#pentagram@irc.freenode.net logs for 29 Oct 2010 (GMT)

Archive Today Yesterday Tomorrow
Pentagram homepage


[00:31:43] --> Kirben has joined #pentagram
[00:31:43] --- ChanServ gives channel operator status to Kirben
[03:19:22] <-- Kirben has left IRC (Ping timeout: 240 seconds)
[03:20:07] --> Kirben has joined #pentagram
[03:20:07] --- ChanServ gives channel operator status to Kirben
[03:20:16] <-- spioH has left #pentagram
[04:43:02] --> Kirben_ has joined #pentagram
[04:43:02] --- ChanServ gives channel operator status to Kirben_
[04:43:50] <-- Kirben has left IRC (Ping timeout: 264 seconds)
[09:00:07] --- Kirben_ is now known as Kirben
[10:32:42] <-- Kirben has left IRC ()
[11:03:50] <dominus> my yearly commit... :)
[11:30:13] <wjp> hm, is this --enable-static-libraries OS X-specific?
[11:31:56] <dominus> no, it's gcc specific
[11:32:16] <dominus> see line 304
[11:33:58] <dominus> (and following in configure.ac)
[11:34:50] <wjp> I kind of doubt that'll work that easily
[11:35:19] <dominus> I have no idea and way to test, I lifted it off Marzo's changes in Exult
[11:36:00] <wjp> it's really too ad-hoc for my taste to have such a general name as --enable-static-libraries
[11:36:36] <wjp> but I just don't know
[11:37:20] <dominus> should I pull it or make it OS X only for now? maybe you can ask marzo when he shows up in #exult
[11:38:42] <wjp> it's from Exult's configure.ac? hmm
[11:38:51] <dominus> yes
[11:39:41] <dominus> line 301 following and 366 following
[11:39:41] <wjp> I think I'd prefer it to be OS X-specific at this point
[11:39:52] <wjp> you can then also reduce it to a single option
[11:40:09] <dominus> yes, I think I can manage this
[11:40:11] <wjp> it needs more research and testing in general
[11:40:15] <wjp> ok, thanks
[11:40:35] <dominus> will do it later today, need to do some other stuff now
[11:42:05] <wjp> the interaction with PENT_CHECK_SDLTTF is also interesting, since that already checks if linking against freetype manually is necessary
[11:43:38] <wjp> and I think the static stuff should be moved to above all library detection code
[11:43:56] <wjp> (and it should be verified that the tests then test if static linking works)
[11:44:23] <dominus> so we check freetype twice with my change. hmm, I have to admit that PENT_CHECK stuff was way over my head mostly
[11:44:28] <wjp> but doing this properly will take quite some effort
[11:45:14] <wjp> (hence doing only the OS X-case for now sounds good)
[11:45:43] <dominus> ok, will do, especially with the OS X only case
[11:47:01] <dominus> freetype is something I need to check special for the OS X building, since you *can* build static on OS X with freetype still linked to dynamic
[11:52:59] <dominus> we might need to talk about this with Exult
[12:42:30] --> spioH has joined #pentagram
[12:42:30] --- spioH is now known as Tori_
[13:22:45] <dominus> my head hurts from configure stuff :)
[13:24:49] <wjp> a common sentiment :-)
[14:14:38] <dominus> the os x with satic libs needs to be below the libs checking or it will fail with the SDL-Config part, since that is defined by PENT_CHECK_SDL
[14:15:05] <-- Colourless has left IRC (Ping timeout: 255 seconds)
[14:15:53] <wjp> for now that's ok
[14:16:24] <dominus> ok, static works, now I need to test normal compile :)
[14:17:01] <wjp> I suppose we may want to change PENT_CHECK_SDL to ask for --static-libs depending on some option
[14:18:33] --> Colourless has joined #pentagram
[14:18:33] --- ChanServ gives channel operator status to Colourless
[14:19:32] <dominus> wjp, does that make sense?:if test x$with_macosx_static_lib_path == x -o x$HOST_SYSTEM != xMACOSX; then
[14:20:19] <dominus> it seems to work at least :)
[14:20:49] <wjp> what do you do after that then?
[14:21:00] <dominus> GUI_LIBS="$GUI_LIBS $SDL_LIBS"
[14:21:20] <wjp> ah, there
[14:21:24] <wjp> yes, sounds ok
[14:21:26] <dominus> so that either when with_static... is not defined or when we are not on os x
[14:22:21] <dominus> ok, then I think I got it beat into shape. testing a bit though before comitting :)
[14:25:07] <wjp> great
[14:38:25] <dominus> and yes, if I don'T change ./configure builtindata settings, data.h gets corectly cleaned off the system
[14:46:40] <dominus> hmm, strange, here it is again. Now I'm puzzled
[14:48:17] <wjp> oh... I think I know
[14:48:59] <dominus> seems like it gets created even when builtindata is set to no
[14:49:15] <wjp> maybe the automatic dependency caching makes make think data.h needs to be recreated
[14:49:53] <wjp> we may want to delete all cached dependencies anyway on 'make clean'
[14:51:24] <wjp> does it only appear the first time you build after disabling builtindata?
[14:51:57] <dominus> yes, I was just about to write that since I couldn't reproduce on a second attempt :)
[15:53:11] <-- Colourless has left IRC (Ping timeout: 255 seconds)
[15:55:49] --> Colourless has joined #pentagram
[15:55:49] --- ChanServ gives channel operator status to Colourless
[17:54:49] <wjp> just tried in Linux, and I indeed get a bunch of linking errors with --enable-static-libraries before your last commit
[17:56:16] <wjp> it does work when I use libtool to link now
[17:56:56] <dominus> we use that with Exult, right?
[17:57:40] <dominus> so the configure option makes sense with Exult but not with Pentagram
[17:58:10] <dominus> btw, good timing, I just returned to the machine for just some minutes :)
[18:04:32] <wjp> :-)
[18:04:53] <wjp> it still should be integrated with the library tests, though
[18:06:31] <dominus> for exult?
[18:07:16] <-- Tori_ has left #pentagram
[18:07:25] <dominus> <- gone again :)
[23:35:34] <-- dominus has left IRC (Quit: Leaving.)
[23:38:46] --> Kirben has joined #pentagram
[23:38:46] --- ChanServ gives channel operator status to Kirben