Previous day | Jump to hour: 01 02 03 04 05 06 07 08 09 10 11 12 13 14 15 16 17 18 19 20 21 22 23 | Next day

Seconds: Show Hide | Joins: Show Hide | View raw
Font: Serif Sans-Serif Monospace | Size: Small Medium Large

Click in the nick column to highlight everything a person has said.
The Logo icon identifies that the person is a core developer (has commit access).

#rockbox log for 2016-11-27

00:08:08 Join Bilgus [0] (ae6611d9@gateway/web/freenode/ip.
00:09:43Bilgusjohnb2: yes pushing Vol+ and Vol- would turn on the backlight as it is an unfiltered button combination, just like play and select would do the same
00:10:29BilgusI could probably block unmapped button combos but I kinda felt like it was a good idea to leave them
00:15:31 Join JanC_ [0] (~janc@lugwv/member/JanC)
00:16:44 Quit JanC (Killed ( (Nickname regained by services)))
00:16:44 Nick JanC_ is now known as JanC (~janc@lugwv/member/JanC)
00:33:55 Quit pamaury (Ping timeout: 246 seconds)
00:42:05 Quit ulmutul (Quit: ChatZilla 0.9.93 [Firefox 50.0/20161104212021])
00:57:39 Quit lebellium (Quit: ChatZilla 0.9.93 [Firefox 50.0/20161104212021])
00:58:40***Saving seen data "./dancer.seen"
01:08:48 Quit paulk-collins (Quit: Leaving)
01:08:53dima_any clue when themes server might be up?
01:10:17[Saint]No. It's up/down sporadically, the issue is known, the factor is time.
01:10:26 Quit atsampson (Ping timeout: 245 seconds)
01:11:40 Join atsampson [0] (
01:21:56 Part robertd1
01:30:43dima_you guys wouldnt happen to have a local copy of the "terminal" theme would you?
01:37:37Bilgusprobably be a good idea to specify the player as well
01:38:33 Join nlogex [0] (
01:38:57dima_oh, does that matter?
01:39:00dima_its for the ipod 5g
01:41:51[Saint]yes, yes it does.
01:42:40 Quit Bilgus (Quit: Page closed)
01:55:58 Quit fujisan (Quit: Textual IRC Client:
01:57:22 Quit ender` (Quit: typeof(string).GetField("Empty").SetValue(null, "Empty");)
02:04:21 Quit ZincAlloy (Quit: Leaving.)
02:04:38 Join ZincAlloy [0] (~Adium@2a02:8108:8b80:1700:cd07:1b58:c282:4df)
02:04:46 Quit ZincAlloy (Client Quit)
02:05:36 Join ZincAlloy [0] (~Adium@2a02:8108:8b80:1700:dde4:f8ed:a838:a32)
02:05:36 Quit ZincAlloy (Client Quit)
02:24:49 Quit krnlyng (Ping timeout: 260 seconds)
02:37:20 Join krnlyng [0] (
02:37:25dima_can't seem to get the cache site working
02:37:34dima_keeps taking me back to the main theme page
02:58:44***Saving seen data "./dancer.seen"
02:58:46 Join Galois [0] (
03:05:24[Saint]it wouldn't be useful anyway, as all you would dig up are URIs to files on a server that isn't responsive.
03:13:09dima_so all the files are hosted on the same server then
03:13:29dima_that's disappointing
03:23:37 Quit megal0maniac (Ping timeout: 260 seconds)
03:23:52*__builtin has run into some design issues with some of these puzzles
03:24:09__builtinespecially this one:
03:28:47[Saint]define design issues?
03:29:10[Saint]I imagine pretty much everything without a DPAD-ish type control setup is fucked with that one.
03:29:14__builtinwell control
03:29:38__builtinthere's no way of controlling that, as you said, with a d-pad
03:29:51[Saint]DPAD to control selection, select to 'pick up' a handle, menu+select to drop it.
03:30:11 Join megal0maniac [0] (~megal0man@unaffiliated/megal0maniac)
03:30:29[Saint]and, of course, DPAD to move it when it is 'picked up'.
03:30:39__builtinwell, I guess
03:30:49__builtinalso there's the issue of fonts
03:31:14__builtinthe game code can request whatever size font it wants
03:31:28__builtinbut there's no easy way of specifying a font size in rockbox, is there?
03:31:32[Saint]heh - I drew a penis.
03:31:48[Saint]and, no, there isn't.
03:32:15__builtinshort of reading the entire fonts/ directory and tabulating every font's specs
03:32:16[Saint]though you could ship a font pack for the plugin specifically and add it to the game addons.
03:32:53[Saint]just do '01-font1_10pt.fnt', '02-font2-11pt.fnt' etc.
03:33:19[Saint]ugly as hell but it keeps it out of the shipped binaries.
03:33:32[Saint](because it would get massive)
03:33:46__builtinheck why not just pack them inside the binary and RLE them and write them out on disk and load them at runtime?
03:34:20__builtinthe plugin is already twice as big as the next biggest one
03:34:24[Saint]because that would make the binary needlessly massive and you know that like, maybe three people will ever care about this?
03:35:28[Saint]there's a very real reason we don't ship fonts except sysfont and the default. they get large, quick.
03:35:46__builtinare they compressed at all?
03:36:54[Saint]we actually made an effort to remove a bunch of themes and fonts because it made the shipped images huge.
03:37:09[Saint]i sincerely doubt anyone's gonna be too happy with you crammin' them back in there.
03:37:45__builtinso IIUC the fonts are just bitmaps stored without any compression?
03:38:18[Saint]Now, if the user can download them /if they want to/ with the game addons pack, that's a different story.
03:38:19[Saint]and yes.
03:38:37[Saint]AFAIK they're just standard bitmap fonts with an arbitrary extension.
03:39:14[Saint](actually, maybe .fnt is standardized? I don't /think/ so...but, maybe.)
03:39:46__builtinno, doesn't look like it
03:40:05__builtinthe magic number is "RB12"
03:41:56[Saint]Aha, you're right. Been a while since I used the bitmap font scheme.
03:42:15[Saint] might give you some more insight.
03:42:30[Saint]it links to more info on FNT which seems to be proprietary.
03:43:40[Saint]you can use AA fonts too, but this isn't going to make things any more friendly with the shipped image size.
03:55:19__builtinwhat's stopping actual truetype support?
04:00:12[Saint]lack of an actual truetype engine, presumably.
04:01:30dima_[Saint]: think i should just go to bed and hope its up tomorrow? i'm not sure how long its usually down so
04:02:22[Saint]Oh, I didn't realize you were waiting for this.
04:02:26[Saint] to not.
04:02:39[Saint]It'll be up sometime between soon and possibly never.
04:03:18dima_alright. ill just make do with what i have now and try again tomorrow. have a good one.
04:03:22 Quit dima_ (Quit: Page closed)
04:21:43 Quit nlogex (Ping timeout: 245 seconds)
04:48:18 Join alexweissman [0] (
04:58:46***Saving seen data "./dancer.seen"
05:09:16 Quit alexweissman (Ping timeout: 240 seconds)
05:56:26 Join alexweissman [0] (
06:08:04 Quit alexweissman (Remote host closed the connection)
06:25:05 Quit PurlingNayuki (Ping timeout: 260 seconds)
06:53:35 Quit TheSeven (Disconnected by services)
06:53:46 Join [7] [0] (~quassel@rockbox/developer/TheSeven)
06:58:49***Saving seen data "./dancer.seen"
07:03:16 Join PurlingNayuki [0] (~Thunderbi@
07:18:38 Join alexweissman [0] (
07:19:35 Quit PurlingNayuki (Ping timeout: 246 seconds)
07:21:07 Join PurlingNayuki [0] (~Thunderbi@
07:22:46 Quit alexweissman (Ping timeout: 250 seconds)
07:31:36 Quit PurlingNayuki (Ping timeout: 250 seconds)
07:33:56 Join PurlingNayuki [0] (~Thunderbi@
07:42:10 Quit PurlingNayuki (Ping timeout: 244 seconds)
07:43:02 Join PurlingNayuki [0] (~Thunderbi@
07:51:09 Join PurlingNayuki1 [0] (~Thunderbi@
07:51:22 Quit PurlingNayuki (Ping timeout: 250 seconds)
07:51:24 Nick PurlingNayuki1 is now known as PurlingNayuki (~Thunderbi@
07:59:21 Join PurlingNayuki1 [0] (~Thunderbi@
07:59:48 Quit PurlingNayuki (Ping timeout: 260 seconds)
07:59:49 Nick PurlingNayuki1 is now known as PurlingNayuki (~Thunderbi@
08:18:37 Quit PurlingNayuki (Quit: PurlingNayuki)
08:18:47 Join PurlingNayuki [0] (~Thunderbi@
08:33:32 Join athidhep [0] (
08:41:30 Join Miles [0] (63fe171e@gateway/web/freenode/ip.
08:41:41 Quit Miles (Client Quit)
08:48:21 Join ipod [0] (47ee5f61@gateway/web/freenode/ip.
08:48:36ipodhey the webpage for the apple docs file redirects to the support. its no longer there
08:48:44ipodcant figure out how to boot back into rockbox on my ipod video
08:52:33 Quit PurlingNayuki (Ping timeout: 268 seconds)
08:55:13 Quit ipod (Quit: Page closed)
08:58:51***Saving seen data "./dancer.seen"
09:05:56 Join PurlingNayuki [0] (~Thunderbi@
09:08:43[Saint]Seven minutes folks.
09:09:02[Saint]I guess he figured out that the manual explains this.
09:09:15[Saint]Or actually bothered rebooting.
09:09:44[Saint]Or, more likely, thinks we're all assholes for not answering immediately and had a hissy fit.
09:12:03 Quit PurlingNayuki (Ping timeout: 244 seconds)
09:21:52 Join PurlingNayuki [0] (~Thunderbi@
09:25:44 Join dima_ [0] (47ee5f61@gateway/web/freenode/ip.
09:26:06dima_[Saint]: i just got a crash or something
09:26:20dima_it says Data abort at 0004950C (0)
09:26:35dima_pc:0004950C sp:4000A7A0
09:26:42dima_a: 0001DF80
09:26:45dima_bt end
09:30:06[Saint]It happens. Those numbers are meaningless sadly.
09:30:26dima_and the themes are back up too
09:30:30dima_glad i stayed up
09:30:36[Saint]Those numbers required a mapfile to correspond them to anything useful.
09:30:51dima_is there any way to force a database rebuild?
09:30:59[Saint]We strip the mapfile from released builds for no obvious reason.
09:31:13[Saint]Sure. Check the database context menu.
09:31:27[Saint]Context, init db.
09:32:13dima_i dont see that
09:32:57 Quit PurlingNayuki (Ping timeout: 256 seconds)
09:32:57[Saint]You know what the context menu is?
09:33:20dima_sorry, last time i used this was literally like 3 years ago
09:33:23[Saint]Hold select on the Database main menu entry.
09:33:39*[Saint] gestures in the direction of our fine manual
09:33:41dima_ah got it
09:33:50dima_i tried reading that but...
09:33:56dima_its big
09:34:13[Saint]Yeah. My time is less valuable. Got it. :P
09:34:55[Saint]It's big, but it kinda has to be. There's a lot of features.
09:35:09[Saint]If you get the PDF version you can search it.
09:35:14dima_yeah, im starting to grasp the scope of this whole project
09:35:37dima_is there an application to edit metadata on the rockbox?
09:36:12[Saint]We don't manipulate metadata or any media outside of file naming.
09:36:27[Saint]Or moving it around.
09:37:12dima_i guess file moving is fine for now. drop it in a folder and mark it for editing next time im at a computer
09:37:39dima_good thing there is playlist support
09:39:14dima_using another program to scan the whole ipod and mass edit files wont corrupt it or anything right?
09:39:58 Join ender` [0] (
09:43:44[Saint]I can't speak for any given program's misgivings.
09:44:16[Saint]The database will deal gracefully with metadata changes.
09:44:32[Saint]And playlists are path based.
09:44:46[Saint]So...on our end. No. It won't hurt.
09:45:03dima_hold on what do you mean path based
09:45:08dima_they dont link over folders?
09:46:39 Join paulk-collins [0] (
09:48:23[Saint]Folders are part of paths.
09:48:40[Saint]Paths are comprised of them even.
09:48:54[Saint]Well...not necessarily. But often.
09:49:23dima_i thought you meant the files had to be in the same directory
09:58:20 Quit dima_ (Quit: Page closed)
10:14:43 Join xorly [0] (
10:16:13 Join pamaury [0] (~pamaury@rockbox/developer/pamaury)
10:23:18 Join PurlingNayuki [0] (~Thunderbi@
10:28:04 Join rela [0] (~x@pdpc/supporter/active/rela)
10:40:23 Quit PurlingNayuki (Ping timeout: 256 seconds)
10:58:45 Join ZincAlloy [0] (~Adium@2a02:8108:8b80:1700:d4a8:865c:8456:a59d)
10:58:55***Saving seen data "./dancer.seen"
11:12:16 Quit xorly (Ping timeout: 260 seconds)
11:15:31 Join lebellium [0] (
11:39:52 Quit ZincAlloy (Quit: Leaving.)
11:56:00 Join ZincAlloy [0] (
12:22:22duo8i wonder if rockbox could work on a typical pc
12:26:32[Saint]We have an SDL port. So, yes.
12:26:46[Saint]I does. Has for about a decade.
12:28:47 Join robertd1 [0] (~root@
12:49:46 Join Senji [0] (~Senji@
12:54:11 Quit Senji (Ping timeout: 246 seconds)
12:56:25duo8i mean bare metal
12:57:16pamauryyou could make it work, it's just a matter of having the proper drivers, which in the case of a PC is a real mess though
12:58:00pamauryfor the screen you could really on the bios, but for storage and audio, that's a whole different story
12:58:57***Saving seen data "./dancer.seen"
12:59:15[Saint]Beep that shit out in 8bit glory.
12:59:40pamaurylol, death to audiophiles, 8bit ought to be enough for everyone
13:03:48 Quit idonob (Ping timeout: 248 seconds)
13:06:04 Join idonob [0] (
13:14:58 Quit paulk-collins (Remote host closed the connection)
13:19:35 Join paulk-collins [0] (
13:19:35 Quit bluebrother (Read error: Connection reset by peer)
13:19:35 Quit fs-bluebot (Read error: Connection reset by peer)
13:22:51 Quit idonob (Ping timeout: 258 seconds)
13:24:46 Join bluebrother [0] (~dom@rockbox/developer/bluebrother)
13:25:07 Join idonob [0] (
13:29:46 Join xorly [0] (
13:36:38 Join fs-bluebot [0] (
13:39:23 Quit ZincAlloy (Quit: Leaving.)
13:42:35 Quit ender` (Quit: Sex is dirty - if you do it right. — Isaac Asimov)
13:51:15 Join ZincAlloy [0] (
13:58:51 Join alexweissman [0] (
14:10:17 Quit girafe (Read error: Connection reset by peer)
14:17:05 Quit alexweissman (Remote host closed the connection)
14:41:34 Join PurlingNayuki [0] (~Thunderbi@
14:42:27 Quit PurlingNayuki (Read error: Connection reset by peer)
14:42:32 Join PurlingNayuki1 [0] (~Thunderbi@2001:da8:215:4ff:79ce:22e4:f0db:fc25)
14:44:53 Nick PurlingNayuki1 is now known as PurlingNayuki (~Thunderbi@2001:da8:215:4ff:79ce:22e4:f0db:fc25)
14:47:31 Join cohokiller673 [0] (
14:55:13cohokiller673whats the latest method to put it on ipod classic 160gb and is that something i want to do if i want to sync with musicbee
14:58:22[Saint]Freemyipod docs clearly state the current method in their guided install documentation
14:58:59***Saving seen data "./dancer.seen"
14:59:17cohokiller673ok, that's where ill go
15:06:44[Saint]But I'm fairly confident mudicbee uses libgpod, no?
15:07:01[Saint]It should be able to sync with iOS
15:12:36cohokiller673i've tried and it works sometimes but most of the time it doesn't, sometimes tracks will sync but the ipod will be really slow and glitchy
15:13:59[Saint]You should probably be aware that Rockbox needs no form of "sync" at all.
15:14:23[Saint]Just drag and drop.
15:15:08[Saint]It'll build the database itself, optionally, by walking the disk for available media.
15:15:24[Saint]Or, there's just a file browser.
15:16:00[Saint]Just use plain UMS for media transfer.
15:16:12[Saint]No spooky magic required.
15:17:26[Saint]You only need to use the crazy iTunes compatibility sync magic if you intend to use the original firmware.
15:17:37[Saint]And...why do that? ;)
15:18:53 Join alexweissman [0] (
15:19:51cohokiller673so tired of itunes
15:20:33pamauryseriously, error 404 has a PURPOSE, display a nice popcorn bowl breaks tools
15:23:43cohokiller673i might be able to use my ipod with my car stereo's usb port once i get rockbox installed
15:25:50cohokiller673will it be a fat32 drive?
15:27:56[Saint]So long as you're not doing this on a Mac. Yes.
15:28:17[Saint]And...maybe, maybe not.
15:29:00 Quit ZincAlloy (Quit: Leaving.)
15:29:12[Saint]Some devices see the iPod USB VID/PID and don't even try to handle it as any other method despite it enumerating as mass storage.
15:29:29[Saint]Fair warning.
15:35:04 Join Senji [0] (~Senji@
15:41:47 Quit xorly (Ping timeout: 265 seconds)
15:47:19 Quit athidhep (Quit: athidhep)
15:52:10 Join xorly [0] (
15:55:16 Quit robertd1 (Ping timeout: 240 seconds)
15:56:30 Join robertd1 [0] (~root@
15:56:36 Quit alexweissman (Remote host closed the connection)
16:27:32 Join alexweissman [0] (
16:30:21 Join ZincAlloy [0] (
16:33:30 Join athidhep [0] (
16:33:49 Quit robertd1 (Ping timeout: 268 seconds)
16:36:09 Quit ZincAlloy (Quit: Leaving.)
16:38:57 Quit alexweissman (Remote host closed the connection)
16:40:51 Join robertd1 [0] (~root@
16:42:04 Quit athidhep (Quit: athidhep)
16:42:27 Join athidhep [0] (
16:43:52 Quit athidhep (Read error: Connection reset by peer)
16:51:42 Quit robertd1 (Ping timeout: 245 seconds)
16:54:36 Join robertd1 [0] (~root@
16:55:41 Quit Ivoah (Quit: Connection closed for inactivity)
16:58:47 Quit robertd1 (Ping timeout: 245 seconds)
16:59:03***Saving seen data "./dancer.seen"
16:59:24 Join robertd1 [0] (~root@
17:07:57 Quit robertd1 (Ping timeout: 245 seconds)
17:12:04 Join robertd1 [0] (~root@
17:19:16 Quit robertd1 (Ping timeout: 248 seconds)
17:33:21 Join ZincAlloy [0] (~Adium@2a02:8108:8b80:1700:45d3:8dbc:8450:ff5)
17:41:32pamaurygevaerts: do you know curl well?
17:43:51*__builtin points to bagder in #rockbox-community
17:49:20*pamaury is ending up rewriting at this point
17:49:40__builtinwhat's wrong with it?
17:52:31pamaurynothing per-se, just it's not very flexible, it was written with bare-metal toolchains in mind
17:52:41pamauryand want to build a linux toolchain
17:53:05__builtinah, that makes sense
17:53:21__builtinjust don't rm -rf all your toolchains this time
17:54:51pamaurywell that was a ct-ng "feature"
17:55:09pamaurybut I'm rewriting the toolchain build without ct-ng
17:55:26pamauryalthough in the process I've come to understand why people use ct-ng ;)
18:05:35 Join wodz [0] (
18:06:17wodzpamaury: So why people use ct-ng? Really, I am curious
18:10:25pamaurywodz: because building a linux toolchain involves GNU/magic
18:11:16pamauryseriously, this is what you need to give to configure to build a first stage GCC (I'm sure it changes at every version):
18:11:16pamaurybuildtool "arm-sony-linux-gnueabi" "gcc" "$gcc_ver" "$gcc_opts −−enable-languages=c \
18:11:16pamaury −−without-headers −−disable-threads −−disable-libgomp −−disable-libmudflap \
18:11:16DBUGEnqueued KICK pamaury
18:11:16pamaury −−disable-libssp −−disable-libquadmath −−disable-libquadmath-support \
18:11:16pamaury −−disable-shared −−with-newlib −−disable-libgomp −−disable-libitm \
18:11:18pamaury −−disable-libsanitizer −−disable-libatomic" "" ""
18:11:49pamauryI still fail to see why there isn't a "−−stage1" option given how common this operation is
18:11:59wodzpamaury: sure, but ct-ng doesn't solve this really. It does not document the steps needed. Last time I looked it was a bunch of 'recipes' which break here and there.
18:12:35pamauryyeah but that's the magic of it: people went to great length to figure out the magic commands at every step, because those are completely undocumented anyway
18:13:04wodzpamaury: exactly this, instead of providing insane set of commands why not to document it AND try to push patch to sanitize this
18:14:29pamauryyeah I don't understand why gcc doesn't have a saner way of doing it, maybe the more recent versions do, but 4.9.4 clearly doesn't. In theory it has the −−without-header option but this is far from enough
18:21:16pamaurymy feeling is that gcc would gain from not being written in GNU/C using GNU/autocrazytool with GNU/crazyconfig with a GNU/unhelpful spirit ;)
18:23:01wodzIt is known fact that gcc is arcane. Thats one of the reason llvm/clang took off (and license thing as well)
18:36:27pamauryit's such a shame that glibc doesn't build with clang
18:36:54pamauryalso we should try again clang, maybe I'll try clang on arm to see if it works
18:47:04wodzpamaury: The problem is that we are basically clueless if it doesn't work.
18:48:28wodzpamaury: btw.
18:49:18pamaurywodz: there was a time it didn't work because clang didn't know about attribute((interrupt)). I think last time it did but I couldn't spot anything obviously wrong in the generated code
18:52:39wodzpamaury: I mean you'd need jtag or something to figure out whats wrong.
18:59:03 Quit Senji (Read error: Connection reset by peer)
18:59:04***Saving seen data "./dancer.seen"
19:00:17 Join Senji [0] (~Senji@
19:00:34 Quit Senji (Read error: Connection reset by peer)
19:01:01 Join Senji [0] (~Senji@
19:04:38 Join girafe [0] (
19:29:16 Join nlogex [0] (
20:03:58 Quit wodz (Ping timeout: 246 seconds)
20:11:59[Saint]speaking of things that don't build things.
20:12:16[Saint]wasn't gcc failing to build itself until fairly recently?
20:12:21[Saint]or possibly still?
20:21:51pamauryI think gcc N and N+1 can always build each other
20:22:08pamaurythe problem is that gcc from now can't build real old gcc
20:24:03 Quit michaelni (Ping timeout: 250 seconds)
20:24:11pamauryand sometimes not even that old, for example GCC 6.1 couldn't build some really recent GCC because of some stupid code that used non-standard C
20:24:25pamaury(that GCC 6.1 finally dropped)
20:36:53 Join michaelni [0] (
20:43:51 Quit alucryd (Remote host closed the connection)
20:44:14 Join wodz [0] (
20:46:07 Join alucryd [0] (~quassel@archlinux/developer/alucryd)
20:59:05***Saving seen data "./dancer.seen"
21:04:55 Join Miles [0] (63fe171e@gateway/web/freenode/ip.
21:07:49 Quit wodz (Ping timeout: 250 seconds)
21:13:04MilesAnybody know about Rockbox's AAC-HE support? (on Clip+) I'm trying to find out if it actually does the spectral band replication or just decodes the LC part of the stream.
21:15:39MilesAlso is it generally considered polished now? Saw a bunch of complaints about it but they were mostly from years ago.
21:32:35 Quit xorly (Ping timeout: 248 seconds)
21:41:38 Quit tuaim (Quit: leaving)
21:48:10 Join saratoga [0] (126f650e@gateway/web/freenode/ip.
21:48:28saratogaMiles: SBR is decoded except on targets with not enough RAM (SBR uses a lot of RAM)
21:48:31 Join Miles_ [0] (63fe171e@gateway/web/freenode/ip.
21:48:53saratogai think of the Sandisk players that is just the Clipv1 and m200v4 which can't do SBr
21:49:48saratogaolder players will try to decode SBR, but often cannot in realtime due to lack of CPU power
21:50:05saratogawe got very close to getting it working on the old PP devices, but i think it was never quite 100%
21:50:49 Quit Miles (Ping timeout: 260 seconds)
21:51:00 Quit Miles_ (Client Quit)
21:51:13 Join Miles [0] (63fe171e@gateway/web/freenode/ip.
21:51:23MilesSeen Bilgus by the way?
21:51:30MilesBeen trying to get a hold of him.
21:52:12saratogaoh you posted those test files right?
21:53:27saratogai looped that folder for 48 hours on the Clip+ but could not reproduce anything weird (unless it happened briefly and resolved itself between times I checked)
21:53:36MilesI wanted to report Bilgus' build with the modified CPU boost behaviour was working well for me.
21:53:51MilesI wonder if he ever managed to reproduce the issue/
21:54:25MilesYou were still able to pause or skip at any point?
21:54:32saratogayeah, UI was normal
21:55:02saratogawell normal when i checked, no idea if it was weird for a bit in the middle of the night or something
21:55:04MilesThanks for trying at least.
21:55:33saratogai still think the best thing to do would be to reset your settings then try to reproduce it with some minimum set of files
21:56:20MilesI guess I could do that with a spare microSD.
21:56:59MilesI actually didn't test to see if my card or the rest of the collection had anything to do with it yet.
21:57:13saratogamost likely you only need one or two files and to set the player to repeat
21:57:41saratogayour settings may also matter
21:58:02saratogabut anyway, if boost matters, it sounds like the codec isn't yielding enough under some conditions
21:58:37saratogawe have slower players than the clip+ as well, i should probably test one of those
21:58:39MilesIf the tweaked firmware works with the same collection and settings as before now though, I'd imagine something is screwy with Rockbox itself somewhere probably right?
21:59:13saratogaif boosting fixes the problem (and assuming the problem is actually replicable enough that we think its fixed), it is probably something in the codec itself
21:59:48saratogaopus doesn't yield at all right now, it just decodes a frame of audio and then returns, its possible under some conditions that takes too long and starves the UI thread
22:00:34saratogai haven't worked on the front end of the codec at all (just the transform parts) so i'm not sure how big of a problem that is
22:00:43MilesOh that's interesting.
22:00:57MilesI use a framesize of 60 to encode my opus when the default is 20.
22:01:08MilesI wonder if that's why my files are a little more crashy.
22:01:45saratogathat would make sense
22:01:57saratogaour test files for opus use 5 and 20 ms
22:02:15saratogai have no idea what that actually changes
22:02:51saratogacan you post a link to a 60 ms frame size file? I can't look now, but i'll try later
22:03:01saratogaframe length rather
22:03:09MilesWell my sample set I shared here before was all 60.
22:03:14MilesI can find the link again.
22:03:18saratogaoh okk
22:03:19saratogai have those
22:03:42saratogaif 60 literally means that it takes 3 times as long as 20 to decode one frame, that could be a problem
22:03:49saratogaalthough i don't know if it really works that way
22:04:06MilesIIRC it does.
22:04:39saratogai remember that decoding 5 ms files was much slower than 20 ms
22:04:57saratogatotal time, not time per frame
22:05:40MilesWell I know filesystem access can bottleneck the thing sometimes.
22:07:01Miles...but it probably wouldn't matter in that case now that I think about it.
22:07:04[Saint]Miles: what pray tell is the 'modified boost behavior'?
22:07:09[Saint]genuinely curious.
22:07:32MilesBilgus gave me a firmware where CPU boost is 10 seconds instead of 1.
22:07:58saratogalooking at the documentation, the celt mode of opus uses 20 ms frames at most, so i guess if you specify longer it just packs 2 or 3 together, which probably is linearly longer to decode
22:08:02[Saint]and, yeah, it's really hard to imagine any codec on any target where IO is a legitimate concern for bottlenecking.
22:08:26saratogaalthough this suggests that if you're using celt mode opus you probably should leave the frame length at 20 ms
22:10:11MilesYeah the 60ms things is mostly out of habit since I use lower bitrates for other stuff. 96 kbps the overhead from frame headers every 20ms should be totally negligible.
22:11:00MilesI will try that later then with the unmodified nightly firmware.
22:11:24MilesAnd a folder with the same songs with 60ms frames.
22:14:25saratogaback of the envelop calculation is that 20 ms uses about 1 million cycles per frame, so 60 would be 3 million cycles, or 75 milliseconds unboosted
22:14:35[Saint]the more I look at it the more it seems that pretty much nothing will gain from the (admittedly very limited) work I've done on intermediate scaling.
22:14:43[Saint]pretty much ready to throw this in the toilet.
22:14:55saratogaor probably a little less since the CPU is more efficient unboosted, but roughly 50-100 ms is a good estimate
22:15:27saratoganot sure why it doesn't glitch for a few milliseconds and then boost though
22:15:46[Saint]there's not really a hell of a lot of targets with room to scale efficiently in a linear way.
22:15:54CtcpIgnored 1 channel CTCP requests in 0 seconds at the last flood
22:15:54*[Saint] grumbles
22:16:04saratogai guess there the reason for that is probably the reason you have this problem and I don't
22:16:12[Saint]not without a whole bunch of futzing around.
22:16:13saratogamaybe some settings ?
22:16:34saratogai wanted to do intermediate scaling for a while, but it is a lot of work for modest gain
22:16:55[Saint]probably you thought of the things I'm finding now slightly quicker.
22:17:06saratogaa simplier way would be to have 2 boost levels but let codecs or plugins specify which 2 (so for example mp3 could request 2 lower speeds, but APE the fastest)
22:17:37saratogaso all you have to do is implement 2 different boost clocks, and an API for switching between them
22:18:47[Saint]hmmm, that's actually pretty clever.
22:19:32[Saint]I would probably also use the 'middle' clock for GUI_BOOST
22:20:02[Saint]very few targets need to boost the GUI as aggressively as it does. A couple do boost it but don't need it at all in fact.
22:20:09saratogathat gets tricky though, since now you need a way to keep track of 3 levels at once
22:20:26saratogawhereas our boost counter assumes only 2
22:20:27[Saint]ipod6g can safely go without GUI_BOOST entirely.
22:20:40saratogathat means you should lower the clock!
22:20:48[Saint]it was just inherited as being enabled for all iPods.
22:21:15saratogadidn't prof_wolff have a patch for the 6g that massively reduced the voltage/clocks ?
22:21:16[Saint]n2g likely doesn't need it either.
22:21:32saratogai vaguely remember that you could get even more ridiculous battery life by better power management
22:21:44[Saint]yes, he did, it was far too aggressive and was hit or miss on individual targets.
22:22:02[Saint]riiiight on the edge of stable.
22:22:07saratogaby the way, buschel wanted this on the old ipod video years ago, so he could do 20 MHz low, and then 100 MHz boost IIRC
22:22:24saratogasince 30 MHz actually wasted a fair amount of power
22:22:35saratogaor maybe it was 16mhz
22:22:36saratogai forget
22:23:24[Saint]actually, what's with that? I have noticed on the iPod video, and maybe others, we /do/ actually have three scaling frequencies.
22:23:32[Saint]but you can only select 24MHz from the boost debug menu.
22:23:45saratogamaybe an idle frequency? i forget
22:23:57saratogawhat are the clocks?
22:23:58[Saint]and it'll go back to 30 immediately when you leave.
22:24:12saratoga30/80 and what is the last? 24?
22:24:19[Saint]24, 30, and...whatever MAX is...80?
22:24:39saratogai vaguely remember that 24 is used when the screen is off and playback is disabled
22:24:49saratogaprobably 24 should just replace 30 now that we have gui boost
22:24:59[Saint]likely so.
22:25:16saratogaMP3 playback is multithreaded, so it only needs ~20 MHz for realtime
22:25:31saratogathe extra 10 are wasted when not driving the GUI
22:28:05 Join xorly [0] (
22:30:09 Quit uwe_ (Ping timeout: 250 seconds)
22:30:28 Join uwe_ [0] (
22:31:15 Quit puckipedia (K-Lined)
22:32:40 Join puckipedia [0] (
22:50:30 Quit saratoga (Quit: Page closed)
22:51:07 Quit Senji (Ping timeout: 250 seconds)
22:55:39 Quit paulk-collins (Quit: Leaving)
22:57:25 Join Miles_ [0] (63fe171e@gateway/web/freenode/ip.
22:59:06***Saving seen data "./dancer.seen"
23:08:59 Quit Miles_ (Ping timeout: 260 seconds)
23:20:53 Quit lebellium (Quit: ChatZilla 0.9.93 [Firefox 50.0/20161104212021])
23:48:43 Join JanC_ [0] (~janc@lugwv/member/JanC)
23:49:57 Nick JanC is now known as Guest29593 (~janc@lugwv/member/JanC)
23:49:57 Quit Guest29593 (Killed ( (Nickname regained by services)))
23:49:57 Nick JanC_ is now known as JanC (~janc@lugwv/member/JanC)
23:58:53cohokiller673Saint, couldn't I change the VID/PID of my device somehow so it wasn't recognized as an ipod?

Previous day | Next day