IRC logs for #opendune on OFTC at 2010-02-06
⏴ go to previous day
00:44:46 <Xaroth> my remark would be: "Doable: yes; when: dunno"
00:45:20 <Xaroth|Work> phone workage, w00p
01:18:14 <glx> after some tests, AB00 content depends on loaded driver so I won't commit anything :)
01:42:39 <glx> ok when using driver 2 (pcsound.drv) I can easily recognize pcsound.drv content in AB00
08:46:50 * Xaroth|Work in car en-route to france
08:48:02 <Xaroth|Work> glx beware, the dutchie is on his way!
12:59:55 <proyvind> someone wants to fix libeastwood/pyeastwood on windows today? ;)
13:01:29 <TrueBrain> still no Windows here :)
13:02:58 <TrueBrain> you should install it yourself I guess :)
13:12:21 <proyvind> I want glx to do it! ;p
13:12:29 <TrueBrain> so ask him nice,and give him something nice for it ;)
13:12:38 <proyvind> glx: 13:59:55 < proyvind> someone wants to fix libeastwood/pyeastwood on windows today? ;)
13:12:59 <proyvind> I think there's just one bug left with the PakFile class on windows..
13:13:38 <glx> the main problem for me is cmake :)
13:14:07 <proyvind> is it really that bad?
13:14:45 <glx> even when I tell it to use mingw it generates a makefile using cl (msvc compiler)
13:16:04 <proyvind> I thought you weren't able to use mingw with python (unless you built python yourself with it)..?
13:16:53 <glx> true, but I wasn't able to build python module either because I were using 64bit python (but now I have working MSVC 64bit compiler)
14:29:25 <glx> and some libs are a pain to compile with msvc (samplerate is one of them)
14:40:11 <glx> ok I finally have a project file
14:42:36 <TrueBrain> glx: how hard would it be to add additional code to the sound drivers (which are loaded in AB00) to add extra functionality to Dune2 itself?
14:42:39 <glx> but VocFile.cpp fails to compile
14:48:38 <glx> of course there's no hearable sound ;)
14:49:30 <glx> like the first change in 2756
14:50:05 <glx> decompiled is 0xAB000000, but when running without JIT it's 0x44AF0000
14:50:19 <TrueBrain> at 44AF most likely is the loader for AB00
14:50:30 <TrueBrain> for the overlay there is code which patches that up
14:50:34 <TrueBrain> but .. AB00 is not in that range ;)
14:51:07 <TrueBrain> what is going on in /decompiled/cs__2B6C.c
14:51:33 <TrueBrain> please apply the patch I gave you yesterday
14:51:43 <TrueBrain> the changes in decompiled.h are not acceptable
14:52:35 <TrueBrain> f__AB00_0000_0003_1E3A is just a wrapper around the real function
14:52:42 <TrueBrain> hmm .. then something is still wrong ..
14:53:00 <TrueBrain> owh, it only sorts the include tree
14:53:46 <TrueBrain> but okay .. any idea about my question?
14:54:31 <TrueBrain> and about 2B6C, as that is just plain weird
14:54:52 <glx> I don't know what happened with 2B6C
14:56:08 <TrueBrain> 609E is never used again
14:56:14 <TrueBrain> smells like a JIT fart
14:56:34 <TrueBrain> memory corruption of some kind
14:56:47 <glx> anyway as I said last night, I don't think it's a good idea to commit all this AB00 stuff
14:57:51 <TrueBrain> what I understood from you, you want to try to implement a audio driver, without implementing the emulation stuff, right?
14:57:57 <TrueBrain> so with this patch, you have enough information to do so, not?
14:58:18 <TrueBrain> then I suggest to commit none of this, as it kind of integrates itself with AB00 :)
14:58:56 <TrueBrain> suggestion for approach: if you want to make a function call native which requires a part of your current patch, you ask me, then I C-ify that function (you can do that too of course :p), and then you can add your handler?
14:59:39 <glx> f__B483_0283_0014_983A() <-- I can name this one :)
15:00:24 <TrueBrain> I am going to check the decompiler, why it keeps decompiling unneeded functions
15:03:19 <glx> is emu_Sound_Play() good enough ?
15:05:19 <glx> could be music too (as sound and music drivers are identical)
15:05:40 <Xaroth|Work> en-route home again
15:05:48 <TrueBrain> Xaroth|Work: hehe .. enjoy :)
15:06:02 <TrueBrain> glx: well ... find out ;)
15:06:06 <TrueBrain> add a printf, and it should show you
15:06:14 <TrueBrain> (turn off/on music for example :))
15:06:27 <Xaroth|Work> just as crap as last time
15:06:44 <Xaroth|Work> foggy as fook as well
15:07:18 <TrueBrain> here it is foggy too :)
15:07:30 <glx> else opendune didn't run without JIT due to some unresolved jumps
15:07:35 <TrueBrain> glx: so I expect working sound for 0.3 ;)
15:09:02 <TrueBrain> glx: anything I can help with, please do let me know
15:09:13 <TrueBrain> but the fact you know which function starts the playback, should make things relative easy :)
15:09:22 <TrueBrain> (as I assume it will be a simple sound-format)
15:09:43 <glx> that's the main unknown thing (the sound format)
15:10:28 <TrueBrain> the AB00 code looks very simple
15:11:00 <TrueBrain> I cant remember the kind of music, but for pcsound I expect frequenties :)
15:11:18 <TrueBrain> anything else, and I expect simple WAV format (without headers, most likely)
15:12:58 <DorpsGek> SVN: truebrain (r856) -Fix: remove now unused @defined functions
15:16:38 <glx> the main problem is to follow how it reads the data, because a big part is done via a timer interrupt
15:16:59 <TrueBrain> yeah, a new note every N time
15:17:21 <TrueBrain> but the timer is not always called at the same rate
15:19:07 <TrueBrain> okay, I found out why the decompiler poops out too many functions .. now a way to fix it .. hmm ..
15:19:17 <glx> f__2756_050F_000B_0871 seems to be some kind of timer interupt handler
15:19:45 <TrueBrain> it is the custom timer handler
15:19:56 <glx> and it execute a chain of handlers
15:26:55 <glx> hmm I could try to convert em_Sound_Play() instead just naming it :)
15:27:18 <glx> as it is already present in trunk
15:32:09 <TrueBrain> if the entry function of an overlay is used, I don't have the information to see if it is called :(
16:11:48 <TrueBrain> for some reason it keeps including useless functions .. and I don't get why :(
16:13:35 <TrueBrain> I was abusing how I had things thought out
16:40:30 <Xaroth> meh, in and out of france :P
16:40:52 <Xaroth> run in, grab what you need, and GTFO
16:41:17 <TrueBrain> finally .. the JIT produces 'sane' output now
16:41:21 <TrueBrain> removed tons of functions
16:41:27 <TrueBrain> not all for the better :p
16:41:58 <Xaroth> the worklist updated with the new numbers? or the 'amount' of functions still somewhat equal?
16:42:17 <TrueBrain> how can the amount of functions ever change? :p
16:42:21 <TrueBrain> it requires new functions ;)
16:42:33 <Xaroth> well, JIT making tons of 'double' functions that are identical
16:42:48 <TrueBrain> euh .. those are long long long gone
16:43:07 <Xaroth> at least we're at 16% conversion rate already :)
16:45:29 <TrueBrain> I hope my latest effort removes a few more
16:47:55 <TrueBrain> glx: LevelEnd always returns 1?
16:48:11 <TrueBrain> so lets remove the return value :)
16:48:50 <glx> hmm I get an unresolved jump in MSVC (debugger) but not in cmd
16:49:06 <glx> and of course the windows close too fast for me to read where
16:52:03 <TrueBrain> wrote off another 3 functions
16:52:23 <Xaroth> right, off to have dinner at in-laws
16:52:33 <TrueBrain> dinner .. there is an idea
16:53:36 <DorpsGek> SVN: truebrain (r857) -Fix: finally fixed the decompiler bug; now only functions which are used are considered in decompiled/
16:54:14 <glx> hmm for some weird reason it loaded a sound driver
16:54:27 <TrueBrain> 35% named, 16% C-ified
16:54:31 <glx> but only insinde the debugger
16:56:40 <TrueBrain> there .. that makes me happy :)
16:56:46 <TrueBrain> now only I want to get ride of @define stuff .. not sure yet
16:57:07 <TrueBrain> I think I scan all files for 'extern void f__'
16:57:10 <TrueBrain> and use that as 'defined'
17:03:19 <TrueBrain> variable_04 most likely contains the sound itself
17:03:52 <glx> I don't exactly what is variable_04
17:11:40 <TrueBrain> fuck ... it no longer wants to start ...
17:12:13 <TrueBrain> of course, I had your dune.cfg loaded :)
17:12:41 <DorpsGek> SVN: truebrain (r858) -Codechange: make the decompiler clever enough to understand 'extern void', which makes all @defines unneeded
17:13:34 <glx> I had the same problem with the wrong working path
17:19:51 <TrueBrain> hmmm .. I tried to reduce '@implements' lines
17:19:55 <TrueBrain> but it took a turn for the worst
17:19:58 <TrueBrain> first .. food .. maybe then I can think :p
17:22:14 <DorpsGek> SVN: glx (r859) -Add: named and C-ified Sound_Play()
17:25:41 <TrueBrain> hmm .. yes .. we did make one 'mistake'
17:25:44 <TrueBrain> if now new code pops up
17:25:47 <TrueBrain> for code we already C-ified
17:25:51 <TrueBrain> there is no way of knowing ;)
17:27:42 <glx> hmm but when we C-ify usually we remove unresolved jumps/calls
17:27:50 <TrueBrain> yes, but this is the other way around
17:28:14 <TrueBrain> for example, a new function, say in your sound driver, starts using f__1A34_21E0_001A_547E
17:28:19 <TrueBrain> what function is that C-ified?
17:29:16 <TrueBrain> ah, House_EnsureHarvesterAvailable :p
17:29:22 <glx> oh but we have function_names for that theorically
17:29:37 <glx> but we tend to remove functions from it once C-ified
17:29:47 <TrueBrain> the decompiler demands it :)
17:29:54 <TrueBrain> but okay .. I don't think we can expect many more new functions
17:30:04 <TrueBrain> else it will be searching :p
17:30:55 <glx> if I understand, the problem is for fully C-ified functions, the ones without emu_ wrapper
17:31:25 <TrueBrain> and if some new piece of code starts using them ;)
17:35:49 <TrueBrain> now to figure out why the decompiler still things a few functions are called, while they are not ...
17:58:38 <Xaroth|Work> she is also waiting for food
18:38:14 <glx> even if this code will never be C-ified, it helps to understand what's going
18:39:48 <DorpsGek> SVN: glx (r860) -Add: named some drivers related functions
18:40:10 <glx> luckily my openttd trunk dir was clean ;)
18:40:30 <glx> commit in the wrong window
18:43:58 <glx> I still need to find size of struct at 6344
18:44:22 <glx> I know it's music driver stuff, but I don't know its size
18:47:21 <glx> ,...emu_push(emu_cs); emu_push(0x0FBA); emu_cs = 0x01F7; emu_Tools_Memcopy();
18:47:32 <glx> hmm maybe an indication of 2 structs ;)
18:48:59 <TrueBrain> 0x26 in size, from 6302 to 6344
18:49:05 <TrueBrain> still has a small grey area between those 2 blocks
19:00:43 <TrueBrain> k .. done having dinner .. lets try to find why the decompiler is acting stupid again :)
19:09:10 <TrueBrain> valgrind reports problems .. that can be the reason :p
19:27:31 <TrueBrain> I have absolutely no idea how it could ever have worked
19:28:00 <TrueBrain> dumb luck comes to mind
19:30:48 <glx> luck is nice sometimes :)
19:31:04 <TrueBrain> now trying to patch it up ... difficult
19:33:10 <TrueBrain> depending how some random memory is initialized I guess, I now get another version of a function :p
19:34:58 <TrueBrain> whoho, finally valgrind shows no problems
19:35:02 <TrueBrain> still a weird function :p
19:45:55 <TrueBrain> and it insists on adding 27 dangling functions when ever I remove tons of @implements lines ....
19:47:03 <glx> f__2756_0A81_0022_9F5C() <-- uint16 emu_CustomTimer_AddHandler(csip32)
19:48:31 <glx> TrueBrain: the fun is when it doesn't add a function or when it adds them twice
19:58:06 <TrueBrain> LOL! Hunting return values
20:01:23 <DorpsGek> SVN: truebrain (r861) -Fix: remove unneeded @implements entries
20:12:13 <DorpsGek> SVN: truebrain (r862) -Fix: don't name functions emu_ if they no longer are
20:21:05 <DorpsGek> SVN: truebrain (r863) -Codechange: remove now unused emu_ functions
20:21:24 <TrueBrain> I now have a script which tells me when there shouldn't be any use of the emu_ function ;)
20:23:05 <TrueBrain> 10,968 done, 93,475 more to go
20:23:38 <glx> sometimes I do a search for them :)
20:23:48 <TrueBrain> I now made it automated ;)
20:24:13 <TrueBrain> and it also means my decompiler cleanup is done :)
20:46:35 <DorpsGek> SVN: glx (r864) -Fix (r863): someone again forgot to run a script
21:01:16 <TrueBrain> had a nice dinner with your inlaws?
21:09:13 <tneo> did level 9 with atreides always had so little cash?
21:09:46 <TrueBrain> there is one house which has more money in L9 .. the updated scenarios (see topic) fixes that, by giving all houses I believe 1000 credits
21:10:04 <tneo> 1000 is too little for L9
21:10:18 <TrueBrain> 2 houses have that in the original :)
21:10:35 <TrueBrain> but .. you need to be fast :)
21:10:41 <TrueBrain> SP is prio number 1 :)
21:10:47 <tneo> can't recall, /me is looking for his Diskettes :P
21:12:16 <Xaroth> 300 for windtrap, 400 for refinery
21:12:31 <TrueBrain> after first harvester return, SP
21:12:36 <TrueBrain> after second, order harvesters
21:12:39 <TrueBrain> always worked for me :)
21:15:12 <TrueBrain> jump point is invalid :(
21:15:15 <TrueBrain> seems like corruption again tneo :(
21:15:42 <tneo> it produced a crash log even ... :-/
21:16:29 <TrueBrain> we should really hurry C-ifying :)
21:16:32 <TrueBrain> then those problems become tracable
21:16:39 <TrueBrain> or at least less annoying :)
21:16:56 <tneo> I just keep posting them until then :-)
21:17:10 <TrueBrain> and if you find a way to reproduce them, let me know
21:29:33 <tneo> aaaaaahhhhhhhhh those freezes... :-(
21:30:29 <TrueBrain> the higher the level, the more they happen ...
21:30:38 <TrueBrain> the reason I suspect Units or Structures beind the issue here
21:30:56 <TrueBrain> or maybe the timing between timer thread and main thread
21:35:25 <TrueBrain> hmm .. yes, if there is somewhere a piece of code which does incorrect locking, it could cause random memory writes
21:41:28 <tneo> they seem to have become less though
21:41:38 <TrueBrain> that is good to know/hear :)
21:51:01 <glx> hmm after some tests in dosbox I think I'll redo the JIT run like if a soundblaster is installed :)
21:51:30 <TrueBrain> haha, that shouldn't be any suprise ;)
21:55:15 <glx> true, but .pcs (pcspeaker) and .c55 (sound canvas) use a similar format and sound canvas doesn't sound that bad
21:55:45 <glx> anyway sound blaster is better
22:19:17 <TrueBrain> mplayer doesn't know those formats :p
22:19:40 <TrueBrain> Input #0, voc, from 'zaffirm.voc':
22:19:42 <TrueBrain> Duration: 00:00:00.89, bitrate: 94 kb/s
22:19:43 <TrueBrain> Stream #0.0: Audio: pcm_u8, 11764 Hz, mono, s16, 94 kb/s
22:20:43 <TrueBrain> (also stored with the chunk method)
22:21:36 <TrueBrain> it again is: FORM<length32><CHUNKNAME><length><CHUNKNAME><length>..
22:21:51 <TrueBrain> first total length field is only calculated wrong :)
22:23:21 <TrueBrain> seems like multiple sounds in one file
22:24:28 <TrueBrain> lands of lore uses the same format
22:26:56 <TrueBrain> well, good luck decyphering :)
22:27:04 <TrueBrain> you will be my hero if you manage to get sane sound out of it :)
22:29:32 <TrueBrain> that format they did not invent
22:30:22 <glx> well adlib uses it too ;)
22:30:37 <TrueBrain> owh, it is Westwood specific
22:30:50 <TrueBrain> weird .. all those others have a clear footprint of Westwood, this not .. yet they did create it
22:30:54 <TrueBrain> or at least the extension of it ;)
22:31:27 <TrueBrain> but yes, you had adplug source code, so :p
22:32:39 <TrueBrain> even active development
22:34:49 <glx> all .adv have miles design signature :)
22:36:05 <glx> first step is run with JIT ;)
22:36:15 <TrueBrain> adl is normally sound to the soundblaster itself
22:36:23 <TrueBrain> which knows how to handle the opcodes it is giving
22:36:29 <TrueBrain> so I think you are better off picking another one
22:36:42 <TrueBrain> else you have to implement an emulator for the card itself :)
22:36:53 <TrueBrain> if at least I read this right :)
22:38:11 <glx> well, I'll see what the JIT gives
22:38:38 <TrueBrain> AdPlug plays sound data, originally created for the AdLib (OPL2) and Sound Blaster (Dual OPL2/OPL3) audio boards, directly from its original format on top of an emulator or by using the real hardware. No OPL chip is required for playback.
22:39:52 <TrueBrain> but sure, give it a spin :)
22:43:11 <TrueBrain> libeastwood seems to implement something that looks like adl stuff too
22:43:14 <TrueBrain> code is ugly as hell
22:43:54 <TrueBrain> even allows using OPL :)
22:44:55 <glx> if you read correctly you'll see it's adplug ;)
22:46:35 <TrueBrain> by the looks ... it will be a bitch :)
22:47:48 <TrueBrain> so yeah .. let me know if you need any help :)
22:49:14 <glx> hmm it seems the driver checks for hardware presence (text is displayed in intro, but it shouldn't as voices are enabled in the config)
22:49:32 <TrueBrain> it seems to require hardware support for it :)
22:50:02 <TrueBrain> I think tomorrow I will pick yet another random file, and start C-ifying it :)
22:50:38 <TrueBrain> we are pretty far along .. 35% named .. that is a lot
22:51:23 <TrueBrain> reading code gets more and more fun
22:51:28 <TrueBrain> you just know what things do .. 22A6 -> draw
22:52:42 <TrueBrain> did you name those functions already? (in SVN that is)
22:53:04 <TrueBrain> please do, if you have the time :)
22:53:15 <glx> I'm not sure about the naming
22:53:22 <TrueBrain> not relevant at this stage
22:53:26 <TrueBrain> any naming that is a bit clear will do :)
22:53:31 <glx> [20:47:24] <@glx> f__2756_0A81_0022_9F5C() <-- uint16 emu_CustomTimer_AddHandler(csip32)
22:54:03 <glx> and for the hard_linked one?
22:54:21 <TrueBrain> well, minus the 'uint16' part btw :p
22:54:28 <TrueBrain> Interrupt_CustomTimer?
22:54:34 <TrueBrain> emu_Interrupt_CustomTimer
22:55:00 <glx> ok I'll name these 2 functions
22:55:07 <TrueBrain> please do :) Else we keep on wondering ;)
22:55:55 <TrueBrain> I even can read from the decompiled: open file .. read block ...
22:55:58 <TrueBrain> this is a bit sick :p
22:58:43 <TrueBrain> found the WSA loader :p
22:58:47 <TrueBrain> (by random looking)
23:00:01 <TrueBrain> hmm ... 1FB5_01FA does a tiny bit more then I expected
23:00:06 <TrueBrain> seemly it also looks in .pak files
23:00:13 <TrueBrain> which means ... that I changed the behavoir
23:00:19 <TrueBrain> savegames no longer can be in .pak files :p
23:03:06 <DorpsGek> SVN: truebrain (r865) -Add: named one more function
23:04:01 <TrueBrain> ah, no .. 1FB5_15B5 seems to be raw access
23:04:08 <TrueBrain> where 01FA is .pak access or something
23:04:15 <TrueBrain> I should name them :)
23:06:21 <DorpsGek> SVN: glx (r866) -Add: named 2 more functions
23:07:39 <TrueBrain> owh, forgot to disable that :p
23:12:52 *** proyvind_ has joined #openDune
23:15:27 <TrueBrain> yeah, I think I am right .. you could have _save000.dat in a pak ..
23:15:31 <TrueBrain> maybe I should just try it :p
23:17:52 <TrueBrain> 6 file functions remaining
23:18:52 <TrueBrain> I guess now it becomes slightly important to C-ify the fopen function, which takes .pak files into account ..
23:19:27 <DorpsGek> SVN: truebrain (r867) -Add: named a bunch of file related functions
23:21:28 <DorpsGek> SVN: truebrain (r868) -Fix (r867): oops .. caused a name conflict
23:24:48 <glx> conflict between 1FB5 and 01F7
23:24:55 <glx> and you forgot to update src
23:29:25 <DorpsGek> SVN: truebrain (r869) -Fix (r867): final fix on simple renaming ;)
23:30:52 <TrueBrain> yeah ... tomorrow I am going to see what I can do for cs 1FB5
23:31:14 <TrueBrain> as when that is done, and we have a pure C 'fopen' which also checks .PAK files, I can take a look at this WSA I found :p
23:31:29 <glx> as it's used everywhere it's a good idea :)
23:31:48 <TrueBrain> also, I need to clean up scenario.c
23:31:54 <TrueBrain> clearly one of the older works
23:32:48 <TrueBrain> what did you mean with those NoInput functions?
23:36:23 <glx> they lock input interrupts
23:37:53 <glx> 353F:986C <-- this is the lock :)
23:39:51 <TrueBrain> k, patched all File_Open calls, to display what they are opening
23:40:20 <glx> not a good idea on windows
23:47:45 <TrueBrain> f__253D_00FA_0010_873D <- reads a file with some headers ... not sure
23:47:59 <TrueBrain> the 01F7 function I couldn't fix
23:48:23 <TrueBrain> anything else you are pointing to with 'wrong'?
23:49:27 <TrueBrain> you understand that patch is huge, and unreadable, right?
23:49:33 <glx> 16C5 is related to teams (and I think it's C-ified)
23:49:58 <TrueBrain> so you want to say it decompiles functions that you didn't touch
23:50:01 <TrueBrain> that I can work with ;)
23:50:06 <TrueBrain> make sure you have up-to-date svn
23:50:28 <glx> JIT found only 2 functions
23:51:14 <glx> but the decompiler did way too much
23:51:37 <TrueBrain> it should have been the latest, but possible I did somthing wrong
23:55:24 <glx> maybe because I use libjit and libtoc you gave to me yesterday
23:55:37 <TrueBrain> they are updated, yes
23:56:57 <DorpsGek> SVN: truebrain (r870) -Add: named a few functions based on the files they were opening
23:57:14 <TrueBrain> there is a global function which opens FORM based files
continue to next day ⏵