#gemrb@irc.freenode.net logs for 12 Jul 2011 (GMT)

Archive Today Yesterday Tomorrow
GemRB homepage


[00:22:48] <-- brad_a has left IRC (Quit: brad_a)
[00:38:54] --> brad_a has joined #gemrb
[00:39:51] <-- brad_a has left IRC (Remote host closed the connection)
[00:39:58] --> brad_a has joined #gemrb
[00:42:59] <-- duckpunch has left IRC (Read error: Connection reset by peer)
[00:43:49] --> duckpunch has joined #gemrb
[00:50:04] <-- brad_a has left IRC (Ping timeout: 240 seconds)
[03:42:31] <-- duckpunch has left IRC (Read error: Connection reset by peer)
[03:43:16] --> duckpunch has joined #gemrb
[06:36:52] --> lynxlynxlynx has joined #gemrb
[06:36:53] <-- lynxlynxlynx has left IRC (Changing host)
[06:36:53] --> lynxlynxlynx has joined #gemrb
[06:36:53] --- ChanServ gives channel operator status to lynxlynxlynx
[06:39:45] --> harijan2 has joined #gemrb
[07:11:47] <-- |Cable| has left IRC (Remote host closed the connection)
[07:29:14] --> bar11 has joined #gemrb
[09:41:33] <-- duckpunch has left IRC (Read error: Connection reset by peer)
[09:42:36] --> duckpunch has joined #gemrb
[10:54:02] --> Tonurics has joined #gemrb
[12:01:56] <-- duckpunch has left IRC (Quit: Lost terminal)
[12:40:00] <-- lynxlynxlynx has left IRC (Ping timeout: 276 seconds)
[13:01:27] --> lynxlynxlynx has joined #gemrb
[13:01:27] <-- lynxlynxlynx has left IRC (Changing host)
[13:01:27] --> lynxlynxlynx has joined #gemrb
[13:01:27] --- ChanServ gives channel operator status to lynxlynxlynx
[13:43:09] --> |Cable| has joined #gemrb
[14:32:03] <-- bar11 has left #gemrb
[15:03:25] --> brad_a has joined #gemrb
[15:44:36] <-- brad_a has left IRC (Ping timeout: 276 seconds)
[15:50:03] --> brad_a has joined #gemrb
[16:27:43] <brad_a> the problem with font colors on SDL 1.3 seems to be that this line sprBuffer->SetPalette( newPal ); in Font::PrintFromLine has no effect
[16:29:30] <brad_a> and SDL doesn not throw an error either. and the corresponding call to sdl in sdlvideo returns with success
[16:31:01] <brad_a> yeah commenting out that line and running on sdl 1.2 provides the same result
[16:34:40] <wjp> hm
[16:34:49] <brad_a> i even tried changing SDL_SetPalette to the SDL 1.3 prescribed SDL_SetPaletteColors but no change
[16:35:03] <wjp> the SDL_compat version just does the same
[16:35:23] <brad_a> yeah i know
[16:35:30] <brad_a> i didnt realize till i tried it tho
[16:35:47] <-- lynxlynxlynx has left IRC (Ping timeout: 240 seconds)
[16:35:53] <brad_a> i went poking in the sdl source to see if i could figure out why
[16:51:53] --> lynxlynxlynx has joined #gemrb
[16:51:53] <-- lynxlynxlynx has left IRC (Changing host)
[16:51:53] --> lynxlynxlynx has joined #gemrb
[16:51:53] --- ChanServ gives channel operator status to lynxlynxlynx
[17:05:44] * wjp builds SDL 1.3
[17:06:40] <-- brad_a has left #gemrb
[17:06:50] --> brad_a has joined #gemrb
[17:15:49] <wjp> now, how do I convince cmake to pick up the right SDL? I deleted the cache, the new sdl-config is the first in my path and the new sdl.pc is in the first dir in my PKG_CONFIG_PATH... grrr...
[17:16:19] <fuzzie> i'm pretty sure cmake ignores sdl-config
[17:16:37] <wjp> argh, it has a hardcoded list of paths
[17:16:49] <wjp> so apparently there's a $SDLDIR
[17:17:01] <fuzzie> the best bit is the hardcoded list of python versions
[17:17:31] <wjp> so how exactly did they decide this was a better idea than autoconf again?
[17:18:12] <fuzzie> i expect it was about the point where they realised they could grep for #include rather than doing actual dependency detection?
[17:18:23] <wjp> ?!
[17:18:54] <fuzzie> i'm pretty sure i have grumped in here before about how cmake does crazy rebuilds because it ignores things like #ifdef UNNECESSARY_INCLUDES
[17:19:52] * wjp is rapidly cycling through states of disbelief, annoyance and frustration
[17:21:16] <wjp> while I'm asking stupid cmake questions...
[17:21:39] <wjp> is there any way we can emulate ./configure --help ?
[17:22:57] --> Avenger has joined #gemrb
[17:22:58] <wjp> libSDL-1.3.so.0 => /home/wjp/pkgs/SDL-1.3hg/lib/libSDL-1.3.so.0 (0x00007f74f139f000)
[17:23:01] <wjp> hurray
[17:23:02] <Avenger> hi
[17:23:06] <wjp> hi
[17:23:27] <wjp> (the hurray was just because I scored a small victory over cmake's attempts to confuse me)
[17:23:32] <Avenger> fuzzie, wjp, anyone: your efforts to support autoconf again would be welcome :)
[17:23:40] <fuzzie> doesn't it work?
[17:23:45] <Avenger> i have my own gripes against cmake too
[17:23:56] <wjp> at some point I got the impression that "everyone" was using cmake
[17:24:07] <wjp> (for gemrb, that is)
[17:24:12] <Avenger> if i create a new file (like a new 2da) it isn't installed until i manually touch cmakefile
[17:24:22] <Avenger> i use cmake ONLY because automake doesn't work
[17:24:29] <fuzzie> well, because we shouldn't be using wildcards, we should be using lists of files
[17:24:54] <Avenger> automake worked with wildcards, i think
[17:25:05] <wjp> brad_a: which font issues do you get with SDL 1.3?
[17:25:24] <Avenger> anyway... that's my only problem with cmake, otherwise it works, on linux
[17:25:28] <-- Avenger has left IRC (Client Quit)
[17:29:55] <brad_a> wjP: font issues are start a dialogue with somebody and all the words are the default color
[17:30:06] <brad_a> even the coices at the bottom that should be red
[17:32:28] <wjp> brad_a: it's fine here
[17:33:30] <brad_a> :\
[17:34:28] <fuzzie> nothing relevant changed in your tree?
[17:35:37] <brad_a> im going to try building the latest sdl 1.3 even though i doubt that will fix it
[17:36:23] <brad_a> nothing relavant between my code and git and this problem has existed since the first iOS build
[17:37:19] <fuzzie> i don't know if there are opportunities for alignment issues in there
[17:38:07] <brad_a> well the thing is that setpallete obviously works otherwise the text would be messed up. the problem seems to be setting it repeatedly or something. im not entirely sure
[17:38:48] <fuzzie> oh, it's probably just bad CreatePalette behaviour actually
[17:39:25] <fuzzie> i have dim memories of a problem there
[17:39:33] <brad_a> i would expect a bad pallete to have sdl either throw an error or return -1 but it doesnt
[17:39:58] <brad_a> not saying you are wrong of course just saying that sdl seems to also have a problem
[17:40:23] <fuzzie> i guess
[17:41:41] <fuzzie> weird
[17:41:47] <brad_a> very weird
[17:42:05] <brad_a> trying newest sdl 1.3 now...
[17:43:09] <brad_a> nope no diffrent
[17:43:28] <brad_a> wjp: are you absolutely sure it is linked against 1.3?
[17:43:42] <fuzzie> you're trying on OS X?
[17:44:39] <brad_a> just iOS. ill try os x thats a good idea
[17:46:38] <wjp> brad_a: I pasted the ldd line above
[17:46:54] <wjp> so yes :-)
[17:47:46] <brad_a> ah didnt see that
[17:51:39] <brad_a> can i assume you were building agains the current git source?
[17:52:24] <wjp> updated gemrb git and SDL 1.3 hg right before building
[17:53:14] <wjp> and for what it's worth, I'm on 64 bit Linux
[17:56:53] <brad_a> i will try both 32 bit and 64 bit on os x
[17:57:40] <brad_a> tho i am please that my horizontal mousewheel scoll events work as i hoped. tho a bit slow
[17:58:02] <brad_a> i get the same problem on 32 bit os x
[17:58:06] <brad_a> trying 64 bit
[17:59:34] <brad_a> same issue
[18:00:39] <brad_a> i have read a couple times it the sdl documentation were things work in linux but nothing else
[18:09:41] <wjp> hm, actually, it's broken here too
[18:10:23] <brad_a> well then
[18:10:25] <wjp> it seems I had two copies of 1.3 around, and the earlier one was half a year or so old
[18:10:35] <brad_a> wow
[18:10:48] <wjp> maybe a bit more, dunno
[18:11:04] <brad_a> so then do we think its an SDL bug or are we doing something "wrong"
[18:13:25] <wjp> I suppose I could go and bisect SDL
[18:14:16] <brad_a> ug lol
[18:19:50] <wjp> not made any easier by their having generated files in hg
[18:21:09] <wjp> hm, assertion failed in gemrb
[18:21:19] <wjp> gemrb: /data/gemrb/git/gemrb/gemrb/core/Scriptable/Scriptable.cpp:514: void Scriptable::ProcessActions(): Assertion `CurrentActionTicks == 0 && CurrentActionState == 0' failed.
[18:24:53] <lynxlynxlynx> that's a recent addition
[18:25:38] <lynxlynxlynx> fuzzie will want to know details
[18:25:50] <wjp> just got it again
[18:27:32] <wjp> fuzzie: if you have my set of BG2 saves, it happens reproducibly when loading the early one called "Sewers"
[18:29:06] <fuzzie> phft
[18:31:44] <fuzzie> how odd
[18:34:05] <lynxlynxlynx> original saves or from gemrb?
[18:34:10] <fuzzie> original
[18:34:19] <fuzzie> i can't run gemrb now
[18:34:19] <lynxlynxlynx> ;/
[18:34:27] <fuzzie> the source all looked ok though
[18:36:11] <fuzzie> except, not
[18:36:28] <fuzzie> AttackReevaluate does the wrong thing and could lead to fail
[18:36:46] <fuzzie> same for AttackOneRound
[18:38:55] <fuzzie> and i guess Resurrect/Die in theory
[18:45:34] <wjp> brad_a: of course the SDL commit that breaks it is a rather large one...
[18:45:45] <wjp> changeset: 5288:d4381f3b0d1e
[18:46:08] <wjp> three things lumped together in one commit too...
[18:49:44] <brad_a> ewww...
[18:51:06] <wjp> hm, this looks fishy
[18:51:09] * wjp tries something
[18:53:02] <wjp> huh, this is stupid
[18:53:13] <brad_a> do tell
[18:53:44] <wjp> let me figure out if "this" is me or SDL first :-)
[18:55:39] --> Beh0lder has joined #gemrb
[18:56:05] <Beh0lder> hi all
[18:56:44] <brad_a> hey
[19:00:59] <wjp> right, it's stupid
[19:01:10] <wjp> when updating the palette it increases the version of the palette
[19:01:18] <brad_a> yeah thats what i saw
[19:01:20] <wjp> but when blitting it only checks _the target's_ palette
[19:01:35] <wjp> so it completely misses the update to the source palette and re-uses the cached blit map
[19:01:57] --> PixelScum has joined #gemrb
[19:01:57] <-- BaldimerBrandybo has left IRC (Read error: Connection reset by peer)
[19:02:02] <brad_a> so easy fix then
[19:02:09] <wjp> it is?
[19:02:12] <brad_a> i dont know
[19:02:14] <wjp> I must be missing something
[19:02:15] <brad_a> you tell me
[19:02:18] <wjp> ah
[19:02:23] <wjp> you were just guessing, ok :-)
[19:02:26] <wjp> pity :-)
[19:02:28] <brad_a> lol
[19:02:33] <brad_a> yeah
[19:02:34] <fuzzie> well, you make it sound so easy
[19:03:09] <brad_a> yeah you made it sound like you just need an additional check or something
[19:03:47] <wjp> well, yes, but what needs to be checked against what? :-)
[19:03:57] * wjp ponders
[19:05:52] * wjp ponders some more
[19:10:26] <brad_a> im guessing we cant alter the version from sdlvideo...
[19:11:13] <wjp> ah, look: http://bugzilla.libsdl.org/show_bug.cgi?id=1224
[19:11:45] <brad_a> hazah!
[19:12:38] <brad_a> i searched for bugs but came up empty. what search did you use to find that?
[19:13:07] <wjp> source palette map :-)
[19:13:13] <fuzzie> filed a month ago? nice to know SDL 1.3 is a nice reliable future :p
[19:13:22] <brad_a> he he
[19:13:39] <wjp> patch works for me
[19:13:43] <brad_a> sweet
[19:13:45] <brad_a> thank you so much
[19:14:22] <brad_a> add yet another hack in SDL for the iOS build lol
[19:15:05] <brad_a> tho this one is hopefully going to be commited and is a good fix unlike the other changes ive had to make
[19:17:08] <wjp> from what I can tell this patch makes sense
[19:17:58] <brad_a> yeah. ive had to do some ugly hacks for landscape mode and retina display support
[19:19:04] <brad_a> tho i think i found out a way to do away with the retina display hack
[19:26:14] <brad_a> that was some good detective work sir
[19:26:21] <brad_a> works beautifuly
[19:49:16] <fuzzie> Lightkey points out that the SDL bug report was filed by bgK, i guess the world of people who care about palettes is a fairly small world
[20:13:45] --> lynxlynxlynx_ has joined #gemrb
[20:14:07] <-- lynxlynxlynx has left IRC (Read error: Operation timed out)
[21:04:57] <-- Beh0lder has left #gemrb
[21:08:37] <-- lynxlynxlynx_ has left IRC (Remote host closed the connection)
[22:51:56] --> jschall has joined #gemrb
[23:14:54] <-- brad_a has left IRC (Quit: brad_a)