#gemrb@irc.freenode.net logs for 6 Apr 2011 (GMT)

Archive Today Yesterday Tomorrow
GemRB homepage

[00:05:57] <-- Maighstir has left IRC (Quit: ~ Trillian Astra - www.trillian.im ~)
[01:04:02] --> pupnik_ has joined #GemRb
[01:06:06] <-- pupnik has left IRC (Read error: Operation timed out)
[01:16:37] <-- ar has left IRC (Ping timeout: 276 seconds)
[01:16:39] --> ar__ has joined #GemRb
[01:16:53] --- ar__ is now known as ar
[01:16:57] <-- ar has left IRC (Changing host)
[01:16:58] --> ar has joined #GemRb
[01:27:35] --> ar__ has joined #GemRb
[01:33:56] <-- ar has left IRC (Ping timeout: 240 seconds)
[01:33:56] --- ar__ is now known as ar
[01:33:57] <-- ar has left IRC (Changing host)
[01:33:57] --> ar has joined #GemRb
[02:17:18] --> Bo_Thomsen has joined #GemRb
[02:18:39] <-- Bo_Thomsen has left IRC (Client Quit)
[03:19:13] --- pupnik_ is now known as pupnik
[05:50:48] <pupnik> how do i clear the SDL keyboard input value in key->keysym.sym
[05:52:59] <pupnik> i have to make the keypress invisible to the following code, if a certain condition is met
[05:53:13] --> Maighstir|phone has joined #GemRb
[05:53:57] <-- Maighstir|phone has left IRC (Client Quit)
[06:50:07] --> lynxlynxlynx has joined #GemRb
[06:50:07] <-- lynxlynxlynx has left IRC (Changing host)
[06:50:07] --> lynxlynxlynx has joined #GemRb
[06:50:07] --- ChanServ gives channel operator status to lynxlynxlynx
[06:59:42] <pupnik> key->keysym.sym = SDLK_AT; // overwrite pressed key value with something else
[07:15:55] --> adominguez has joined #GemRb
[07:24:58] --> kettuz has joined #GemRb
[07:30:23] <edheldil> pupnik: either it's an event in which case you simply consume it, or it's a key state in which case clearing it makes no sense, no?
[07:33:48] <pupnik> well the rest of the code interprets keys with a complex set of kemaps
[07:33:56] <pupnik> which i can't understand
[07:34:06] <pupnik> so i'm catching the key events ahead of it
[07:34:09] <pupnik> and modifying
[07:34:20] <pupnik> and it works, yaay
[07:39:19] --> lubos has joined #GemRb
[08:44:46] <pupnik> http://i.imgur.com/g8D2u.jpg want the ramen? you CAN'T HANDLE THE RAMEN!
[08:49:03] <-- |Cable| has left IRC (Remote host closed the connection)
[08:51:40] <edheldil> I don't get the point
[08:52:18] <pupnik> i have lots of ramen
[08:53:22] <pupnik> actually have a question re gemrb - has anyone done a mod to the client to render the map background to an opengl (openglES) surface?
[08:54:48] <fuzzie> it isn't that easy
[08:58:17] <edheldil> do you have to have a rect for each tile or can you use tiles as multiple textures for a single mesh?
[08:58:19] <pupnik> never is i guess
[08:59:03] <pupnik> i figured you could render all the tiles of the map region once
[08:59:10] <pupnik> texture memory is huge nowadays
[08:59:34] <pupnik> then get variable zoom with one call
[08:59:47] <edheldil> do you mean creating a single huge texture for the whole map? I would not be so sure it would work
[08:59:57] <pupnik> i'm just askin
[09:00:33] <edheldil> I suspect you would hit texture size limits, at least on older cards
[09:01:25] <edheldil> but I have only tiny experience with OpenGL
[09:02:20] <edheldil> Maybe fuzzie will RE it for you ;-))))
[09:03:09] <fuzzie> openglES is not likely to be in a situation with a lot of memory :p
[09:04:09] <edheldil> but in worst case, I see nothing bad with a rect mesh for each tile
[09:05:58] <fuzzie> we don't cope very well in some places too
[09:06:33] <fuzzie> when i was looking at opengl, the first thing i looked at was MOS
[09:06:40] <pupnik> MOS?
[09:06:50] <fuzzie> which is used for the world map, for example
[09:06:58] <pupnik> yeah you're probably right about gles and texture mem
[09:07:00] <fuzzie> but we render the whole thing into one huge 32bpp buffer
[09:07:18] <pupnik> i was just playing 0.6.4 on the 3.7" screen and wished for seamless scaling
[09:07:35] <fuzzie> and if you take say the 4776*3000 worldmap from the BP-BGT worldmap, that is 57mb just for the worldmap
[09:08:43] <pupnik> wh we only have 24MB on the powervr sgx
[09:08:54] <fuzzie> or, 201mb, if you pick the huge one
[09:09:06] <edheldil> for this, an opengl plugin would internally slice the map into squares and use them as textures for parts of the mesh. Still would eat memory, though
[09:09:13] <fuzzie> well, no
[09:09:19] <fuzzie> for this, we fix the MOS importer to be less stupid :P
[09:09:56] <fuzzie> but that is as far as my experimentation got :/
[09:10:14] <edheldil> eh, MOS is tiled. right?
[09:10:19] <fuzzie> yep
[09:10:23] <pupnik> cool to know it's kickin around in your brain fuzzie
[09:10:44] <fuzzie> 256-color tiles
[09:12:01] <edheldil> so a possible plan would be to extend MOS importer to act as a tile factory and change the map control to use tiles instead of a single sprite ...
[09:12:13] <fuzzie> yes, that is probably a very sensible approach
[09:12:43] <fuzzie> i was hoping you could just render paletted textures directly, on opengles
[09:12:48] <fuzzie> but apparently all the drivers are really slow at it
[09:15:24] <fuzzie> so you'd probably need the opengl plugin to be smart about how much it keeps around
[09:16:51] <fuzzie> and this is about where it gets to my 'i hate everything' stage
[09:17:06] <pupnik> yes
[09:17:23] <pupnik> blitting to textures on a lot of these things is slow, they're not designed for it
[09:17:39] <fuzzie> they don't seem designed for very much
[09:18:07] <fuzzie> apparently preserving colour buffers is hard too
[09:18:41] <pupnik> powervr has 'fixed' this for the next gen
[09:20:32] <fuzzie> powervr being such nice trustworthy open people
[09:20:39] <fuzzie> who would never lie through their teeth about their hardware
[09:49:28] --> boriskr has joined #GemRb
[09:54:12] <lynxlynxlynx> heh
[10:32:51] <pupnik> i dunno, i'm pretty impressed with 3D on a battery powered device
[10:33:23] <pupnik> they just designed the sgx around a pretty specific use-case
[10:48:23] <-- kettuz has left IRC (Quit: Leaving)
[11:23:23] --- ChanServ gives channel operator status to wjp
[13:06:26] --> Bo_Thomsen has joined #GemRb
[13:19:36] <-- boriskr has left IRC (Read error: Connection reset by peer)
[14:56:31] <-- MikeChelen has left IRC (Read error: Connection reset by peer)
[15:03:26] <-- pupnik has left IRC (Quit: Reconnecting)
[15:03:41] --> pupnik has joined #GemRb
[15:03:41] <-- pupnik has left IRC (Changing host)
[15:03:41] --> pupnik has joined #GemRb
[15:11:45] --> test32894789234u has joined #GemRb
[15:15:13] --- Gekz_ is now known as Gekz
[15:52:16] <-- lubos has left IRC (Quit: Leaving.)
[16:03:19] --> Maighstir has joined #GemRb
[17:07:55] <-- adominguez has left IRC (Quit: Saliendo)
[17:26:44] --> boriskr has joined #GemRb
[17:46:17] --> |Cable| has joined #GemRb
[18:05:31] <boriskr> is it possible to continue a saved game from gemrb (ubuntu) in original game?
[18:08:30] --> kettuz has joined #GemRb
[18:13:27] <lynxlynxlynx> maybe
[18:13:48] <lynxlynxlynx> depends on the game type and if we broke something
[18:14:19] <lynxlynxlynx> we try to be compatible though, so just try it
[18:17:44] <boriskr> i suppose save procedure is broken. original game hangs at 50% loading
[18:20:15] <boriskr> but in gemrb yesterday build seems to work fine, monsters dont respawn any more
[18:21:08] <lynxlynxlynx> :)
[19:01:33] <-- test32894789234u has left #GemRb
[19:14:57] <-- Maighstir has left IRC (Quit: ~ Trillian Astra - www.trillian.im ~)
[19:50:34] --> edheldil_ has joined #GemRb
[19:54:01] <-- kettuz has left IRC (Quit: Leaving)
[19:54:38] --> kettuz has joined #GemRb
[20:50:44] --> Avenger has joined #GemRb
[20:50:44] --- ChanServ gives channel operator status to Avenger
[20:50:57] <Avenger> tomprince, the latest cleanup broke compilation on msvc6
[20:51:38] <Avenger> CachedFileStream doesn't compile anymore
[20:52:34] <fuzzie> huh
[20:53:09] <fuzzie> really?
[20:53:38] <Avenger> first problem is with: _FILE* str;
[20:53:47] <fuzzie> oh, wait
[20:53:56] <fuzzie> that's because CachedFileStream is gone
[20:54:03] <fuzzie> i wonder why git didn't delete the file
[20:54:07] <Avenger> oh as in completely remove?
[20:54:09] <Avenger> ok
[20:54:29] <Avenger> this is because when i copy over files to windows i don't delete old files
[20:55:02] <Avenger> what about CachedFileStream .h?
[20:55:04] <fuzzie> but we didn't merge the latest fixes into trunk i see, i forgot
[20:55:13] <fuzzie> CachedFileStream is completely gone now
[20:55:23] <Avenger> the h file wasn't deleted, i think
[20:55:31] <fuzzie> and the rest should build with msvc6 because tomprince tried it
[20:55:44] <Avenger> MemoryStream is gone too?
[20:55:49] <fuzzie> yeah
[20:56:11] <fuzzie> i guess CachedFileStream.h must still be around
[20:56:13] <Avenger> oh wait, the .h file was probably touched by me
[20:57:06] <fuzzie> no, it is gone too.
[20:57:44] <fuzzie> but someone should merge the last fixes
[20:57:53] <lynxlynxlynx> as-is?
[20:57:57] <fuzzie> right now it breaks if you have a .sav with duplicate files in it
[20:57:59] <fuzzie> yes, i think so
[20:58:20] <Avenger> wtf is CacheCompressedStream
[20:58:43] <fuzzie> it takes a compressed stream, and makes a cached file out of it
[20:58:45] <Avenger> and where is it
[20:58:55] <Avenger> i see references only, meh
[20:59:04] <fuzzie> oh i guess you're all out-of-date for new files too
[20:59:25] <lynxlynxlynx> trying them out
[20:59:29] <Avenger> ahh FileCache.cpp
[20:59:29] <tomprince> core/FileCache.{h,cpp}
[20:59:46] <lynxlynxlynx> Avenger: are you from budapest?
[20:59:53] <fuzzie> you also want SlicedStream.cpp
[21:00:02] <Avenger> i found that one
[21:00:05] <Avenger> any other?
[21:00:25] <Avenger> yes lynx
[21:00:47] <lynxlynxlynx> i'll be there for a few days in july
[21:01:01] <Avenger> cool ;)
[21:01:39] <lynxlynxlynx> err, make that june
[21:02:12] <lynxlynxlynx> we could be all awkward and meet :)
[21:03:35] <lynxlynxlynx> re saves: yep, that fixed it
[21:04:34] <fuzzie> well, it's still a worrying bug, why would you get an area in a .sav twice?!
[21:04:35] <Avenger> yep, write me a mail where you will be
[21:04:53] <Avenger> in the same sav? whoa
[21:05:03] <Avenger> is it duplicated in cache too?
[21:05:23] <fuzzie> right now gemrb explodes, but with the fixes it overwrites like it used to
[21:05:26] <Avenger> if yes, then blame case sensitivity
[21:05:52] <fuzzie> we strlwr() in the DecompressSaveGame code
[21:06:03] <fuzzie> but yes, i think it must be case sensitivity
[21:06:23] <CIA-52> GemRB: 03tom.prince * r86e63bc4b397 10gemrb/gemrb/ (core/FileCache.cpp plugins/BIFImporter/BIFImporter.cpp):
[21:06:23] <CIA-52> GemRB: FileCache: Add better error handling and reporting.
[21:06:23] <CIA-52> GemRB: If a file is already in the cache, we fall down trying to extract it
[21:06:23] <CIA-52> GemRB: again, and end up reading bogus data.
[21:06:23] <CIA-52> GemRB: Signed-off-by: Tom Prince <tom.prince@ualberta.net>
[21:06:24] <fuzzie> not good
[21:06:33] <CIA-52> GemRB: 03tom.prince * r74d6b5415714 10gemrb/gemrb/ (3 files in 2 dirs):
[21:06:33] <CIA-52> GemRB: BIFImporter: Overwrite files in cache, when extracting save game.
[21:06:34] <CIA-52> GemRB: Signed-off-by: Tom Prince <tom.prince@ualberta.net>
[21:06:46] <CIA-52> GemRB: 03lynxlupodian * r18ecca8d5f9e 10gemrb/gemrb/ (3 files in 2 dirs): Merge branch 'save-bugs' of https://github.com/tomprince/gemrb
[21:06:50] <fuzzie> but those two patches bring us back to how it used to work :P
[21:11:08] <Avenger> oh well, no how missiles for me today then, i gotta sync my sources
[21:11:12] <-- Avenger has left IRC (Quit: ChatZilla [Firefox 3.6.16/20110319135224])
[21:17:27] <CIA-52> GemRB: 03avenger_teambg * r0b3749c61f0b 10win32/MSVC6/GemRB/core/Core.dsp: synced project file for msvc6
[21:21:23] * tomprince whispers cmake and then runs away. :)
[21:39:41] <-- boriskr has left IRC (Read error: Connection reset by peer)
[21:47:58] <lynxlynxlynx> cool, managed to recover pst from a backup
[21:57:41] --> PixelScum has joined #GemRb
[22:00:12] <edheldil_> ehh, I spent time implementing mouse button repeat only to find it's already there, just usually disabled. Grr
[22:00:25] <-- BaldimerBrandybo has left IRC (Ping timeout: 252 seconds)
[22:00:52] <lynxlynxlynx> sure :)
[22:01:18] <lynxlynxlynx> we use it for distributing thief skills, where it is most notable
[22:01:32] <lynxlynxlynx> we don't do it like the original though and the discoverability is low
[22:04:21] <edheldil_> why not enable it on a per-control basis?
[22:15:13] <lynxlynxlynx> that's what we do now
[22:20:51] <edheldil_> how? I noticed only global setting being switched on & off
[22:21:28] <fuzzie> unset GEM_RK_DISABLE on the control
[22:21:36] <fuzzie> via SetRKFlags
[22:21:57] <fuzzie> oh, no, that is the wrong one isn't it
[22:22:50] <fuzzie> ok, well we *should* have that on the control :P
[22:23:50] <edheldil_> hehe that's what I wanted to implement :)
[22:24:06] --> BaldimerBrandybo has joined #GemRb
[22:24:07] <edheldil_> SetRepeatClickFlags() is global
[22:27:17] <-- PixelScum has left IRC (Ping timeout: 258 seconds)
[22:28:07] --> SiENcE has joined #GemRb
[22:28:11] <-- SiENcE has left IRC (Client Quit)
[22:44:25] <-- Bo_Thomsen has left IRC (Quit: Leaving.)
[23:38:47] <-- kettuz has left IRC (Quit: Leaving)