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 2012-07-06

00:01:00Tornehm, yeah, VLOG() just uses __FILE__ and then there's a helper function that matches it against globs at runtime
00:01:14Torneso you can enable, say, */chromeos/*
00:01:51 Join perrikwp_ [0] (
00:01:57gevaertsThe best way would be to not use absolute paths in the build system, but that doesn't sound like a lot of fun to change
00:02:16Tornewait, we're using absolute paths?
00:02:33Tornewell, no, not particularly.
00:02:59 Quit perrikwp (Ping timeout: 244 seconds)
00:04:04saratoga__i when I amend something to gerrit, do i have to git add file1.c file2.c again?
00:04:22saratoga__hmm no i guess not
00:04:28Torneyou have to add anything you change, whether you're making a new commit or amending one
00:04:36Tornethat's how it knows what should become part of hte commit
00:04:46 Join [Saint__] [0] (~Saint]@unaffiliated/saint/x-8516940)
00:04:53Torne−−amend just makes it *also* include the content of the previous commit as well as whatever you have added to the index since
00:04:57saratoga__so if it was added to the previous commit i have to readd it after amend?
00:05:28Tornei'm not sure what you mean
00:05:42Torneyou have to add everything you change, always, no matter whether you are amending or not
00:05:53 Join perrikwp [0] (
00:05:58saratoga__i mean does the previous commit unadd things?
00:06:13saratoga__for instance on SVN if i did svn add file.c i don't ever have to do that again
00:06:18saratoga__but i guess on git i do?
00:06:37Torne`git add` is not about creating files, it's about adding *changes*
00:06:42Torneanything you change, you have to add to the index
00:06:48Tornein order for it to be included in a commit
00:07:09Torneif you've changed something since the commit you already made, you need to add it to the index before amending
00:07:14saratoga__so its not about tracking a single file, but rather it actually copies the contents out of the file when i do that?
00:07:16Torneor else it won't include that in the new amended commit
00:07:23saratoga__ok that makes more sense
00:07:30Tornethe index is "the next commit"
00:07:34Tornemore or less.
00:07:35 Quit [Saint_] (Ping timeout: 240 seconds)
00:07:48Torne`git commit` just saves the index as a commit on the current branch
00:07:54saratoga__also, is it possible to put a comment after a specific commit to gerrit besides "Uploaded patch set 5."
00:08:05Tornewrite a comment yourself
00:08:28 Quit perrikwp_ (Ping timeout: 264 seconds)
00:08:43Torneyou can just edit the commit message, also
00:08:58saratoga__yeah but that changes the message at the top of the gerrit task, correct?
00:11:28 Quit prof_wolfff (Ping timeout: 264 seconds)
00:13:32 Quit XavierGr ()
00:16:13saratoga__awesome, latest push to gerrit is working nicely
00:16:28saratoga__errors get logged when the disk next spins up, and warnings get compiled out
00:17:21saratoga__i wonder if it makes sense to include a time stamp
00:18:02 Quit domonoky (Read error: Connection reset by peer)
00:20:52 Quit bertrik (Quit: And That, My Liege, Is How We Know the Earth to Be Banana Shaped)
00:23:07 Quit perrikwp (Read error: Connection reset by peer)
00:23:48 Quit Torne (Quit: process in uninterruptible sleep in do_exit :()
00:25:27 Quit ender` (Quit: How does Moses make his tea? HEBREWS IT. Then he sells it for prophet.)
00:25:48 Join Torne [0] (~torne@rockbox/developer/Torne)
00:26:21 Quit pamaury (Remote host closed the connection)
00:26:50 Join perrikwp [0] (
00:27:26 Quit [Saint__] (Remote host closed the connection)
00:32:22 Join perrikwp_ [0] (
00:33:56 Quit kevku (Quit: KVIrc 4.2.0 Equilibrium
00:34:22 Join perrikwp__ [0] (
00:34:55 Quit perrikwp (Ping timeout: 255 seconds)
00:37:10 Quit perrikwp_ (Ping timeout: 246 seconds)
00:37:27 Join perrikwp [0] (
00:39:52 Quit perrikwp__ (Ping timeout: 256 seconds)
00:42:20 Join perrikwp_ [0] (
00:42:35 Quit perrikwp (Ping timeout: 240 seconds)
00:51:03 Quit n1s (Quit: Ex-Chat)
00:52:26 Quit lebellium (Quit: ChatZilla [Firefox 14.0/20120628060610])
00:57:26funmanwhy not, we could use the tick number
01:14:24saratoga__funman: don't we have an actual RTC on most targets?
01:14:38funmantrue but the tick has more precision
01:15:02saratoga__i was mostly thinking of a time stamp to let someone know that an error they're seeing in the log is from today not 3 weeks ago
01:15:24saratoga__i guess most users might accumulate very large log files over time
01:16:52saratoga__also, i wonder if it makes sense to offer the LOGF style viewer in the debug screen
01:17:11saratoga__i guess not if you can just view the log with the text viewer plugin
01:17:50saratoga__some kind of message if the buffer overflows letting people know that log entries were lost would be nice too
01:27:09***Saving seen data "./dancer.seen"
01:38:33 Join perrikwp [0] (
01:41:17 Join perrikwp__ [0] (
01:41:28 Quit perrikwp_ (Ping timeout: 264 seconds)
01:43:20 Quit perrikwp (Ping timeout: 252 seconds)
01:46:00 Join perrikwp [0] (
01:48:38 Quit perrikwp__ (Ping timeout: 250 seconds)
01:50:01 Join perrikwp_ [0] (
01:50:37 Quit perrikwp (Ping timeout: 265 seconds)
01:52:15 Join perrikwp [0] (
01:54:42 Quit perrikwp_ (Ping timeout: 252 seconds)
01:56:28 Join perrikwp_ [0] (
01:59:02 Quit perrikwp (Ping timeout: 250 seconds)
02:13:39 Quit shamus (Read error: Connection reset by peer)
02:13:52 Join shamus [0] (
02:16:11 Join perrikwp [0] (
02:18:52 Join perrikwp__ [0] (
02:18:54 Quit perrikwp_ (Ping timeout: 252 seconds)
02:19:17 Quit perrikwp (Read error: Operation timed out)
02:27:21 Quit MethoS- (Ping timeout: 265 seconds)
02:28:19 Quit perrikwp__ (Ping timeout: 265 seconds)
02:31:40JdGordonsaratoga__: stopping logging if the buffer is full is probably not the best idea
02:32:02JdGordonlinux uses the buffer as a simple ringbuf so the last items will always be in the log which is far more useful
02:40:06 Quit liar (Ping timeout: 245 seconds)
02:40:54 Quit soap (Ping timeout: 252 seconds)
02:54:44 Join saratoga [0] (98032941@gateway/web/freenode/ip.
02:55:01saratogaJdGordon: yeah I was thinking of using a ring buffer, but didn't get around to it
02:55:20saratogaprobably won't matter though, since hopefully you're not filling the buffer that fast
02:56:04JdGordon60MB buffer can hold an hour of audio right?
02:56:35saratogai don't suppose you're going to be interested in going through the theme engine and adding logging ?
02:56:41saratogain PCM?
02:58:18JdGordonno, compressed buffer
02:58:27JdGordonand yeah, I'll for sure add logging
02:58:58saratogayeah its a little over an hour at 128k
02:59:46saratogai'd like to see the playback, buffering, and buflib systems log more of their state
03:00:00saratogaand see if we find more rare bugs and error conditions
03:00:04JdGordonso thats potentially an hour of logs without being flushed
03:00:20saratogayeah true
03:00:40saratogai was thinking about having a variant that spins up the disk and flushes immediately
03:00:45saratogain case you're about to crash
03:01:52saratogafor debug use though we can just make the log buffer really big, and for everyday users they should rarely generate log info unless they have a broken file or something
03:07:50JdGordonwhy doesnt __FILE__ work with the build system?
03:09:02[Saint]damn you GNU Unifont!
03:12:48*[Saint] would really like to be able to use different font sizes (and/or styles) in a theme without hardcoding them.
03:13:36[Saint]Or, to be able to apply a font conditionally.
03:13:40saratogaJdGordon: file returns the entire path you gave gcc, which in our case is an absolute path that changes for each build system
03:14:02[Saint]GNU Unifont is *such* an ugly font, I really don't feel good about using it by default.
03:14:05saratogaso if we compile that into each file the build size will vary by a rather large amount depending on the path each person uses for their build client
03:14:22saratoga[Saint]: make a better default font
03:14:23JdGordonsaratoga: ah right, ok... can you add __LINE__ do the macro thouhg?
03:14:45saratogagevaerts was working on that, he thought it would require tweaking the build system
03:14:57saratogaor make files rather
03:15:07saratogathere is nothing built into GCC that does what we want tho
03:15:15[Saint]saratoga: I've made plenty of beautiful ttf fonts, that look really great on target, but the "problem" is character coverage.
03:15:15JdGordonalso, how should I use it in the skin engine? if only ERROR is enabled for release builds, and they arent really errors...
03:15:31[Saint]GNU Unifont is the only font that covers all langs.
03:15:38[Saint]but its hideous.
03:15:41saratoga[Saint]: can you mix what you have with gnufont?
03:15:57[Saint]come again?
03:16:05saratogabasically fix up the characters you are familiar with and use what it provides for the character sets you haven't worked with
03:16:20saratogaalso, maybe cut out some of the more uncommon character sets so its not so gigantic
03:16:35saratogaJdGordon: hmm i'm not sure
03:16:44saratogamy intention was that we'd only log errors
03:16:52saratogawhat kind of stuff were you thinking?
03:17:06saratogaperhaps warnings by default makes sense too?
03:17:09JdGordonwell in your header ERROR is for fatal errors, which broken skins arenyt
03:17:13[Saint]saratoga: I've been avoiding doing so, as its quite a daunting task.
03:17:29JdGordonalso, I dont want to compile in the skin debugging code which checkwps has
03:17:30[Saint]I would really like a better font system :)
03:18:05saratogaJdGordon: i don't think it has to be a fatal error
03:18:20saratogai figured decoder errors would trigger that, even if they can be recovered by skipping through the bitstream
03:18:31[Saint]So, instead of a theme hardcoding a specific font, it could just specify a size, and the user can select which font they use.
03:18:40[Saint]that would be sweet...
03:18:51JdGordonsaratoga: ok
03:18:54saratogacan't you already change the font (provided the size doesn't change)?
03:19:01JdGordonturns out i misread the header... serious != fatal
03:19:01saratogaperhaps i should clarify the description
03:19:11[Saint]saratoga: only the userfont.
03:19:21[Saint]a theme can hardcode ~9 other fonts.
03:19:33saratogado you know c? fix it ;)
03:19:46saratogawe could also add more warning levels too if you think its worthwhile
03:19:46[Saint]I don't know *enough* C for this, no :)
03:20:34JdGordon[Saint]: what you want is just a matter of programming :p
03:20:39JdGordonand how to do the ui for it
03:20:54[Saint]"just" :)
03:21:56JdGordonsaratoga: actually, I think if the buffer is full and it is an ERROR it should flush the buffer and then add the new one, if its < ERROR then ignore it
03:22:13JdGordonan extra disk spinup isnt worse than losing logging
03:22:40saratogai was worried about a situation where someone has a bad file on a playlist and keeps spinning up the disk for no reason to log the same info
03:22:52saratogai kind of dislike the idea of allowing this system to increase power use
03:22:59saratogaif you're worried about that, lets just make the buffer bigger
03:23:16JdGordonanother option is to use buflib handles for each message
03:23:23saratogawhats the most log you think you could reasonably make in an hour?
03:23:27JdGordonbut then we dont complelty trust buflib to be bug free either
03:23:42saratogahell on 64MB targets we could probably make the buffer 50KB and not worry :)
03:23:58saratogaits really only the 2MB AMS players where this gets tight
03:24:07JdGordoncan you at least add a "dropped messages" counter so we know for the future if its somethign we should be dealing with?
03:24:16saratogayeah i'll at least do that
03:24:20 Join RipStik_ [0] (4bd6ff9d@gateway/web/freenode/ip.
03:24:28saratogai was going to make it print "MESSAGE TRUNCATED" if it gets full
03:24:42saratogaso you know to make the buffer bigger, or force more flushes by accessing the file system
03:27:11***Saving seen data "./dancer.seen"
03:28:18JdGordonlast thing.. can you add the log level of the message to the output please?
03:28:39JdGordoni.e... "E some_function:123 This died"
03:29:58 Quit RipStik_ (Ping timeout: 245 seconds)
03:31:50saratogaif you want, feel free to pull the source and change the formatting :)
03:32:34saratogathe patch is so simple i'm kind of amazed we didn't have this a decade ago
03:32:36JdGordonno, complaining is easier :)
03:33:05saratogaarguably this system is more simple to implement then logf, let alone logf over USB or remote
03:55:02[Saint]Is there a way to generate all voice files (as in, one for each language) for a given target?
03:55:15[Saint]I really don't want to have to do it by hand.
03:59:05 Quit Wardo (Quit: Blarglarg)
04:00:09saratoga__JdGordon: ok added the warning letter
04:02:13[Saint]Ok....this is...suck!
04:02:40JdGordonsaratoga__: cool, also, can you get rid of the end of line spaces? (fix your editor :) )
04:02:49JdGordonand what does HOME_DIR become?
04:02:51[Saint]Unless I'm mistaken, or blind, it seems I *will* have to build all the voices one-by-one, by hand. :'-9
04:02:54 Join TheSphinX_ [0] (
04:03:03JdGordon[Saint]: script it
04:03:38JdGordonshouldnt be too hard
04:03:52[Saint]JdGordon: If I do that, I may as well "do it right" and add "all" as a parameter to configure
04:06:15 Quit TheSphinX^ (Ping timeout: 245 seconds)
04:09:27 Quit capitalthree (Ping timeout: 244 seconds)
04:12:23saratogaJdGordon: on the clip+ its just the root of the file system
04:12:29saratogai guess its different on hosted targets or something
04:15:03[Saint]Hmmm...for some reason I can't build an Afrikaans voicefile for RaaA
04:15:29[Saint]there's no error, it just finishes immediately and produces a ~500 byte file
04:15:57saratoga[Saint]: the font thing seems like a great way for you to get better a c
04:15:58saratogaat c
04:16:01 Join amiconn_ [0] (amiconn@rockbox/developer/amiconn)
04:16:01 Quit amiconn (Disconnected by services)
04:16:04 Nick amiconn_ is now known as amiconn (amiconn@rockbox/developer/amiconn)
04:16:20 Quit pixelma (Disconnected by services)
04:16:21 Join pixelma_ [0] (pixelma@rockbox/staff/pixelma)
04:16:23 Nick pixelma_ is now known as pixelma (pixelma@rockbox/staff/pixelma)
04:17:04 Quit saratoga__ (Ping timeout: 245 seconds)
04:17:09 Join capitalthree [0] (
04:17:20saratogaJdGordon: ok fixed endlines and added a message if the log overflows
04:17:50[Saint]I never use them...where do voice files live in the /.rockbox/foo/bar/baz dir structure?
04:18:09[Saint]in /.rockbox/langs?
04:18:28saratogahmm that file had a lot of trailing white spaces
04:19:48saratogaanyway, as far as i'm concerned this patch does what it needs to, so if you want to try using it for something go ahead, i probably will just do a bit more clean up and then wait to make sure no one hates it before i commit
04:26:33 Quit anewuser (Read error: Connection reset by peer)
04:29:33saratoga8KB ought to be a reasonable log size even on 64MB targets, you'd have to work pretty hard to hit 50-100 messages in about an hour of playback
04:29:41 Quit Rower85 (Quit: Hmmm...)
04:31:05 Join anewuser [0] (~anewuser@
04:31:05 Quit anewuser (Changing host)
04:31:05 Join anewuser [0] (~anewuser@unaffiliated/anewuser)
04:32:40amayeris the log file to log user errors or system errors?
04:33:21amayerive been listening all day(almost). at first i thought it was for the compiler but then i read something about a missing playlist song
04:34:05*[Saint] wonders how the player would log user errors...
04:34:13saratogaits a generic log for recording whatever output we need
04:34:35saratogabut probably errors and unexpected conditions mostly, as well as debug info
04:34:49amayeroh ok. thats what i ment debug(not system)
04:35:28amayeri was talking like missing song files vs "function write_play_count() failed to access database"
04:35:44[Saint]"log: 01:05:1992 - Didn't pass second level conversational French"
04:35:44saratogai originally wanted this for recording codec errors when people find problem samples
04:35:51[Saint]logging user errors would be creepy :)
04:36:12amayerhaha i guess
04:36:24amayerwhen they were talking about playlists thats what made me wonder
04:36:42saratogaif i want to have a patch that depends on the new logging system, but gets its own gerrit task, do I just branch my current branch and work on them separately ?
04:37:34[Saint]Oh...that's a fun one.
04:38:23saratogai assume if i'm in a branch, and then make a new branch, the new one descends from that branch and I should be able to merge them later ?
04:43:27JdGordonshould do
04:46:11saratogaso if the two branches don't touch any of the same files, its just a simple "git merge" later to combine them?
04:59:32 Quit [7] (Disconnected by services)
04:59:41 Join TheSeven [0] (~quassel@rockbox/developer/TheSeven)
04:59:51 Part amayer
05:02:13 Join perrikwp [0] (
05:21:21saratogacool got codecs working with the new log system
05:23:48[Saint]saratoga: nice.
05:23:59[Saint]What task is this (if any?) on gerrit?
05:24:05[Saint]saratoga: ^
05:24:21saratogai didn't push it to gerrit yet since i haven't really done much with it
05:27:14***Saving seen data "./dancer.seen"
05:29:10saratogai guess i can if you want to try it, but right now i just changed a single DEBUGF to ERRORF :)
05:44:47 Join BHSPitMonkey [0] (~stephen@unaffiliated/bhspitmonkey)
05:51:37[Saint]saratoga: I'd primarily like to see how its being implemented
05:52:06saratogathe codeclib stuff? i just added it to the codec struct and then stuck some preprocessor crap to make it work
05:52:49saratogaor do you mean this
05:54:08 Quit anewuser (Read error: Connection reset by peer)
05:57:10[Saint]Wow...for some reason, it takes about three times as long as any other language to generate a voice file for Hebrew
06:00:45JdGordonodd... maybe rtl issues?
06:06:56[Saint]maybe, donnu.
06:07:15[Saint]*dunno, even.
06:18:45[Saint]Hahaha....its *soooooooo* lucky that we can chuck themes/fonts/voice files/etc. on the sdcard for RaaA
06:19:44[Saint]Otherwise, the 480x800 cabbiev2 full touch port, with fonts and voice files included so it "just works" for all langs out of the box would make a ~120MB .apk :)
06:20:33[Saint]a ~10MB .apk and a ~10MB .zip file is much, much better :-)
06:21:33[Saint]that ~10MB .zip file re-inflates to around ~110MB...but, who cares as long as its on the sdcard, right?
06:22:33saratogawhat is all that space for, the bmps?
06:25:35 Quit BHSPitMonkey (Read error: Connection reset by peer)
06:26:00saratogai don't suppose there is anyway we can use the fonts already in android?
06:26:06[Saint]the particular build I'm putting together now is *massive*, though, as it includes all the voice files as well because I want people to be able to test localization
06:26:28[Saint]and if something in the menu messes up, voice is a good way to be able to find your way home.
06:26:59[Saint]saratoga: yeah....nah, unfortunately.
06:27:26saratogai guess the OS doesn't expose its font files?
06:27:36[Saint]I can use the Android fonts insofar as I can strip them out of the Android OS and generate .fnt files from them ;)
06:27:57saratogacould we just ask the graphics layer to render a given character and take a bitmap of it?
06:28:11saratogaseems like there must be someway to use the text already in android
06:28:42[Saint]Quite probably, but, that's way too low-level stuff for me.
06:29:22CIA-5Commit 68ee7aa in rockbox by Jonathan Gordon: skin_engine: Make pressing the setting bar touch region work
06:29:28saratogayeah i have no idea how android actually works
06:30:05 Quit Thra11_ (Ping timeout: 252 seconds)
06:32:19CIA-568ee7aa build result: 4 errors, 0 warnings (Jonathan Gordon committed)
06:33:14saratogais amiconn's build machine broken?
06:43:55[Saint] /me is rather impressed that a debian image, running chroot'ed from the sdcard of his GTab even gets any builds at all.
06:44:04[Saint]heh, fail.
06:45:24[Saint]But, ...yeah. I'm quite impressed that [Saint] manages to contribute at all. Even though it only seems to get wps' and bootloaders abd everything else gets killed off because the round finishes...I'm contributin' somethin' :)
06:45:42[Saint]*and everything
06:49:03 Quit Scromple (Read error: Connection reset by peer)
06:57:29*Mir waves at saratoga
06:57:48Mirjust wave
06:58:01saratogaif you want to chat, check out rockbox-community
07:03:11 Join Keripo [0] (
07:09:32 Join Ward [0] (
07:09:56 Nick Ward is now known as Guest33137 (
07:12:37 Quit fyrestorm (Ping timeout: 245 seconds)
07:27:17***Saving seen data "./dancer.seen"
07:54:49 Quit saratoga (Quit: Page closed)
08:06:09[Saint]Hmmmm, Russian beats Hebrew in the voice file compiling time highscore game.
08:08:27 Join liar [0] (
08:17:46 Quit mshathlonxp (Ping timeout: 252 seconds)
08:22:52amiconnsaratoga_: Hmm, maybe it's from the ia32-libs -> multilib transition? Since Google earth doesn't really work on linux anyway and I thought it's the only thing needing 32 bit libs I removed them.
08:23:16*amiconn probably needs to reinstall some 32 bit libs
08:29:41 Join kevku [0] (
08:30:13 Quit Guest33137 (Quit: Blarglarg)
08:30:59amiconnsaratoga_: Should be fixed now. Apparently only libstdc++6:i386 was missing, according to ldd. aapt is starting again now
08:33:10amiconnOf course I meant multiarch, not multilib
08:41:50 Join fyrestorm [0] (
08:46:09[Saint]Hahahaha....all voice files == ~75MB
08:49:05 Join ender` [0] (
08:52:14 Quit [Saint] (Remote host closed the connection)
08:54:31 Join Thra11 [0] (~thrall@
09:01:36 Quit perrikwp (Read error: Connection reset by peer)
09:02:50 Join perrikwp [0] (
09:17:24 Join mgottschlag [0] (~quassel@reactos/tester/phoenix64)
09:27:19***Saving seen data "./dancer.seen"
09:34:43 Quit markun (Remote host closed the connection)
09:39:55 Quit shamus (Read error: Connection reset by peer)
09:40:09 Join shamus [0] (
10:03:13kugelhm, current simulator builds are quite CPU hogs!
10:04:35kugel90%+ CPU usage
10:10:16TanguyI have just bought and rockboxed a Sansa Clip Zip, which works quite well.
10:12:44TanguyI just noticed something quite puzzling: with the OF, when plugged to a computer it shows a certain FAT filesystem, while with RB, it shows another one, with two additional directories (##MUSIC## which contains the bundled example music and ##PORT##).
10:13:26TanguyThe most funny is that the OF does not expose them by mass storage, but can play the music it contains...
10:20:20TanguyDo you know if the exposed FAT actually exists or if it is a construction that hides a different internal structure?
10:30:08gevaertsTanguy: the filesystem as exposed by rockbox is exactly as it exists
10:30:42gevaertsThe OF does a bit of trickery to hide things like ##MUSIC##, which it uses for audio uploaded over MTP
10:31:43TanguyWell, what the OF does seems quite hazardous to me.
10:32:19TanguyModifying the FAT on the fly between what the computer sees and what is actually there…
10:37:46 Join soap [0] (~soap@rockbox/staff/soap)
10:46:56 Join dfkt [0] (dfkt@unaffiliated/dfkt)
10:50:45kugeltwo threads hog the CPUb
10:51:22kugelthe main thread (both sigalt and sdl threads) and an external thread created by pulseaudio
10:52:06 Join einhirn [0] (
10:53:04kugelthe main thread does idle in pthread_cond_wait()/SDL_SemWaitTimeout() though
10:58:36TanguyThat means that the OF is designed so that the player is used in mass storage or MTP but not both, since they appear to completely separate them.
10:59:27gevaertsTanguy: basically, yes
11:02:56funmankugel: a very small timeout maybe ?
11:05:12kugelfunman: out tick rate
11:07:19Tanguygevaerts: Quite a bizarre choice. I theoretically appreciate MTP, but implemented like that… Anyway, I am using RB now. :-)
11:08:57funmankugel: ah of course, perhaps profiling can show which code is using most CPU
11:09:37kugelcan you reproduce?
11:09:53kugelperhaps it's just a bug in some library
11:09:58gevaertsTanguy: Same here. MTP allows a lot of fun stuff (no corruption on unclean unplug, continued playback while copying, ...), but in the real world people seem not to want to do all that
11:10:45TanguyAnd file system abstraction.
11:11:50TanguyLike a file transfer protocol, or perhaps a network file system (I am not sure of how it works internally), instead of a block device protocol.
11:12:45TanguyBy the way, what file systems does RB support?
11:12:51gevaertsAh yes, forgot about that one for a moment
11:12:53gevaertsOnly FAT
11:13:01gevaertsWell, 16 and 32
11:13:13TanguyFor the internal memory, I shall certainly not change the FS because the boot loader would not work, but for the external SD…
11:13:18gevaertsFilesystem implementations use RAM :)
11:13:24gevaertsAnd doing them takes work
11:13:26TanguyOh, too bad. FAT then.
11:13:29TanguyYes, sure. :-)
11:13:43*Tanguy would have loved to use UDF, but FAT will do it.
11:14:24gevaertsFor the specific case of an audio player, the FAT limitations tend not to be much of a problem anyway
11:14:38TanguyWhile I am at it, the wiki page about virtual keyboard tells to put the kbd file in .rockbox to use it permently.
11:15:18TanguyBut I have two questions about that: 1. how should I name it and 2. is there a way to navigate to .rockbox since RB hides it?
11:15:41*gevaerts can answer (2)
11:16:21TanguyWell, what annoys me with FAT is the file name limitations. My rsync from my computer failed until I renamed my files to work around the FAT limitations. :-)
11:16:24gevaertsYou can change the "show files" setting to show it. I'd do that sort of stuff over USB though
11:16:42gevaertsTrue. That one can be a bit annoying
11:16:46TanguyOver USB? What do you mean, like editing the config file?
11:17:42gevaertsNo, I mean I don't care if I see .rockbox on the player, when I need to do something there I use my laptop
11:18:13TanguyYes, but loading a keymap is done from RB.
11:18:50gevaertsAh, yes
11:19:03gevaertsAnyway, change the "show files
11:19:11TanguyFor what I have read it implies selecting it from the file browser.
11:19:18TanguyOkay, this is what I missed.
11:19:19gevaerts" setting. Apart from that, I've never used the virtual keyboard
11:21:12TanguyAlso, to all Rockbox developpers, thanks and cheers, it rocks. And it is faster than the OF. :-)
11:23:51 Quit factor (Remote host closed the connection)
11:27:21 Join Thra11_ [0] (
11:27:22***Saving seen data "./dancer.seen"
11:27:45 Quit Thra11 (Ping timeout: 246 seconds)
11:46:37 Join pamaury [0] (
11:46:37 Quit pamaury (Changing host)
11:46:37 Join pamaury [0] (~quassel@rockbox/developer/pamaury)
11:47:32funmansometimes it drains the battery faster than the OF too :-)
11:48:37 Quit Keripo (Quit: Leaving.)
11:53:52 Quit liar (Read error: Connection reset by peer)
11:54:35 Join liar [0] (
11:59:02 Quit einhirn (Quit: Miranda IM! Smaller, Faster, Easier.
12:06:54 Join [Saint] [0] (~Saint]@
12:06:55 Quit [Saint] (Changing host)
12:06:55 Join [Saint] [0] (~Saint]@unaffiliated/saint/x-8516940)
12:08:00[Saint].voice files don't need to be in the internal storage for RaaA, do they?
12:08:23[Saint]I mean, I can just put them in /mnt/swdcard/rockbox/lang/, yes?
12:08:40kugel[Saint]: good question. I never tried it
12:09:05kugelmight be that rockbox doesnt find the langs anymore then
12:09:49[Saint]I can't test it on device presently either, not until later this evening.
12:10:04[Saint]It just occured to me that it may not work, so I thought I'd ask.
12:11:37[Saint]I don't want to have to put ~70MB of voice files in internal storage to make an "international" version (an odd request from another party), but I've got no problems putting that much data on /sdcard
12:13:51 Join Topy [0] (
12:17:20[Saint]Hummm....afrikaans.lang has not a single voice string
12:18:04 Quit T44 (Ping timeout: 264 seconds)
12:18:27*[Saint] thought that voice file built a little too quickly... :)
12:18:58gevaertsSome of them are not worth including at all
12:18:59[Saint]gevaerts: I don't know Afrikaans :)
12:19:10 Join MethoS- [0] (~clemens@
12:19:28gevaertsI'm not suggesting you translate it. I'm suggesting you look at the numbers before deciding to use it
12:21:17[Saint]On the plus side, that is an impressive amount of fully supported locales
12:29:55 Quit kevku (Ping timeout: 272 seconds)
12:41:47 Join kevku [0] (
12:53:13 Join Rower85 [0] (
13:01:35 Quit Topy (Read error: Connection reset by peer)
13:03:39 Join anewuser [0] (~anewuser@
13:03:39 Quit anewuser (Changing host)
13:03:39 Join anewuser [0] (~anewuser@unaffiliated/anewuser)
13:20:04*[Saint] has an interesting request for anyone interested in slightly poking at the skin engine: "Only display the scroll bar when number_of_lines_in_list x list_item_height > ui_viewport_height"
13:20:33[Saint]JdGordon: ^ ...sound sane?
13:20:51[Saint](its not often I have a simple, sane idea :))
13:22:02kugel[Saint]: isnt that the status-quo?
13:22:29kugelor are you talking about skinned lists only?
13:22:31[Saint]kugel: Nope, the space is reserved whether it is displayed or not.
13:23:23kugelyes, but the bar isnt displayed
13:23:44kugelyour request was to not display it
13:23:44[Saint]Right, I should have clarified this.
13:24:13[Saint]"Don't display the scrollbar, *and* return the list to the "no scrollbar" position" :)
13:24:36kugelit's not reservered if you put it onto the right side
13:25:08kugelI think I would be super annoyed if the lists would jump back and forth depending on the content
13:25:20[Saint]I figure this is more useful for small screens, where it can get some space back for long menu entries so they maybe don't scroll, or scroll less at least.
13:26:34[Saint]ANd yes, I know, but the scrollbar on the right looks weird with long/scrolling menu entries, and weird in general with the line selector (IMO (and yes, I know its subjective/matter of opinion))
13:27:26***Saving seen data "./dancer.seen"
13:28:49[Saint]ANyway, regardless of what is "right" or "wrong", or "ugly" or "pretty", ...I think it would be nice if the list returned to the "no scroll bar" position if the scroll bar doesn't need to be displayed, when the scroll bar is positioned to the left.
13:29:03[Saint]bah...sticky shift key.
13:30:02 Quit kevku (Ping timeout: 248 seconds)
13:31:56 Quit MethoS- (Ping timeout: 265 seconds)
13:42:19 Join kevku [0] (
13:44:54 Join benedikt93 [0] (~benedikt9@unaffiliated/benedikt93)
14:14:14 Quit soap (Read error: Operation timed out)
14:14:54 Join soap [0] (~soap@rockbox/staff/soap)
15:10:01 Join amayer [0] (
15:27:28***Saving seen data "./dancer.seen"
15:33:35 Quit mgottschlag (Ping timeout: 240 seconds)
15:38:39 Quit soap (Ping timeout: 246 seconds)
15:42:16 Quit hype (Quit: ["Textual IRC Client:"])
15:50:39 Join PRETTY_FUNCTION [0] (~sigBART@
15:58:45 Join einhirn [0] (
16:00:30 Join yuriks [0] (~yuriks@
16:06:35 Quit einhirn (Ping timeout: 240 seconds)
16:14:59 Join einhirn [0] (
16:20:00 Quit pamaury (Ping timeout: 244 seconds)
16:27:33 Join MethoS- [0] (~clemens@
16:31:17 Quit PRETTY_FUNCTION (Remote host closed the connection)
16:33:54 Join PRETTY_FUNCTION [0] (~sigBART@
16:36:33 Join mshathlonxp [0] (msh@
16:51:33 Join mgottschlag [0] (~quassel@reactos/tester/phoenix64)
16:56:02 Join XavierGr [0] (~xavier@rockbox/staff/XavierGr)
16:57:03 Join nosa [0] (
16:58:05 Quit nosa-j (Ping timeout: 240 seconds)
16:58:06 Nick nosa is now known as nosa-j (
17:06:00 Quit einhirn (Read error: Connection reset by peer)
17:08:27 Join einhirn [0] (
17:12:48 Join domonoky [0] (~Domonoky@rockbox/developer/domonoky)
17:23:29 Quit anewuser ()
17:27:29***Saving seen data "./dancer.seen"
17:47:50 Quit mgottschlag (Ping timeout: 244 seconds)
17:58:36 Join Horscht [0] (~Horscht@xbmc/user/horscht)
18:02:10 Quit MethoS- (Read error: Connection reset by peer)
18:02:55 Join MethoS- [0] (~clemens@
18:03:15 Quit PRETTY_FUNCTION (Read error: Connection reset by peer)
18:03:38 Join PRETTY_FUNCTION [0] (~sigBART@
18:07:51 Quit PRETTY_FUNCTION (Remote host closed the connection)
18:17:25 Quit einhirn (Quit: Miranda IM! Smaller, Faster, Easier.
18:23:29 Join lebellium [0] (
18:46:40 Join PRETTY_FUNCTION [0] (~sigBART@
18:56:00amayershouldnt it say "Play status, 0-8, in the order: ..." instead of "Play status, 0-4, in the order: ..."
18:56:02amayeror does that mean each of them has a 0-4 value?
18:58:29 Join Wardo [0] (
19:11:02 Quit PRETTY_FUNCTION (Remote host closed the connection)
19:14:50 Join soap [0] (
19:14:50 Quit soap (Changing host)
19:14:50 Join soap [0] (~soap@rockbox/staff/soap)
19:15:30 Quit Horscht (Quit: Verlassend)
19:18:07amayerwell i have to go. ill check back in later.
19:18:49 Part amayer
19:27:30***Saving seen data "./dancer.seen"
19:28:25 Quit bluebrother^ (Read error: Operation timed out)
19:28:26 Quit fs-bluebot (Ping timeout: 245 seconds)
19:29:43 Join fs-bluebot [0] (
19:30:48 Join bluebrother [0] (~dom@rockbox/developer/bluebrother)
19:41:55preglow_commit mailer still down?
19:42:39 Nick preglow_ is now known as preglow (
19:53:08 Join prof_wolfff [0] (
19:55:26saratoga_preglow: you still doing any rockbox stuff?
20:00:24preglownot really, no
20:00:28preglowno big itches to scratch :>
20:02:46preglowwell, apart from it not exactly behaving too nice on my android phone
20:02:53preglowbut i just can't be bothered to learn android
20:03:31saratoga_you could help me finish our new resampler
20:03:45preglowwhat's to do?
20:04:28saratoga_i want to add an oversampling option, and then write ASM optimized versions
20:04:29 Join bertrik [0] (
20:04:29 Quit bertrik (Changing host)
20:04:29 Join bertrik [0] (~bertrik@rockbox/developer/bertrik)
20:06:55saratoga_i played around (in matlab) with a ~ 9 order FIR oversampling filter, it works much nicer for the case where you have high frequency content and are down sampling
20:07:20saratoga_its also more accurate even when you have no higher freqeuncy stuff to alias, less distortion in general
20:07:38saratoga_on armv5e the DSP instructions lets you do FIR taps almost for free
20:07:42saratoga_so might as well
20:08:24bertrikwhich DAPs have an armv5e or better?
20:08:42funmanyou want the full list ?
20:08:50saratoga_most newer ones
20:08:53 Join mgottschlag [0] (
20:08:54 Quit mgottschlag (Changing host)
20:08:54 Join mgottschlag [0] (~quassel@reactos/tester/phoenix64)
20:09:01bertrikdoes the FIR impulse have any symmetry?
20:09:13saratoga_symmetry how?
20:09:42preglowlinear phase
20:09:52bertriklike identical taps, or positive taps that have the same magnitude as negative ones
20:10:00pregloweven symmetry around the middle of the impulse response
20:10:04saratoga_oh, probably yes
20:10:09saratoga_or at least we could design one that did
20:10:11preglowthey usually do, for some reason
20:10:29preglowbut linear phase isn't always the best choice
20:10:30bertrikthat could save a couple of multiplies (and replace them by an add or sub)
20:10:35saratoga_i was using 16 bit values in matlab, so the taps are already so small they'd fit in registers anyway
20:11:08preglowbertrik: you mainly just save a memory access, the rest isn't significant
20:11:24bertrikbecause the muls are almost for free already, you mean?
20:11:25saratoga_yeah, add and mul are the same speed
20:11:28preglowmultiplies being so cheap
20:11:58preglowyou can't necessarily do it anyway, depending on your code
20:12:11preglowyou'd basically have to iterate the input array from both sides at the same time
20:13:28saratoga_imx21 is also ARM9E
20:17:47saratoga_yeah just trying a least squares solver, it gives symmetric coefficients
20:18:01 Quit kugel (Ping timeout: 245 seconds)
20:18:04saratoga_so i guess you really only have to store 5 16 bit values
20:23:31preglowsolving for what?
20:24:04preglowthere are nice much-used closed form solutions you can use for designing resampling filters
20:28:55 Join kugel [0] (~kugel@
20:28:56 Quit kugel (Changing host)
20:28:56 Join kugel [0] (~kugel@rockbox/developer/kugel)
20:29:14 Join fyre^OS [0] (
20:31:44 Quit fyrestorm (Ping timeout: 249 seconds)
20:33:38saratoga_i guess i should really try and preweight the FIR frequency response so that it compensates for the hermite response
20:37:32bertrikwhat's the idea exactly? Do some hermite interpolation then follow it up by the FIR (or the other way around)?
20:37:37preglowbut you were talking about oversampling too, if i'm not mistaken
20:46:32saratoga_yeah, oversample then hermite
20:47:01saratoga_i was testing 48>44.1 as 48 ->96 > (hermite) 44.1
20:47:15saratoga_also, 192k as the second frequency
20:48:11preglowif oversampling then hermite, you shouldn't really need to compensate much
20:48:44preglowthe freq response of the hermite should be pretty flat in the passband
20:50:50saratoga_heh, i just checked and it is indeed flat
20:50:55saratoga_only rolls off about 1 dB
20:52:55saratoga_but yeah, i don't see why we can't do resampling for a few MHz on ARM9E and above that is completely transparent even for the hard cases
20:54:04saratoga_yeah, so on the ATRAC3 FIR i benchmarked 3.5 cycles per tap, and that was for a long filter where i couldn't store the taps in memory
20:54:17saratoga_so figure less then 3 cycles per tap for a short one
20:56:06saratoga_so IIUC, the hermite will take ~4 input samples from the FIR, each of which will need ~7 taps, so thats about <100 cycles per output sample or about 8 MHz
20:59:58preglowcouldn't store the taps in memory?
21:01:12saratoga_for atrac?
21:01:20saratoga_i mean in registers
21:01:24saratoga_it was a really long filter
21:01:37preglowyou sure having the filter that short is worth it?
21:01:49saratoga_the frequency response in matlab looks pretty good
21:02:11saratoga_gives >35dB attenuation of all aliased content
21:02:18saratoga_on top of what the hermite gives
21:03:14saratoga_we could go longer though
21:03:33saratoga_was actually thinking a ~15 tap filter and 192k intermediate frequency might make sense too
21:04:45preglowyeah, with such short filters least-squares might actually make sense
21:05:03saratoga_i picked least squares just because it seemed simple
21:05:09saratoga_and i guess its also linear phase if that matters
21:05:23saratoga_i don't know if there is something better
21:05:35preglowpeople usually do windowed sinc with a kaiser window
21:05:50pregloweasily parameterizable
21:09:00saratoga_parameterizable meaning?
21:27:33***Saving seen data "./dancer.seen"
21:45:30 Quit TheSphinX_ (Ping timeout: 244 seconds)
21:50:33 Join TheSphinX^ [0] (
21:51:12 Quit dfkt (Quit: -= SysReset 2.55=- Sic gorgiamus allos subjectatos nunc.)
21:52:01 Join einhirn [0] (
21:56:22preglowsaratoga_: you can recompute the filter for a given cutoff frequency easily
21:59:20 Quit einhirn (Read error: Connection reset by peer)
22:02:03saratoga_ah yes
22:02:22saratoga_i was just going to hard code the 48k to 44.1 k case since the rest aren't so hard
22:03:39 Quit ender` (Quit: Just as the absurd acts which prohibited the sale of game were virtually repealed by the poacher, just as many absurd revenue acts have been virtually repealed by the smuggler, so will this law be virtually repealed by piratical booksellers. -- Thomas B)
22:03:42 Quit MethoS- (Ping timeout: 272 seconds)
22:07:58 Join ender` [0] (
22:23:33 Join factor [0] (
22:30:24 Quit kevku (Quit: KVIrc 4.2.0 Equilibrium
22:30:43 Join kevku [0] (
22:33:21 Join n1s [0] (
22:33:22 Quit n1s (Changing host)
22:33:22 Join n1s [0] (~n1s@rockbox/developer/n1s)
23:03:37 Quit benedikt93 (Quit: Bye ;))
23:27:34***Saving seen data "./dancer.seen"
23:31:44 Quit n1s (Ping timeout: 265 seconds)
23:32:24 Quit preglow (Quit: leaving)
23:34:30 Join n1s [0] (
23:34:31 Quit n1s (Changing host)
23:34:31 Join n1s [0] (~n1s@rockbox/developer/n1s)
23:39:34 Join preglow [0] (thomj@rockbox/developer/preglow)
23:49:06 Quit Xerion (Quit: )
23:51:02 Quit kevku (Quit: KVIrc 4.2.0 Equilibrium

Previous day | Next day