| 00:17:00 | | Quit jacobk (Ping timeout: 276 seconds) |
| 00:17:13 | | Join jacobk_ [0] (~quassel@47-186-65-73.dlls.tx.frontiernet.net) |
| 00:26:31 | | Join WebGuest26 [0] (~WebGuest2@2603:8000:502:4f7c:3ffc:7420:22bd:b57a) |
| 00:27:48 | | Quit WebGuest26 (Client Quit) |
| 01:00 |
| 01:55:53 | *** | Saving seen data "./dancer.seen" |
| 02:00 |
| 02:13:51 | | Join aaabbb_ [0] (sitku@bitcoinshell.mooo.com) |
| 02:19:07 | | Quit jacobk_ (*.net *.split) |
| 02:19:07 | | Quit sch (*.net *.split) |
| 02:19:08 | | Quit akaWolf (*.net *.split) |
| 02:19:09 | | Quit LjL (*.net *.split) |
| 02:19:10 | | Quit aaabbb (*.net *.split) |
| 02:19:10 | | Quit dbohdan (*.net *.split) |
| 02:19:10 | | Quit braewoods__ (*.net *.split) |
| 02:19:14 | | Quit Malinux (*.net *.split) |
| 02:23:33 | | Join akaWolf [0] (~akaWolf@akawolf.org) |
| 02:24:18 | | Join jacobk_ [0] (~quassel@47-186-65-73.dlls.tx.frontiernet.net) |
| 02:24:19 | | Join LjL [0] (~ljl@user/ljl) |
| 02:24:19 | | Join dbohdan [0] (~dbohdan@user/dbohdan) |
| 02:24:19 | | Join braewoods__ [0] (~braewoods@user/braewoods) |
| 02:24:19 | | Join Malinux [0] (~malin@2001:4641:4dfa::12c:c4a7) |
| 02:28:01 | | Quit amiconn (Quit: http://quassel-irc.org - Chat comfortably. Anywhere.) |
| 02:28:01 | | Quit pixelma (Quit: .) |
| 02:29:05 | | Join amiconn [0] (jens@p4fe76649.dip0.t-ipconnect.de) |
| 02:29:05 | | Join pixelma [0] (marianne@p4fe76649.dip0.t-ipconnect.de) |
| 03:00 |
| 03:55:57 | *** | Saving seen data "./dancer.seen" |
| 04:00 |
| 04:04:14 | | Quit jacobk_ (Ping timeout: 260 seconds) |
| 04:04:50 | | Join jacobk [0] (~quassel@47-186-65-73.dlls.tx.frontiernet.net) |
| 04:27:50 | | Join sch [0] (a2fe4b5ecb@irc.cheogram.com) |
| 04:37:28 | | Quit jn (Ping timeout: 245 seconds) |
| 04:37:58 | | Join jn [0] (~quassel@user/jn/x-3390946) |
| 04:46:29 | | Quit bertrik (Ping timeout: 248 seconds) |
| 05:00 |
| 05:30:48 | | Quit Maxdamantus (Ping timeout: 252 seconds) |
| 05:32:55 | | Join Maxdamantus [0] (~Maxdamant@user/maxdamantus) |
| 05:56:00 | *** | Saving seen data "./dancer.seen" |
| 06:00 |
| 06:26:19 | | Join othello7 [0] (~Thunderbi@pool-100-36-176-164.washdc.fios.verizon.net) |
| 06:44:05 | | Quit othello7 (Quit: othello7) |
| 07:00 |
| 07:56:02 | *** | Saving seen data "./dancer.seen" |
| 08:00 |
| 08:32:56 | | Quit tobbez (Ping timeout: 252 seconds) |
| 08:34:38 | | Join tobbez [0] (tobbez@user/meow/tobbez) |
| 09:00 |
| 09:56:03 | *** | Saving seen data "./dancer.seen" |
| 10:00 |
| 10:53:36 | _bilgus | speachy I think I finally have the asan warnings fixed pushing it through a function apparently makes ASAN happy, anyway with as fraught with reverts as this has been can you check it on your machine? |
| 10:53:44 | | Quit Arsen (Ping timeout: 252 seconds) |
| 10:53:59 | _bilgus | g#6197 |
| 10:54:03 | rb-bluebot | Gerrit review #6197 at https://gerrit.rockbox.org/r/c/rockbox/+/6197 : libmad remove warnings in sim by William Wilgus |
| 10:54:08 | | Join Arsen [0] (arsen@gentoo/developer/managarm.dev.Arsen) |
| 11:00 |
| 11:05:29 | _bilgus | actually making mad_f_mul or MLA/MLZ functions works too but this is less invasive / cleaner |
| 11:38:04 | _bilgus | adding FORCE_INLINE brings back the warnings so something about that interface makes it so UBSAN doesn't fall apart |
| 11:56:07 | *** | Saving seen data "./dancer.seen" |
| 12:00 |
| 12:33:59 | | Nick amiconn is now known as Guest1429 (jens@p4fe76649.dip0.t-ipconnect.de) |
| 12:33:59 | | Join amiconn_ [0] (jens@p200300ea8703b300305e95fffec66ff3.dip0.t-ipconnect.de) |
| 12:33:59 | | Quit Guest1429 (Killed (calcium.libera.chat (Nickname regained by services))) |
| 12:33:59 | | Nick amiconn_ is now known as amiconn (jens@p200300ea8703b300305e95fffec66ff3.dip0.t-ipconnect.de) |
| 12:34:44 | | Join pixelma_ [0] (marianne@p200300ea8703b300305e95fffec66ff3.dip0.t-ipconnect.de) |
| 12:34:44 | | Nick pixelma is now known as Guest5396 (marianne@p4fe76649.dip0.t-ipconnect.de) |
| 12:34:44 | | Quit Guest5396 (Killed (calcium.libera.chat (Nickname regained by services))) |
| 12:34:44 | | Nick pixelma_ is now known as pixelma (marianne@p200300ea8703b300305e95fffec66ff3.dip0.t-ipconnect.de) |
| 12:54:01 | _bilgus | so I did a little digging into playback in the sim looking at SDL_AudioSpec obtained from SDL_OpenAudioDevice() it returns back obtained.size at double the expected, we pass 2048 and 2048 *2 = 4096 but it shows 8192! |
| 12:54:57 | _bilgus | when I instead pass 1025 for wanted_spec.samples obtained.size returns 4096 and playback sounds good |
| 12:58:54 | _bilgus | 1024 sorry |
| 13:00 |
| 13:09:08 | | Join chris_s [0] (~chris_s@2a09:bac3:2f02:2478::3a2:d) |
| 13:11:42 | | Join lebellium [0] (~lebellium@2a01cb0405d07f00d33a0da62bfc2206.ipv6.abo.wanadoo.fr) |
| 13:21:29 | _bilgus | g#6198 is my working hack |
| 13:21:32 | rb-bluebot | Gerrit review #6198 at https://gerrit.rockbox.org/r/c/rockbox/+/6198 : [BugFix] SDL_OpenAudioDevice returns double the sample size expected WIP by William Wilgus |
| 13:23:36 | user890104 | should I split G#6195 to code changes and text changes, or it's fine as is? |
| 13:23:40 | rb-bluebot | Gerrit review #6195 at https://gerrit.rockbox.org/r/c/rockbox/+/6195 : Convert non-ASCII characters to UTF-8 or C hex literals by Vencislav Atanasov |
| 13:24:05 | paulcarroty | _bilgus: yep, 4x replugging helps, no ideas why. |
| 13:24:54 | paulcarroty | lifepo 18650 ftw |
| 13:25:00 | _bilgus | speachy has got a fuze+ up testing maybe he will find something there otherwise I'll keep trying with otherdevices as I have a chance |
| 13:26:19 | speachy | wait, so it's not the stock battery. LiFePO chemistry wasn't a thing at the time the fuze+ was built either, the curves are probably all wrong and the charger might also not be doing the right thing. |
| 13:27:04 | _bilgus | user890104, I guess that depends what chance you think it has of a revision IMO if you think its beyond possibility leave it if theres a chance I'd split it to keep it saner |
| 13:27:54 | _bilgus | paulcarroty, might try running a battery bench and plug in the new values if we don't see anything else |
| 13:28:24 | speachy | _bilgus: samples is not the buffer size, size needs to be samples * bytes_per_sample * channels |
| 13:29:36 | speachy | so.. the proper solution is to use the "obtained" size properly. I'll deal with this tonight |
| 13:29:55 | _bilgus | that would explain the doubled size but why then does it work? |
| 13:30:26 | speachy | probably has to do with how we fill the buffers |
| 13:30:47 | user890104 | _bilgus: I'll split it, since the code changes are far less than the documentation/comment changes |
| 13:32:00 | speachy | and stuff under plugins/ I'd rather not deviate from upstream |
| 13:32:44 | speachy | (where applicable) |
| 13:33:28 | _bilgus | https://github.com/emscripten-ports/SDL2/blob/324df6865ae3c7d194ed233a86867c48ec96c6a3/src/audio/SDL_audio.c#L1678C18-L1678C54 |
| 13:36:16 | _bilgus | I bet the pointer is too small, maybe it expects the pinter to be 32 bit? |
| 13:36:55 | _bilgus | does some stupid case macro then division by *ptr https://github.com/emscripten-ports/SDL2/blob/324df6865ae3c7d194ed233a86867c48ec96c6a3/src/audio/SDL_audiocvt.c#L569 |
| 13:38:48 | speachy | I highly doubt the problem is in SDL itself. |
| 13:39:28 | _bilgus | I'd lean towards we have something off but IDK tbh its odd |
| 13:39:29 | speachy | remmeber "obtained" is under control of the SDL audio *driver* itself |
| 13:39:53 | speachy | basically under (eg) pulse audio it wants a larger buffer than we request, and we don't handle that properly. |
| 13:40:04 | _bilgus | it also has a line where wanted is updated with the size obtained but that doesn't appear to happen |
| 13:40:21 | _bilgus | alsa does the same though |
| 13:40:41 | speachy | for me, alsa does the right thing (obtained == requested), ditto pipewire |
| 13:41:26 | _bilgus | extra weird |
| 13:41:26 | speachy | our actual decode/playback buffers are fixed IIRC. |
| 13:42:54 | speachy | I want to say I went through something like this with the hosted (alsa) driver. |
| 13:56:09 | *** | Saving seen data "./dancer.seen" |
| 14:00 |
| 14:14:58 | | Quit chris_s (Quit: Client closed) |
| 14:51:11 | _bilgus | speachy this sound familar?? https://github.com/libsdl-org/SDL/issues/2068 |
| 14:59:15 | _bilgus | I'm at 2.30 but it sounds like we can do the same pass NULL for obtained and force it to our desired format |
| 15:00 |
| 15:00:08 | | Quit toruvinn (Ping timeout: 252 seconds) |
| 15:13:08 | | Join toruvinn [0] (~toruvinn@KD119106027116.ppp-bb.dion.ne.jp) |
| 15:14:18 | _bilgus | no that doesn't help the items it checks for still match |
| 15:56:13 | *** | Saving seen data "./dancer.seen" |
| 16:00 |
| 16:18:23 | | Quit speachy (Quit: WeeChat 4.4.3) |
| 16:25:42 | | Join lebellium_ [0] (~lebellium@2a01cb0405d07f0064799244f1929087.ipv6.abo.wanadoo.fr) |
| 16:28:45 | | Quit lebellium (Ping timeout: 260 seconds) |
| 16:32:00 | | Quit jacobk (Ping timeout: 276 seconds) |
| 17:00 |
| 17:43:18 | | Join speachy [0] (~speachy@hurricane.shaftnet.org) |
| 17:43:19 | | Quit speachy (Changing host) |
| 17:43:19 | | Join speachy [0] (~speachy@rockbox/developer/speachy) |
| 17:43:19 | Mode | "#rockbox +v speachy" by ChanServ (ChanServ@services.libera.chat) |
| 17:43:40 | speachy | _bilgus: there doesn't seem to be anything amiss in our SDL code, from what I can tell. |
| 17:44:09 | speachy | ...just exhibits some pathological behavior with some audio backends.. |
| 17:45:59 | speachy | shrinking the requested sample count to 1K makes pulseaudio happy. the 2K default is about 5ms at 44KHz fwiw. |
| 17:50:46 | speachy | we default to 1K on hosted devices, bumped up to 2K on some more pathological ones |
| 17:50:59 | speachy | default is 256 (!) |
| 17:56:15 | *** | Saving seen data "./dancer.seen" |
| 17:56:58 | user890104 | speachy: if it's about the SDL target (200) I can test on pipewire (Ubuntu 24.04), if you tell me how to change the sample count |
| 17:57:19 | speachy | oh, pipewire seems to be happy with whatever |
| 17:57:36 | user890104 | I had choppy sound the last time I tested |
| 17:58:09 | user890104 | 256 is 625 uS, which sounds low to me |
| 17:58:19 | speachy | 256 is the default for bare metal |
| 17:58:42 | speachy | well, that's the "mixer buffer" size |
| 18:00 |
| 18:08:04 | | Quit lebellium_ (Quit: Leaving) |
| 18:09:27 | speachy | user890104: choppy sound, but what was the backend it was using? (it's now printed at startup btw) |
| 18:09:53 | rb-bluebot | Build Server message: New build round started. Revision e9b0aa495b, 345 builds, 9 clients. |
| 18:09:53 | rb-bluebot | SDL: special-case pulseaudio backend with a smaller PCM buffer size by Solomon Peachy |
| 18:10:30 | user890104 | I'll let you know as soon as I rebuild it, and remember what combination of PREFIX and make target I was using so it loads everything correctly |
| 18:10:52 | user890104 | I couldn't make it run from the build dir with the themes, plugins, codecs, etc. |
| 18:15:47 | user890104 | I'm making an SDL build, not a simulator build |
| 18:16:29 | user890104 | and the timing sounds like it's a bit slower, so about 1 second of delay for every 10-15 seconds of track (that's on mp3) |
| 18:16:48 | user890104 | basically the song is streched in time, without pitch change |
| 18:19:39 | | Nick aaabbb_ is now known as aaabbb (sitku@bitcoinshell.mooo.com) |
| 18:19:42 | | Quit aaabbb (Changing host) |
| 18:19:42 | | Join aaabbb [0] (sitku@user/aaabbb) |
| 18:23:25 | rb-bluebot | Build Server message: Build round completed after 812 seconds. |
| 18:23:26 | rb-bluebot | Build Server message: Revision e9b0aa495b result: All green |
| 18:33:20 | _bilgus | speachy works well here |
| 19:00 |
| 19:11:01 | speachy | EXCELLENT. :d |
| 19:11:07 | speachy | silly capslock |
| 19:45:34 | _bilgus | speachy look there at samples -> https://wiki.libsdl.org/SDL2/SDL_AudioSpec |
| 19:45:53 | _bilgus | Uint16 samples; /**< Audio buffer size in sample FRAMES (total samples divided by channel count) */ |
| 19:46:33 | speachy | yeah? |
| 19:46:40 | speachy | that's how we're using it |
| 19:46:53 | speachy | the callback function uses bytes though, not samples or frames. |
| 19:47:05 | _bilgus | but FRAMES |
| 19:47:31 | _bilgus | don't they have a macro to calculate that for you? |
| 19:49:44 | _bilgus | https://github.com/libsdl-org/SDL/issues/8265 |
| 19:56:18 | *** | Saving seen data "./dancer.seen" |
| 19:56:56 | | Join phantom_ [0] (~phantom@2600:1702:5b51:96e0:30c:262d:b238:fd48) |
| 19:59:34 | _bilgus | SDL_AUDIO_FRAMESIZE(stream->src_spec) |
| 20:00 |
| 20:00:19 | _bilgus | but I want to say src_spec is returned by SDL_GetAudioDeviceSpec but I may be off base there |
| 20:00:44 | speachy | we know what we asked for; if we didn't get it we automatically set up a coversion to create what we asked for |
| 20:01:04 | _bilgus | yeah and oddly it doesn't activate the conversion |
| 20:01:12 | speachy | because we get what we ask for. |
| 20:01:31 | speachy | we only allow samples to be changed |
| 20:02:11 | speachy | our code really doesn't care what 'samples' is; we just want to keep it as low as possible for latency reasons. |
| 20:03:12 | _bilgus | yeah it looks like it mallocs it when it asks anyway so IDK weird, almost wonder if someone chnaged something in anticipation of SDL3 issues that has screwed us in pulseaudio |
| 20:03:43 | speachy | pulseaudio has been iffy for a while now |
| 20:04:12 | speachy | with us, I mean |
| 20:04:36 | speachy | it's annoying that it doesn't prefer pipewire over pulseaudio |
| 20:04:42 | | Join massiveH [0] (~massiveH@2600:4040:a982:5400:501a:9d06:7095:afbf) |
| 20:07:01 | _bilgus | I mean we could make it prefer it.. |
| 20:11:10 | _bilgus | https://wiki.libsdl.org/SDL2/SDL_AudioInit |
| 20:21:05 | speachy | we'd then need special casing for every platform we run on. :D |
| 20:21:17 | speachy | let's see how this shakes out |
| 20:22:27 | | Join jacobk [0] (~quassel@2603:8080:b200:7b02:77cb:6304:f9db:dda1) |
| 20:22:46 | speachy | also rather leery of "don't use this function" |
| 20:31:14 | | Quit jacobk (Ping timeout: 260 seconds) |
| 20:44:08 | | Quit phantom_ (Quit: WeeChat 3.5) |
| 20:55:58 | | Join othello7 [0] (~Thunderbi@pool-100-36-176-164.washdc.fios.verizon.net) |
| 20:58:27 | | Join jacobk [0] (~quassel@47-186-65-73.dlls.tx.frontiernet.net) |
| 21:00 |
| 21:09:43 | | Quit jackie (Ping timeout: 252 seconds) |
| 21:10:20 | | Join jackie [0] (~jackie@banana-new.kilobyte22.de) |
| 21:32:40 | | Join dconrad [0] (~dconrad@152.117.104.232) |
| 21:39:28 | | Quit speachy (Quit: WeeChat 4.4.3) |
| 21:40:49 | | Quit rb-bluebot (Ping timeout: 248 seconds) |
| 21:41:24 | | Quit bluebrother (Ping timeout: 276 seconds) |
| 21:42:56 | | Join bluebrother [0] (~dom@user/bluebrother) |
| 21:46:31 | dconrad | _bilgus: looks like your changes to g#6147 work as advertised, at least when everything is on the same volume |
| 21:46:57 | dconrad | I have yet to try with things moving around to different partitions |
| 21:55:11 | | Join rb-bluebot [0] (~rb-bluebo@rockbox/bot/utility) |
| 21:56:22 | *** | Saving seen data "./dancer.seen" |
| 22:00 |
| 22:09:33 | | Quit winsim-buildbot (Ping timeout: 244 seconds) |
| 22:13:08 | dconrad | it took a minute to get my partitioning tool to let me do 2 partitions on an sd card but I got it |
| 22:16:20 | dconrad | seems like everything works as expected with 2 drives - except some weirdness in a corner case |
| 22:17:18 | | Join winsim-buildbot [0] (~lamest@85-218-172-116.norlyscustomer.net) |
| 22:17:52 | dconrad | oh no wait I wasn't testing what I thought I was testing, never mind |
| 22:18:03 | dconrad | yeah it seems like it's working as advertised in all respects |
| 22:20:31 | dconrad | interesting that <microSD0> acts like a "root drive", with <microSD1> underneath it, but on my main disk that only has one partition, I still have a <microSD0> for some reason? |
| 22:28:02 | dconrad | ah, it's due to my redirect file I suppose |
| 22:45:14 | dconrad | question, do we claim support for playlists which point to files/paths with non-ascii (utf of some flavor, I suppose?)? I came across this combination testset where I can play the file directly, but the playlist doesn't work (put the test set on the root of the drive): https://drive.google.com/file/d/1vKi8xRGrUvovUmvJqXDoL-s6kIyaMHqS/view?usp=sharing |
| 22:46:04 | dconrad | and the display of the title is messed up in the playlist viewer |
| 22:47:04 | dconrad | if it seems like a bug I'll put a bug report up |
| 22:52:44 | | Quit dconrad () |
| 23:00 |
| 23:06:28 | | Quit massiveH (Quit: Leaving) |
| 23:10:45 | _bilgus | I mean yes but there are going to be corner cases I imagine |
| 23:11:17 | _bilgus | dconrad if it works for you feel free to push it |
| 23:12:44 | _bilgus | speachy the rest of the text makes it sound less dire 'and should not be used unless you have a specific need to designate the audio driver you want to use.' |
| 23:16:09 | _bilgus | also was that not wanting to cange from upstream in plugins referring to libmad? we are upstream as far as whats in there unless you think I should replace upstream in the FSM_DEFAULT case with upstream? it works fine there and we could leave all the special stuff |
| 23:16:52 | _bilgus | tired of the wall of UBSAN it makes it hard to find the actual warnings |
| 23:56:23 | *** | Saving seen data "./dancer.seen" |
| 23:57:37 | | Quit _bilgus (Read error: Connection reset by peer) |
| 23:58:45 | | Join _bilgus [0] (~bilgus@syn-162-154-213-134.res.spectrum.com) |
icon identifies that the person is a core developer (has commit access).