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 2005-06-13

00:00:37crashd200 dolllars
00:00:38preglowporting musepack wouldn't be hard with an arm...
00:01:05BagderI guess ipod qualifies as an arm based player
00:05:27preglowit has to be 1) programmable, 2) non-ipod
00:05:55t0maspreglow: why not iriver H3xx ?
00:05:59preglowi'm tired of everybody having ipod
00:06:01t0masor the new archos things?
00:06:03preglowt0mas: that's coldfire based
00:06:03t0maswith video?
00:06:14t0maspreglow: jup, but why the arm thing?
00:06:21t0masrockbox won't run on it for some months...
00:06:22crashdpreglow: h10 h10!
00:06:28preglowi wanna learn arm asm :PP
00:13:44 Join Road_Runner [0] (
00:14:30Road_RunnerHey. Question: how can you delete the playlist that is now playing? I mean, starting from fresh ?
00:16:34preglowhave you tried reading the manual?
00:16:39Road_Runnerho yeah.
00:17:15Road_Runnertalks about insert & queue.
00:17:29Road_Runnersometimes I can get it but i'm not sure how.
00:18:57Road_RunnerI would expect the 'playlist options' screen would have a function for clearing the o.t.f. playlist, but I guess the problem starts because there isn't an o.t.f. playlist.
00:25:15 Quit tvelocity ()
00:26:57 Quit markun ()
00:31:37 Quit Harpy (Read error: 110 (Connection timed out))
00:32:44Road_Runner1 way I found to have a fresh playlist is to play the last file on the playlist & then removing it from list - > "END OF SONG LIST" - > list is empty
00:33:02Road_Runnerbut isn't there a better way?
00:39:21preglowno idea
00:41:46Road_Runnerdon't you use playlist or what..?
00:47:23preglowi do
00:47:30preglowbut i haven't really been using rockbox for long
00:47:50preglowi'm mostly just queueing stuff
00:54:42 Join courtc_ [0] (
00:55:32 Quit courtc (Read error: 110 (Connection timed out))
00:56:46preglowoh well
00:56:51 Quit preglow ("vel")
00:57:01 Join bg_ [0] (
00:57:20Road_Runnernighty night
01:02:52 Quit bg_ ("Chatzilla [Firefox 1.0.4/20050515]")
01:04:38 Join hardeep [0] (hardeeps@SDF.LONESTAR.ORG)
01:04:57 Part MoosCamaro
01:11:17 Join Seedy [0] (
01:12:35 Quit Seed (Nick collision from services.)
01:13:36 Nick courtc_ is now known as courtc (
01:34:50 Quit gromit` (Remote closed the connection)
01:37:34***Saving seen data "./dancer.seen"
01:43:31 Quit ShockerEngr (Read error: 145 (Connection timed out))
01:46:31 Quit Stryke` (Read error: 110 (Connection timed out))
01:47:48 Quit ]RowaN[ ()
02:25:48 Join ashridah [0] (
02:27:30 Quit Sucka ("a bird in the bush is worth two in your house")
02:40:10 Quit hicks (Remote closed the connection)
02:40:34[solid]vorbis playback passed 10hrs and is still playing :)
02:42:13Rori"Something is afoot" replied the leg
02:42:38Rori[solid] tis a time to rejoice!
02:43:39RoriNow I need to rerip some of my collection to ogg or lame. shame most of my mp3 collection is in non-lame format :P
02:43:52Roriand I don't own the cd's to rerip them
02:43:52[solid]i need to get some sleep ;o
02:45:03Rorihow will you know when it stops playing?
02:46:21yngwiit won't :-)
03:33:25 Join amiconn_ [0] (
03:37:35***Saving seen data "./dancer.seen"
03:47:36 Quit yngwi ("Chatzilla 0.9.68a [Firefox 1.0.4/20050511]")
03:51:44 Quit amiconn (Read error: 110 (Connection timed out))
03:51:45 Nick amiconn_ is now known as amiconn (
03:53:31 Quit Rori (Read error: 104 (Connection reset by peer))
03:56:07 Join Rori [0] (
03:56:28Roripc crashed during Rockbox disk to disk copy :P
03:58:48 Quit Road_Runner ("CGI:IRC (EOF)")
04:00:20HCloh my fucking god gta san andreas is sweet
04:00:42*HCl stole a jet powered plane from the airport and stored it in his personal hangar
04:01:38*HCl sleeps
04:05:43 Join QT_ [0] (as@area51.users.madwifi)
04:13:12RoriI don't get this. I was playing a track and 2/3 through it stops but the time still says the song has not finished and suddenly it starts playing back a track from a different folder that I had not playlisted or anything
04:13:34Roriweird bug
04:17:34 Quit QT (Read error: 113 (No route to host))
04:21:42Roridoes anyone know what version of Lame -nogap was introduced?
04:33:00Rorigapless on lame does not seem to work
04:43:13 Join asdsd____ [0] (
04:47:57 Part asdsd____
05:02:04 Join barbita [0] (
05:06:34 Part barbita
05:15:12Rorihmmm. seems it may be the way people rip and encode in lame perhaps? trying my own rip in EAC with the −−nopgap option
05:15:27Rori−−nogap rather
05:16:20RoriI hear conflicting stuff about how it should be done. some say you should do −−nogap track1.wav track2.wav track3.wav in one command while others state this is false information
05:19:29 Join david88 [0] (
05:20:52 Quit david88 (Client Quit)
05:22:43Rori−−nogap appears to be broken in eac
05:34:26crwlusing −−nogap with all the decoded wavs at the same command line is quite difficult, i don't know of any ripper/encoder tools that support it natively
05:35:11crwli thought that "gapless support with lame tag information" means something like that lame writes the exact samples that need to be decoded in a header and a decoder that supports the header, can play those files gaplessly
05:35:29crwland files encoded with −−nogap is something that any mp3 decoder can support
05:36:21crwlbut i don't actually know :)
05:36:31crwlthat's just what i'm *hoping*
05:37:39***Saving seen data "./dancer.seen"
05:41:08ashridahcrwl: the man page for lame here suggests that it should be given multiple contiguous files
05:41:36ashridahie, it's only interested in doign gapless analysis for a single album, not from randomsong1 -> randomsong2
05:41:58crwlbut that's pretty cumbersome
05:42:27crwlwouldn't storing the sample exact lenght of the source file somewhere be enough, as long as the decoder supports that+
05:42:49crwlbah, the easiest way is to ditch lame and use vorbis all the way :P
05:42:53crwlwork to do ->
05:46:41 Quit hardeep ("BitchX: born to raise hell")
05:48:38 Join DMJC [0] (
05:53:54RoriwinLame does
05:54:34Roritesting now
05:55:00Roriyou rip with eac to wav then encode in winLame with the nogap option built in
05:55:07Rorineed to tag afterwards though methinks
06:03:01Roriit does not work. I think you are right. Ogg kills Lame for gapless.
06:11:22DaKi][erand in compression to quality ratio as well, but it would still be handy to have a true gapless for all mp3's that works for thoes that have their libaries of music already encoded this way
06:13:25ashridahcan't squeeze blood from a stone
06:26:59 Join mishter [0] (
06:28:25 Quit mishter (Client Quit)
06:44:26 Join elinenbe_ [0] (
07:37:42***Saving seen data "./dancer.seen"
08:05:12 Nick Seedy is now known as Seed (
08:24:27 Join mrlala [0] (
08:31:04 Quit ashridah (Read error: 60 (Operation timed out))
08:33:21 Join ashridah [0] (
08:45:58 Join scriptkitty [0] (
08:48:39 Join gromit` [0] (~gromit`
08:57:37 Quit DMJC ("Leaving")
09:00:45 Part courtc ("Leaving")
09:02:11 Quit mrlala (Read error: 110 (Connection timed out))
09:09:40 Join bobTHC [0] (
09:17:07 Join Harpy [0] (
09:19:01 Join dapureplaya [0] (
09:19:43dapureplayahi everyone
09:19:45dapureplayadoes Musepack playback yet?
09:19:58bobTHCmornin' folks !
09:20:06dapureplayagood evening ;)
09:20:40*Bagder waves
09:20:52*dapureplaya waves back
09:23:54Bagderthe linux port threads on misticriver don't stop
09:24:03BagderI love them ;-)
09:27:37 Join StrathAFK [0] (
09:31:25linuxstbdapureplaya: No, Musepack doesn't work yet. But it's getting closer.
09:31:44dapureplayaah right just curious
09:32:07dapureplayai wasn't sure coz i haven't been keeping up to date with CVS activity on the site. i thought i may have missed out on some news.
09:32:12Bagderwe should keep a list of codecs and their status in the wiki...
09:32:32linuxstbThere is one (I think it's at the end of SoundCodecs)
09:32:59*Bagder looks
09:33:19dapureplayaof course
09:33:25dapureplayabut i thought it wasn't updated.
09:33:40dapureplayaie. Wavpack says it's not playable on iRiver yet hehe :-p
09:34:20dapureplayabut i guess MPC isn't "ready" yet.
09:34:43linuxstbI hadn't noticed that Wavpack commit. Anyone tested it?
09:34:51dapureplayalistening to it now
09:35:05dapureplayalossy format at avg 500 kBit
09:35:15dapureplayai'm going to try lossless format now
09:35:42dapureplayaruns very well. the hard drive doesn't get much of a workout
09:35:45linuxstbSo should I add the tick for WavPack in the SoundCodecs page?
09:35:54dapureplayafor sure
09:37:47***Saving seen data "./dancer.seen"
09:38:11linuxstbWell, that's 6 codecs supported (8 if you split MPEG into 1,2 and 3). Truely a Multi-Codec Jukebox.
09:38:56Bagderamazingly quick
09:39:05dapureplayai would say that the codec status would be a bit more than 50% now :)
09:39:35dapureplayayou guys been progressing rapidly thruout the last 2 weeks.
09:42:43BagderI'm only using rockbox these days
09:43:01dapureplayayep. Same. Haven't used iRiver FW for a while.
09:43:15dapureplayawell since FLAC playback anywayz
09:43:27 Join oxygen77 [0] (
09:43:47linuxstbI've never used the IRiver FW for more than about 10 minutes - I just couldn't understand it.
09:44:19Bagderhehe, I fully sympathize with that
09:44:21dapureplayathe gapless on lossy wavpack os flawless at the moment
09:44:59linuxstbdapureplaya: That doesn't surprise me - only MP3 is a problem to get gapless, everything else "Just works".
09:45:39dapureplayayeah true. Hey wasn't there a commit recently to try and fix the MP3 gappless? how does it run now? haven't tried it
09:45:50 Quit Strath (Read error: 110 (Connection timed out))
09:46:22linuxstbIt's still not working perfectly - but the code is written, it just needs debugging.
09:47:14dapureplayaah. Hey, quick change of topic. i can't find the codec to encode AC3 files for dbPowerAmp. What can i use? Any tips?
09:47:46 Quit oxygen77 (Client Quit)
09:48:09linuxstbThe main purpose of AC3 was to allolw you to rip audio from DVDs (and some digital TV) and play it on the iRiver without re-encoding.
09:49:14dapureplayaoh i see. Is it a DVD-audio format?
09:49:16linuxstb"ffmpeg" comes with an AC3 encoder, but I don't know how that works with Windows.
09:49:28linuxstbAlmost - it's an audio format for DVD-Video.
09:49:44linuxstbDVD-Audio uses either uncompressed PCM or "Meridian Lossless Packing".
09:52:23dapureplayaah rightio
09:53:02dapureplayaare tehre any plans on implementing APE at the moment?
09:53:03linuxstbDVD-Audio is my other interest outside Rockbox - see my project at
09:55:51dapureplayaah nvm i just foudn out APE isn't open source
09:56:37linuxstbAnd the other (possible) problem is that it has a reputation for being "hard" to decode - i.e. takes a lot of processor power. So I'm not sure it would even be possible.
09:57:27DaKi][erisnt much of a market out there for APE, it only advantage over FLAC is a few % better compression, but FLAC is far more popular
09:58:16dapureplayaah i see.
09:58:22linuxstbI am quite interested in adding a Shorten decoder - now that there is a GPL'd decoder available. But it's a low priority, I convert all my SHNs to FLAC anyway.
09:58:22dapureplayathat's fair enough\
09:59:21dapureplayayeah never heard of shorten before until i came across rockbox
09:59:51linuxstbI think Ape could have become more popular if released under a different license - it started to gain momentum, but then stopped when no open source projects adopted it.
10:00:09linuxstbOr maybe it's simply the case that FLAC is good enough.
10:00:23dapureplayamuch more popular aswell.
10:00:26DaKi][erand a bit more widely supported
10:04:46 Join tvelocity [0] (
10:05:17 Join rasher [0] (
10:08:38 Nick Lynx_awy is now known as Lynx_ (Lynx@
10:08:54dapureplayalinuxstb: your project is pretty clever. not a bad idea at all.
10:13:25linuxstbThanks. It's working great for me.
10:13:36dapureplayayou working on it solo?
10:13:51dapureplayaor you got a team of minions on the case aswell? ;)
10:13:59linuxstbYep. A couple of people have written "front-ends" (one Windows, one Mac), but dvda-author itself is just me.
10:14:24linuxstbNo minions have shown any interest...
10:28:37 Quit tvelocity ("Leaving")
10:30:49 Quit [solid] ("leaving")
10:33:28 Join mrlala [0] (
10:36:45 Quit dapureplaya (Read error: 145 (Connection timed out))
11:10:05 Quit mrlala ("-=SysReset 2.53=-")
11:19:56 Join DMJC [0] (
11:27:27*Bger is surprised of FLAC compressing 3CDs at average compression ratio 0.25 ...
11:30:43ashridahi expected flac encoding to be more cpu intensive when set to −−best
11:31:39ashridahi'm also vaguely considering converting my mp3s to flac, even tho i don't have the originals. it's not like they're going to DEcrease in quality.
11:31:59ashridahhm. only problem is, that's a fairly large increase in size
11:32:09ashridahwon't be able to keep as much pron on my player :)
11:32:24Bgerbut i don't think that you'll have any advantage of doing it
11:32:59Bgerbut 25% from the size of the original .wav ....
11:33:21zewhy would you reencode from mp3 to flac?
11:33:32ashridahze: elitism
11:33:37zeits dumbism
11:33:41ashridahwhy else :)
11:34:04zenobody'll think you're elite for transcoding already lossy files into a bigger lossless format
11:34:31zeyou've already lost the quality so the only effect of making them lossless now is to take up more space for no reason
11:34:43ashridahah, you're misconstruing the context. i get to be elite when i tell them all my music is in flac, not how it got there :)
11:35:12ashridahze: elitism has never been about having no flaws, it's about cleverly hiding them
11:35:37ashridahotherwise you wouldn't need it, since you'd have no ego problems to begin with
11:35:58ashridahthe reality is, i really couldn't be fucked sitting through a marathon reencoding session
11:36:01zewhat makes you think true elitism is about ego problems?
11:36:21zethats more likely just the overt pseudoelitism
11:36:42zethats nothing but ostensible elitism
11:37:01zeand says nothing about us underspoken elitists who're just better than most people but prefer not to go around bragging about it
11:37:21*ashridah shrugs
11:37:31zeok there was supposed to be humor in that
11:37:33zebut nevermind
11:37:36ashridahyou are your own biggest critic.
11:37:43ashridahoh, no, i got it
11:37:48***Saving seen data "./dancer.seen"
11:38:39ashridahmp3's make me feel dirty, like i'm using gifs on a webpage or something
11:38:45zewell yeah you should always be your own biggest critic though
11:38:58zebut there's a distinct difference between what you know you're capable of and what you've manifested so far
11:39:06zewhich is where your self critism comes from
11:39:16zeat least in some cases
11:39:25zeyeah i know what you mean
11:40:17zemy recomendation is to track down or otherwise obtain honest ogg and/or flac rips to replace them with though
11:41:06ashridahze: my biggest problem is that i was robbed a few years back. i don't have cds for stuff i had back then
11:41:16ashridahsadly, that was before i discovered the wonders of ogg
11:41:48zeand employ a large breed of horse to anally rape internet explorer for intentionally not supporting png alpha channel
11:42:40zeashridah: well ogg and flac have become popular enough now, you can often find copies of things available for download in them
11:42:47ashridahheh. fortunately, last time i had to do anything web-based, the lowest version of IE i had available was 5.5, which can be hacked around with javarsescript
11:42:55ashridahze: yeah, that'd imply i had bandwidth tho
11:43:02zehey man
11:43:13zei was downloading 2-disc movie rips back when i was on modem
11:43:30ashridahyeah, except i've got a tv episode habit to keep up with :)
11:43:51ashridah3 350MB episodes a week is a bitch when you're using torrents
11:44:11linuxstbashridah: Why don't you just _tell_ people all your music is FLAC?
11:44:23zehaha linuxstb has a point
11:44:29ashridahlinuxstb: hahaha
11:44:37zewhat're they gonna do, analyze the files on your player?
11:44:44linuxstbBut FLAC support is very cool...
11:45:08zei got some flac copies of a few things some years ago
11:45:18zeand ended up getting a flac-capable player before i got around to transcoding them
11:45:37zenow if only the karma had opensource firmware...
11:45:43 Join preglow [0] (
11:45:53preglowit's not cooler than wavpack, though
11:45:55preglowi love the hybrid mode
11:45:58ashridahthe sad part is i know some local artists who refuse to keep their original recordings in lossless formats.
11:46:07zeashridah: that is sad
11:46:23zethey should burn the whole original session to cd/dvd's
11:46:25ashridahtrying to justify it by claiming '320kbps doesn't lose anything noticable'
11:46:29ashridahthen remixing it half a dozen times :)
11:46:39zein the original seperate tracks and everything
11:46:41zecomplete backup
11:47:29ashridahyou'd think so
11:47:31zethen when they wanna transfer it to 6DOF 3D blueray ultradefinition audio or whatever
11:47:45zeit'll only sound sorta crappy rather than goddamn horrible
11:48:06ashridah... blueray is horrible?
11:48:22zethink a vinyl from the 20's put on CD
11:48:25*ashridah doesn't know his disk formats well
11:48:33zenot blueray itself
11:48:51ashridahah, right.
11:48:53zeblueray is just higher-density optical storage
11:49:24ashridahyou could probably do a decent reconstruction with a laser and several copies of the same vinyl these days
11:49:27ashridahshould average out the defects
11:49:41ashridahdepending on how the defects normally form.
11:50:05zei was just talking about the transfer from original tracks vs transfer from lossy versions
11:50:23zea lot of the defects are from wear, which is probably pretty consistent among different copies
11:50:28linuxstbpreglow: Am I right in assuming Rockbox doesn't support Wavpack correction files?
11:50:29 Join Aison [0] (
11:50:47ashridahze: well, you'd attempt to find an unused copy
11:50:57zebut a laser scan of an unplayed or barely-played and well taken care of copy would probably sound pretty good
11:51:07 Join tvelocity [0] (
11:51:14zethere's probably some pristine masters for most things
11:52:35zeof course that doesn't make up for defects in recording equipment or its usage
11:52:40zeor poor mixing and mastering skills
11:53:17zethough of course i dunno if mixing and mastering were really things that were done much in the 20's
11:53:46zeoh well
11:53:52zeanyway how'd it get to be 3am... bah
11:55:29preglowlinuxstb: indeed
11:55:34preglowlinuxstb: and it'll be hard to support as well
11:56:11preglowlinuxstb: implementing it in the library we use would be pretty easy, i think, but buffering two files at once wont be so easy
11:56:44linuxstbYep, it kind of destroys our streaming codec architecture.
11:57:13preglowlinuxstb: luckily, for me at least, much of the point in having hybrid mode is the fact that i can just toss the compressed files minus the correctional files to my portable
11:57:29preglowthe correctional file would pretty much just waste space
11:57:35linuxstbIt would probably be possible with some kind of interleaving in the audio buffer - but probably only David could attempt that.
11:57:40preglowbut it's nice to have in case i ever want a true lossless copy
11:57:55preglowwe COULD interleave, yeah
11:57:58preglowthat's a nice point
11:58:14preglowbut indeed
11:58:24linuxstbBut you can always re-encode a hybrid file to a single lossless file.
11:58:25preglowwe're lucky to have a codec developer on the team
11:58:28ashridahit's how video/audio streams are usually dealt with. just have key data in the streams that keep the sample counts in check
11:58:50preglowit's nice not to have to worry about one particular codec, since the guy who happens to know the most about it is working on it anyway ;)
11:59:05preglowwhich isn't exactly the case with the other codecs
11:59:25linuxstbYes - he's done a very good job integrating it into Rockbox.
11:59:35preglowcodecwavpack is nice as well
11:59:40preglowsupports fast seeking and all
11:59:53preglowand people say it's gapless
12:00:06preglowthen again, a non-gapless lossless format would be kind of silly :P
12:00:46linuxstbAh yes, military intelligence, friendly fire, non-gapless lossless...
12:01:10preglowwe should make some kind of sample rate negotiation in the codec api
12:01:23linuxstbI think Apple have managed non-gapless lossless though...
12:01:38preglowapple can be as stupid as they want, as long as they look pretty
12:02:06BagderI liked their early ipods, that never stopped spinning the disk when you played a song larger than the amount of ram
12:02:33preglowonly thing i like about the ipod is the navigation wheel mechannism
12:02:39preglowi don't even think it looks very pretty
12:02:50linuxstbI'm not sure where the resampling should go - it can be argued that it should be done within the codecs because it could save yet another copy operation.
12:03:10ashridahcrap! it's got that really great game where you shoot down helicopters and dudes in parachutes. it's clearly better than anything!
12:03:26ashridahinfact, that game's better than all games that run on osX combined!
12:03:27preglowlinuxstb: i think the codecs themselves shouldn't insert buffers, that'll have to go
12:03:47preglowlinuxstb: they should just tell the codec api 'here is the pointer to a frame, do what you need with it'
12:03:47linuxstbpreglow: What do you mean?
12:04:07preglowlinuxstb: then the codec loader does whatever processing is needed, then inserts it into the audio buffer
12:05:01linuxstbThe problem is that the codec libraries all deliver the uncompressed data in different formats. This needs to be converted to big-endian interleaved samples.
12:05:11preglowwe will need a common format, yes
12:05:23preglowhmm, shit, that'll break fast interpolation
12:05:31preglowno it wont
12:05:42linuxstbMy idea is that the resampling could happen at the same time as this conversion. But it may be one optimisation too far.
12:05:50preglowi hate interleaved data
12:05:54preglowit's bad for processing speed, heh
12:06:06preglowthat's not a bad idea, actually
12:06:45linuxstbThanks ;-). I don't think it's obvious where the resampling should go.
12:07:10preglowthe interpolation coefficients can be chosen to do the shifting
12:07:16preglowas long as the shift isn't too big
12:07:26preglowbut yes
12:07:33preglowit's crucial we design this so the codecs don't do it
12:07:41preglowthat'll mean tons of code duplication
12:08:02Bagderunless we provide a nice set of functions
12:08:16linuxstbBut the problem is that the "input format" varies with every codec.
12:08:18preglowBagder: of course, but why not call those from a central place, then?
12:08:35BagderI guess that depends on the data
12:08:40preglowit doesn't vary too much
12:08:47preglowresolution varies
12:08:58preglowinterleaved/noninterleaved varies
12:09:10preglowi believe the codec loader should handle that as well
12:09:10linuxstbFlac has one array of 32-bit integers per channel, MAD uses it's own fixed-point format....
12:09:16preglowpoint is, the possibilities aren't limitless
12:09:27linuxstbWAV is little-endian interleaved....
12:09:30preglowlinuxstb: well, yeah, but there's nothing to it other than a shift of variable size
12:09:43preglowlinuxstb: endianness should be converted on load
12:09:54linuxstbIf you can imagine the format, one of the codecs uses it.
12:10:22Bagderand probably a few additional ones we couldn't imagine ;-)
12:10:32preglowyes, but the possibilities aren't limitless, like i say, you've got either 16 bit or 32 bit data, either might have a variable size shift, and interleaved/noninterleaved
12:10:42preglowthat's very manageable
12:11:00preglowdithering also should be done by rockbox, not the codecs
12:11:05linuxstbOr 24-bit in the case of libmad...
12:11:13preglowlinuxstb: no, it's 32 bit
12:11:22linuxstbOK, sorry.
12:11:29preglowlinuxstb: precision is 24 bit, though, but that's not the point
12:12:36linuxstbMaybe the resampling should go in playback.c for now, and if performance becomes an issue, we can think about moving it into the codecs.
12:12:53preglowwhy would it be any less of an issue there?
12:13:08linuxstbBecause it can be combined with other operations.
12:13:20preglowthey can be combined in playback.c as well
12:13:54preglowas long as the codec delivers its buffers to playback.c with bit depth, precision and interleavedness specified
12:13:58preglowwhich should not be a problem
12:13:59preglowbut of course
12:14:11preglowif we put them in the codecs, certain parameters code be inlined to the code
12:14:19preglowbut then we'd need these routines in a header somewhere
12:14:24preglowcodeclib or whatever
12:15:01linuxstbI think we've standardised on 16-bit, Stereo, big-endian interleaved at the moment. Some (but not all) the codecs do mono to stereo conversion.
12:15:24preglowthat too is a job that doesn't belong there, imho
12:16:04rasherI see someone has been making battery tests.. why aren't the results on I ask
12:16:12linuxstbI agree. audio_insertbuffer should probably take "nchannels" and "frequency" parameters.
12:16:12preglowrasher: ask [solid]
12:16:43preglowlinuxstb: codecs should also be aware for what configuration they're generating data
12:17:02preglowlinuxstb: liba52, for instance, has its own routines for downmixing which could be affected
12:17:30preglowi don't really believe we'll see portable with 5.1 configuration soon, though ;)
12:19:01preglowlinuxstb: then again, output precision isn't fixed, uda1380 does support more bit depths than 16
12:19:06linuxstbI posted a message to the linux-audio-dev list the other day - there are a few replies there with pointers to resampling code that may (or may not) be helpful.
12:19:14preglowlinuxstb: url, please
12:19:42linuxstb^Search for "resampling" in that page.
12:20:10 Join hicks [0] (~hicks@193975afca3afbf3.session.tor)
12:20:56preglowsinc resampling is out of the question for us, imho
12:20:58preglowit's too memory intensive
12:23:29preglowso i'm still stuck with linear resampling for the case of 48->44.1
12:23:42preglowfor bigger rate changes i can cook up some filter trickery
12:24:09rasherdoes anyone know if Linus got the ihp-110?
12:24:25preglowhe did
12:24:38preglowhe'll be away for a while, though
12:25:15rasherAlrighty then.
12:25:44 Quit rasher ("CGI:IRC")
12:32:45 Join nick_ [0] (~irc@56e4593e34a71906.session.tor)
12:43:46 Join MoosCamaro [0] (
12:48:06 Join MO-Pantsu [0] (
12:55:14 Join Patr3ck [0] (
12:56:29 Quit Rori (Read error: 145 (Connection timed out))
13:12:20 Quit edx (Read error: 110 (Connection timed out))
13:25:12Slasherii am now at telecomm lab and i could take some power measurements from iriver IF i could find a molex connector that will fit the pcb
13:37:49***Saving seen data "./dancer.seen"
13:41:37 Join spiralout [0] (
13:52:38HClbad idea to play san andreas till 4 am
13:52:49*HCl thinks he finally caught up on all the sleep he needed
13:54:23preglowplease stop mentioning san andreas :/
13:54:49HClwhy? :/
13:55:59preglow'cause i wanna play it very, very badly
13:56:05HClwhy don't you
13:56:11preglowgotta finish school business
13:56:15HClk k
13:56:21preglowand i'm bored out of my brain
13:58:19BgerHCl congrats for the sleep :)
14:01:36preglowi could boot windows and play for just a couple of hours...
14:01:49HCldoesn't sound smart :x
14:01:58HClits pretty addictive
14:02:04preglowi happen to know that
14:02:17preglowi sat up to 4 am the other day myself
14:02:24HCloh, right.
14:02:35preglowi'm addicted to motorcycle riding
14:02:37preglowit's so damn fast :/
14:02:39HCldidn't know you've played it already
14:02:47HClwait till you get to the planes
14:02:51HClor the hovercrafts
14:03:01preglowi've tried the hovercraft
14:03:03preglowand i'm a master of planes
14:03:06preglowi played it on ps2
14:03:10HCloh, right
14:03:23linuxstbMmm. I'm listening to a wavpack-encoded album, and half way through one song, it's started playing the second half of the previous song....
14:03:38linuxstbBut I had been playing with seeking earlier in the album.
14:03:47preglowthere are buffering bugs still
14:03:50preglowi had that happen with mp3s
14:03:56preglowafter toying around with seeking
14:04:09preglowwavpack too is very robust when it comes to seeking, so it probably went along with it
14:04:29linuxstbI'll stop seeking then...
14:05:29linuxstbWavpack encoding is VERY fast though. Seems like a good candidate for encoding recordings in realtime.
14:05:37preglowit is
14:05:40linuxstbUnless it needs floats...
14:05:42preglowencoding == decoding, for wavpack
14:05:49preglowthat's what david explained to me, at least
14:06:01preglowso encoding should actually be comperable to decoding in complexity
14:06:05preglowand wavpack is incredibly efficient as it is
14:06:08linuxstbIt would also increase buffering time if we did real-time lossless encoding when recording.
14:07:26preglowshouldn't be a problem with wavpack, no
14:09:12HClso we can encode to wavpack, at least?
14:16:12*MO-Pantsu can't seem to get Lame gapless to work that well. So I have dumped it for Ogg at least on my own rips
14:16:59MO-Pantsudoes FF/RW work on Ogg yet? Not checked
14:17:33linuxstbMO-Pantsu: No.
14:18:45MO-PantsuI've kinda given up on gapless mp3 for now. it was not implemented into Lame until 3.89 anyhow which is why I'd prefer a generic gapless dsp
14:19:14 Join Jim__ [0] (
14:19:34preglowlame 3.89 has been out for ages
14:20:09MO-PantsuI have a lot of old encodes I downloaded that are not Lame encoded. Some are even crappy Xing :P
14:20:51 Join Patr3ck_ [0] (
14:22:11zeoh man, xing?
14:22:35zeman how did you not know better
14:23:02Jim__hey! can anyone tell me what the image at the bottom of the screen when playing a song reprasents? Below how far into the song you are...
14:23:15MO-Pantsuhey when mp3 started everyone was using Fraunhoffer but then people complained about encoding speed and alll the n00bs found Musicmatch with Xing :P
14:23:40preglowJim__: peak meter, i think
14:23:42zeyeah but encoding speed has never mattered except for prerelease rip groups
14:23:48preglowJim__: not yet operational on iriver, if that's what you mean
14:23:54zeits all about the files you get
14:24:16MO-PantsuI can't find many Ogg encodes :P
14:24:18Jim__yer on iriver, thanks!
14:25:07preglowlinuxstb: there's another thing we need to do while we're traversing the data we get from codecs, peak meter
14:25:26 Quit Jim__ ("Chatzilla [Firefox 1.0.4/20050511]")
14:26:33MO-Pantsuwhen Rockbox finally gets recording with meters it will become a usefull recording device. I hate guessing.
14:27:00preglowyes, indeed
14:27:11MO-Pantsuthat's a way off though methinks :)
14:27:25preglowmight be
14:27:33preglowwe've got a guy recording with rockbox
14:27:35preglowon iriver
14:28:50MO-Pantsuthe thing is will Rockbox do a live level check to meters in realtime with any degree of accuracy? :)
14:29:02preglowit'll be as accurate as we want
14:29:11preglowthere's probably a clipping warning as well
14:29:24MO-Pantsuthat would rule
14:30:00MO-Pantsuit would become 'the' defacto DAP for taking to concerts etc
14:30:03preglowautomatic gain reduction on clip wouldn't be too hard either
14:30:04preglowif you want that
14:30:49HCli really need to stop playing san andreas
14:30:51HClcheck out rockbox
14:30:54HCland work on the searchengine..
14:30:57MO-Pantsumaybe. I prefer to find a level and sit with it coz it stops the pumping effect
14:31:00preglowDUMB <-
14:31:04HClthat too
14:31:17preglowMO-Pantsu: yes, sure, me too, but in some situations you don't have time to calibrate
14:31:56MO-Pantsunever liked compression (Except for fat kick drums) :)
14:32:27MO-Pantsucompression & limiting can be usefull for some stuff
14:32:38preglowwell, it's not compression
14:32:45preglowit's just a one-shot gain reduction
14:32:45MO-PantsuI know just mentioning it
14:33:06preglowif it were to constantly readjust recording gain, it might sound like compression
14:33:12preglowbut i don't like that, heh
14:33:36 Quit Patr3ck (Read error: 110 (Connection timed out))
14:37:02linuxstbpreglow: Is the peakmeter something for the firmware to take care of? Maybe the hardware can give us that information back.
14:37:50preglowsincerely doubt it
14:38:25linuxstbI'm sure that's how it works on the Archos, so maybe some hardware can do it, even if the iRiver can't. Just a thought.
14:40:11linuxstbDoes the famous iRiver "recording glitch" happen with both analogue and digital inputs, or is it just a digital problem?
14:40:30MO-Pantsupreglow: you know you was saying the other day you could not get rid of the glitch in lame gapless mp3's and could not figure out why?
14:40:40preglowMO-Pantsu: yes
14:40:46preglowlinuxstb: some people report both
14:40:58MO-Pantsudid you try encoding using deprecating commandline in lame?
14:41:33preglowwhat depracated commandline?
14:41:36MO-PantsuIE: lame −−nogap −−alt-preset standard track1.wav track2.wav track3.wav
14:41:46preglowi don't care about −−nogap, that should work
14:41:50preglow−−nogap is a hack
14:42:08MO-Pantsuapparently −−nogap don't work too well unless eaither successive track is encoded in one commandline
14:42:14preglowof course
14:42:20preglowit doesn't work at all
14:42:25MO-Pantsuoh ok
14:42:33preglowlame specifically says you HAVE to give it _all_ wav files
14:42:34MO-Pantsujust checking coz I can't get it to work either
14:42:45linuxstbI think −−nogap is a good idea. Problem is with rippers that rip and and encode one track at a time. But even those could be made to work (by moving samples itself).
14:42:48preglowif you just give lame one wav, −−nogap doesn't do shit
14:42:51MO-PantsuI used winLame that supposedly does that for you
14:43:10preglowlinuxstb: i think it's a horrible hack, i want my track transitions to be where they're supposed to be
14:43:13MO-Pantsubut it still glitched so shrugs
14:43:16linuxstbIt's like "sector boundary errors" with CD audio - you need to burn CD audio in multiples of 2352 bytes, and the same limitation applies to MP3.
14:43:24MO-Pantsuyes it sucks
14:43:39MO-Pantsuit all sucks
14:43:52preglowMO-Pantsu: it might glitch now with my current code
14:43:57preglowMO-Pantsu: it didn't glitch before
14:44:25preglowwe'll have it working, don't worry
14:44:31preglowbut don't expect mp3 playback to just work nmow
14:44:35preglowit's under development
14:44:39linuxstbBut moving the transitions a few ms won't make any difference - in fact it makes no difference at all if you are playing back the album sequentially.
14:44:40MO-Pantsuno worries
14:44:52preglowlinuxstb: 'course not, but still
14:45:09preglowlinuxstb: it's still a hack, lame info header gapless makes it redundant
14:45:40linuxstbOnly for players that support it. The −−nogap option makes MP3s that play gaplessly anywhere - assuming the player doesn't add extra silence between tracks.
14:46:01linuxstbThat's why I think it's a useful solution.
14:46:32preglowa lot of players do add extra slence
14:46:39preglowiriver, ipod, creative players
14:46:42preglowso it doesn't matter
14:46:51preglowworks for archos, of course
14:47:12linuxstbThat's true - −−nogap is the only way to get gapless on the Atchos.
14:48:04MO-Pantsuprobably be easier to add cue file support and if people want gapless in mp3 rip it as a single file
14:48:18 Join edx [0] (
14:48:52linuxstbYes, I'm very keen on cue files - I even wrote a patch to Rockbox a few years ago, but it never made it into the official code.
14:49:13preglowcue sheet support should be a breeze this time around
14:49:19preglowif you don't demand too much precision
14:49:51MO-Pantsutrue that I note when you reburn a cd from a cue file the points are not the same
14:50:09linuxstbI think we should aim for sample-accurate seeking for those codecs that support it.
14:50:21MO-Pantsudo any?
14:50:33preglowmp3 is the only format that benefits from cue support
14:50:41preglowand sample-accurate mp3 is a serious pain in the ass
14:50:45MO-Pantsump3 only goes to the next frame
14:51:16MO-Pantsuto make it work accurate would need to be after the decoder stage
14:51:30preglownah, it needs to be in the decoder
14:51:32DaKi][erwould it show the segments of the whole file in the file browser or just the .mp3 and .cue in there?
14:51:52preglowthe decoder has to walk the bitstream until it finds the frame it needs, then it needs to decode until it finds the sample it needs
14:51:56MO-Pantsuwell when I have edited mp3's in mp3directcut I can never get to the exact point
14:52:00preglowthis is painfully slow even on my atlon64
14:52:29preglowmp3 is a dirty bitch
14:52:42MO-Pantsuslut even
14:52:45preglowoh, yes
14:52:47linuxstbI would be more than happy for Rockbox to have a "slow but accurate seeking" mode that scans the whole file. But I think others would be happier with a "fast but inaccurate" mode.
14:53:05preglowlinuxstb: would probably be nice to be able to save the resulting seektable
14:53:31linuxstbI agree. Or even calculate it off-line.
14:54:07linuxstbDoes anyone know anything about Matroska? Would that give us a seektable if we wrapped MP3 files in a Matroska container?
14:54:44MO-PantsuI am thinking about attacking some of my MP3 collections where I don't own the original CD and using winamp with a output stacker and realtime them out to wav files using the ds output to achieve gapless on those non-lame files. dirty. don't like re-encoding and it would take a while but should work.
14:55:12linuxstbMy requirement is that I make lots of recordings from radio (in MP2 format), and often index them using a cuefile. I know I can split them, but would prefer a single long file.
14:56:00MO-Pantsuyou can embed cue in the file (at least in mp3 I know you can)
14:56:34tvelocityin a tag? that sucks
14:56:52linuxstbMO-Pantsu: It makes no difference where the cue information is stored, the work is to make Rockbox actually use them properly.
14:56:59MO-Pantsuyeah well would be if you want to reburn it
14:57:01preglowMO-Pantsu: rockbox might very well end up supporting the winamp gapless method
14:57:08preglowdon't jump to any conclusions
14:57:22MO-PantsuI would kiss your booty if it did ;)
14:58:19linuxstbAre there any standard ways to store seektables in MP3 files? Does the Xing VBR header do that?
14:58:41MO-Pantsubut this is my - how many days now? 3rd or 4th day talking about gapless. Beginning to sound like a spoilt brat. I shall endeavour not to mention it again for a while at least
14:59:10MO-PantsuI am re-ripping the CD's I own to Ogg as I type
15:00:40preglowlinuxstb: yes, there are
15:00:53preglowlinuxstb: only for vbr files, though, but then again cbr SHOULD be easily seekable
15:01:04linuxstbMO-Pantsu: Why don't you rip to FLAC? Then you can re-encode to any lossy format in the future without having to re-rip.
15:01:06tvelocityspeaking of ogg, how is rockbox handling multiple streams in an ogg file?
15:01:06preglowlinuxstb: the xing header contains a hundred seek points, the less common vbri header contains a variable amount
15:01:25MO-Pantsulinuxstb it's all about the storage space :)
15:01:39linuxstbMO-Pantsu: DVD-R?
15:01:55MO-PantsuIt's OK I am happy with this method
15:01:58linuxstbpreglow: Does Rockbox use the seekpoints?
15:02:11preglowlinuxstb: it uses xing seek points, it does not yet use the vbri ones
15:02:41 Nick mbr_ is now known as mbr (
15:02:47*MO-Pantsu wants his new psu to arrive. Fedup of the noisey failing psu currently in there
15:03:12linuxstbI've written my own software to record DVB radio to MP2 files (under Linux) - I may think about adding those headers to the output files.
15:03:32preglowthose files are cbr, no?
15:03:52linuxstbYes, but they can (in theory) have a padding byte in some of the frames.
15:03:57hicksAnyone here use gentoo?
15:04:01preglowhicks: aye
15:04:14linuxstbHowever, I don't think that's used by any of the radio stations I record from.
15:04:22preglowlinuxstb: what's the point?
15:04:28linuxstbOf what?
15:04:31hicksHow did you setup the cross compiler? Did you use crossdev, or a portage overlay or manual installed it in your home dir?
15:04:31preglowlinuxstb: padding
15:04:38preglowhicks: i compiled it by hand
15:04:43linuxstbI have no idea.
15:04:44preglowhicks: in my home dir
15:05:06hicksok,cheers :)
15:05:36MO-PantsuI want a DVR with network support and no copy protection
15:05:55linuxstbFreevo, MythTV ?
15:05:59MO-PantsuCopy the MPEG2 stream directly without loss. That would be so nice.
15:06:02preglowwavpack deecodes lossy files much, much slower
15:06:18linuxstbIs it still realtime?
15:06:28preglowlinuxstb: what bitrate/sample rate do they transmit?
15:06:47linuxstbGenerally 192kbps/48KHz. Sometimes Stereo, sometimes Joint-Stereo
15:07:05 Join Sucka [0] (
15:07:14preglowi've got a 192kbps 44.1khz mp2 file, and it sounds completely decent
15:07:18preglowi can't spot any artifacts
15:07:41pregloweven though they say mp2 required 128kbps per channel for transparency
15:07:58preglowi think subband codecs sound nicer than more complex transform codecs anyway
15:08:02linuxstbOnly problem is the 15KHz-16KHz cut-off - they do sound "flatter" than analogue radio. I like high-frequency noise...
15:08:06preglowthat's why i'm a bit partial to musepack
15:08:25preglowi can't hear frequencies above 16khz anyway, heh
15:09:33linuxstbIsn't musepack a "fork" of MP2?
15:09:39preglowkind of
15:09:48preglowit's based on the same concepts
15:10:09preglowi like it, it's simple and neat, no windows or anything
15:10:41preglowjust splits the signal into different frequency bands and quantises the different bands according to what we can hear
15:11:15preglowmp3 does basically the same thing, but uses a nasty mdct per filter band as well
15:11:24 Quit DMJC ("Leaving")
15:11:29preglowwhich gives the "nice" artifacts at low bitrates
15:12:02linuxstbIt just seemed too much work to keep up to date with all the various lossy codecs, so I just chose FLAC.
15:12:29preglowhaven't got enough disk space for that, i'm afraid
15:12:52preglowkeeping all my music ripped as flac would require me buying another disk, heh
15:12:55linuxstbI also sometimes actually listen to CDs...
15:13:02preglowsure, i listen to cds all the time
15:13:13preglowi occasionally listen to vinyl as well
15:13:28preglowi like physically handling my music from time to time ;)
15:13:42tvelocityi only touch my CD's to rip them
15:14:19tvelocityor maybe to admire it's booklet's etc
15:14:25tvelocitybut never to listen to it:P
15:16:46 Join Chamois [0] (
15:17:27 Quit ashridah ("sleep")
15:23:56 Join DMJC [0] (
15:24:51MO-PantsuI listen to DTS CD's
15:31:30 Quit scriptkitty (Read error: 110 (Connection timed out))
15:37:53***Saving seen data "./dancer.seen"
15:39:28 Join cYmen [0] (
15:52:27preglowlinuxstb: it's quite obvious it's the lack of seeking that's killing mpc, btw
15:52:32preglowlinuxstb: the very first thing it does it try to seek
15:52:46preglowlinuxstb: completely disregarding it's own can_seek function
15:53:54linuxstbDo you know where it tries to seek to?
15:54:09preglowto the first proper mpc frame, it seems
15:54:30linuxstbSo is it just seeking forwards?
15:55:28preglowit first seeks to the start of the file
15:55:34preglowit fails there already
15:55:38preglowthen it tries to seek again
15:55:46preglowi wonder why the hell they implemented can_seek
15:56:25linuxstbMaybe we can add two special cases to the seek function in codecmpc - seeking forward (replace by read) and seeking to ci->curpos (to take account of the initial seek).
15:57:59 Join ehntoo [0] (
15:58:44preglowyou could try, of course
15:59:54preglowi don't think it even needs to seek, provided you don't have id3v2 data patched in front of the stream
16:00:02preglowbut it does of course check seeks return value nonetheless
16:01:16linuxstbIs the offset parameter to seek_impl the offset from the start of the file?
16:02:11preglowbelieve so
16:02:14preglowi made it return true
16:02:17preglowbut still no dice
16:02:22preglowit just tries longer before it fails
16:02:37linuxstbI'm just trying my hack now.
16:03:19preglowmy files has no id3
16:03:32linuxstbI have implemented seeks forward, and seeks to the current position.
16:03:56preglowit does a seek forward, i see
16:04:38linuxstbNo. :-( It just takes longer to fail.
16:05:35preglowtry logging in seek
16:05:41preglowto see if it ever tries seeking backwards
16:06:00linuxstbHow do I log from a plugin? I don't think logf is exported.
16:06:12preglowfdprintf :-)
16:06:25preglowlogf not being supported is a crying shame
16:06:32preglowi'm considering just implementing it locally
16:06:45linuxstblogf should be added to the plugin API.
16:07:11preglowbagder didn't want it to
16:07:19preglowi can't see why, but there it is
16:07:59preglowthere'll be a function call overhead if its not used
16:08:14preglowbut anywho, i'll enable it locally
16:08:20preglowit really helps debugging
16:09:28linuxstbIt could be disabled via #ifdefs - it's only used by developers anyway.
16:09:57 Join DMJC-L [0] (
16:10:18preglowyou'd have to wrap every logf usage with #ifdef DEUG
16:10:20preglowDEBUG, even
16:10:36linuxstbUse a LOGF macro
16:10:39preglowor better yet
16:10:41preglowLOGF macro, yes
16:10:54preglowLOGF(rb, "lol %s", hehe);
16:11:43preglowwe could also constantly keep logf() at the end of the plugin api struct, so it'll never bother anyone
16:12:11preglowi'll go outside and enjoy the weather for a bit, brb
16:17:53preglowseems like summer finally arrived
16:20:59 Join B4gder [0] (
16:21:12B4gderfor the reconrd, I _do_ want logf() in the plugin API
16:21:31B4gderor logf functionality at least
16:21:59preglowB4gder: aight, then someone has misrepresented your opinion ;)
16:22:05*t0mas doesn't see a record function on iriver yet... sorry,,, can't record you ;-)
16:22:29HClSlasheri: awake?
16:30:12t0masawake too? :)
16:30:40t0mas(or amiconn? / preglow? / Linus? / Zagor?)
16:31:04preglowB4gder: but yeah, with a #ifdef HAVE_LOGF \n define logf() entry at the end of the plugin api struct, and a LOGF macro, shouldn't we be ok?
16:31:41B4gderlogf() is a macro if built without logf
16:31:45preglowahh, yes
16:31:48preglowbut with a wrapper function
16:31:59preglowplugin_logf, or something
16:32:08B4gderyeps, I'm fine with that
16:32:21t0masok, next design issue :)
16:32:26t0masthe wps on remote...
16:32:40t0masshouldn't we separate all displaying functions in tree, menu and wps?
16:32:52t0masand make 1 sort of lib/api for it?
16:33:07preglowif we wrap the logf function pointer in the api struct with ROCKBOX_HAVE_LOGF, we shouldn't need it, since the compiler will never see logf() then
16:33:12t0masso we can easily use it on other size screens (portability) and on the remote?
16:33:40B4gderpreglow: but there is no logf() function if built without it
16:33:43SlasheriHCl: yes :)
16:34:53B4gderI'm off again
16:34:55 Quit B4gder ("CGI:IRC")
16:35:30 Quit elinenbe_ (" Want to be different? HydraIRC -> <-")
16:45:45linuxstbpreglow: codecmpc firstly seeks to 0 (with ci->curpos==0) and then seeks to 0 again (but when ci->curpos==10).
16:46:01linuxstbSo it's the second seek that now fails.
16:46:33preglowcould you try commenting out the seeks? :-)
16:46:53linuxstbFrom where? libmusepack?
16:47:27linuxstbI'll have a look.
16:49:00 Join tksu [0] (
16:50:10tksuhi, great work guys! I just flashed my player and am really hyped... gotta start learning that c now
16:50:30tksuis the splitedit plugin removed from iriver builds on purpose?
16:50:59tksuokay.. just said that iriver is supported in the wiki and it had button mappings and everything
16:51:10tksubut, i'll wait :)
16:51:47preglowdo that
16:53:51 Part tksu ("congrats to all")
16:54:38 Nick StrathAFK is now known as Strath (
16:59:59 Quit thegeek_ (Read error: 131 (Connection reset by peer))
17:00:10linuxstbI may be wrong, but I think splitedit relies on the peakmeter - which isn't implemented on the iRiver yet.
17:01:55t0maswasn't it working with the mas chip?
17:02:03t0masand iriver doesn't have one...
17:03:26crwlwhat about the bass/treble settings, are they easy to implement on iriver (ie. on hardware side)
17:04:22preglowcrwl: yes, i've already done it, but i'm waiting for some stuff from austriancoder
17:04:34preglowit's really simple to do, but i don't want to crash his fixes as well
17:05:24linuxstblibmusepack is seek-crazy when opening a file...
17:06:11crwlpreglow, okey
17:07:07preglowdamn, how easy musepack is determined to be on us
17:09:23linuxstbI can't get it to work - we need a proper solution. But I can't do any more now.
17:09:57linuxstbIt keeps wanting go to back a few bytes. I think more time needs to be spent fixing libmusepack to remove those seeks properly
17:10:02preglowseems like the coder is going to drop by afterwards
17:22:33 Quit tvelocity ("Leaving")
17:25:32 Quit spiralout (Read error: 104 (Connection reset by peer))
17:25:42 Nick MoosCamaro is now known as Moos- (
17:26:33 Nick Moos- is now known as Moos (
17:30:47t0maslol didn't like opening 15 documents at once
17:32:03bipaksry, wrong window
17:32:20Slasherit0mas: Hehe. I hope you are not editing any source code with openoffice ;)
17:33:18t0masofcourse I am
17:33:26t0masdon't you like different size sourcecode?
17:33:32t0maswith tables... and a print playout?
17:33:39t0masoh and nice changeable colors...
17:33:41preglowSlasheri: you got any idea what the seeking bugs are caused by?
17:33:44t0masand every few lines a new font
17:33:51Slasheripreglow: yep, forward seeking is fixed :)
17:33:58Slasherior it should be, please test
17:34:06preglowSlasheri: the bug where you suddenly are skipped right into the middle of another song, yes?
17:34:13preglowooh, lookie there
17:34:16t0mas(but no, I was just opening a word document... and it choked on some MS shit)
17:34:31Slasherii think the latest commit should fix those, but i can't be sure
17:34:44preglowyes, it's not an mp3 specific fix?
17:35:00preglowgreat, 'cuz it happened for wavpack as well ;)
17:35:08preglowmight test afterwards
17:35:18Slasheriit was a problem with all file types that support seeking
17:35:51crwlwhat about that queue/insert playlist/modify playlist when the next tracks are already buffered thing? ;)
17:36:32Slashericrwl: i don't know yet about that.. :/
17:37:01Plugh_I recorded 18 hours on my FMR using a build from last week, and a lot of the tracks (74 min splits) are unplayable on anything but the recorder. Does anyone know of some good vbr fixing programs that deal with the buggy recording?
17:37:30Plugh_the mp3 are showing in the WPS as fixed bitrate mp3, which I know isn't true
17:37:43preglowdoes seek_buffer work together with read_filebuf?
17:37:54Slasheripreglow: yes it should
17:37:58***Saving seen data "./dancer.seen"
17:38:11preglowi'll try hacking that into musepack, then
17:38:26 Join hardeep [0] (hardeeps@
17:38:32Slasheribut remember that backward seeking is only partially implemented
17:39:06hardeephi all
17:39:27hardeepnice work on the iriver, it's working great
17:39:34preglowwe need to seek backwards a couple of bytes, i think
17:39:48Slasherimaybe some configuration options should be added to codec interface to support files better that require more seeking
17:40:04Slasheriok, currently that is a problem if those bytes are not buffered
17:40:27preglowi've got musepack
17:40:34linuxstbI don't think libmusepack _requires_ seeking - it's just how it works at the moment. i.e. I think it could be rewritten to remove those seeks.
17:40:45Slasheri(buffer will be cleared if seeking more bytes that are currently available in the buffer)
17:40:46preglowit's so incredibly not realtime
17:41:09linuxstbpreglow: What did you have to do?
17:41:23preglowlinuxstb: just call seek_buffer from musepacks seek callback
17:41:40linuxstb:-). Simple.
17:41:49preglow return ci->seek_buffer(offset);
17:42:18preglowbut yeah
17:42:24preglowthe codec is so incredibly inefficient as it is
17:42:42Slasheripreglow: btw, currently seek_buffer returns only true or false.. =)
17:42:57preglowthe fact that it works even partially is a miracle deserving of jumping and loud shouting
17:42:58SlasheriMaybe it should be fixed to return the position
17:43:06preglowSlasheri: that's how musepack seek likes it
17:43:11Slasheriah, ok
17:43:47preglowcurse this bloody codec to hell
17:44:12preglowi'll just do a nasty hack and see how the audio turns out
17:46:12 Join markun [0] (
17:47:00markunt0mas, you there?
17:52:46 Quit nick_ (" ")
17:55:19hardeepSlasheri: implementing audio_flush_and_reload_tracks() would fix the problem with buffered tracks and modifying the playlist
17:55:27bipakcould someone update the daily build changelog? :>
17:56:02Slasherihardeep: ah, thanks :)
17:56:21Slasherithen that should be implemented :)
17:57:20Slasherihardeep: when audio_flush_and_reload_tracks is called, will playlist_peek(0) return the right track to start buffering with?
17:57:37hardeepSlasheri: yeah
17:57:52hardeepalthough in the mpeg code, we only flushed the tracks that weren't being played
17:58:27Slasheriah, so everything doesn't need to be reloaded from disk
17:59:02Slasherii will implement that :)
17:59:25 Quit Patr3ck_ ()
17:59:39crwlwhee ;)
17:59:50Slashericrwl :)
18:02:04 Join Shagnar [0] (
18:02:35HClSlasheri: is that track event thing done yet?
18:03:57SlasheriHCl: Not yet.. You would need exact information when the track is changed?
18:04:15SlasheriI can do that.. I will try to do it soon ;)
18:05:24t0masmarkun: yes I am
18:07:38markunt0mas: I tried to enable seeking with the remote, but it didn't work, do you want to take a look at it?
18:07:56t0masI'm a little busy...
18:08:02t0masso maybe you can email it?
18:08:17HClSlasheri: well, i need events on when a track gets played
18:08:18HCland its filename
18:08:22preglowi lose 17 bits of accuracy to make musepack work
18:08:26t0mastomas and Salfischberger
18:08:38preglowless then sixteen bits remain
18:08:40t0masmarkun: and you can do it in dutch...
18:09:48markunI know, but It's not very kind to the other people in the forum..
18:10:20t0masforum is ok too
18:10:28t0masjust drop it somewhere so I can read it tonight...
18:11:15markunok, I'll mail it
18:20:14CoCoLUSpreglow, about the bass/treble settings...
18:20:22CoCoLUSyou said you'll wait for austriancoder?
18:21:20CoCoLUScould you tinker some unofficial patch for the curious? :)
18:23:26preglowno time, sorry
18:24:28CoCoLUSwhen do you think austriancoder will be ready?
18:25:18CoCoLUSsorry for being annoying but thats the only thing keeping me from using rb during normal usage :)
18:25:42preglowi don't know, he said he should have commited it yesterday or earlier
18:25:45preglowbut it didn't happen
18:25:47preglowbrb for real now ;)
18:30:34markunpreglow: about replaygain, I think we should tell the decoder to scale down if the peak value is too high to prevent clipping and then recalculate the replaygain value and adjust the volume.
18:31:31 Nick QT_ is now known as QT (as@area51.users.madwifi)
18:45:32HCldidn't austriancoder lose his iriver?
18:46:02 Join tvelocity [0] (
18:46:16markunhe lost it?
18:46:23HCldamage it
18:46:27preglowyeah, he did
18:46:38*HCl prods Slasheri
18:46:47preglowmarkun: we need to think about that, i'd really like it if the codec loader did the scaling
18:47:30markunThe scaling to prevent clipping or also the replaygain?
18:48:39markunI think it would be best to keep the resolution as high as possible and use the mixer to do the replaygain, no?
18:50:03markunOr would you like to 32-bit samples from the codec (in the case of Tremor) and then have a central place that does the output conversion and clipping prevention?
18:50:17markun.. like to get ..
18:53:15pregloweverything i've seen outputs 32 bit samples
18:53:33preglowso a central place would be great
18:53:39preglowwe don't only need to do scaling
18:53:45preglowwe also need to do shifting and resampling
18:53:48preglowand possibly eq
18:53:50preglowand other effects
18:55:41preglowso the codecs could instead of talking directly to the pcm buffer driver, talk to some codec loader function
18:55:53linuxstbA quick audit: FLAC outputs 32-bit host-endian non-interleaved integers
18:56:16linuxstbWAV is little-endian 16-bit interleaved integers
18:56:21preglowwavpack outputs 32 bit little endian, i think
18:56:38preglowso yeah
18:58:03linuxstbA52 is a 32-bit fixed-point representation (non-interleaved)
18:58:27preglowboth libmad and tremor output 32 bit fixed point in host endian
18:58:33preglowlibmad non-interleaved
18:59:00linuxstbVorbis is 16-bit host-endian interleaved
18:59:59preglowbut yeah
19:00:07preglowwe obviously need the function to handle both cases
19:00:26SlasheriHCl: Hmm, you don't need the exact time when the track is changed? (however, i can do also that later)
19:00:32linuxstbYep - the data returned by ov_read is passed directly to the audio_insertbuffer function.
19:00:56SlasheriHCl: If +- few seconds is ok, i can do it very soon
19:02:04linuxstbpreglow: It could be useful to support 24-bit FLAC files - they are becoming more common in live music trading circles.
19:02:23crwl24-bit FLAC = FLAC encoded from 24 bit source?
19:02:33linuxstbcrwl: Yes.
19:03:28HClSlasheri: yea, it is
19:03:50HCltiming isn't too important, as long as it gets the filename of the track that finished playing
19:04:07HCland as long as it only gets called once for each time it finished playing
19:04:45hicksAnyone else get a compilation error with the latest cvs? For h100 target failing on "Make in dumb" (dumb) CC src/it/itread.c
19:04:57 Quit hardeep ("BitchX: all the things phone and hop won't include")
19:05:00hickssaying insn does not satisfy its constraints.
19:06:07markunlinuxstb: in vorbisfile.c the samples are converted to 16 bit, we cound change the code to output the original 32 bit.
19:06:15HClnope, it compiled here fine
19:06:25SlasheriHCl: ok, i can do that :)
19:06:37HCl :)
19:06:42preglowlinuxstb: that shouldn't be a problem, just specify a bigger shift factor to the output function
19:06:46preglowthat is, if we choose to do it that way
19:06:48hickshmm. Are you using binutils 2.16 and gcc 3.3.4?
19:07:10preglow3.3 wont work
19:07:17linuxstbmarkun: OK - that could be useful.
19:07:45hicksThat would probably explain it then. I'll rebuild using 3.4
19:07:48markunlinuxstb: If you want, it's at about line 1574
19:08:02 Quit Shagnar (Remote closed the connection)
19:08:51linuxstbIt depends if we change the codec API. At the moment it only accepts 16-bit, big-endian interleaved samples.
19:09:02hickspreglow, 3.4.0 or 3.4.4?
19:09:43crwli've got 3.4.4
19:10:33hicksok, I'll grab that version.
19:11:20HClcan the uda output higher than 16bits sound?
19:11:28*amiconn used rockbox on iriver "for real" the first time
19:11:42linuxstbI think 20-bits is the limit.
19:12:11linuxstbI think the DAC can do 24-bit, but there's another bottleneck somewhere (the Coldfire?).
19:12:55crwlamiconn, i've done that for two days now :)
19:13:29amiconnI used it in my car, and believe it or not, I still prefer using one of my archoses
19:13:49amiconnThe most annoying thing in the car is the lack of voice ui
19:16:14crwli don't do that
19:16:27crwli've used it instead of the original firmware mostly because rockbox does gapless
19:16:34crwland that's really something :)
19:16:48markunI just noticed gstreamer has some tag readers for different formats. ape, vorbis, id3
19:16:49amiconnYes it is, definitely
19:16:55preglowamiconn: well, given the fact that you're an old archos user, that doesn't really surprise me...
19:16:57*amiconn notices markun
19:17:25amiconnmarkun: I read that you didn't understand why menu navigation from the remote wasn't implemented
19:17:59markunamiconn: was there a good reason?
19:18:17amiconnThere is a reason for this - as the archos remote doesn't have a button to enter the menu, it doesn't make much sense to have menu navigation on the remote
19:18:38markunYes, I figured that would be the problem..
19:18:46amiconnYour commit has the side effect that it's now possible to navigate the menu on archos too
19:19:00markunwell, now you can at least leave the menu with the remote if you like :)
19:19:15markunor change some things once you are in, but it's a bit useless..
19:20:31markunamiconn: we could remove the defines of the archos remotes again if you think it's better.
19:20:46amiconnIt's not a big problem since the code size increase is minimal though
19:21:38amiconnIt would also be easy to disable remote navigation for the menu - just remove the MENU_RC_* #defines from menu.h for the archos keypads
19:21:44preglowHCl: i'm sitting on some heavily fixed and updated dumb code
19:21:53markunI'm not so sure about my button layout for the iriver remote, but it can be changed later.
19:22:37markunI made 'play' the run button, so keeping it pressed opens the context menu..
19:22:45HClpreglow: mm?
19:22:47amiconnI'm not so sure TREE_RC_RUN was really redundant
19:22:49HClfrom where did that come
19:22:58preglowfrom the guy who made the foo_dumb plugin
19:23:31HClany floating point things gone?
19:23:35markunamiconn: I removed TREE_RC_ENTER, but we could put it back if we assign a different key to it.
19:23:36amiconnIt looks like it was, maybe to make it similar to the main TREE_RUN / TREE_ENTER dfines
19:23:40preglowdont think so, i'll have a look
19:23:48*HCl goes back to sleep :/
19:24:35amiconnmarkun: The reason for having 2 defines there for the main unit buttons is the Ondio, which only supports one of them
19:24:48markunamiconn: another thing: the 'center-scrolling' code needs to be implementd in so many cases that I started to wonder if we shouldn't share the code..
19:25:08amiconnAs long as we don't have to support more different remotes, I think it can stay as it is now
19:25:38markunIt's now in tree.c and menu.c but should also be in playlist_viewer.c
19:26:03markun't we have a generic list browser or something?
19:27:24amiconnWell, one of the main principles of rockbox is KISS, and a generic list browser would put just another level of abstraction
19:27:39amiconnIt might be very complicated to get it that generic.
19:28:32linuxstbIs FLAC playback broken for anyone else? It doesn't seem to be buffering enough data - it plays for about a second, then stops to read data from disk, then plays a second.... etc
19:28:38amiconnThere are many special cases, e.g. in the tree browser, there is the partial load for databse browsing
19:28:48markunHm, maybe. Also rockboy has a menu..
19:29:17amiconnYes, and rockboy (as some other plugins) should use the core menu functions for that
19:29:32amiconnlinuxstb: FLAC has a performance problem with some files
19:29:45linuxstbThese are the same files that played perfectly in the past.
19:29:46preglowHCl: there are still floats everywhere
19:30:16linuxstbIt does look like a buffering problem (from the LED action)
19:30:19preglowbut this one is much more accurate, they say
19:30:31 Quit einhirn (Read error: 131 (Connection reset by peer))
19:30:45preglowface it, flac just barely runs
19:31:48linuxstbBut up until today, I had no problems at all.
19:32:10amiconnmarkun: I thought about the center scrolling configuration. There are problems with either approach.
19:32:53amiconn(1) We could set it in percent, but this creates the problem that many values will lead to the same amount of 'line margin'
19:33:27markunamiconn: I've implemented it now using percents and made it configurable in steps of 10%
19:33:41amiconn(2) We could also set it in lines (the more logical way imho), but then we need to clamp too high settings in the code, depending on th etotal line count
19:34:09markun0% is old behaviour, 100% is center-scrolling. I have it on 80% now.
19:34:21amiconnThe code within each browser function should be just one line per direction
19:35:12markunI like it this way, do you want to try a patch?
19:35:30markunas allways it's not finished :)
19:35:31amiconnWhen using percent, it might be more logical to allow a range of 0..50 (50 would be center)
19:36:09markunamiconn: Yes, I thought of that too. Now I devide by 200..
19:36:53amiconnI think I should get asm memset() finished, then start working on the graphics api
19:37:03t0mas3 days to go
19:37:03t0mas 19:36:24 up 297 days, 10:21, 0 users, load average: 0.00, 0.00, 0.00
19:37:05amiconnasm memcpy() has to wait...
19:37:23t0masamiconn: graphics api?
19:37:59***Saving seen data "./dancer.seen"
19:38:17t0mas(have you read my idea about separating the display code?)
19:40:19*amiconn is still reading up today's log
19:40:40linuxstbI've changed the ci->configure lines in codecflac to the same as codecwavpack, and that seems to have fixed the problem.
19:42:42 Quit bobTHC ("Smoke Weed Every Day !")
19:43:22hicksIt compiles now, it was because of my gcc version after all. Only problem now is the built rockbox.iriver file hangs on boot showing result: 0
19:48:15 Join muphicks_ [0] (~hicks@810058c306a5aef0.session.tor)
19:48:17 Nick muphicks_ is now known as hicks2 (~hicks@810058c306a5aef0.session.tor)
20:03:35 Quit hicks (Nick collision from services.)
20:03:46 Nick hicks2 is now known as hicks (~hicks@810058c306a5aef0.session.tor)
20:07:59 Join hardeep [0] (hardeeps@
20:11:16amiconnPlugh_: Still there?
20:17:38 Join Tangleding [0] (
20:18:05TangledingHello Rockbox
20:18:08 Join hubbel [0] (
20:19:40MoosHi Tang
20:20:33 Join scriptkitty [0] (
20:20:49Mooscongratulations for your new H140 ;)
20:21:16t0masamiconn: done reading?
20:22:12TangledingThankls Moos
20:22:28Tangledingi've updated with Rbx just as iv'e received it
20:22:57Moosgood reflexe ;)
20:23:01TangledingThe Rbx progress is amazing really
20:23:18Tangledingcongratulation to every contributors
20:23:44Tangledingi'm waiting after my travel to paris
20:23:56Tangledingbut if i have money enough when i come back
20:24:07Tangledingwill be glad to donate another time :)
20:24:46MoosParis wait you :)
20:25:21preglowdamn, we gotta find something to do with all this money
20:29:56t0maspreglow: pay hosting for :)
20:30:19t0masand buy some more H3xx's for the hardware people
20:30:38Tangledingdo what you want guys
20:30:43t0masso that port will continue too... and maybe do that with 2 bdm's ?
20:31:08t0masso some other core dev can do things with a bootloader too?
20:31:16Tangledingyou have some money loss with all buying:
20:31:24TangledingH3xx, H100,
20:31:37Tangledingnext M3, M5 and X5 ;)
20:31:52preglowhosting is sponsored, afaik
20:32:05t0masyes, and linus's next volvo?
20:32:12preglowi want a car too :/
20:32:20*t0mas wants a license..
20:32:23t0masvalid target too?
20:32:24Moosahah :)
20:32:39MO-Pantsuwhat codec is .wv?
20:32:57t0mas9 months... left... and then noone will ever dare to drive near Gouda again :P
20:33:21amiconnMO-Pantsu: wavpack
20:33:37MO-Pantsuah...supported? or to be supported?
20:33:44MO-PantsuI may just convert it otherwise
20:34:03preglowi've got the license, just need the car
20:34:14preglowMO-Pantsu: it's supported
20:34:28t0maspreglow: why not buy a very cheap one? 12th hand?
20:34:28preglowas of today
20:34:38preglowt0mas: i will once i move
20:34:41Tangledinglol Rockbox is planned to be ported on volvo to?
20:34:43MO-Pantsucool! you iz fast
20:34:52t0masTangleding: if I get a car for that: yes
20:35:17t0masbut seriously: Linus was looking at some way to connect his iriver to the volvo sound system
20:35:24t0masso maybe he'll find out something intresting
20:35:59Tangledingokay (was joking of course)
20:36:12t0masI know :)
20:36:45*t0mas is bouncing around his room
20:36:46*t0mas np: Various Artists - Every Single Day (0:24) []
20:40:55 Join Stryke` [0] (
20:46:39 Join webguest70 [0] (
20:48:41webguest70is R-box's mp3 playback now better than irivers mp3 playback ?
20:48:54t0masit's more accurate
20:49:00Tangledingi'm checknig if gapless
20:49:02t0masbut it lags eq
20:49:03Tangledingif yes
20:49:08t0masso no extra bass/treble
20:49:12Tangledingit will be "Hell YES!!!!
20:49:53webguest70hmm, a bit of bass eq would be nice
20:50:17preglowit's not often gapless
20:50:22 Join xam [0] (xam@2001:770:102:2:201:faff:fe7e:c00d)
20:50:38webguest70I've got old ears, need all the help they ca get
20:50:45crwlit seems to be quite gapless now at least with lame encoded files
20:51:05webguest70good all my mp3's are encoded in lame
20:51:06crwland i have noticed no more skipping to next track (few hundred milliseconds) too early
20:51:39Tangledingwill be quite soon i guess
20:51:53preglowi tried to make lame gapless work
20:51:58preglowbut something's wrong with libmad
20:52:06preglowor my code...
20:52:52*webguest70 hands preglow a frosty beer
20:53:54Tangledingi see
20:54:02webguest70now to see if I can avoid bricking my player :)
20:54:05Tangledingwish you suceed preglow
20:56:07 Quit xam ("Chatzilla [Firefox 1.0.4/20050517]")
20:56:40 Join xam [0] (
21:04:48 Quit webguest70 ("CGI:IRC")
21:04:56 Join webguest70 [0] (
21:08:05Tangledingbye all
21:08:11 Quit Tangleding ("Chatzilla 0.9.68a [Firefox 1.0.4/20050511]")
21:17:43webguest70heh, outstanding, love the file tree browser
21:18:36webguest70dir buffer is full ?
21:19:40preglowlarge directory?
21:20:00webguest70yes 37 gigs worth
21:20:14preglowyes, there is such a thing as a too large directory
21:20:19preglowthe limit can be raised in preferences
21:21:08preglowgeneral settings>system>limits i guess is the place
21:21:54webguest70heh, still trying to figure out how to navigate, need some time
21:24:32webguest70upsidedown mode, neat
21:24:54*HCl is bored :/
21:25:54 Join DomZ [0] (
21:26:26bipakHCl: hack some new features ;)
21:29:26hicksHCl your not the HCl from pulltheplug are you?
21:31:28HCli am.
21:31:58hicksThought it was unusual to see that nick elsewhere :P
21:32:11HClheh :p
21:32:35hicksI was gonna say HCi and see your response instead of asking :P
21:32:51HClo.o how would that have helped? :p
21:32:57HCli would've said hcL or so
21:33:12 Join thegeek [0] (
21:33:23HClptp was a long time ago, i'm glad i left that place.
21:34:17hicksI havn't been there for ages. Used to enjoy playing around on mainsource, but when it went down a few years ago, I never went back.
21:34:27HCli see
21:34:54HClthere's a 486-mainsource-copy under a cabinet in my room, but. i found it too bloody boring to set it back up completely
21:35:21hicks:) I think I've got all the level solutions on a backup cd somewhere.
21:35:28HClhehe :)
21:38:02***Saving seen data "./dancer.seen"
21:39:17bipakis there a .rockbox folder in cvs?
21:39:42bipakor do i have to copy all necessary files into it? my self
21:40:05 Quit xam (Read error: 54 (Connection reset by peer))
21:40:49preglowdo a make zip
21:40:54hicksbipak, if you're building from cvs do a "make zip" and you'll have a zip with the bin and .rockbox folder
21:40:56preglowif you're building yourself
21:41:42bipakah nice
21:42:11 Join xam [0] (
21:42:57bipakcan't open dir fonts at /home/bip/tools/rockbox_cvs/rockbox/tools/ line 90.
21:43:34hicksdo you have a fonts dir? If not check out the fonts module from cvs
21:44:09bipaki have a font dir in the firmware folder
21:44:24bipakso i think i need the fonts module
21:44:35bipakcause there is just one bdf file
21:44:55preglowcheck out the fonts dir
21:45:21bipakdone :>
21:45:27preglownow you should be good to go
21:45:46bipakyep, worked
21:49:10webguest70is it possible to delete tracks ?
21:49:26preglowkeep the button pushed and press delete in the menu
21:53:13webguest70hmm, it can delete whole dir, but I cant see a menu for deleting a single track within a dir
21:55:27webguest70cancel that^^, my blindness, am I'm not even drunk
21:56:08bipaknice work, cant say it often enough
21:57:03 Part xam
21:57:32hubbelsomeone should tell Motorola to fix Figure 3-5 "Exception Stack Frame Form" in the MCF5249 User's Manual.. It's fucked
21:58:33 Join Pete [0] (
21:58:42bipakfeel free to do it ;)
21:59:41hubbelbah, they should have their own proofreaders
21:59:52bipakhehe :>
22:03:55Petehow long should a scandisc on iriver ihp 120 take?
22:06:14bipakis it right that gapless means... to play file after file without any delay?
22:08:23 Join bipak_ [0] (
22:08:33bipak_got the answer :)
22:11:07 Quit Pete ("CGI:IRC")
22:11:48 Join matsl [0] (
22:12:16preglowhubbel: it's not the first erronous part of the manual i've seen
22:12:54webguest70the pop between track selections is quite loud
22:13:39preglowwe know
22:13:53preglowhubbel: so, what's up with recording?
22:16:57hubbelpreglow: i've got dma working
22:17:40preglowand it's working well?
22:18:12hubbelpreglow: One small glitch but I think I've found it now
22:18:48preglowdo you know if we can have full duplex audio, btw?
22:18:53hubbellinus had an error in the default interrupt function, which was supposed to tell which interrupt really occured - but it masked the wrong bits
22:19:08hubbelpreglow: yes, it should be possible
22:19:24hubbelpreglow: like time-shifted recording/playback =)
22:19:43preglowi'm thinking about realtime special effects, heh
22:19:59hubbelhum.. it would be cool to make realtime vocoder or some darth-vader filter =)
22:20:05bipak_lol :D
22:21:06preglowi'm still thinking about the vocoder, heh
22:21:27hubbelrealtime pitch change filter is cool
22:21:50Plugh_amiconn: I'm here
22:21:52preglowyeah, that shouldn't be too hard either
22:22:24Plugh_amiconn: you have some input on my vbr probs?
22:22:47HClheheh. sweet.
22:23:05HClmy university is gonna upgrade from 1gbit to 10gbit within this year
22:23:24preglowwhat necessity
22:23:27markunHCl: what a nice university you have :)
22:23:27HCl :P
22:23:30HClmarkun: :p
22:23:37preglowi already have over 10gbit, though
22:23:46preglowcan't remember how much
22:24:03preglowthat is, _i_ only have 100mbit :/
22:24:55markunjust found out the metadata in wavpack is apev2. fb2k writes replaygain info to mp3 files also in apev2.
22:25:05HClwe can have 100mbit per computer, multiple computers = more.
22:25:12HClwith a maximum of 300mbit per flat
22:25:42 Quit bipak (Read error: 110 (Connection timed out))
22:25:58preglowhaven't tried that
22:26:02preglowdepends on the switch ports, i guess
22:26:12preglowi've only got one ethernet socket
22:26:23preglowthere's another one in the kitchen, though
22:27:02preglowso i guess there's a theoretical maximum of 400mbit per flat here
22:27:15HCl :)
22:27:29preglowi shall leech like a mad man the days before i move
22:27:31*HCl goes to try his gta mission again..
22:27:40HClthen it'll be too late :)
22:27:43HClleech now, burn on dvd :)
22:27:46preglowno time now
22:27:50HClwhy not?
22:27:53preglowlots of disk space
22:28:02preglowneed to boot windows to leech, can't work in windows
22:28:25HCllook at the bottom pic there
22:28:30HCltotal stats of our university over time
22:29:01HCleven though we're well under 800mbit, the administator person of our isp says it "warranted an upgrade to 10gbit" :P
22:29:52HClall because we hit the 1gbit flatline earlier this week :p
22:30:05HClour university ftp server pulled 850mbit on its own :P
22:30:12hubbelAnother error in MCF5249 UM, there's no bit in INTERRUPTEN or INTERRUPTEN3 to enable/disable PDIR3UNOV (int 0x56)
22:30:15*HCl has root access on that thing :)
22:34:08webguest70I have pasted wps code to a .text file, can I put that in .rockbox's dir ?
22:34:19Stryke`change the extension to .wps
22:34:38Stryke`then place it in .rockbox
22:37:59 Join yngwi [0] (
22:38:08yngwihi all
22:40:32yngwitoday i had my h140 out with me and rockbox running... worked great, had to reset only once.. :-)
22:40:56BagderWear Your Pin (tm)
22:41:02yngwithe first "real life" test
22:41:35yngwiyeah, i have a special "pencil" don't know the word, which fits into the reset hole
22:42:13yngwi"propelling pencil"
22:42:26preglowHCl: strongly enhanced version of dumb
22:42:50preglowi just used a piece of wood i found
22:43:31MO-PantsuFF is still broken on MP3. Someone wanna let Miika know? :)
22:43:43preglowbroken how?
22:43:50yngwiit sounds really good, besides the known problems :-)
22:44:05MO-Pantsuit still thinks the track has finished before it has if you FF
22:44:36MO-Pantsuif you FF to near end of some tracks it ends the track prematurely
22:44:41preglowthat isn't the bug he fixed today
22:44:44MO-Pantsuoh ook
22:44:57pregloweverything that has to do with track length is really shady thus far
22:44:57SlasheriMO-Pantsu: yep, i found that problem just recently when i was testing track flushing and reloading
22:45:19preglowi don't even know where codecmpa gets it's current tracklength from
22:45:27preglowunless it's got a lame header
22:45:58markunwhere do we want to do all the tag reading?
22:46:03MO-PantsuYeah I was testing the gap between tracks and found it's jumping ahead before the track ends so have to listen to the entire track to test that
22:47:24preglowseeking is not very accurate
22:47:28preglowknown behaviour, etc, lol
22:47:34Slasherimarkun: Hmm, something called metadata.c or similar might be good place :)
22:48:04preglowbut yeah, do we want rockbox to have loaders for all metadata types?
22:48:24preglowit's not that bad an idea, really, there aren't THAT many metadata types
22:48:55Slasheripreglow: codecs could read the metadata but that will cause problems to the next track display etc.
22:48:56linuxstbYes, I think it should just be one function - e.g. "get_metadata" that gets called for every type of file.
22:48:59markunid3, id3v2, apev1, apev2, vorbicomment?
22:49:33markundoes mpc still use apev1?
22:49:45preglowapev2, i guess
22:55:10markunShould the metadata of a file be read when it's added to the playlist?
22:55:37Bagderit should be read when loaded into the audio buffer
22:55:48Bagderif it should work like existing rockbox
22:56:07linuxstbOr earlier if the first track is larger than the audio buffer. Otherwise we have no "next track" info.
22:56:28Bagderright, but that's a limit we've accepted
22:56:48Bagderso next track info only appears when we actually have that info around
22:56:56linuxstbI don't feel strongly about it. My WPS doesn't even contain the next track...
22:57:12Bagderrather common for Archos' 1.6MB buffer ;-)
22:57:13preglowi don't care either
22:58:04preglowlinuxstb: that resampler thread quickly evolved away from our needs ;)
22:58:24linuxstbpreglow: That's linux-audio-dev for you - we were lucky to get two or three on-topic replies.
22:59:29linuxstbThey've spent the last week discussing compilers...
22:59:52preglowi wonder how picky people are about resamplers
23:00:04preglowperhaps a simple linear interpolator will do for other resampling as well...
23:00:32linuxstbI think we should just get something working - it's easy to improve later.
23:00:34Bagderseems clever to at least start with the easy way
23:02:01linuxstbWe need to agree on the format codecs will give Rockbox the uncompressed data in. A common format, or their own format?
23:02:03preglowi wonder how large buffers we'll ever need
23:02:25preglowtheir own format plus a shifting coef would be nice
23:02:36preglowthen again
23:03:31linuxstbIt seems easier to do it the way we do now - possibly with the exception of mono->stereo conversion.
23:04:32preglowand all the other processing we need to do
23:06:21preglowi don't we'll win much apart from code duplication if we keep such processing in the codecs
23:06:26preglowi don't think, even
23:08:17linuxstbBut every codec's different.
23:09:01linuxstbBut I am almost convinced.
23:09:04preglowin what regards?
23:09:15preglowa lot of them share the fact that they serve data in 32 bit ints, noninterleaved
23:09:20preglowso they have to interleave in the codec now
23:09:30preglowif we ever get replaygain support, they all need to gain their data
23:09:32preglowand eq it
23:10:28preglowthe only notable difference is data type size and interleavedness, imho
23:10:31 Join looksaus [0] (
23:10:48hubbelThis is unreal.. I've get a crash when I write "Done" to DSR1 in my DMA1 handler (recording).. This works fine in the DMA0 handler (playback).. I've double-checked the address of DSR1 and I can read out the status without problem
23:11:25preglowperhaps this is why iriver uses interrupts ;)
23:12:47hubbelThe thing is if I use DMA without the interrupt-when-finished feature, it works fine, but you don't know exactly when the transfer has finished
23:16:14Slasherii managed to fix the seeking problem :)
23:16:21Slasherinow testing flushing
23:16:21preglowSlasheri: leetness
23:16:35preglowhave you tested playing silence between track changes, btw? :PP
23:16:45preglowthat click after skipping a track is my last major annoyance
23:16:50Slasherino i havent :)
23:17:13SlasheriHmm.. I don't get any click when track changes..
23:17:23hubbelpreglow: aaaaaaaaa
23:17:27hubbelpreglow: now it works!
23:17:30SlasheriWith mp3 if the track contains some silence
23:17:48preglowhubbel: what wrong?
23:17:49hubbelpreglow: changed DSR = 1 to DSR |= 1
23:18:16preglowSlasheri: you dont? i get clicks at the start of all tracks
23:18:44DomZi can confirm this :)
23:18:48Slasheripreglow: ah, yes with skipping.. Now i got a very small near unhearable click
23:18:57preglownear unhearable?
23:19:00preglowmine clicks like mad
23:19:06Slasherihmm :/
23:19:14DomZnot too much for me
23:19:21preglowi'm convinced it's dc
23:19:25preglowcan't think of anything else
23:20:43markunI have the same problem. You don't get the click if you skip during silence
23:21:51preglowwell, of course not, if the dac holds a zero you wont hear anything
23:21:57preglowif it holds a non-zero sample, though
23:22:09 Quit Chamois (" HydraIRC -> <- IRC for those that like to be different")
23:22:11preglowlinuxstb: your oasis ac3 skips like mad now, btw
23:22:35markunyes, it was my proof for the dc theory
23:22:44looksausis anyone toying with FLAC support on the iRiver-iHP1x0?
23:23:13Bagderlooksaus: it is already there
23:23:14preglowlooksaus: toying? it's already supported
23:23:19looksausit seems all kinds of things are moving around it, but the progress is somewhat scarcely documented
23:23:29 Quit matsl (Remote closed the connection)
23:23:36preglowwell, the wiki says it's supported
23:23:42looksausis there a battery life comparison somewhere?
23:23:43preglowthat's pretty much all the documentation you're going to see
23:23:57looksauspreglow, sorry, I read the FAQ and the main page
23:24:57preglowbut yeah, read the url linuxstb pasted, there you've got all the file formats we currently support
23:25:13looksaushey, this is really nice
23:25:48looksausso, any comparisons of battery life with the original firmware versus rockbox on the iHP-120?
23:26:03preglownone really official, no
23:26:10preglowanyone know what solid ended up on for vorbis?
23:26:14linuxstbpreglow: I think the oasis ac3 skipping is just buffering problems - it looks like the buffer is becoming empty.
23:26:29preglowlinuxstb: it didn't skip yesterday
23:26:41 Join ashridah [0] (
23:26:42linuxstbAnd neither did my FLACs
23:27:01looksauspreglow, WOW!! someone is working on MIDI on the iRiver thing?
23:27:21preglowlooksaus: yes
23:27:26looksaussorry if I sound so excited, but as a music teacher, this is REALLY great news
23:27:40preglowit's even working
23:27:44preglowbut it's too slow at the moment
23:28:00preglowand the author wont have time to hack at it for a little while
23:28:03bill20r3mod files are next
23:28:19preglowlooksaus: you're seldom required to apologize for enthusiasm here ;)
23:28:35preglowyes, indeed, but such is life
23:28:39ashridahheh, the midi module will be a pain to buffer in with other music
23:28:45preglownot really
23:29:00ashridahpreglow: well, given the potential size of the wave table
23:29:08preglowthe buffer will just have to be flushed sooner than usual in worst case
23:29:40*HCl prods Slasheri
23:29:52linuxstbpreglow: Increasing the size of FILEBUF_CHUNKSIZE in the codeca52 fixes the oasis problem.
23:29:53SlasheriHCl: Not done yet ;)
23:30:16SlasheriHCl: I am trying to get seeking & flushing working
23:30:20 Quit gromit` (Remote closed the connection)
23:30:24preglowlooksaus: but yeah, don't hold your breath on midi, there's a lot of work to be done for that kind of file yet, and we're still not done with ordinary music playnack
23:30:30preglowit'll get done, though
23:32:13looksaushm, I'm just discovering rockbox and asking about things that are probably out of reach, but still...
23:32:16HCl ;p okay :)
23:32:20HClthats fine
23:32:25HCli was just wondering whether you had forgotten or not :)
23:32:57looksauswould it be possible to do something like gstreamer's speed plugin does
23:33:18SlasheriHCl: i will do it tomorrow if i have some time :)
23:33:21looksauslike, slowing some music down a little
23:33:22HClk :)
23:33:31HClno idea
23:33:44preglowlooksaus: 'course
23:33:46HCldo we have the power to resample to change pitch and stuff? *looks at preglow*
23:33:51 Quit Harpy (Read error: 145 (Connection timed out))
23:34:04looksausI saw pitch mentioned
23:34:05preglowlooksaus: depends on if you want pitch to be changed or not
23:34:32looksauspitch: no, speed: yes
23:34:43preglowthat's time stretching, it's not trivial, but can be done
23:34:45looksaussorry if I'm asking dumb questions
23:35:18preglowi might look at it once, since that's more my area of interest
23:36:58looksauswould be nice :)
23:37:48looksausenjoy hacking on rockbox... I'm off to sleep
23:38:03***Saving seen data "./dancer.seen"
23:38:12 Part looksaus ("Bezig met verlaten")
23:40:16HClheh o.o.
23:42:01preglowthe uda's headphone output isn't decoupled
23:42:06preglowsmall wonder we get pops and clicks, then
23:44:21hubbelpreglow: hm.. did I forget to set some UDA register to compensate for that?
23:45:09hubbelhum.. Speex project looks interesting.. wonder if it can encode in realtime
23:45:16 Quit tvelocity (Read error: 60 (Operation timed out))
23:46:21preglowhubbel: only thing i can think of the uda driver can do, is mute and unmute the digital part before any other parts are powered up/down
23:46:31preglowas described in 8.12 in the data sheet
23:46:47preglowright now every part of the uda is initalisied at once, i think
23:47:19 Join gromit` [0] (~gromit`
23:47:33hubbelpreglow: ok, going to try the 8.12 stuff later
23:50:40 Join courtc [0] (

Previous day | Next day