/sc2code/comm/yehat/yehatc.c:148: warning: missing initializer
src/sc2code/comm/yehat/yehatc.c:148: warning: (near initialization for `yehat_desc.AlienAmbientArray[13].Bloc
kMask')
src/sc2code/comm/yehat/yehatc.c:165: warning: missing initializer
src/sc2code/comm/yehat/yehatc.c:165: warning: (near initialization for `yehat_desc.AlienTransitionDesc.BlockM
ask')
src/sc2code/comm/yehat/yehatc.c:172: warning: missing initializer
src/sc2code/comm/yehat/yehatc.c:172: warning: (near initialization for `yehat_desc.AlienTalkDesc.BlockMask')
src/sc2code/comm/yehat/yehatc.c:174: warning: missing initializer
src/sc2code/comm/yehat/yehatc.c:174: warning: (near initialization for `yehat_desc.AlienTextTemplate')
gcc -o obj/debug/./src/sc2code/comm/zoqfot/zoqfotc.o -c  -I/opt/include/SDL -I/opt/include/SDL -g -O0 -W -Wal
l -DDEBUG_TRACK_SEM -DGFXMODULE_SDL -DSOUNDMODULE_SDL -I/usr/local/include -I src -I src/sc2code -I src/sc2co
de/libs src/sc2code/comm/zoqfot/zoqfotc.c
src/sc2code/comm/zoqfot/zoqfotc.c:62: warning: missing initializer
src/sc2code/comm/zoqfot/zoqfotc.c:62: warning: (near initialization for `zoqfot_desc.AlienAmbientArray[0].Blo
ckMask')
src/sc2code/comm/zoqfot/zoqfotc.c:69: warning: missing initializer
src/sc2code/comm/zoqfot/zoqfotc.c:69: warning: (near initialization for `zoqfot_desc.AlienAmbientArray[1].Blo
ckMask')
src/sc2code/comm/zoqfot/zoqfotc.c:76: warning: missing initializer
src/sc2code/comm/zoqfot/zoqfotc.c:76: warning: (near initialization for `zoqfot_desc.AlienAmbientArray[2].Blo
ckMask')
src/sc2code/comm/zoqfot/zoqfotc.c:84: warning: missing initializer
src/sc2code/comm/zoqfot/zoqfotc.c:84: warning: (near initialization for `zoqfot_desc.AlienTransitionDesc.Bloc
kMask')
src/sc2code/comm/zoqfot/zoqfotc.c:91: warning: missing initializer
src/sc2code/comm/zoqfot/zoqfotc.c:91: warning: (near initialization for `zoqfot_desc.AlienTalkDesc.BlockMask'
)
src/sc2code/comm/zoqfot/zoqfotc.c:93: warning: missing initializer
src/sc2code/comm/zoqfot/zoqfotc.c:93: warning: (near initialization for `zoqfot_desc.AlienTextTemplate')
src/sc2code/comm/zoqfot/zoqfotc.c:124: warning: `SelectAlienFOT' defined but not used
mkdir: obj/debug/./src/sc2code/libs/decomp/: File exists
make: *** [obj/debug/./src/sc2code/libs/decomp/lzdecode.o] Error 1
==
warnings? What warnings?
there was no warnings in 0.2. so?

ok,
==
gcc -o obj/debug/./src/sc2code/planets/solarsys.o -c  -I/opt/include/SDL -I/opt/include/SDL -g -O0 -W -Wall -
DDEBUG_TRACK_SEM -DGFXMODULE_SDL -DSOUNDMODULE_SDL -I/usr/local/include -I src -I src/sc2code -I src/sc2code/
libs src/sc2code/planets/solarsys.c
src/sc2code/planets/solarsys.c: In function `ExploreSolarSys':
src/sc2code/planets/solarsys.c:1896: warning: empty body in an if-statement
gcc -o obj/debug/./src/sc2code/planets/surface.o -c  -I/opt/include/SDL -I/opt/include/SDL -g -O0 -W -Wall -D
DEBUG_TRACK_SEM -DGFXMODULE_SDL -DSOUNDMODULE_SDL -I/usr/local/include -I src -I src/sc2code -I src/sc2code/l
ibs src/sc2code/planets/surface.c
mkdir: obj/debug/./src/sc2code/ships/androsyn/: File exists
make: *** [obj/debug/./src/sc2code/ships/androsyn/androsyn.o] Error 1
==
same shit.

ok, we built uqm[-debug] binary! gratz to us! ;p

ok, we can even run it and play some melee...
==
Warning: Couldn't find thread to signal!
Warning: Couldn't find thread to signal!
Warning: Couldn't find thread to signal!
Warning: Couldn't find thread to signal!
Warning: Couldn't find thread to signal!
Warning: Couldn't find thread to signal!
Warning: Couldn't find thread to signal!
Warning: Couldn't find thread to signal!
Warning: Couldn't find thread to signal!
==
what the hell is that? Was no in 0.2. doesn do anything bad, but spams a lot.

but, if you try to play game and enter starbase...
==
SoundDecoder_Decode(): looping ipanims/space.mod
Warning: Couldn't find thread to signal!
Freeing Clock Semaphore in 'Starcon2Main' set by thread 'flash solar system menu'!
Warning: Couldn't find thread to signal!
Warning: Couldn't find thread to signal!
Warning: Couldn't find thread to signal!
       'urquan.con' -- 155 bytes
       'lbm/player.fon' -- 1 bytes
       'comm/urquan/urquan.ani' -- 1207 bytes
       'comm/urquan/urquan.fon' -- 1 bytes
       'comm/urquan/urquan.txt' -- 14565 bytes
Found timestamp file: comm/urquan/urquan.ts
       'comm/urquan/urquan.mod' -- 75885 bytes
_GetMusicData(): loading comm/urquan/urquan.mod
   decoder: MikMod, rate 44100 format 170202
SpliceTrack(): loading comm/urquan/urqua027.ogg
Warning: Couldn't find thread to signal!
Warning: Couldn't find thread to signal!
       'comandr.con' -- 223 bytes
       'lbm/player.fon' -- 1 bytes
       'comm/comandr/comandr.ani' -- 2731 bytes
       'comm/comandr/comandr.fon' -- 1 bytes
       'comm/comandr/comandr.ct' -- 292 bytes
       'comm/comandr/comandr.mod' -- 67076 bytes
_GetMusicData(): loading comm/comandr/comandr.mod
   decoder: MikMod, rate 44100 format 170202
       'comm/comandr/comandr.txt' -- 15788 bytes
Found timestamp file: comm/comandr/comandr.ts
SpliceTrack(): loading comm/comandr/coman003.ogg
zsh: segmentation fault (core dumped)  ./uqm-debug
==
SIGSEGV. why?

==
(gdb)
582                     ClearSemaphore (_MemorySem);
(gdb)

Program received signal SIGTRAP, Trace/breakpoint trap.
ClearSemaphore (sem=0x8401650) at src/sc2code/libs/threads/thrcommon.c:548
548             char *sem_name = NULL;
(gdb)
550             for (i = 0; i < numSems; i++)
(gdb)
551                     if (SemMon.Sem == sem)
(gdb)
550             for (i = 0; i < numSems; i++)
(gdb)
551                     if (SemMon.Sem == sem)
(gdb)
553                             sem_name = SemMon.Name;
(gdb)
554                             if (SemMon.Thread && SemMon.Thread != NativeThreadID ())
(gdb)
568                             SemMon.ThreadName[0] = 0;
(gdb)
573                             SemMon.Thread = 0;
(gdb)
574                             break;
(gdb)
576             if (semval != 0)
(gdb)
579             NativeClearSemaphore ((NativeSemaphore) sem);
(gdb)
580     }
(gdb)
mem_simple_unaccess (h=3356) at src/sc2code/libs/memory/w_memlib.c:583
583                     return (1);
(gdb)
0x080c7cbc      588     }
(gdb)
GetStringAddress (String=3738908) at src/sc2code/libs/strings/strings.c:241
241             return (StringAddr);
(gdb)
0x080ed030      242     }
(gdb)
construct_response (buf=0x8337ec0 "No, this is the starship ", R=58) at src/sc2code/commglue.c:265
265                     len = wstrlen (buf);
(gdb)
269                     buf += len;
(gdb)
274                     name = va_arg (vlist, UNICODE *);
(gdb)
279                     if (name)
(gdb)
294                             len = wstrlen (name);
(gdb)

Program received signal SIGSEGV, Segmentation fault.
==
Breakpoint 1, construct_response (buf=0x8337ec0 "", R=58) at src/sc2code/commglue.c:242
242             va_start (vlist, R);
(gdb) next
253                     S = SetAbsStringTableIndex (CommData.ConversationPhrases, R - 1);
(gdb) print R
$1 = 58
(gdb) print name
$2 = (UNICODE *) 0x807e464 "U\211+?\203+?\030+?+-y+¬\203+?oj"
(gdb) next
260                     wstrcpy (buf, (UNICODE *)GetStringAddress (S));
(gdb) next
265                     len = wstrlen (buf);
(gdb) next
269                     buf += len;
(gdb) print len
$3 = 25
(gdb) print buf
$4 = (UNICODE *) 0x8337ec0 "No, this is the starship "
(gdb) next
274                     name = va_arg (vlist, UNICODE *);
(gdb) next
279                     if (name)
(gdb) print name
$5 = (UNICODE *) 0x775007f8 <Address 0x775007f8 out of bounds>
==
looks like construct_response gets shit instead of *"Vindicator" as its n-th param. clues? going do gdb it.

Ttthatts all folks!

Err... Should not we go to... err... some *place* with our developer talks?

Hmm... on other platforms you don't have to specify the libraries that are indirectly used. Strange really. What does 'ldd /x11/usr/lib/libvorbis.so' say?


This is not serious. It's testing for what's available. The game will work without OpenGL. But I wonder why it's printing the ld error message. That should have been redirected to /dev/null.


Hmm... this is a serious problem. It looks like this message comes from sed. That would explain the '@BLA@' stuff in your config.h. I think it doesn't accept '@' in the format string. I'd need some more information on what sed your system has.


Well, it *is* checking whether the dir exists (see Makefile.build). Maybe your 'mkdir' handles '-p' strangely? Or maybe your 'test' has problems? Could you investigate?


No, there were a LOT of warnings in 0.2. Only they weren't shown by default. In the cvs build, gcc compiles with '-W -Wall' if you make a debugging compile. You can even choose more strict checking in the config menu. Anyhow, this is just a fraction of the warnings that 0.2 generated. We'll get to 0 warnings yet.


This went rather smoothly compared to the first MacOS compiles, really.


This is actually new code. I'm not so sure if this warning is serious. McMartin will know.


Because of a bug, I guess.
Unfortunately, the gdb info you sent doesn't help much.


Ah, cool. It's hard to do anything about from just the info you showed (though a stack trace wouldn't hurt).


#sc2 on irc.freenode.net is where the developers hang out. We've also got a bugzilla bug database (http://uqm.stack.nl/cgi-bin/bugs/index.cgi) set up. And there's also a mailing list at 'sc2-devel at lists.sourceforge.net', but it's rarely used (though we'll read it if you send mail there).
As for the building problems, you might as well send them directly to me (svdb at stack.nl). I'm the one who made the unix build scripts.

Hmm... on other platforms you don't have to specify the libraries that
are indirectly used. Strange really. What does 'ldd
/x11/usr/lib/libvorbis.so' say?
==
That's a good question with a very interesting answer!

localhost% ldd /x86/usr/lib/libvorbis.so
zsh: command not found: ldd

i'll try to obtain it (unsure:where) and will try again...

==
Hmm... this is a serious problem. It looks like this message comes from
sed.
==
i guess, it comes from printf:
                       if menu_have_choice "$MENU" "$OPTION"; then
                               printf "  %*i. %s %s\n" "${#NUM_OPTIONS}" "$I" \
                                               "$SELECTED" "$TEMP_TITLE"
                       else
                               printf "  %-*s  %s (N/A) %s\n" "${#NUM_OPTIONS}"
"-" \
                                               "$SELECTED" "$TEMP_TITLE"
                       fi
and i can understand him: i failed to understand, what this format should do, even after 10 minutes of meditating on it...

==
That would explain the '@BLA@' stuff in your config.h. I think it
doesn't accept '@' in the format string. I'd need some more information
on what sed your system has.
==

ok, messing with scripts:
should not string
           VALUE=$(echo "$VALUE" | sed 's,\([\&/]\),\\\1,g')
have s/sed/$SED/? ;p

ok, now:
config.h.*.sed:)
s/@WORDS_BIGENDIAN@/#undef WORDS_BIGENDIAN/g
s/@HAVE_strupr@/#define HAVE_strupr/g
s/@HAVE_stricmp@/#define HAVE_stricmp/g
s/@HAVE_getopt_h@/#undef HAVE_getopt_h/g
s/@CONTENTDIR@/\/usr\/local\/games\/lib\/uqm\/content/g
funny?

so, we may consider, that _tr_ is not always brain alive and do s/tr a-z A-Z/tr '[a-z]' '[A-Z]'/

and my system do have GNU sed, as i c.

==
Well, it *is* checking whether the dir exists (see Makefile.build).
Maybe your 'mkdir' handles '-p' strangely? Or maybe your 'test' has
problems? Could you investigate?
==
i failed to reproduce it form commandline
it really never screams when with -p and really Makefiles do contain -p and really it still screams when building... ;(
4 times - for each uqm_SRCDIR's in src/sc2code/Makeinfo

sadly, i cant gdb sh-scripts and make files... ;(

==
This went rather smoothly compared to the first MacOS compiles, really.
==
i guess. but do someone really still uses it? Non-X MacOS?

i will think about bug - later...

251                     fprintf(stderr, "DEBUG: len = %d\n",len);
253                     S = SetAbsStringTableIndex (CommData.ConversationPhrases, R - 1);
255                     fprintf(stderr, "DEBUG: Returned from call to SetAbsStringTableIndex\n");
256                     fprintf(stderr, "DEBUG: GetStringAddress(S) = %s\n",(UNICODE *)GetStringAddress(S));
257                     fprintf(stderr, "DEBUG: buf = '%s' (%d) \n", buf, *buf);
258                     fprintf(stderr, "DEBUG: strlen(buf) = %d\n", strlen(buf));
260                     wstrcpy (buf, (UNICODE *)GetStringAddress (S));
262                     fprintf(stderr, "DEBUG: buf = '%s' (%d) \n", buf, *buf);
263                     fprintf(stderr, "DEBUG: strlen(buf) = %d\n", strlen(buf));
265                     len = wstrlen (buf);
267                     fprintf(stderr, "DEBUG: len = %d\n",len);
269                     buf += len;
271                     fprintf(stderr, "DEBUG: buf = '%s' (%d) \n", buf, *buf);
272                     fprintf(stderr, "DEBUG: strlen(buf) = %d\n", strlen(buf));
274                     name = va_arg (vlist, UNICODE *);
276                     fprintf(stderr, "DEBUG: returned from va_arg call\n");
277                     fprintf(stderr, "DEBUG: name = %s\n",name);

Program received signal SIGSEGV, Segmentation fault.
0xb03396c2 in ?? ()
(gdb) print name
No symbol "name" in current context.
(gdb) stack
Undefined command: "stack".  Try "help".
(gdb) bt
#0  0xb03396c2 in ?? ()
#1  0xb033919a in ?? ()
#2  0xb03352fe in ?? ()
#3  0x08053df8 in construct_response (buf=0x83384f9 "", R=58)
   at src/sc2code/commglue.c:277
#4  0x08087ad0 in Intro () at src/sc2code/comm/comandr/comandr.c:633
#5  0x08052906 in HailAlien () at src/sc2code/comm.c:2044
#6  0x08052d26 in InitCommunication (which_comm=350231303)
   at src/sc2code/comm.c:2162
#7  0x0807cd78 in VisitStarBase () at src/sc2code/starbase.c:441
#8  0x0807e94d in Starcon2Main (blah=0x8337510) at src/sc2code/starcon.c:926
#9  0x080edcf3 in ThreadHelper (startInfo=0x8401e64)
   at src/sc2code/libs/threads/thrcommon.c:196
#10 0xb823088f in ?? ()
#11 0xb8230a7e in ?? ()

that looks like some Arilou abduction ;(
cause:
                       construct_response (
                                       shared_phrase_buf,
                                       no_but_well_help0,
                                       GLOBAL_SIS (ShipName),
                                       no_but_well_help1,
                                       0
                                       );
yes, GLOBAL_SIS (ShipName)==GlobData.SIS_state.ShipName IS "Vindicator\0\0\0\0\0"...
why may va_arg fail?

==
Intro () at src/sc2code/comm/comandr/comandr.c:626
626                     SET_GAME_STATE (STARBASE_VISITED, 1);
628                     NPCPhrase (ARE_YOU_SUPPLY_SHIP);
629                     if (GLOBAL_SIS (ElementAmounts[RADIOACTIVE]))
633                             construct_response (
(gdb) print GlobData.SIS_state.ShipName
$1 = "Vindicator\0\0\0\0\0"
(gdb) print &GlobData.SIS_state.ShipName
$3 = (UNICODE (*)[16]) 0x833880a
(gdb) step
construct_response (buf=0x83384e0 "", R=58) at src/sc2code/commglue.c:240
240             fprintf(stderr, "DEBUG: We've entered into construct_response!\n");
242             va_start (vlist, R);
244             fprintf(stderr, "DEBUG: R = %d\n", R);
251                     fprintf(stderr, "DEBUG: len = %d\n",len);
253                     S = SetAbsStringTableIndex (CommData.ConversationPhrases, R - 1);
255                     fprintf(stderr, "DEBUG: Returned from call to SetAbsStringTableIndex\n");
256                     fprintf(stderr, "DEBUG: GetStringAddress(S) = %s\n",(UNICODE *)GetStringAddress(S));
257                     fprintf(stderr, "DEBUG: buf = '%s' (%d) \n", buf, *buf);
258                     fprintf(stderr, "DEBUG: strlen(buf) = %d\n", strlen(buf));
260                     wstrcpy (buf, (UNICODE *)GetStringAddress (S));
262                     fprintf(stderr, "DEBUG: buf = '%s' (%d) \n", buf, *buf);
263                     fprintf(stderr, "DEBUG: strlen(buf) = %d\n", strlen(buf));
265                     len = wstrlen (buf);
267                     fprintf(stderr, "DEBUG: len = %d\n",len);
269                     buf += len;
271                     fprintf(stderr, "DEBUG: buf = '%s' (%d) \n", buf, *buf);
272                     fprintf(stderr, "DEBUG: strlen(buf) = %d\n", strlen(buf));
274                     name = va_arg (vlist, UNICODE *);
276                     fprintf(stderr, "DEBUG: returned from va_arg call\n");
(gdb) print name
$4 = (UNICODE *) 0x7ad007f8 <Address 0x7ad007f8 out of bounds>
(gdb) print &name
$5 = (UNICODE **) 0x7f84ae4
==


==
(gdb) bt
#0  construct_response (buf=0x8338531 "", R=2056) at src/sc2code/commglue.c:276
#1  0x08087ad0 in Intro () at src/sc2code/comm/comandr/comandr.c:633
#2  0x08052906 in HailAlien () at src/sc2code/comm.c:2044
#3  0x08052d26 in InitCommunication (which_comm=350231303) at src/sc2code/comm.c:2162
#4  0x0807cd78 in VisitStarBase () at src/sc2code/starbase.c:441
#5  0x0807e94d in Starcon2Main (blah=0x8337510) at src/sc2code/starcon.c:926
#6  0x080edcf3 in ThreadHelper (startInfo=0x8401e64) at src/sc2code/libs/threads/thrcommon.c:196
#7  0xb823088f in ?? ()
#8  0xb8230a7e in ?? ()
==


i may even 'set variable name' in gdb:
==
DEBUG: buf = 'No, this is the starship ' (78)
DEBUG: strlen(buf) = 25
DEBUG: len = 25
DEBUG: buf = '' (0)
DEBUG: strlen(buf) = 0
DEBUG: returned from va_arg call
DEBUG: name = Vindicator
DEBUG: ------------------------------------
DEBUG: We are in the special va_args loop!
DEBUG: Current critical variables read:
DEBUG: ------------------------------------
DEBUG: name = Vindicator
DEBUG: buf = '' (0)
DEBUG: len = 25
DEBUG: strlen(name) = 10
DEBUG: strlen(buf) = 0
DEBUG: ---- continuing ----
DEBUG: len = 10
DEBUG: buf = 'Vindicator' (86)
DEBUG: strlen(buf) = 10
DEBUG: buf = '' (0)
DEBUG: Calling va_arg
DEBUG: Return from call to va_arg
DEBUG: R = 2056
DEBUG: name = Vindicator
DEBUG: len = 10
DEBUG: Returned from call to SetAbsStringTableIndex
DEBUG: GetStringAddress(S) = Uh, I meant a different red thing. Never mind.
DEBUG: buf = '' (0)
DEBUG: strlen(buf) = 0
DEBUG: buf = 'Uh, I meant a different red thing. Never mind.' (85)
DEBUG: strlen(buf) = 46
DEBUG: len = 46
DEBUG: buf = '' (0)
DEBUG: strlen(buf) = 0
DEBUG: returned from va_arg call
DEBUG: name =
[1]  + segmentation fault (core dumped)  ./uqm-debug
==

ok, i can even get fwiffo, fly to mercury and get some radioactives, but:
==
DEBUG: We've entered into construct_response!
DEBUG: R = 64
DEBUG: len = 0
DEBUG: Returned from call to SetAbsStringTableIndex
DEBUG: GetStringAddress(S) = I am Captain
DEBUG: buf = '' (0)
DEBUG: strlen(buf) = 0
DEBUG: buf = 'I am Captain ' (73)
DEBUG: strlen(buf) = 13
DEBUG: len = 13
DEBUG: buf = '' (0)
DEBUG: strlen(buf) = 0
DEBUG: returned from va_arg call
DEBUG: name =
[1]  + segmentation fault (core dumped)  ./uqm-debug
==
on the second phrase...

ok, what do work wrong in our commglue? or their va_*?

If you don't have it, then QNX probably does things differently. I'd need to know how though.


It's all in the manual page. printf works like the libc printf() function. The first argument is the format string specifying how to interpret the rest of the options.
But you're right. It does look like this is what generated the 'format errors', not sed. Why it only does that for the debug menu, I don't know.
What would be helpful would be if you set the environment variable 'SH' to 'sh -x', and then rerun the config script. The debug output that it generates then will be very useful to me. It would be very long, so you shouldn't post that here, but instead mail it to me directly.


You're right. It's corrected in CVS now.


I see. Appearantly, [a-z] is the System V way, and a-z is the original BSD way, which is now part of the POSIX standard (though BSD and GNU tr also support [a-z]. I suspect [:lower:] is the most portable, so I've changed the line to
tr "[:lower:]" "[:upper:]"


For sh-scripts, setting SH to "sh -x" in your environment will help.


Well, I was talking about MacOS X.

As for the construct_response, I think the easiest way to debug this is by looking what exactly is on the stack as the function is run.

If you don't have it, then QNX probably does things differently. I'd
need to know how though.
==
i have got "ldd in shell script"
so, we may see:
localhost% ldd libSDL.so
       libm.so.2 =>    /lib/libm.so.2
       libph.so.2 =>   /usr/lib/libph.so.2
       libGL.so.2 =>   /usr/lib/libGL.so.2
       libasound.so.2 =>       /lib/libasound.so.2
       libc.so.2 =>    /proc/boot/libc.so.2
localhost% ldd libvorbis.so
       libc.so.2 =>    /proc/boot/libc.so.2
localhost% ldd libogg.so
       libc.so.2 =>    /proc/boot/libc.so.2

so, we can see, while 3rd-party made libSDL is linked properly, libs that came with distribution are linked really strangely.
so, we will have to have that QNX) part in config...

==
It's all in the manual page.
==
:p QNX does not have man by default... :p

==
printf works like the libc printf() function. The first argument is the
format string specifying how to interpret the rest of the options.
==
yes. and what do these 2 strings mean? i am curious, and printf is curious too...
" ? %*i. %s %s\n"
" ? %-*s ? %s (N/A) %s\n"

==
But you're right. It does look like this is what generated the 'format
errors', not sed. Why it only does that for the debug menu, I don't
know.
==
I have changed these format strings and everything went ok.
why only in menu? because, only in menu there are such strings. have not tried menus other than debug...

oh, btw. I have got OpenGL devel packages and made it with OpenGL support.
It DOES run and work but it goes TERRIBLE ;(

and, our bug:
==
Breakpoint 1, construct_response (buf=0x8339f80 "", R=58) at src/sc2code/commglue.c:240

240             fprintf(stderr, "DEBUG: We've entered into construct_response!\n");
(gdb) info stack
#0  construct_response (buf=0x8339f80 "", R=58) at src/sc2code/commglue.c:240
#1  0x08088140 in Intro () at src/sc2code/comm/comandr/comandr.c:633
#2  0x08052f76 in HailAlien () at src/sc2code/comm.c:2044
#3  0x08053396 in InitCommunication (which_comm=350231303) at src/sc2code/comm.c:2162
#4  0x0807d3e8 in VisitStarBase () at src/sc2code/starbase.c:441
#5  0x0807efbd in Starcon2Main (blah=0x8338fb0) at src/sc2code/starcon.c:926
#6  0x080ef513 in ThreadHelper (startInfo=0x8403e64) at src/sc2code/libs/threads/thrcommon.c:196
#7  0xb823088f in ?? ()
#8  0xb8230a7e in ?? ()
(gdb) next
242             va_start (vlist, R);
(gdb) next
244             fprintf(stderr, "DEBUG: R = %d\n", R);
(gdb) print vlist
$1 = 0x7f84aee "o\a@\201\b\b\200\2373\b:"
(gdb) ptype R
type = short unsigned int
==
what is strange:
1)vlist is junk
2)type of R is short unsigned int, but later program will try to R = va_arg(vlist, int);
3)i fail to understand what does work wrong...

Ok, then I'll add that. Is the -I/usr/local/include and -L/usr/local/lib necessary in QNX, or did you just copy them from the line from FreeBSD (on Linux, these paths are used automatically).


%*i is like %i (take an integer), but it reads the length of the field from an additional argument. The minus sign in the second line means left-align instead of right-align.
If any of those what your printf chokes on, I can easilly work around it.
Just let me know which one (or maybe both).


The main menu also is a menu, that's why I was confused. I forgot for a moment it was a different kind of menu.


OpenGL only helps if you have hardware acceleration. In Linux, only a few video drivers support it. I suspect it's even worse on QNX.


It's not. It's not supposed to be point to a string. It's pointing to somewhere on the stack.


That itself is not a problem. The number fits in a short.
What I'm wondering though if whether a complete has really been pushed on the stack. (see below)


I suspect it is a variable size issue of some sort. The easiest way to check would be to check what's on the stack.
Could you do the following from gdb, from within construct_response:
(gdb) x/40xb &R
and also:
(gdb) p sizeof no_but_well_help0

Also, I'd like to know what architechture this is? IA-32?

I noticed you visited the IRC channel, but left again after a minute.
I was there, but like many other people, we're often working on something, and don't check the IRC session every minute. It's still the best place to find us, and probably the fastest way to solve this.

Ok, then I'll add that. Is the -I/usr/local/include and -L/usr/local/lib
necessary in QNX, or did you just copy them from the line from FreeBSD
(on Linux, these paths are used automatically).
==
You dont need. Bacause, QNX is very strange on paths anyway:
libs should be taken from /<arch>/usr/lib/ and /usr/local/lib (libvorbis would be here by default)
includes - from /opt/include

thats do work without -I -L as i see...

==
%*i is like %i (take an integer), but it reads the length of the field
from an additional argument. The minus sign in the second line means
left-align instead of right-align.
If any of those what your printf chokes on, I can easilly work around
it.
Just let me know which one (or maybe both).
==
they _both_ does not work.
but, all menus looks correctly with:
==
                       if menu_have_choice "$MENU" "$OPTION"; then
                               printf "  %i. %s %s\n" "$I" \
                                               "$SELECTED" "$TEMP_TITLE"
                       else
                               printf "  %s  %s (N/A) %s\n" "-" \
                                               "$SELECTED" "$TEMP_TITLE"
                       fi
==
=>
==
 -= Main menu =-
 1. Include debugging information        Debug info + strict compile checks
 2. OpenGL graphics support              Include OpenGL graphics support
 3. Sound backend                        Use MixSDL for sound
 4. Installation paths                   [...]

Press a number plus <ENTER> if you want to change something,
or just <ENTER> if everything is ok: 1

 -= Include debugging information =-
 1.     No debugging information
 2.     Debugging information
 3. --> Debug info + strict compile checks

Select the option you want by typing a number plus <ENTER>
or just <ENTER> if everything is ok:

 -= OpenGL graphics support =-
 1.     Don't include OpenGL graphics support
 2. --> Include OpenGL graphics support

Select the option you want by typing a number plus <ENTER>
or just <ENTER> if everything is ok:

 -= Sound backend =-
/usr/bin/ntox86-ld: cannot find -lopenal
collect2: ld returned 1 exit status
OpenAL not found.
 1. --> Use MixSDL for sound
 -      (N/A) Include OpenAL support (experimental)

Select the option you want by typing a number plus <ENTER>
or just <ENTER> if everything is ok:
==
We does not have more than 10 options, and not going to, so we can use KISS method.

==
OpenGL only helps if you have hardware acceleration.
==
Yea. We have it and it does work. How does it work - it's not our problem, SDL version looks good...
BTW, SDL itself is still beta under QNX.
"We call it beta cause it's betta when nothing"... :p
==
In Linux, only a few video drivers support it. I suspect it's even worse on QNX.
==
Sure. Qhile it's enough for Q3, btw ^^
==
(gdb) x/40xb &R
0x7f84aea:      0x3a    0x00    0x1c    0x4b    0xf8    0x07    0x40    0x81
0x7f84af2:      0x08    0x08    0x80    0x9f    0x33    0x08    0x3a    0x00
0x7f84afa:      0x00    0x00    0xaa    0xa2    0x33    0x08    0x3b    0x00
0x7f84b02:      0x00    0x00    0x00    0x00    0x00    0x00    0x00    0x00
0x7f84b0a:      0x00    0x00    0x3c    0x4e    0xf8    0x07    0x64    0x2f

(gdb) p sizeof no_but_well_help0
$1 = 4

so, it's not that short...

but...
==
(gdb) ptype no_but_well_help0
type = enum {NULL_PHRASE, GLAD_WHEN_YOU_COME_BACK, GIVE_FUEL, GIVE_FUEL_AGAIN, ARE_YOU_SUPPLY_SHIP,
   DO_YOU_HAVE_RADIO_THIS_TIME, HERE_IS_ANOTHER_LANDER, THE_WHAT_FROM_WHERE, ABOUT_TIME,
   MESSAGE_GARBLED_1, MESSAGE_GARBLED_2, HERE_IS_A_NEW_LANDER, THIS_MAY_SEEM_SILLY, OK_THE_NAFS,
   OK_THE_CAN, OK_THE_UFW, OK_THE_NAME_IS_EMPIRE0, OK_THE_NAME_IS_EMPIRE1, FUEL_UP0, FUEL_UP1,
   WHAT_KIND_OF_IDIOT, DONT_KNOW_WHO_YOU_ARE, THATS_IMPOSSIBLE, ASK_AWAY, RADIOS_ON_MERCURY,
   THANKS_FOR_HELPING, STARBASE_IS, HAPPENED_TO_EARTH, URQUAN_LEFT, BASE_ON_MOON, ACKNOWLEDGE_SECRET,
   ABOUT_BASE, GOOD_LUCK_WITH_BASE, DEALT_WITH_BASE_YET, HERE_COMES_ILWRATH, VERY_IMPRESSIVE,
   IT_WAS_ABANDONED, YOU_REALLY_FOUGHT_BASE, IM_GLAD_YOU_WON, IM_SURE_IT_WAS_DIFFICULT, THAT_WAS_PROBE,
   DEEP_TROUBLE, GOOD_NEWS, SURE_HOPE, ABOUT_BASE_AGAIN, COOK_BUTTS, OVERTHROW_ALIENS, KILL_MONSTERS,
   GOOD_LUCK_AGAIN, STARBASE_WILL_BE_READY, overthrow_evil_aliens, annihilate_those_monsters,
   cook_their_butts, where_get_radios, well_go_get_them_now, we_will_transfer_now, what_will_you_give_us,
   before_radios_we_need_info, no_but_well_help0, no_but_well_help1, yes_this_is_supply_ship,
   what_slave_planet, i_lied, plumb_out, we_are_vindicator0, we_are_vindicator1, we_are_vindicator2,
   first_give_info, we_must_go_now, where_can_i_get_radios, ok_i_will_get_radios, whats_this_starbase,
   what_about_earth, where_are_urquan, our_mission_was_secret, we_are_here_to_help, tell_me_about_base,
   we_will_take_care_of_base, tell_me_again, base_was_abandoned, we_fought_them, oh_yes_big_fight,
   i_lied_it_was_abandoned, i_cant_talk_about_it, name_1, name_2, name_3, name_40, name_41,
   i_lost_my_lander, i_lost_another_lander, need_fuel, need_fuel_again, what_was_red_thing, it_went_away,
   we_destroyed_it, what_probe, take_care_of_base_again, goodbye_commander}
(gdb) p sizeof no_but_well_help0
$3 = 4
(gdb) ptype R
type = short unsigned int
(gdb) p sizeof R
$4 = 2
==
Something mystical... Reminds me about Arilou again...

==
Also, I'd like to know what architechture this is? IA-32?
==
Sure ;) Cause trying to get alpha version of game with beta libs on "still testing" platform is kinda tricky ;)
And i have seen non-x86 pc 2 or 3 times in my life ;p

==
I noticed you visited the IRC channel, but left again after a minute.
I was there, but like many other people, we're often working on
something, and don't check the IRC session every minute. It's still the
best place to find us, and probably the fastest way to solve this.
==
Sure ;p
I got there only to look around... I will be there, sometimes... ;p
Sorry for leaving irc; i had my phone line dead ;(