| 01:00 |
| 01:00:43 | | Quit pixelma (Quit: .) |
| 01:00:44 | | Quit amiconn (Quit: http://quassel-irc.org - Chat comfortably. Anywhere.) |
| 01:01:41 | | Join amiconn [0] (jens@p4fe76649.dip0.t-ipconnect.de) |
| 01:01:41 | | Join pixelma [0] (marianne@p4fe76649.dip0.t-ipconnect.de) |
| 01:02:22 | _bilgus | so my audio is still not quite right with the upstream but it is better |
| 01:02:31 | _bilgus | (on sim) |
| 01:02:58 | _bilgus | less snap crackle |
| 01:19:55 | _bilgus | so here on arm the optimized code is 2% faster 24.99mhz vs 25.04 |
| 01:54:10 | *** | Saving seen data "./dancer.seen" |
| 01:58:17 | user890104 | MarcAndersen: once there is something that makes sense to be built and installed on a Nano 3G/4G, there will be an announcement on the mailing list, with instructions and a video |
| 01:58:56 | user890104 | currently it's just bits and pieces, the "main" part of the patch is not submitted yet |
| 02:00 |
| 02:00:18 | user890104 | I've submitted nearly everything *else* of the patch, therevare 2 or 3 files left to review, test, translate the comments from Spanish and verify if they make sense |
| 02:03:36 | | Quit chris_s (Quit: Client closed) |
| 02:11:55 | _bilgus | So enabling OPT_SPEED beats what we have on my ARM |
| 02:15:00 | _bilgus | .. in 640 more bytes |
| 02:19:49 | paulcarroty | _bilgus: yep, the same stuff. random discharging after 50% battery, you can force to charging it by 2-4 cable replugging. also battery levels sometimes are 15-20% lower when stop charging manually - maybe it's the root of the problem. |
| 02:22:49 | _bilgus | the question I guess Is why its discharging the battery |
| 02:23:22 | _bilgus | that I can't reproduce mine stayed plugged all day and stayed at 100% with discharging displayed |
| 02:26:12 | paulcarroty | you don't need all day, 30-50m is enough |
| 02:32:10 | edhelas_ | Is the battery remaining time estimation in the info menu quite good ? |
| 03:00 |
| 03:54:13 | *** | Saving seen data "./dancer.seen" |
| 04:00 |
| 04:04:36 | | Quit jacobk (Ping timeout: 276 seconds) |
| 04:04:53 | | Join jacobk [0] (~quassel@47-186-65-73.dlls.tx.frontiernet.net) |
| 05:00 |
| 05:54:17 | *** | Saving seen data "./dancer.seen" |
| 07:00 |
| 07:40:38 | MarcAndersen | Anyone who can figure this error out? https://pastebin.com/3w4LrCPB |
| 07:47:00 | MarcAndersen | If I skip puzzles everything compiles here. What am I missing? |
| 07:54:21 | *** | No seen item changed, no save performed. |
| 08:00 |
| 08:14:36 | | Join akaWolf [0] (~akaWolf@akawolf.org) |
| 08:21:36 | braewoods__ | MarcAndersen, it's a linker error. either something isn't being linked in, or in the wrong order. |
| 08:22:46 | MarcAndersen | I get that, but it's the only place it fails. I think puzzles are pretty new in the source tree |
| 08:25:19 | braewoods__ | Yea, probably needs patching. You can try grepping for the missing symbols to see where they occur. |
| 08:25:30 | braewoods__ | More specifically, what source code even defines it. |
| 08:30:34 | speachy | MarcAndersen: the linker doens't like weak symbols under windows, that's hit us before, but the underlying problem is that the puzzleshelp stuff isn't being generated. |
| 08:32:40 | speachy | this is going to take a bit of spelunking to fix. |
| 08:32:58 | speachy | __builtin: any thoughts on ^^^ ? |
| 08:33:48 | MarcAndersen | What should I do about it, I don't want to make an M build unless you would like me to do so |
| 08:45:10 | speachy | no.. should be straightforward to resolve. |
| 08:45:27 | speachy | is it just sgt-slide breaking, or others too? |
| 08:48:17 | MarcAndersen | I'm not sure, it stops at that error. Maybe we should try to fix slide |
| 09:00 |
| 09:05:07 | MarcAndersen | When I try to run it though once it starts playing audio I get Unknown sample format obtained: 33056 and it crashes. |
| 09:37:40 | braewoods__ | speachy, would it be a worthwhile project to port to SDL 2.0? |
| 09:41:46 | MarcAndersen | Port what? I compiled this simulator with sdl 2.30.10 |
| 09:54:24 | *** | No seen item changed, no save performed. |
| 10:00 |
| 10:00:09 | paulcarroty | tried to find any guides for aarch64 app compilation, but failed |
| 11:00 |
| 11:54:25 | *** | No seen item changed, no save performed. |
| 12:00 |
| 12:24:15 | dook | _bilgus: Nice work with the tmo additions to the %t tag. Been on my feature wishlist for a while |
| 12:41:30 | _bilgus | just be aware that it takes more ram so try to only use the %t(n) form unless you need longer timeouts %t(n,0) or actually want to hide it %t(n,h) |
| 12:43:20 | _bilgus | I actually frre around 700 kb from running cabbie so in theory you can add 10 of them and still be under previous ram usage |
| 12:43:38 | _bilgus | (in the prior commits) |
| 12:44:51 | _bilgus | thanks :0 |
| 12:44:53 | _bilgus | :) |
| 13:00 |
| 13:01:21 | dook | Good to know! I think it will be a very specific use, but I'll keep an eye on ram if I use it more |
| 13:14:42 | | Join lebellium [0] (~lebellium@2a01cb0405d07f00d33a0da62bfc2206.ipv6.abo.wanadoo.fr) |
| 13:30:45 | | Quit jacobk (Ping timeout: 276 seconds) |
| 13:31:58 | | Join jacobk [0] (~quassel@47-186-65-73.dlls.tx.frontiernet.net) |
| 13:54:28 | *** | Saving seen data "./dancer.seen" |
| 14:00 |
| 14:14:38 | _bilgus | 700 bytes no kb :/ |
| 14:57:20 | | Quit JanC (Read error: Connection reset by peer) |
| 14:59:04 | | Join JanC [0] (~janc@user/janc) |
| 15:00 |
| 15:54:32 | *** | Saving seen data "./dancer.seen" |
| 16:00 |
| 16:41:14 | | Quit jacobk (Ping timeout: 260 seconds) |
| 16:44:47 | | Quit COMPL_EXE (Read error: Connection reset by peer) |
| 16:45:07 | | Join COMPL_EXE [0] (~compl.exe@aosc/dev/origincode) |
| 16:47:53 | tobbez | so the erosq (hw4) hosted port has a *really* bad issue (destroy your ears bad!) |
| 16:48:07 | tobbez | unplugging headphones, adjusting volume, then plugging in headphones again causes extremely high (max?) audio volume (until volume is adjusted again) |
| 17:00 |
| 17:49:52 | __builtin | speachy: MarcAndersen: you're correct, the issue is weak symbols. The "unfinished" sgt-puzzles don't have help text, so I made dummy/nullhelp.c to provide dummy help text and allow the games to compile. |
| 17:50:40 | MarcAndersen | What about the crash i had with the sample rate? |
| 17:52:19 | __builtin | The best way to fix this would be to modify the code in puzzles/rockbox.c to not reference the help text variables when compiling an unfinished plugin. But this is difficult because rockbox.o is complied only once, and then linked individually against each of the game source files (in src/). |
| 17:53:15 | __builtin | A possible solution to this would be introducing a new build rule to compile puzzles/rockbox.c two times - both with and without a "NO_HELP" macro. |
| 17:54:06 | __builtin | A stopgap would be to just disable the unfinished puzzles on windows. Comment out src/unfinished/*.c in puzzles/SOURCES.games. |
| 17:54:35 | *** | Saving seen data "./dancer.seen" |
| 17:55:48 | __builtin | MarcAndersen: what are you doing to cause the sample rate issue? Running a plugin or just playing back audio? |
| 17:57:14 | MarcAndersen | The sim crashes as soon as sdl tries to play some audio. If I turn off the voice I can navigate the menus until I try to play the file then I get Unknown sample format obtained: 33056 and it crashes |
| 17:57:46 | __builtin | sdl? |
| 17:58:12 | __builtin | do you mean you're just playing an audio file in rockbox? or playing audio in an sdl plugin? |
| 17:58:21 | __builtin | (I only know about the latter) |
| 17:59:57 | MarcAndersen | It is the windows simulator. I start it up, go in and choose a file, then crash. If a voice is loaded it crashes at startup so it happens as soon as the sim tries to play audio. |
| 18:00 |
| 18:00:07 | __builtin | ah, looks like it originates from firmware/target/hosted/sdl/pcm-sdl.c: DEBUGF("Unknown sample format obtained: %u\n", |
| 18:00:50 | __builtin | I'd debug from there. 33056 is 0x8120 in hex |
| 18:02:01 | MarcAndersen | I have no idea about that. I would just like to get it working so I can update the sims on my page. I'm not good at c at all :) |
| 18:03:07 | __builtin | it seems to be a non-trivial bug, from a quick skim of the SDL PCM driver. |
| 18:03:17 | | Quit lebellium (Quit: Leaving) |
| 18:03:48 | MarcAndersen | After the crash gdb says Program received signal SIGFPE, Arithmetic exception. |
| 18:07:17 | MarcAndersen | Ok. I will try to download the latest build from my own page and see if that works |
| 18:09:40 | MarcAndersen | Revision 025841cfb5 which was the latest ones I built works, but I think that was sdl 1? |
| 18:20:17 | MarcAndersen | Could someone else try to compile a windows sim to see if it's a problem at my end? |
| 18:24:17 | speachy | braewoods__: We're already on SDL2. |
| 18:25:01 | speachy | __builtin: I'm leaning towards just not doing building those two plugins under windows. |
| 18:25:27 | speachy | and whatever's going on with that crash, it's isolated to SDL2+Windows. |
| 18:27:16 | MarcAndersen | speachy, Can you try to compile one, except if you are sure that is the problem? |
| 18:28:13 | speachy | you mean a winsim build? not going to happen immediately, but I'll add it to the list |
| 18:28:48 | MarcAndersen | Yes. Just to make sure it's not my setup that's messed up |
| 18:33:45 | | Quit jn (Ping timeout: 252 seconds) |
| 18:34:18 | | Join jn [0] (~quassel@user/jn/x-3390946) |
| 19:00 |
| 19:15:20 | | Join Burak__ [0] (~Burak@185.25.123.34) |
| 19:17:22 | __builtin | speachy: yes, agreed. Can you make that change or should I? |
| 19:17:50 | | Quit Burak_ (Ping timeout: 252 seconds) |
| 19:18:43 | | Quit q3k (Remote host closed the connection) |
| 19:19:03 | | Join q3k [0] (q3k@boston-packets.hackerspace.pl) |
| 19:19:09 | | Quit q3k (Changing host) |
| 19:19:09 | | Join q3k [0] (q3k@hswaw/infra/q3k) |
| 19:21:16 | __builtin | I'd just do it based onn #ifndef __MINGW32__ |
| 19:26:19 | speachy | some of the SOURCES files use WIN32 |
| 19:26:54 | speachy | the mismash of windows defines always comsfused me. |
| 19:28:26 | rb-bluebot | Build Server message: New build round started. Revision da8bde0860, 345 builds, 9 clients. |
| 19:28:27 | rb-bluebot | puzzles: Disable "unfinished" puzzles for Windows builds by Solomon Peachy |
| 19:29:20 | MarcAndersen | Now that's out of the way, we just need to fix the crash. |
| 19:37:38 | | Join massiveH [0] (~massiveH@2600:4040:a982:5400:6c89:3e6d:b20a:4aa3) |
| 19:42:01 | rb-bluebot | Build Server message: Build round completed after 815 seconds. |
| 19:42:02 | rb-bluebot | Build Server message: Revision da8bde0860 result: All green |
| 19:49:22 | | Quit jn (Ping timeout: 248 seconds) |
| 19:50:48 | | Join jn [0] (~quassel@2a0a-a549-e4cd-0-20d-b9ff-fe49-15fc.ipv6dyn.netcologne.de) |
| 19:50:49 | | Quit jn (Changing host) |
| 19:50:49 | | Join jn [0] (~quassel@user/jn/x-3390946) |
| 19:54:39 | *** | Saving seen data "./dancer.seen" |
| 19:55:39 | MarcAndersen | I think I've made some progress. 0x8120 is AUDIO_F32LSB which is not defined in the switch statement that starts on line 106 in pcm-sdl.c. |
| 20:00 |
| 20:01:49 | | Join jacobk [0] (~quassel@2603:8080:b200:7b02::b71) |
| 20:03:14 | speachy | so.. it's saying that it _only_ supports 32-bit floating point data? that's insane |
| 20:03:34 | MarcAndersen | Can it be fixed? |
| 20:04:04 | speachy | https://discourse.libsdl.org/t/wasapi-backend-returns-audio-f32lsb-even-if-i-request-audio-s16sys/23007 |
| 20:05:10 | __builtin | SDL_audio offers on-the-fly interpolation/type conversion |
| 20:05:59 | __builtin | which the existing Rockbox SDL PCM code already uses |
| 20:06:11 | speachy | yeah, this is rather stupid honestly. |
| 20:06:23 | speachy | there's a way to make it "no, really give me what I want" |
| 20:06:23 | MarcAndersen | So then we just add it to the switch statement? |
| 20:06:36 | | Quit desowin (Ping timeout: 252 seconds) |
| 20:06:55 | __builtin | I believe there is - you can pass *obtained = NULL to SDL_OpenAudio |
| 20:07:05 | __builtin | but that probably breaks other assumptions in the PCM SDL driver |
| 20:07:37 | __builtin | I would just try adding AUDIO_F32* to the switch statement (with pcm_channel_bytes=4) |
| 20:09:49 | speachy | Returns a valid device ID that is |
| 20:09:51 | speachy | 0 on success or 0 on failure |
| 20:09:53 | speachy | |
| 20:09:55 | speachy | .... seriously? |
| 20:10:19 | speachy | no, the right thing to do is swithc to SDL_OpenAudioDevice |
| 20:11:15 | | Join desowin [0] (~desowin@rockbox/developer/desowin) |
| 20:12:59 | speachy | let me see if this still works.. |
| 20:16:49 | __builtin | speachy: I forgot to tell you that you to disable the unfinished puzzles, you should edit resync.sh to comment them out, as SOURCES.games is auto-generated by that. |
| 20:16:59 | __builtin | otherwise a future resync will overwrite your changes |
| 20:17:29 | speachy | btw the genhelp script goes boom quite badly |
| 20:18:36 | __builtin | you need a custom build of halibut installed locally for it to work. It needs to be a version with a custom patch I wrote to help parse the help text: https://www.franklinwei.com/git/halibut |
| 20:18:55 | __builtin | I need to update resync.sh to reflect the new URL |
| 20:19:14 | __builtin | (and perhaps document all this better in the wiki at some point!) |
| 20:19:17 | speachy | ah. in this case it was going boom due to header stuffs. |
| 20:19:40 | __builtin | yeah, the patch to halibut modifies how it outputs section headers, which are then parsed by genhelp.sh |
| 20:19:51 | __builtin | so it's very likely that. Do you have a vanilla halibut installed? |
| 20:20:05 | __builtin | (halibut = simon's homegrown documentation generator, by the way) |
| 20:21:06 | speachy | yes |
| 20:22:00 | __builtin | it's just this patch: https://www.franklinwei.com/git/halibut/commit/?id=c985a69bd38c244327c70af80ab9d04b07fbe8b3 |
| 20:22:24 | speachy | just ran into the libmad thing that bilgus hit, huh. asan doesn't like it. |
| 20:30:07 | speachy | now I'm getting a voice stack overflow. hah! |
| 20:31:05 | MarcAndersen | I just tried to add it to the switch statement. Now it runs but there's no audio. |
| 20:34:27 | rb-bluebot | Build Server message: New build round started. Revision b3dac27aa6, 345 builds, 9 clients. |
| 20:34:27 | rb-bluebot | SDL: Convert to SDL2's AudioDevice() API by Solomon Peachy |
| 20:35:32 | MarcAndersen | Guess what? I got it to work by setting set SDL_AUDIODRIVER=directsound ! |
| 20:36:15 | MarcAndersen | So I don't know if we need that commit now or? |
| 20:36:23 | speachy | try it with this ^^ patch |
| 20:36:59 | MarcAndersen | You mean the commit you just made? |
| 20:37:17 | speachy | yes |
| 20:37:33 | MarcAndersen | Ok. A few minutes then |
| 20:37:59 | speachy | apparnely SDL defaults to using a WASAPI backend that only speaks F32* |
| 20:38:41 | MarcAndersen | Then I have to remember to clear that environment var. |
| 20:47:45 | rb-bluebot | Build Server message: Build round completed after 799 seconds. |
| 20:47:46 | rb-bluebot | Build Server message: Revision b3dac27aa6 result: All green |
| 21:00 |
| 21:03:30 | | Join othello7 [0] (~Thunderbi@pool-100-36-176-164.washdc.fios.verizon.net) |
| 21:04:40 | MarcAndersen | It works, but it studders as hell |
| 21:05:54 | speachy | unfortunately we can't select which driver is used from within rockbox |
| 21:06:58 | rb-bluebot | Build Server message: New build round started. Revision 65d94ecd08, 345 builds, 9 clients. |
| 21:06:58 | rb-bluebot | SDL: Allow the audio device used to be specified on the command line by Solomon Peachy |
| 21:07:23 | MarcAndersen | What's that? |
| 21:07:58 | MarcAndersen | How did you put that in so quickly? |
| 21:07:59 | speachy | allows a choice of which _device_ gets used instead of "the default" |
| 21:08:29 | speachy | but it won't let you select which _driver_ is used |
| 21:09:09 | speachy | it does however print the list of allowed drivers, and of the selected driver, the list of detected devices. |
| 21:09:30 | MarcAndersen | I think that's it for now. Thank you very much for all the help! I will set up a new build server in the coming days and build fresh sims. |
| 21:09:43 | speachy | I can add help text that says sometihng like "if you have stuttering audio use SDL_AUDIODRIVER=<...>" |
| 21:10:09 | | Quit jacobk (Ping timeout: 260 seconds) |
| 21:10:51 | MarcAndersen | It was fun to track down the sdl bug. |
| 21:16:47 | speachy | it sucks how the default linux and windows SDL_AUDIODRIVER is so pathological for us |
| 21:19:54 | speachy | MarcAndersen: thanks for pointing us in the right direction. We probably have numerous other lurking platform variance differences lurking.. |
| 21:20:21 | rb-bluebot | Build Server message: Build round completed after 803 seconds. |
| 21:20:23 | rb-bluebot | Build Server message: Revision 65d94ecd08 result: All green |
| 21:20:33 | rb-bluebot | Build Server message: New build round started. Revision d483281fea, 345 builds, 9 clients. |
| 21:20:33 | rb-bluebot | SDL: Allow SDL_OpenAudioDevice() to override sample count by Solomon Peachy |
| 21:20:48 | MarcAndersen | I love rockbox! Thanks to you too. |
| 21:23:58 | | Join jn_ [0] (~quassel@2a0a-a549-eb45-0-20d-b9ff-fe49-15fc.ipv6dyn.netcologne.de) |
| 21:23:59 | | Quit jn_ (Changing host) |
| 21:23:59 | | Join jn_ [0] (~quassel@user/jn/x-3390946) |
| 21:24:07 | | Quit jn (Ping timeout: 264 seconds) |
| 21:28:38 | MarcAndersen | We have a problem in sims.pl: You don't appear to have sdl-config. It is renamed to sdl2-config |
| 21:32:23 | rb-bluebot | Build Server message: Build round completed after 711 seconds. |
| 21:32:25 | rb-bluebot | Build Server message: Revision d483281fea result: 70 errors 0 warnings |
| 21:32:34 | rb-bluebot | Build Server message: New build round started. Revision 1328464c86, 345 builds, 9 clients. |
| 21:32:35 | rb-bluebot | puzzles: Update resync script to work around WIN32 issues by Solomon Peachy |
| 21:33:15 | MarcAndersen | Oh no! |
| 21:43:47 | rb-bluebot | Build Server message: Build round completed after 674 seconds. |
| 21:43:49 | rb-bluebot | Build Server message: Revision 1328464c86 result: 30 errors 0 warnings |
| 21:45:36 | MarcAndersen | How could 40 errors disappear by that? That makes no sense to me |
| 21:54:43 | *** | Saving seen data "./dancer.seen" |
| 21:57:47 | _bilgus | WASAPI IIRC you have to set up a conversion stream thru the API to take your format to whatever WASAPI wants sample rate/ format |
| 21:58:18 | _bilgus | its not very fun to use but it IME is beter than anything here in linux land once you get it working |
| 21:59:24 | _bilgus | I wrote some autoit libraries so I could use the loopback to record tracks |
| 21:59:31 | | Join jacobk [0] (~quassel@47-186-65-73.dlls.tx.frontiernet.net) |
| 21:59:39 | _bilgus | https://www.autoitscript.com/forum/topic/197769-wasapi-audio-capture-w-loopback-what-u-hear-udf/ |
| 22:00 |
| 22:04:05 | _bilgus | speachy I have a patch up for libmad to test it on other device as is its 600bytes larger and 2-10% faster than ours I think if that carries with the other platforms itd be worth it to go with upstream |
| 22:04:14 | _bilgus | it sounds better too |
| 22:05:31 | _bilgus | In my mind that its faster probably negates the buffer you are losing |
| 22:07:58 | _bilgus | also in the same patch are the fixes for the casts to shut up asan, the size differeing one it only in the sim and I think is bc the asm has the wrong size var (also why I think its clicky staticky on the sim) |
| 22:33:10 | | Join jn [0] (~quassel@user/jn/x-3390946) |
| 22:33:12 | | Quit jn_ (Ping timeout: 272 seconds) |
| 22:33:30 | | Join chris_s [0] (~chris_s@2a09:bac3:2b55:126e::1d6:123) |
| 23:00 |
| 23:02:05 | | Quit massiveH (Quit: Leaving) |
| 23:22:18 | speachy | uh... I guess some of the builders have a very old version of SDL2. |
| 23:22:49 | speachy | pipewire seems to be the best of most worlds on Linux-land. |
| 23:23:51 | speachy | _bilgus: the platform that are hurting are the underpowered OG ipods and coldfire-based devices |
| 23:25:10 | speachy | yeah.. freemyipod-user890104 is using GCC7 (!) and presumably an equally ancient SDL2 |
| 23:26:55 | _bilgus | IDK I think they should try a test_codec with that patch its faster on this underpowered arm |
| 23:27:33 | chris_s | _bilgus: left you comments on the patch, unfortunately sounds distorted on iPod4g/video (using the upstream variant) |
| 23:28:11 | chris_s | kind of "warbly" |
| 23:28:22 | speachy | I think it was added in SDL2.0.6 ... in 2018 |
| 23:32:28 | speachy | _bilgus: At this point I'm more concerend with correctness across the board than speed. |
| 23:34:27 | _bilgus | #if 0 would be the upstream one correct? |
| 23:34:45 | chris_s | yeah |
| 23:35:04 | speachy | don't forget arm (and m68k) has a lot of asm optimizations that are likely unique to us, might be interdependent |
| 23:35:24 | rb-bluebot | Build Server message: New build round started. Revision da3d658880, 345 builds, 9 clients. |
| 23:35:25 | rb-bluebot | SDL: Work around build problems with SDL2 < 2.0.6 by Solomon Peachy |
| 23:35:28 | _bilgus | ours just sounds more muddled like the top end is lost |
| 23:35:34 | speachy | I don' tthink we have any mips optimizations |
| 23:35:57 | speachy | "muddled" .. in sims or also on the device? :) |
| 23:36:12 | _bilgus | I'm ok with either TBH but its for sure better on the sim (size mismatch of buffer I bet) |
| 23:36:14 | speachy | I mean mp3 ought to be easy to validate correctness. |
| 23:36:34 | _bilgus | on device its muddled more with head |
| 23:36:36 | speachy | same input should yield the same output |
| 23:36:41 | _bilgus | and sim is noiser too |
| 23:36:42 | chris_s | it sounds very obviously wrong on the iPods, i.e. not a subtle difference |
| 23:37:09 | _bilgus | well that sounds like a not better results |
| 23:37:51 | _bilgus | ok I'll push that with the proper casts this time |
| 23:43:31 | speachy | this "better" is independent of the SDL audio driver shenanigans, correct? |
| 23:47:49 | rb-bluebot | Build Server message: Build round completed after 746 seconds. |
| 23:47:51 | rb-bluebot | Build Server message: Revision da3d658880 result: All green |
| 23:54:47 | *** | Saving seen data "./dancer.seen" |
icon identifies that the person is a core developer (has commit access).