[Back to Index]

[00:03] <snover> that is unfortunate.
[00:03] <m_kiewitz> criezy: can you update the graphics driver?
[00:03] <m_kiewitz> or maybe try another SDL1 version?
[00:04] <criezy> Update the graphics driver on a mac? Is that allowed? :P
[00:04] <m_kiewitz> i don't own a mac,no idea
[00:05] <m_kiewitz> on Windows I would have done this
[00:05] <m_kiewitz> try to get another SDL1 version
[00:05] <m_kiewitz> unless it's built into ScummVM on Mac
[00:06] <snover> but it does remind me i need to set up a PR for discussion on a change related to pixel format switches, to figure out what will be the best way to deal with the fact that phant2 plays 16bpp videos in between 8bpp game screens
[00:07] <snover> this currently destroys and recreates the window, which causes a bunch of bad stuff to happen. for SDL2 i can fix it to not destroy the window, but no such luck for SDL1, and i dont know what would happen to some of the other backends
[00:08] <criezy> I compile it myself, I can use any SDL 1.2 version I want.
[00:08] <m_kiewitz> snover: on Windows I think it works just fine already
[00:08] <m_kiewitz> criezy: try around with another SDL version
[00:08] <criezy> It's just going to take a while to compile another SDL version...
[00:09] <snover> criezy: is yours a G5 system?
[00:09] <m_kiewitz> snover: at least I think so. afaik there is another video format already playing in16bpp and I never had window recreation
[00:11] <criezy> It's actually a Intel Core Duo with roseta to run ppc applications.
[00:11] <criezy> I also have a G3, but there it takes days to compile anything that has more than a few lines of code.
[00:12] <snover> m_kiewitz: im writing it up now, one sec :)
[00:13] <m_kiewitz> what a bunch of crap
[00:13] <m_kiewitz> i figured out a SAP kernel performance problem
[00:14] <m_kiewitz> reading filenames from directories with lots of files inside of them (1000+) gets really slow
[00:14] <m_kiewitz> and I just wrote my own application to read directories
[00:14] <m_kiewitz> now call that one and don't use SAP kernel anymore
[00:14] <m_kiewitz> previously job took 50-60 seconds,even when there was nothing to do
[00:14] <m_kiewitz> now it's 8 seconds :P
[00:15] <m_kiewitz> no idea what they are doing probably some cache problem
[00:15] <snover> m_kiewitz: for some reason i am reminded of http://thedailywtf.com/articles/The-Speedup-Loop
[00:16] <m_kiewitz> lol
[00:16] <m_kiewitz> this silly issue happens for ages
[00:17] <m_kiewitz> i finally contacted SAP some weeks ago
[00:17] <snover> also https://lonesysadmin.net/2007/08/17/use-dir_index-for-your-new-ext3-filesystems/
[00:17] <m_kiewitz> got through to someone in India
[00:17] <m_kiewitz> who was quite happy that he finally got a ticket
[00:17] <m_kiewitz> it seems I got it into kernel statistics
[00:18] <m_kiewitz> then it went to another component and now our base support created another ticket and this time support just went "oh we would have to rewrite parts of the transaction" (although it has nothing to do with transaction, it's just a kernel issue)
[00:18] <m_kiewitz> and now i was fed up and solved it myself
[00:19] <m_kiewitz> i guess they have an internal cache and when the directory has more files than that cache it has to do some really weird stuff
[00:19] <m_kiewitz> like maybe reasking windows multiple times to search through the directory
[00:19] <m_kiewitz> the kernel call is even fullybroken, you can give it file spec, but only "*" and "*.*" works
[00:20] <m_kiewitz> so I have to filter everything by myself
[00:20] <m_kiewitz> although filtering isn't the performance problem (well it surely may be another issue)
[00:29] <-- SylvainTV left irc: Read error: Connection reset by peer
[00:30] --> GitHub55 joined #scummvm.
[00:30] <GitHub55> [scummvm] csnover opened pull request #937: [do not land] SDL: Do not recreate SDL2 window when switching pixel formats (master...sdl2-keep-window) https://git.io/vSH4I
[00:30] GitHub55 (GitHub55@192.30.252.41) left #scummvm.
[00:30] <snover> m_kiewitz: so here is the list of troubles https://github.com/scummvm/scummvm/pull/937
[00:34] <m_kiewitz> one would have to check all sorts of different platforms
[00:44] <snover> running at 24bpp and doing the conversion internally would be the other option i can think of, but then that would lose any gpu acceleration, wouldnt it?
[00:45] indrora (indrora@zaibatsutel.net) left #scummvm ("Closing Window").
[00:49] <m_kiewitz> changing all of that just for one game? idk
[00:49] <m_kiewitz> and that would be tons of changes
[00:50] <m_kiewitz> and would create other problems, yes
[00:50] <m_kiewitz> doesn't SDL run a 8bpp window with 16bpp or 24bpp internally already?
[00:50] <m_kiewitz> it's really weird that theycan't change it properly on the fly.
[00:54] --> Uhfgood joined #scummvm.
[00:54] <snover> Im not sure. The state of the various graphics backends is not terribly clear to me right now.
[00:55] <m_kiewitz> snover: does the current code work far enough?
[00:55] <m_kiewitz> i could check what SDL1 is doing on Windows
[00:56] <snover> m_kiewitz: which code are you referring to about current code?
[00:56] <m_kiewitz> current code in main branch
[00:56] <m_kiewitz> does Phantasmagoria 2 work somewhat with that including the videos?
[00:57] <snover> yes, it works, it just is frustrating to try to play because of those four things mentioned in the PR description
[01:01] <m_kiewitz> as i said -other 16bpp videos are playing just fine for me
[01:01] <m_kiewitz> so it may be something platform-specific
[01:03] <m_kiewitz> will look into the keyboard code first
[01:03] <m_kiewitz> i need to rewrite an interface for work first though, it's just converting KIX to REXX, so nothing major
[01:05] <snover> no worries, i expect a few days at least to hear anything from anyone else about it :)
[01:09] <snover> i had that tab open in my browser with half-written description for like 2 or 3 weeks already, so i felt like i really ought to go complete it.
[01:18] --> klusark joined #scummvm.
[01:27] <tsoliman> snover: if I fullscreen the launcher and then start some of the games, the effect you describe in PR#937 happens .. along with some weirdness .. the games are 7th Guest, SQ6, and some others ... I tried your patch and it didn't fix them .. SDL2 on macOS
[01:27] <tsoliman> to be fair it happens with or without your patch
[01:30] <-- Dominus left irc: Ping timeout: 260 seconds
[01:31] <tsoliman> in discworld2 it beeps 4 times and redoes the fullscreening animation but then goes to normal - with your patch it doesn't go to normal and instead is in a small window
[01:32] --> Dominus joined #scummvm.
[01:32] <tsoliman> so basically it is a regression for DW2 - but it wasn't perfect to begin with
[01:35] <tsoliman> nevermind - this weirdness happens half the time without your patch as well ... can't tell what is going on or what causes it to sometimes work and sometimes not work
[01:38] <-- m_kiewitz left irc: Quit: technology isn't intrinsically good or evil. It's how it's used. Like the Death Ray.
[01:48] <snover> tsoliman: are those games changing resolutions?
[01:48] <snover> i mean& sq6 shouldnt ever, it always runs at 640x480.
[01:49] <tsoliman> Just on startup. Not in the middle of the game or anything
[01:49] <snover> ah. ok. that makes sense.
[01:49] <snover> (well, *more* sense.)
[01:50] <tsoliman> yeah :(
[01:50] <tsoliman> FWIW I am doing x3 scaling in the launcher
[01:51] <tsoliman> HQ3x
[01:51] <tsoliman> with AR correction on
[01:51] <snover> ah, ok. then it does make sense, since that increases the resolution of the launcher
[01:51] <snover> the engine ignores scaling and runs at 1x
[01:52] <tsoliman> ok so I turn off scaling and the same problem happens
[01:52] <tsoliman> steps: start scummvm .. option-enter to fullscreen .. start DW2
[01:53] <tsoliman> watch as 4 rapid system beeps ... it unfullscreen-animates itself .. and then fullscreens again
[01:53] <snover> oh, sorry. i am still speaking of the sci engine :)
[01:53] <tsoliman> SQ6 has the same problem but not older SCI games
[01:53] <snover> yes, that is the same behaviour
[01:54] <snover> for some reason after turning off global scaling, scummvm needs to be restarted for it to stop scaling the launcher
[01:54] <tsoliman> but when it fullscreens again (DW2 or T7G) *sometimes* (most of the time?) it just shows a scaled down version in the upper left
[01:54] <snover> that sounds even more unpleasant.
[01:55] <tsoliman> unfullscreening and fullscreening again fixes it
[01:55] <snover> im not sure if i have seen that behaviour before or not
[01:56] <tsoliman> this problem doesn't happen with openGL
[01:56] <tsoliman> it flickers a lot but then starts normally
[01:56] <tsoliman> well what happens for you when you start the launcher, fullscreen it .. and then start SQ6 or DW2 or T7G?
[01:58] <snover> im not sure on the latter two, but with sq6 with the patch from that PR it works the way you would want it to, no bloops or animations or flickers
[02:00] <tsoliman> maybe I need a make clean
[02:00] <snover> without the patch i get a few alert noises and ugly sliding animations
[02:01] <tsoliman> this is on startup right?
[02:01] <snover> yes.
[02:04] <snover> oh, and maybe double-check that you are compiling with SDL2
[02:06] <tsoliman> I did .. and I got it to work right
[02:06] <tsoliman> I had to pick <DEFAULT> in the scaling modes
[02:06] <tsoliman> not "normal - no scaling"
[02:07] <tsoliman> ok it has to be exactly 2x to work
[02:07] <snover> what happens when you choose the normal option sure is& interesting.
[02:07] <tsoliman> 1x and 3x break it
[02:08] <tsoliman> normal seems to be 1x
[02:08] <tsoliman> default = 2x
[02:08] <tsoliman> (or seems to be)
[02:08] <tsoliman> 2x and HQ2x both work fine
[02:09] <snover> yeah, so when configured that way the launcher resolution is the same as the game resolution so the window is not recreated
[02:09] <tsoliman> hmm .. so DW2 and T7G are also fixed now
[02:09] <tsoliman> let me try once more without the patch
[02:10] <snover> i dont know why it took me this long, but i did just realise that SDL2 *has* an API for changing window size
[02:10] <tsoliman> yep .. your patch definitely has an effect .. it fixes them
[02:11] <snover> so i suspect that the resolution changes could also be OK, by switching from destroying & creating a window, to calling SDL_SetWindowSize
[02:11] tsoliman is excited!
[02:11] <tsoliman> I kind of prefer 3x because this screen is big .. normally I use 2x
[02:12] <tsoliman> (and by big I mean high dpi .. higher res)
[02:12] <tsoliman> by big I mean small :D
[02:13] <snover> i want to say part of this problem is because SDL2 support was added to be backwards-compatible to the old SDL1.2, which was still the default at the time
[02:14] <snover> and in SDL1.2 you could only call SDL_SetVideoMode, there was no separate API for window & surface management
[02:16] <tsoliman> I maintain a platform that doesn't have SDL2 support (maemo) .. so I appreciate the backwards compatibility .. but like SDL1 on macOS Sierra has issues
[02:23] <snover> thanks for talking about it, it gave me a moment of clarity to see that the window in SDL2 can probably be fixed for all scenarios including resizes
[02:25] <tsoliman> no thank *you* for looking into it
[03:04] <-- snover left irc: Ping timeout: 245 seconds
[03:09] --> snover joined #scummvm.
[03:09] #scummvm: mode change '+o snover' by ChanServ!ChanServ@services.
[03:17] <-- snover left irc: Ping timeout: 240 seconds
[03:19] --> vinayakvivek joined #scummvm.
[03:22] --> snover joined #scummvm.
[03:22] #scummvm: mode change '+o snover' by ChanServ!ChanServ@services.
[03:30] <-- snover left irc: Ping timeout: 260 seconds
[03:36] --> snover joined #scummvm.
[03:36] #scummvm: mode change '+o snover' by ChanServ!ChanServ@services.
[03:36] <snover> what in the world is going on with my irc connection tonight&
[03:40] <-- Uhfgood left irc: Quit: Leaving
[03:52] <-- snover left irc: Ping timeout: 260 seconds
[04:00] --> csnover joined #scummvm.
[04:00] #scummvm: mode change '+o csnover' by ChanServ!ChanServ@services.
[04:13] --> Joefish joined #scummvm.
[04:15] <Joefish> morning
[05:43] --> somaen joined #scummvm.
[05:43] #scummvm: mode change '+o somaen' by ChanServ!ChanServ@services.
[06:05] --> waltervn joined #scummvm.
[06:05] #scummvm: mode change '+o waltervn' by ChanServ!ChanServ@services.
[06:06] <waltervn> morning
[06:16] --> ny00123 joined #scummvm.
[06:51] <-- Lightkey left irc: Ping timeout: 255 seconds
[06:57] <logix> criezy: probably unrelated, but I have seen similar sdl color corruption with dosbox (but I think I have wrong colors there in windowed mode, fullscreen is ok)
[06:57] <logix> but that's on intel osx
[06:58] <logix> I don't know if there's any relation to bundle/non-bundle build, this is a non-bundle build (./configure && make)
[07:02] <-- Littleboy left irc: Ping timeout: 240 seconds
[07:04] --> Lightkey joined #scummvm.
[07:34] <-- TMM left irc: Quit: Ex-Chat
[07:40] --> Strangerke|work joined #scummvm.
[07:44] <-- LittleToonCat left irc: Remote host closed the connection
[08:34] --> _sev|work joined #scummvm.
[08:34] #scummvm: mode change '+o _sev|work' by ChanServ!ChanServ@services.
[08:39] <-- _sev|work left irc: Client Quit
[09:04] --> WooShell joined #scummvm.
[09:06] <WooShell> good meowning =^.^=
[09:14] --> _sev|work joined #scummvm.
[09:14] <-- _sev|work left irc: Changing host
[09:14] --> _sev|work joined #scummvm.
[09:14] #scummvm: mode change '+o _sev|work' by ChanServ!ChanServ@services.
[09:20] --> m_kiewitz joined #scummvm.
[09:20] #scummvm: mode change '+o m_kiewitz' by ChanServ!ChanServ@services.
[10:05] <-- Strangerke|work left irc: Ping timeout: 260 seconds
[10:39] --> chungy joined #scummvm.
[10:43] <chungy> Heya, the wiki is write protected. I was looking on http://wiki.scummvm.org/index.php/Platforms for any sort of word about a Nintendo 3DS port. There is one available on GitHub, seems to be inactive and based on 1.8.1: https://github.com/Cruel/scummvm/tree/3ds
[10:43] <chungy> Thought I'd let you guys know so it might be listed.
[10:46] <_sev|work> hi chungy
[10:46] <_sev|work> actually, that port was merged, but the porter kind of vanished sincse
[10:46] <_sev|work> *since
[10:48] <criezy> logix: Thanks. The SDL bug mentionned there (https://bugzilla.libsdl.org/show_bug.cgi?id=1235) looks like what I am getting.
[10:48] <chungy> hmm, oh
[10:48] <chungy> hypothetically I could build a 3DS version from the current master?
[10:48] <_sev|work> yes
[10:49] <chungy> I might try that. will need to setup devkit :P
[10:55] <logix> criezy: might well be the same bug, the link there goes to the dosbox tracker and there it's mentioned that it doesn't happen in fullscreen mode
[11:01] <-- _sev|work left irc: Quit: This computer has gone to sleep
[11:02] <criezy> The dosbox tracker ticket actually mention two issues, one in windowed mode and one in fullscreen mode (that causes a color shift to green).
[11:02] <-- user9 left irc: Quit: user9
[11:02] <criezy> The SDL bug is for the second one. The first one looks like a dosbox bug.
[11:02] --> user9 joined #scummvm.
[11:03] --> _sev|work joined #scummvm.
[11:03] <-- _sev|work left irc: Changing host
[11:03] --> _sev|work joined #scummvm.
[11:03] #scummvm: mode change '+o _sev|work' by ChanServ!ChanServ@services.
[11:05] <logix> criezy: ah, indeed
[11:06] chungy (chungy@2001:0:53aa:64c:3c4f:1dce:b689:1b18) left #scummvm ("Leaving").
[11:09] --> ajax16384 joined #scummvm.
[11:09] #scummvm: mode change '+o ajax16384' by ChanServ!ChanServ@services.
[11:13] --> Littleboy joined #scummvm.
[11:13] #scummvm: mode change '+o Littleboy' by ChanServ!ChanServ@services.
[11:32] <-- user9 left irc: Ping timeout: 240 seconds
[11:35] <-- demonimin left irc: Remote host closed the connection
[11:37] <-- waltervn left irc: Quit: Leaving
[11:39] --> demonimin joined #scummvm.
[11:42] --> waltervn joined #scummvm.
[11:42] #scummvm: mode change '+o waltervn' by ChanServ!ChanServ@services.
[11:59] --> Strangerke|work joined #scummvm.
[12:05] <-- L0ngcat left irc: Remote host closed the connection
[12:06] --> L0ngcat joined #scummvm.
[12:46] --> user9 joined #scummvm.
[12:49] <criezy> snover: I see another TO_LE_16 that looks suspicious in audio.cpp:250
[12:49] <criezy> Should this be removed as well? And do you know which game use this deDPCM16 so that I can test?
[12:49] <criezy> And I get noise instead of music in the phantasmagoria intro. Any idea where I should look in the code for that one?
[12:52] <criezy> Ah sorry, that deDPCM16 looks OK. I didn't see the s was from a static array and not read from the stream.
[12:54] <-- user9 left irc: Quit: user9
[12:56] --> user9 joined #scummvm.
[12:57] <criezy> And for phantasmagoria, by 'intro' I meant the sierra logo at the very start.
[12:59] --> demonimin_ joined #scummvm.
[12:59] <-- demonimin_ left irc: Changing host
[12:59] --> demonimin_ joined #scummvm.
[13:01] --> Strangerke_ joined #scummvm.
[13:03] <-- Strangerke left irc: Ping timeout: 240 seconds
[13:03] Nick change: Strangerke_ -> Strangerke
[13:03] <-- demonimin left irc: Ping timeout: 260 seconds
[13:04] <wjp> is that with a dos/win version of the game, on a BE machine?
[13:13] --> Strangerke_ joined #scummvm.
[13:15] <-- ny00123 left irc: Quit: Leaving
[13:16] <-- Strangerke left irc: Ping timeout: 268 seconds
[13:16] Nick change: Strangerke_ -> Strangerke
[13:20] <-- user9 left irc: Ping timeout: 240 seconds
[13:21] --> user9 joined #scummvm.
[13:27] <criezy> The GoG version on a PPC mac (so BE indeed).
[13:32] Nick change: Korak_ -> Korak
[13:34] <-- Strangerke left irc: Ping timeout: 240 seconds
[13:36] --> Strangerke joined #scummvm.
[14:00] <-- _sev|work left irc: Read error: Connection reset by peer
[14:03] --> _sev|work joined #scummvm.
[14:03] #scummvm: mode change '+o _sev|work' by ChanServ!ChanServ@services.
[14:37] <criezy> snover: I only just realized I was running without the fix from yesterday... It fixes sound for the sierra logo in phantasmagoria as well.
[14:41] <criezy> And it fixes the speech in the game as well.
[15:50] <-- _sev|work left irc: Quit: This computer has gone to sleep
[15:51] <-- Strangerke|work left irc: Quit: Bbl
[15:53] --> _sev|work joined #scummvm.
[15:53] #scummvm: mode change '+o _sev|work' by ChanServ!ChanServ@services.
[15:54] <-- _sev|work left irc: Client Quit
[15:56] --> _sev|work joined #scummvm.
[15:56] #scummvm: mode change '+o _sev|work' by ChanServ!ChanServ@services.
[15:59] <csnover> criezy: glad to hear that everything is working OK. audio.cpp is used for SCI16 and it decompresses the entire audio at once and sticks it in a raw stream with a little-endian flag instead of being implemented as a proper audio stream
[15:59] Nick change: csnover -> snover
[15:59] <snover> so thats why that works there.
[16:01] <-- _sev|work left irc: Quit: This computer has gone to sleep
[16:02] --> Henke37 joined #scummvm.
[16:04] <-- tsoliman left irc: Ping timeout: 246 seconds
[16:09] <-- ajax16384 left irc: Quit: Leaving
[16:12] --> Uhfgood joined #scummvm.
[16:23] <snover> m_kiewitz: can you tell me which sound resource in KQ5 is the corrupt one, or how to get to the place where it is loaded?
[16:24] --> LittleToonCat joined #scummvm.
[16:26] <snover> m_kiewitz: oh, never mind! i found it on the wiki. 👍
[16:30] --> _sev|work joined #scummvm.
[16:30] #scummvm: mode change '+o _sev|work' by ChanServ!ChanServ@services.
[16:32] --> ajax16384 joined #scummvm.
[16:32] #scummvm: mode change '+o ajax16384' by ChanServ!ChanServ@services.
[16:49] --> GitHub100 joined #scummvm.
[16:49] <GitHub100> [scummvm] sev- pushed 1 new commit to master: https://git.io/vSQaC
[16:49] <GitHub100> scummvm/master 4873330 Eugene Sandulenko: DIRECTOR: Lingo: Added stub for generic event handlers
[16:49] GitHub100 (GitHub100@192.30.252.40) left #scummvm.
[16:59] <-- _sev|work left irc: Quit: This computer has gone to sleep
[17:16] <-- ajax16384 left irc: Read error: Connection reset by peer
[17:20] --> ajax16384 joined #scummvm.
[17:20] #scummvm: mode change '+o ajax16384' by ChanServ!ChanServ@services.
[17:22] <snover> this sound doesnt seem to crash, but i guess i dont know if i actually have the version of the game with the corrupt resource
[17:29] --> waltervn_ joined #scummvm.
[17:29] #scummvm: mode change '+o waltervn_' by ChanServ!ChanServ@services.
[17:29] <-- waltervn left irc: Disconnected by services
[17:29] Nick change: waltervn_ -> waltervn
[17:30] --> waltervn_ joined #scummvm.
[17:30] #scummvm: mode change '+o waltervn_' by ChanServ!ChanServ@services.
[17:31] <waltervn> Joefish: according to "the internet", the group policy settings for disabling auto updates on windows 10, do no longer work on Windows 10 Pro.
[17:44] --> tsoliman joined #scummvm.
[17:44] #scummvm: mode change '+o tsoliman' by ChanServ!ChanServ@services.
[17:57] <m_kiewitz> snover: it's not the KQ5 first floppy
[17:57] <m_kiewitz> it may even be the latest non speech one, idk for sure, have to check
[18:02] <snover> hm. tricky.
[18:14] --> Vampire0_ joined #scummvm.
[18:14] <m_kiewitz> i need to look, for the credits it was a full remix of basically all the other songs of the game
[18:15] <m_kiewitz> and that makes the corruption even worse, because I would have loved to listen to the whole song.
[18:15] <-- Vampire0 left irc: Ping timeout: 260 seconds
[18:17] <-- Strangerke left irc: Ping timeout: 260 seconds
[18:19] --> m-kiewitz joined #scummvm.
[18:19] <-- m_kiewitz left irc: Read error: Connection reset by peer
[18:19] #scummvm: mode change '+o m-kiewitz' by ChanServ!ChanServ@services.
[18:31] <-- waltervn_ left irc: Read error: Connection reset by peer
[18:53] --> t0by joined #scummvm.
[18:53] #scummvm: mode change '+v t0by' by ChanServ!ChanServ@services.
[19:00] <Joefish> waltervn: ah that sucks.. guess I can never update my version then :P
[19:34] <m-kiewitz> snover: i just tried the French KQ5
[19:34] <m-kiewitz> it has that broken track
[19:34] <m-kiewitz> French KQ5 is on one of the KQ collections
[19:36] <snover> does it crash if you run startsound 699?
[19:36] <snover> m-kiewitz: ^
[19:37] <m-kiewitz> i think i'm not on latest version
[19:38] <m-kiewitz> im getting tons of warnings (I added those checks back then)
[19:38] <m-kiewitz> WARNING: Invalid size inside sound resource 699: track 0, channel 7!
[19:38] <m-kiewitz> WARNING: Invalid offset inside sound resource 699: track 0, channel 8!
[19:38] Last message repeated 2 time(s).
[19:38] <m-kiewitz> WARNING: Invalid size inside sound resource 699: track 1, channel 7!
[19:40] <m-kiewitz> and i actually have tons of saves
[19:40] <m-kiewitz> i had to fix the broken endings of German + French versions, both froze right before the end
[19:42] <m-kiewitz> snover: we should really create some debug command, that goes through all views, pictures and music of the current game and loads/draws/plays all of them automatically
[19:42] <m-kiewitz> to check for corrupt resources
[19:43] <m-kiewitz> because testing the games will surely never cover every single animation etc and if there are some in one of the games, then we would break that game
[19:43] <-- Harekiet left irc: Ping timeout: 252 seconds
[19:43] <m-kiewitz> it's even worse because testing the games would mean testing the English release, but sometimes Sierra had bigger issues in localized games too
[19:43] <m-kiewitz> and localized versions are really rare, i own a few, but not nearly all of them
[19:51] <snover> it seems like a reasonable idea though i am not sure offhand what the best way to actually do that is
[20:03] <m-kiewitz> enum everything, draw everything to screen
[20:04] <m-kiewitz> play every music resource once
[20:04] <m-kiewitz> no idea how difficult it is to draw views/backgrounds for sci32 manually
[20:04] <m-kiewitz> for sci16 it's a piece of cake
[20:11] <snover> i was just thinking about how there is no way to recover from an error and continue validating, but maybe that isnt a problem
[20:13] <Lightkey> twitch.tv/gogcom playing Maniac Mansion: Day of the Tentacle Remastered!
[20:13] <Lightkey> a specific part of it anyway :-)
[20:28] --> Harekiet joined #scummvm.
[20:39] <-- Uhfgood left irc: Quit: Leaving
[20:50] <Joefish> https://www.twitch.tv/gogcom streaming Maniac Mansion
[20:53] --> Strangerke joined #scummvm.
[20:53] #scummvm: mode change '+o Strangerke' by ChanServ!ChanServ@services.
[21:01] <-- ajax16384 left irc: Read error: Connection reset by peer
[21:14] --> zxtheproto joined #scummvm.
[21:14] <-- vinayakvivek left irc: Quit: Connection closed for inactivity
[21:34] <-- zxtheproto left irc: Ping timeout: 260 seconds
[21:41] <t0by> :(
[21:42] <Strangerke> grml :(
[21:44] <Joefish> ? :/
[21:46] <t0by> Eh, I'm sorry about the disappearance of WinterGrapsch.
[21:48] <Strangerke> ditto
[21:48] <Joefish> oh.. didn't want to open up any wounds :/
[21:53] <t0by> I'm gonna call it a night. Good night, you fine folks.
[21:53] <-- t0by left irc: Quit: Bye!
[21:53] <Joefish> good night t0by
[21:56] <Joefish> ... game crashes with segfault and I look through the bt, read the code, try to see behind the structures.. then I glanced over the creature type index of 16711692.. sure that's right :D
[21:57] <Strangerke> index 16711692 0_o
[21:58] <Joefish> yea, either going out of bounds or the dangerous looking structure cast from u16 array went wrong
[21:59] <Joefish> https://github.com/scummvm/scummvm/blob/master/engines/dm/group.cpp#L1288
[22:00] <Joefish> but overall the code is pleasant to read, considering it comes from C asm ifdef hell :P
[22:01] <Joefish> or maybe I've been staring the last days at supernova disasm too long so anything with some structure looks pleasant now
[22:09] --> zxtheproto joined #scummvm.
[22:10] <logix> Joefish: ff000c, if I had to guess I'd say there's a sign error and a cast error (16->24 bit type of some sort) somewhere
[22:11] <logix> scratch the sign error, more like something that's initialized as -1 (0xffffff) and then the lower 16 bit get overwritten
[22:16] <-- zxtheproto left irc: Ping timeout: 260 seconds
[22:28] <Strangerke> Which game are you testing btw? The Atari version or the Amiga version?
[22:28] <Strangerke> *one of the...
[22:28] <Joefish> amiga demo
[22:28] <Strangerke> hum... ok.
[22:29] <Joefish> does it work for you?
[22:29] <Strangerke> the code isn't strictly identical so maybe the structure is wrong
[22:29] <Strangerke> not afair
[22:29] <Joefish> "good" :)
[22:30] <Strangerke> iirc, it was crashing when opening the door, or just before? so the game was broken
[22:30] <Joefish> yea when opening the door and the screen fades out
[22:48] <-- waltervn left irc: Quit: Leaving
[23:18] <Joefish> alright. enough for today
[23:19] <Joefish> good night :)

[23:36] <-- m-kiewitz left irc: Quit: technology isn't intrinsically good or evil. It's how it's used. Like the Death Ray.
[23:38] <snover> Joefish: im having a hard time believing that those casts are not UB.
[23:39] --> Strangerke_ joined #scummvm.
[23:41] <-- Strangerke left irc: Ping timeout: 260 seconds
[23:41] Nick change: Strangerke_ -> Strangerke
[23:46] <Joefish> snover: I don't think so either as *dungeon and *group point to the same memory and violate strict aliasing if I'm not mistaken
[23:48] <Joefish> but I don't think that everything else looks sane except for one struct member so I assume it's either a parsing error or data structures between versions differ
[23:49] --> omer_mor joined #scummvm.
[23:49] <snover> beyond that, as far as i can tell, the classes are not packed, so there are almost certainly padding/alignment problems
[23:50] <Joefish> ah right.. shouldn't make any assumptions about alignment. didn't think of that
[23:52] <-- omer_mor_ left irc: Ping timeout: 252 seconds
[23:54] <-- WooShell left irc: Quit: If you understand or if you don't, if you believe or if you doubt - There's a universal justice, and the eyes of truth are always watching you.
[00:00] --- Sat Apr 15 2017