#pentagram@irc.freenode.net logs for 20 Mar 2005 (GMT)

Archive Today Yesterday Tomorrow
Pentagram homepage

[04:21:40] <-- Darke has left IRC (Read error: 110 (Connection timed out))
[06:05:53] <-- Lightkey has left IRC (Read error: 60 (Operation timed out))
[06:14:34] --> Lightkey has joined #Pentagram
[08:32:26] --> Colourless has joined #Pentagram
[08:32:26] --- ChanServ gives channel operator status to Colourless
[13:11:58] --> servus has joined #pentagram
[13:12:39] --- servus is now known as servus
[13:40:04] <-- Kirben has left IRC ("System Meltdown")
[14:32:13] --> Darke has joined #pentagram
[16:12:34] <watt> hmm... me wonders if the minimap should save it's info to something else.. closing and reopening as well a map changes clear it
[16:17:13] <Colourless> hey
[16:17:18] <Colourless> it's experimental :-)
[16:17:48] <Colourless> when it becomes less experimental the minimap data is probably going to be saved into CurrentMap
[16:18:16] * watt shrugs.. yeah. good point.
[16:20:44] * Colourless is currently doing something unheard of... working on getting sound into pentagram...
[16:21:01] <Colourless> did ya get it... sound... unheard of *waits for collective groans*
[16:21:44] * watt scratches head... groans or applause?
[16:22:41] <watt> Any opposition to making the instance of GUIApp in pentagram.cpp not be dynamically allocated?
[16:24:43] <Colourless> nope
[17:01:10] <watt> Is there a way for me to even directly call operator delete(void  ptr, Allocator  a)?
[17:02:10] <watt> I though "delete (someAllocator) class" would do it.. but, I'm wrong
[17:18:25] <Colourless> nope you can't
[17:19:12] <Colourless> only time a non default operator delete is called is when there is an unhandled exception during construction of the object
[17:19:43] <Colourless> call is a 'quirk' of C++
[17:19:51] <Colourless> s/is/it/
[17:29:29] <Colourless> well, probably 3/4 finished and pentagram is still linking so that is good
[17:30:03] <Colourless> but crashed on startup..
[17:30:57] <Colourless> ah, much better now
[17:37:28] <watt> I think it may be good to change SimplePool/SimpleAllocator to be SegmentedPool/SegmentedAllocator.... seems like a better name.
[18:06:00] <-- Lightkey has left IRC (Operation timed out)
[18:14:42] --> Lightkey has joined #Pentagram
[18:40:05] <Colourless> omg... i... i'm getting some sound
[18:40:11] <wjp> nice :-)
[18:40:44] <Colourless> but it's kind of jittery
[18:45:11] <wjp> hm, is the decompression code fast enough?
[18:45:34] <wjp> it's not overly optimized as it is
[18:45:39] <Colourless> i don't know...
[18:45:43] <Colourless> i'm trying a release build
[18:46:13] <Colourless> doing decompression and cubic resampling on the fly so might get a little slow
[18:47:26] <Colourless> SonarcAudioSample::decode_LPC() seems like it would be quick enough but SonarcAudioSample::decode_EC() is quite complex so not sure
[18:48:44] <Colourless> actually silly me, it's quite fast enough
[18:48:56] <Colourless> only the sfx are jittering, the music plays fine
[18:49:01] <Colourless> so there is a 'bug' somewhere
[18:50:03] * wjp shudders at the decode_EC code... there was a time when I understood that code :-)
[18:50:19] <wjp> (at least I hope so, considering that I wrote it :-) )
[18:50:29] <Colourless> hehe
[18:56:21] <Colourless> oops....
[18:56:27] * Colourless notices a stupid logic error
[19:01:40] <Colourless> yep working good now
[19:05:23] <Colourless> oh, one thing, i modified decode_LPC() so it wouldn't need the order bytes of 0 at the beginning of the buffer
[19:06:34] <Colourless> just makes it easier if we don't have to bother with a buffer that is larger than the amount of samples actually being produced
[19:07:20] <Colourless> now i probably should get rid of the memory leaks
[19:07:34] <watt> that'd be nice
[19:07:43] <watt> ;-)
[19:08:31] <wjp> yes, I see how getting rid of those would make things easier
[19:09:53] <wjp> oh, this is a first; we have a "support request" in our SF tracker
[19:10:50] <Colourless> we do?
[19:11:00] <wjp> https://sourceforge.net/tracker/index.php?func=detail&aid=1166624&group_id=53819&atid=471707
[19:11:10] <Colourless> hmm... from the mess of sound that is the devastation spell, still don't something wrong
[19:11:59] <Colourless> obviously they did not properly setup the paths in the ini :-)
[19:22:06] <Colourless> ok, i can confirm the sonarc decoding is working fine, it's my resampling that is broken
[19:23:01] <watt> commited some changes to pooling.
[19:34:06] <Colourless> oops...
[19:34:10] * Colourless found another error
[19:35:27] <Colourless> yay... everything sounds normal now
[19:38:19] <wjp> nice :-)
[19:38:40] * wjp is 'fixing' non-english intro in the meantime
[21:48:39] <Colourless> well, i think this is in a state that can be committed
[21:50:31] <wjp> great :-)
[21:56:03] <Colourless> committed... hopefully it's functional
[21:57:48] <wjp> let's have a look :-)
[22:01:21] <wjp> audio/AudioMixer.h:55: error: type specifier omitted for parameter `Uint8'
[22:01:35] <Colourless> sdl header
[22:02:15] <Colourless> you'll need to #include <SDL.h> in that file
[22:02:18] <wjp> does the callback have to be in the header?
[22:02:42] <Colourless> no probably not
[22:03:38] <Colourless> you'll need to make AudioMixer::MixAudio() function public though
[22:04:13] <Colourless> or... make it uint8 instead of Uint8
[22:06:37] <wjp> ok, it linked
[22:07:27] <wjp> and crashed
[22:07:32] <Colourless> crashed...
[22:07:36] <Colourless> not good
[22:07:46] <wjp> interesting, it segfaulted twice in a row :-)
[22:08:15] <wjp> (it segfaulted again right after the SDL parachute)
[22:08:45] <wjp> oh, maybe it doesn't like midi being disabled
[22:09:09] <wjp> yes, starts now
[22:09:26] <Colourless> yeah crashed here too...
[22:09:29] <Colourless> lets see what i did wrong
[22:09:45] <Colourless> getting sound?
[22:09:50] <wjp> 0x0827c1aa in Pentagram::AudioMixer::MixAudio(short*, unsigned) (
[22:09:50] <wjp> this=0x85dc450, stream=0x876ce90, bytes=16384) at audio/AudioMixer.cpp:221
[22:09:53] <wjp> 221 if (midi_driver->isSampleProducer())
[22:09:55] <wjp> yes, the fire is happily crackling :-)
[22:10:09] <wjp> "cheater! hahahaha!"
[22:10:31] <Colourless> just add in midi_driver!=0 &&
[22:11:22] <wjp> hm, footsteps will be interesting
[22:11:52] <wjp> I wonder how exactly they depend on the surface you're on
[22:12:22] <Colourless> could be a type flag... maybe
[22:13:13] <wjp> lots of places need to have hardcoded SFX, I'm afraid
[22:13:35] <wjp> will take some time before we get them all :-)
[22:13:50] <wjp> let's see... actor animations have an SFX field
[22:14:02] <wjp> that should take care of weapon sounds
[22:20:46] <Colourless> i'm off now. have fun :-)
[22:21:07] <-- Darke has left IRC (Read error: 110 (Connection timed out))
[22:21:22] <wjp> hm, I have swing sounds working :-)
[22:21:25] <wjp> night
[22:21:40] <-- Colourless has left IRC ("casts improved invisibility")
[22:24:05] --> Darke has joined #pentagram
[22:37:43] <wjp> I think a 4096 sample buffer is a bit high; latency is quite noticable
[23:07:52] <watt> neat!
[23:09:05] <watt> umm.. and is middle clicking thing supposed to delete them? I never noticed that before...