#uwadv@irc.freenode.net logs for 25 Sep 2003 (GMT)

Archive Today Yesterday Tomorrow
Underworld Adventures homepage


[01:22:54] <servus> Do you guys ever want some more models? I have some unreleased ones, with bones. :)
[07:52:23] --> eldron has joined #uwadv
[15:05:41] --> Strangle has joined #uwadv
[15:05:53] <Strangle> anyone here ?
[15:08:45] <-- Strangle has left #uwadv ()
[16:21:11] --> wjp has joined #uwadv
[16:21:11] --- ChanServ gives channel operator status to wjp
[19:31:28] --> Dragst has joined #uwadv
[19:31:31] <Dragst> aye
[19:31:36] <wjp> hi
[19:36:23] --> vividos has joined #uwadv
[19:36:23] --- ChanServ gives channel operator status to vividos
[19:36:49] <vividos> hi
[19:37:29] <wjp> hi
[20:13:10] <vividos> I'm currently doing a test scenario where the bug still occurs
[20:13:30] <wjp> stripping down configure to a minimal test case?
[20:15:00] <vividos> yes
[20:16:09] <wjp> msys devels should be very interested in that
[20:16:18] <vividos> yes
[20:19:23] <wjp> which windows are you running, btw?
[20:19:33] <vividos> win98
[20:58:33] <vividos> I more and more think that there's a problem with zziplib's configure.ac
[20:58:40] <vividos> I cut down uwadv's to almost nothing
[21:02:44] <wjp> a problem in configure.ac causing this bug you mean?
[21:02:48] <wjp> I don't think so
[21:03:03] <wjp> the configure source in which the error was occurring looked ok
[21:04:02] <vividos> hmm ...
[21:04:34] <vividos> or a bug in the acinclude.m4
[21:04:37] <vividos> zziplib's
[21:05:32] <wjp> but the behaviour it was displaying was just broken... $ac_feature only contained letters and it triggered a test for non-alphanumerics
[21:10:14] <vividos> I'm testing to further cut down the configure.ac
[21:11:02] <wjp> why not try to cut down configure itself?
[21:11:12] <wjp> I'm thinking just the loop that processes the options?
[21:11:25] <wjp> (and then cut out all options except --enable and --prefix?)
[21:12:52] <vividos> configure itself is too complicated for me :)
[21:17:21] <wjp> wow, this is interesting
[21:17:29] <wjp> after cutting out the option parsing part it gives the same error here
[21:17:47] <wjp> s/cutting out/cutting it down to/
[21:19:09] <vividos> hmm
[21:20:53] <wjp> 43 lines left, currently
[21:21:10] <wjp> 29...
[21:21:43] <wjp> oh... silly me
[21:22:07] <vividos> didn't regenerated the configure script?
[21:22:15] <wjp> no, I cut out too much :-)
[21:22:20] <wjp> I'm not using autoconf
[21:22:25] <wjp> hacking up configure itself
[21:22:27] <vividos> ah right :)
[21:25:16] <wjp> try running http://www.math.leidenuniv.nl/~wpalenst/configure
[21:27:00] <wjp> (it won't actually do anything, so you can run it from anywhere)
[21:34:03] <vividos> how should I call it?
[21:34:21] <wjp> with the same options that broke zziplib's configure
[21:35:22] <vividos> ok
[21:35:40] <vividos> with /bin/sh './configure' --prefix=/usr/local '--enable-uadebug' --cache-file=/dev/null --srcdir=.
[21:35:44] <vividos> it does nothing
[21:36:16] <wjp> ok, that's the correct behaviour
[21:36:34] <wjp> I was kind of hoping it'd reproduce the bug, but I did cut out a bit much :-)
[21:41:10] <vividos> it seems AC_SET_RELEASEINFO_VERSIONINFO is the culprit
[21:42:06] <vividos> just retesting to be sure
[21:49:43] <wjp> if it fixes it I'd like to see a diff of the resulting configure script :-)
[21:50:28] <vividos> ok :)
[22:04:09] <vividos> the diff is quite large, since AM_PROG_LIBTOOL is commented out for a working script
[22:05:23] <wjp> ah, can you send me the new configure then?
[22:06:18] <vividos> the working one?
[22:06:59] <wjp> yes
[22:07:28] <vividos> I'll send you both
[22:09:18] <wjp> heh, that .works configure is 4 times as small as the .worksnot
[22:11:40] <wjp> hm, the .works has about 150 less lines before the option parsing
[22:12:39] <wjp> including some really weird ones
[22:12:56] <wjp> try running the .worksnot with '--no-reexec' as the first paramater
[22:14:03] <vividos> the size difference comes from AM_PROG_LIBTOOL
[22:14:21] <wjp> most of the size difference is after the option parsing, though
[22:14:47] <wjp> removing everything after that produces a .works of 21226 bytes, and a .worksnot of 27121 bytes
[22:15:21] <vividos> trying the .worksnot one
[22:17:17] <vividos> starting /bin/sh './configure' --no-reexec --prefix=/usr/local '--enable-uadebug' --c
[22:17:19] <vividos> ache-file=/dev/null --srcdir=.
[22:17:26] <vividos> gives the same old error
[22:17:31] <vividos> configure: error: invalid feature name: uadebug
[22:22:58] <wjp> hmm
[22:23:10] <wjp> you're in bash when running this, right?
[22:23:25] <vividos> just sh
[22:24:12] <wjp> try 'export echo_test_string=foo' before running configure.worksnot
[22:25:00] <vividos> doesn't fix it
[22:26:22] <vividos> hmm now it works, very strange
[22:30:12] <wjp> what did you do?
[22:30:28] <vividos> the export thing
[22:31:08] <wjp> that skips a check with the comment "# find a string as large as possible, as long as the shell can cope with it"
[22:31:30] <vividos> hmmmmmmm
[22:31:41] <wjp> maybe it's overflowing something internally by that?
[22:31:41] <vividos> will try that in a new shell with uwadv
[22:33:08] <vividos> may be
[22:37:58] <wjp> I added that test to http://www.math.leidenuniv.nl/~wpalenst/configure
[22:38:22] <vividos> *sigh* now even uwadv's configure gives up with that error
[22:39:06] <wjp> it keeps getting stranger... :/
[22:39:32] <vividos> yes
[22:40:37] <vividos> calling /bin/sh './configure' --prefix=/usr/local '--enable-uadebug' --cache-file=/dev/null --srcdir=.
[22:40:44] <vividos> on the new configure gives no error
[22:41:58] <vividos> I'm giving up now
[22:42:05] <vividos> should go to bed anyway
[22:42:12] <wjp> yeah, me too, I guess
[22:42:32] <vividos> I'm not online for some time, btw
[22:42:42] <vividos> (relocating home)
[22:42:55] * wjp quickly checks if uwadv compiles, then :-)
[22:43:03] <vividos> *g*
[22:43:56] <wjp> undefined reference to `ua_start_menu_screen::ua_start_menu_screen' while linking
[22:44:39] * wjp tries adding screens/start_menu.cpp to Makefile.common
[22:44:51] <vividos> yes, that one is missing
[22:44:56] <wjp> doesn't compile, though
[22:45:20] * wjp tries to parse the error message
[22:46:00] <wjp> img_screen.get_image().get_palette()->rotate(64,64,false); gives a parse error
[22:46:20] <wjp> screens/start_menu.cpp:221: request for member `rotate' in `*(+(+(this +
[22:46:20] <wjp> 168)->ua_image_quad::get_image()())->ua_image::get_palette()())->ua_smart_ptr<T>::operator->()
[22:46:20] <wjp> const [with T = Uint8[256][4]]()', which is of non-aggregate type `
[22:46:20] <wjp> Uint8[256][4]'
[22:46:59] <vividos> ua_palette256 was changed from Uint8[256][4] to a class; might be some dependencies not recompiled
[22:47:11] <vividos> index.cpp I guess
[22:47:34] <vividos> hmm maybe I should commit that file? *bg*
[22:47:37] <wjp> maybe :-)
[22:50:54] <vividos> ok all should be commit'ed
[22:52:49] <wjp> ok, builds and runs now
[22:52:58] <wjp> it gets to the menu
[22:53:22] <wjp> buttons don't do much, but I guess that's expected
[22:54:17] <vividos> yes
[22:54:28] <vividos> ok, got to go. night!
[22:54:31] <wjp> night
[22:55:25] <-- vividos has left IRC ("Leaving")
[23:05:26] <-- wjp has left IRC ("Zzzz...")