#gemrb@irc.freenode.net logs for 29 May 2011 (GMT)

Archive Today Yesterday Tomorrow
GemRB homepage


[00:48:24] <-- Maighstir has left IRC (Quit: ~ Trillian Astra - www.trillian.im ~)
[07:16:46] --> mihairu has joined #gemrb
[07:21:06] <-- mihairu has left IRC (Client Quit)
[07:21:23] --> mihairu has joined #gemrb
[07:34:20] --> lynxlynxlynx has joined #gemrb
[07:34:20] <-- lynxlynxlynx has left IRC (Changing host)
[07:34:20] --> lynxlynxlynx has joined #gemrb
[07:34:21] --- ChanServ gives channel operator status to lynxlynxlynx
[07:46:01] --> Maighstir has joined #gemrb
[08:03:10] <-- zpmorgan has left IRC (Remote host closed the connection)
[08:30:27] --> test32894789234u has joined #gemrb
[08:39:41] <lynxlynxlynx> woah
[08:39:50] <lynxlynxlynx> thanks kde
[08:42:15] <lynxlynxlynx> found a memory tweak and it halves the end memory usage of the minimal test case
[08:48:37] <lynxlynxlynx> works good for real data too
[08:48:52] <fuzzie> :)
[08:57:08] --> Avenger has joined #gemrb
[08:57:09] --- ChanServ gives channel operator status to Avenger
[08:57:11] <Avenger> hi
[08:58:40] <lynxlynxlynx> http://sprunge.us/UjBg?diff and http://pastebin.com/xH3bKwSF
[08:58:41] <lynxlynxlynx> oj
[08:58:57] <lynxlynxlynx> the first being just the gist of it, need to wrap it up nicely
[09:00:18] <lynxlynxlynx> considering the numbers, a smaller factor may be even better
[09:07:34] <lynxlynxlynx> using just one pagesize as a treshold shaves of additional 15m, but i'm not sure that's such a good idea
[09:09:10] <Avenger> this is more useful in small memory machines, i guess?
[09:10:06] <Avenger> do mallopt exist in every system?
[09:10:29] <fuzzie> no
[09:11:07] <fuzzie> i assume M_TRIM_THRESHOLD is glibc-only
[09:12:37] <Avenger> ok, i guess it exists where mallopt exist
[09:17:36] <Avenger> fuzzie, do you have a memory map for gui objects too?
[09:18:29] <Avenger> i started mapping the various hardcoded windows. like vtab_guiconn is at aaeb50
[09:19:03] <fuzzie> no, I haven't even tried looking at those
[09:19:05] <Avenger> it would be nice to have a hint about them
[09:19:22] <fuzzie> other than the game logic in the ScreenWorld
[09:19:35] <Avenger> actually, any vtab stuff is welcome
[09:19:41] <fuzzie> ask me on tuesday maybe :)
[09:19:47] <Avenger> ok
[09:19:56] <fuzzie> exam period at the moment
[09:20:14] <Avenger> ok, go fuzzie, beat the exam first :)
[09:29:00] <lynxlynxlynx> how come you're looking at windows? anything in particular you're after?
[10:03:00] <Avenger> i'm trying to find out how hotkeys/quickspells and such are done in the original
[10:03:19] <Avenger> also, this part of the engine i know the less about
[10:03:52] <Avenger> there are big structures that are totally unexplored
[10:04:00] <CIA-39> GemRB: 03lynxlupodian * r720a67b0ed18 10gemrb/ (CMakeLists.txt cmake_config.h.in gemrb/GemRB.cpp):
[10:04:00] <CIA-39> GemRB: significantly reduce the memory footprint on systems with glibc
[10:04:00] <CIA-39> GemRB: glibc only resizes the heap if it has more than 128KB of free space at the
[10:04:00] <CIA-39> GemRB: end, but many objects are smaller, so fragmentation occurs. This change
[10:04:00] <CIA-39> GemRB: tunes the limit down, resulting in much more reclaim.
[10:04:14] <lynxlynxlynx> hopefully nothing ugly hides in there
[10:05:29] <lynxlynxlynx> the last thing i was looking at was chunking and it was odd
[10:05:48] <Avenger> the graphic stuff?
[10:05:50] <lynxlynxlynx> the animation looks like each cycle has to be played independently
[10:06:08] <lynxlynxlynx> we also have it as a separate animation id
[10:06:10] <Avenger> yes, the cycles are random fragments with separate trajectory
[10:07:14] <fuzzie> and the actor's own animation id is also set, right?
[10:07:28] <Avenger> the animation id is weird, i think the lowest bits of the animation id determine the cycle
[10:07:43] <Avenger> so, a chunk is 4-5 animations
[10:07:54] <-- mihairu has left IRC (Read error: Connection reset by peer)
[10:08:09] <lynxlynxlynx> from what i remember of the original, there are plenty of chunks flying
[10:08:31] <lynxlynxlynx> so my guess is that just all the cycles are ran simultaneously
[10:08:37] <Avenger> actually 10
[10:09:40] <Avenger> you can simply make it x, where x is the cycle count of spchunks
[10:09:59] <Avenger> i thought the particles class already handles this
[10:10:41] <Avenger> only unknown part is the trajectory
[10:10:52] <Avenger> i have difficulties calculating it
[10:11:04] <lynxlynxlynx> maybe the fountain type is used for this
[10:11:13] <Avenger> yes, i planned that
[10:11:28] <Avenger> though, that one itself is half assed
[10:11:40] <Avenger> i think i improved it, but i'm not entirely satisfied
[10:12:42] <lynxlynxlynx> i'll try it
[10:12:45] <Avenger> anyway, for first attempt i would do it: 10 particles, one for each cycle in spchunks. Path is fountain
[10:13:31] <lynxlynxlynx> there's also spblood, which looks like it is used when the actor was frozen
[10:13:59] <Avenger> pst has no spchunks
[10:14:06] <Avenger> all other engine has it
[10:14:18] <lynxlynxlynx> for the stoned ones i guess the palette change persists and spchunks is good
[10:14:58] <Avenger> blood color has nothing to do with it?
[10:15:42] <Avenger> ah yes, there IS a chunk color
[10:15:47] <Avenger> separate from blood color
[10:16:09] <Avenger> we need to simply add it to avatars.2da similarly to the blood color
[10:16:28] <Avenger> i like that patch approach i did with the blood
[10:16:54] <lynxlynxlynx> ugh
[10:17:13] <Avenger> you didn't?
[10:17:40] <Avenger> it is easy to create a chunkclr.2da then do exactly the same
[10:17:53] <Avenger> i can dig up the color values
[10:18:44] <Avenger> btw, frozen/stoned also alters/overrides the blood color
[10:18:44] <fuzzie> the monster stuff seemed to only have two different values for colour chunks
[10:18:45] <lynxlynxlynx> i didn't know there was a separate one
[10:19:05] <Avenger> well, lynx, it was a new discovery
[10:19:12] <Avenger> fuzzie did it :)
[10:19:58] <Avenger> fuzzie: what do you mean by 2 different values
[10:20:17] <Avenger> i thought there is one gradient for blood, and another for chunks
[10:21:00] <fuzzie> Avenger: i mean, the monster anim (7xxx) code seemed to only use two different values
[10:21:12] <fuzzie> i think i probably missed a table or function or something
[10:21:36] <Avenger> ahh, sure you did
[10:21:36] --> budlust has joined #gemrb
[10:21:41] <Avenger> the slimes are all different
[10:21:52] <fuzzie> well, i also didn't check the old anim code
[10:21:56] <fuzzie> i think you called that 'c7xxx' or something
[10:21:58] <Avenger> i got the bloodclr.2da from studying the tob code
[10:22:14] <fuzzie> yeah, i can see the blood colours
[10:22:29] <Avenger> c7 is not 'old' but the clown colored types
[10:22:34] <fuzzie> it is old :P
[10:22:39] <Avenger> well, but it is used
[10:22:42] <fuzzie> sure
[10:22:46] <fuzzie> it's just called 'old' :)
[10:23:33] <fuzzie> it confused me because the normal 7xxx anim code handles all animations, but it gets all the 'old' ones wrong!
[10:23:43] <Avenger> yes
[10:23:49] <fuzzie> very annoying
[10:23:59] <Avenger> it is
[10:24:20] <Avenger> you should always make sure which path is selected
[10:24:37] --> mihairu has joined #gemrb
[10:24:37] <Avenger> and it is a complex code that determines it
[10:24:41] <Avenger> totally lame
[10:24:50] <fuzzie> yes
[10:25:19] <fuzzie> it calls 'CGameAnimationTypeMonsterOld::NewSupport' and then decides based on that, completely crazy
[10:25:31] <fuzzie> i guess they intended to move everything to the new code, and didn't finish it
[10:25:48] <Avenger> yeah, they got no fuzzie to refactor the whole mess
[10:26:08] <fuzzie> but it means that i don't have full information because i didn't find time to decode everything of the old stuff..
[10:26:34] <Avenger> i dug up the blood info relatively easily
[10:26:35] <fuzzie> just that in the *new* 7xxx stuff, it sets loads of different blood colours but only has some flag for chunk colours :)
[10:27:48] <Avenger> well, i couldn't find out what animation is used in pst, and i don't remember how it looks like either
[10:28:05] <fuzzie> well, pst is easy, because 95% of the anims aren't hardcoded
[10:28:13] <Avenger> i meant the chunks
[10:28:17] <fuzzie> ah!
[10:28:21] <fuzzie> i could look, but i guess not now :/
[10:28:26] <fuzzie> i thought i saw the chunk stuff
[10:28:34] <fuzzie> or, maybe that there wasn't chunk stuff?
[10:29:06] <Avenger> see, that's what i meant by 'dont remember' :)
[10:29:14] <fuzzie> ok, so i am useless too :P
[10:29:35] <fuzzie> i don't really want to go back to that stuff while scripting is still broken anyway
[10:31:47] <-- budlust has left IRC (Ping timeout: 240 seconds)
[10:35:00] <fuzzie> and the performance stuff / hashes / etc still need work I guess?
[10:42:13] --> budlust has joined #gemrb
[10:55:52] <lynxlynxlynx> no idea
[12:52:11] <-- Gekz has left IRC (Ping timeout: 246 seconds)
[12:58:45] --> Gekz has joined #gemrb
[12:58:45] <-- Gekz has left IRC (Changing host)
[12:58:45] --> Gekz has joined #gemrb
[13:15:43] --> barra_home has joined #gemrb
[13:30:54] --> budlust_ has joined #gemrb
[13:32:44] <-- budlust has left IRC (Ping timeout: 259 seconds)
[13:40:36] <-- |Cable| has left IRC (Read error: Connection reset by peer)
[13:40:49] --> |Cable| has joined #gemrb
[15:19:45] <-- Avenger has left IRC (Quit: ChatZilla 0.9.86.1 [Firefox 4.0.1/20110413222027])
[15:35:17] --> Avenger has joined #gemrb
[15:35:17] --- ChanServ gives channel operator status to Avenger
[15:35:53] <Avenger> heh, these gui objects have their own keymap, one dword array and one byte array, both 91 in size
[15:36:54] <Avenger> sub_9E3438 does the mapping on construction (at least something i would do the same way)
[15:43:05] <-- Gekz has left IRC (Ping timeout: 264 seconds)
[15:50:12] --> Gekz has joined #gemrb
[16:12:47] <-- mihairu has left IRC (Remote host closed the connection)
[16:30:58] <-- budlust_ has left IRC (Ping timeout: 255 seconds)
[16:33:00] --> budlust has joined #gemrb
[17:02:05] <-- Gekz has left IRC (Ping timeout: 246 seconds)
[17:26:56] --> Gekz has joined #gemrb
[17:26:56] <-- Gekz has left IRC (Changing host)
[17:26:56] --> Gekz has joined #gemrb
[17:31:30] <-- Avenger has left IRC (Quit: ChatZilla 0.9.86.1 [Firefox 4.0.1/20110413222027])
[18:05:42] --> fuzzie_ has joined #gemrb
[18:06:25] <-- fuzzie has left IRC (*.net *.split)
[18:06:25] <-- devurandom has left IRC (*.net *.split)
[18:08:00] --> devurandom has joined #gemrb
[18:09:15] --> mihairu has joined #gemrb
[18:10:03] --- fuzzie_ is now known as fuzzie
[18:10:08] --- ChanServ gives channel operator status to fuzzie
[18:32:59] <CIA-39> GemRB: 03avenger_teambg * r7f1236954656 10gemrb/ (4 files in 3 dirs): Merge branch 'master' of ssh://gemrb.git.sourceforge.net/gitroot/gemrb/gemrb
[18:33:01] <CIA-39> GemRB: 03avenger_teambg * ra45e0332e7ca 10gemrb/gemrb/plugins/GUIScript/GUIScript.cpp: reinitialize the button animation if it was played once
[18:37:10] <-- lynxlynxlynx has left IRC (Ping timeout: 260 seconds)
[18:38:34] <CIA-39> GemRB: 03avenger_teambg * r7ab6a9e94773 10gemrb/gemrb/GUIScripts/bg2/GUIMG.py: play the flash animation/sound in the spell book window
[18:52:33] <-- mihairu has left IRC (Quit: mihairu)
[18:52:45] --> mihairu has joined #gemrb
[18:57:10] <-- test32894789234u has left #gemrb
[18:58:40] <-- mihairu has left IRC (Remote host closed the connection)
[19:08:59] <-- budlust has left IRC (Quit: Lost terminal)
[21:55:25] <-- barra_home has left IRC (Quit: Verlassend)
[22:04:32] --> lynxlynxlynx has joined #gemrb
[22:04:36] <-- lynxlynxlynx has left IRC (Changing host)
[22:04:36] --> lynxlynxlynx has joined #gemrb
[22:04:36] --- ChanServ gives channel operator status to lynxlynxlynx
[22:18:25] <-- lynxlynxlynx has left IRC (Remote host closed the connection)