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 2010-03-14

00:00:55 Quit ender` (Quit: Who is this General Failure and why is he reading my hard drive?)
00:11:13bluebrotheris there a generic macro that indicates if the target is big endian or little?
00:11:19 Quit Llorean (Quit: Leaving.)
00:12:42*bluebrother might be able to use __i386__ and __powerpc__
00:12:43AlexPJdGordon_: Ta for the fix
00:13:54 Join perfectdrug_ [0] (
00:14:18saratogai'm too lazy to build the exact svn versions right now, but on PP the new mdct is something like 40% faster then the stock one
00:16:48 Join moos [0] (moos@rockbox/staff/moos)
00:17:27 Quit perfectdrug (Ping timeout: 265 seconds)
00:19:23 Quit Zagor (Quit: Clint excited)
00:24:13 Join tmzt_ [0] (
00:26:31***Saving seen data "./dancer.seen"
00:34:21CIA-5New commit by bluebrother (r25160): Make voicefont produce proper files on big endian machines. ...
00:37:03 Quit kugel (Remote host closed the connection)
00:38:05bluebrotherpixelma: my last commit hopefully fixes archos voicefiles on ppc. I'm uploading new rbutil binaries the next couple of minutes. Hope you can confirm it working the next time you get a chance to test :)
00:46:24 Join planetbeing_ [0] (
00:49:30 Quit planetbeing (Ping timeout: 240 seconds)
00:49:31 Nick planetbeing_ is now known as planetbeing (
00:50:06 Quit DataGhost (Ping timeout: 258 seconds)
00:50:08 Quit Casainho (Remote host closed the connection)
01:05:55 Quit CGL (Quit: Saliendo)
01:10:30 Quit Farthen (Ping timeout: 240 seconds)
01:10:35 Quit Schmogel (Quit: Miranda IM! Smaller, Faster, Easier.
01:11:28 Join robin0800 [0] (
01:18:57 Quit bertrik (Quit: De groeten)
01:31:51 Part froggyman
01:31:52arbingordonI kinda missed what the talk about USB audio was about - so in short, is it something like hooking in your PMP, and being able to push the all sound from the OS running on the computer out through USB to the PMP, and have rockbox play it in realtime?
01:32:35arbingordon(sorry if this is "offtopic")
01:33:12gevaertsthat's more or less it, yes, although more is probably possible
01:33:37saratogaemailed the tremor mailing list asking what xiph wants for us to upgrade their decoder
01:34:56arbingordonthat's great. would that mean that the computer would still use the internal SPU? it'd be neat to get rid of the low hum that my integrated sound card adds to everything I play.
01:35:32gevaertsthat would depend on the OS you use I guess
01:35:43*gevaerts doesn't know the details there
01:35:50saratogadepends what you mean by "SPU"
01:36:08saratogabut hum is an analog effect, which does not occur in digital interfaces like USB
01:43:07 Join Llorean [0] (~DarkkOne@rockbox/user/Llorean)
01:46:30 Join planetbeing_ [0] (
01:48:30 Quit planetbeing (Ping timeout: 240 seconds)
01:48:30 Nick planetbeing_ is now known as planetbeing (
01:49:35 Quit robin0800 (Remote host closed the connection)
01:59:19pamauryHum, either iso transfers are resulting in garbage, either the audio format is not the one I guess ...
02:06:49pamauryDoes there exist different variant of PCM ?
02:07:38arbingordonyou mean like ADPCM?
02:08:48pamauryI don't know, it's just that if I take the PCM data that usb audio is streaming and put it in a WAV file, then the output is weird
02:09:09pamaurySo it's not exactly PCM data, there is a trick somewhere
02:12:37 Quit karashata (Quit: The fluffy dragon has left completely!)
02:13:18saratogapamaury: theres dozens of combinations of plain PCM
02:13:49saratogalike 8 bit, 16 bit, 24 bit, big endian, little endian, stereo, mono, etc
02:13:52pamauryThen could someone help understand the one used by usb audio ? The spec says:
02:13:53pamauryThe PCM (Pulse Coded Modulation) format is the most commonly used audio format to represent audio data streams. The audio data is not compressed and uses a signed two’s-complement fixed point format. It is left-justified (the sign bit is the Msb) and data is padded with trailing zeros to fill the remaining unused bits of the subframe. The binary point is located to the right of the sign bit so that all values lie within the range [-1,+1).
02:14:38saratogadump it to a file and try playing with sox
02:14:50saratogashould only take a few guesses if you know its stereo with a given sampling rate
02:15:26pamaurywhat is sox ?
02:15:39saratogastandard tool for converting raw pcm to wav
02:15:44saratogaruns on windows/linux
02:15:52saratogai use it for debugging codecs
02:16:33 Join ArmandiuxGS [0] (
02:16:38pamauryok, thanks
02:17:21ArmandiuxGSoh, excuse me guys, I have a little problem with my iPod Nano 2G...
02:21:22ArmandiuxGSWhen I play music and I start the fft plugin (or another plugin, maybe), I see a "Noise Like" video distortion in the screen, and a Error Message appear: Prefetch abort ... at 080002FC ... FSR 0x3FF ... (domain 15, fault 15).
02:22:00saratogawhich plugins exactly?
02:22:01 Join planetbeing_ [0] (
02:23:40saratogai would file a bug report if theres not already one for that plugin
02:25:28 Join FlynDice [0] (~FlynDice@
02:25:33ArmandiuxGSI was playing a '.ogg' file with loudest volume
02:26:34***Saving seen data "./dancer.seen"
02:28:43ArmandiuxGSforget it, I'll try to install the current build (r25160)
02:29:08saratogagenerally if you're not running a current build we don't want to hear about problems :)
02:30:49pamaurysaratoga: on which things can I play to try to find the right format ? I know it's stereo (normally !) at 44100Hz (normally !)
02:31:08saratogaendianness and number of bits probably
02:32:03saratogacheck out the sox switches, but −−bits and −−encoding are likely to be worth looking at
02:32:12pamauryI also now it's 16-bit per sample (well that's what I specified)
02:32:30saratogasomething like sox −−rate 44100 -c 2 −−bits 16 -s in.raw out.wav
02:34:14saratogai think that should be right for 16 bit signed integer at 44100 Hz with stereo
02:34:44pamauryThat's what I tried and that's not great !
02:35:04saratogawhats it sound like?
02:40:39pamauryI can recognize the sound (partially) but it's noisy and saturated
02:41:27saratogaendianess might be wrong
02:41:30ArmandiuxGSWhy when I play music, some plugins (demos), which use much rendering, are constantly slow or fast?
02:41:38saratogaor it may actually be a padded 24 or 32 bit sample format
02:42:13saratogaArmandiuxGS: probably because the CPU clock on your nano keeps changing and those plugins haven't been fixed to work with it yet
02:42:38ArmandiuxGSOMG!, a Message Error: Undefined instruction at 09F8029C
02:42:40pamaurysaratoga: I requested 16-bit per sample so normally this is right but perhaps it's padded and it's less 16-bit in reality.
02:42:51linuxstbpamaury: So are you playing audio from a PC, and recording it in Rockbox?
02:43:24pamauryI'm playing with a PC and my usb code write raw data to a file because the output is not good, or the format is wrong
02:44:04linuxstbThat's what I mean. So can't you compare the dumped data with the original PCM data from the PC?
02:44:59saratogai suppose plotting a couple hundred samples would immediately make it clear if the format is wrong
02:45:02linuxstbOr create a test file with easy to spot values - e.g. 0x1234 for the left channel, 0x5678 for the right
02:45:32pamauryHum, I think I will try this
02:45:43pamauryEasy way to generate this or by hand (program) ?
02:46:00saratogamatlab makes it pretty easy if you've got that
02:46:06linuxstbI would write a little C program to create the raw PCM file, then sox to convert to wav.
02:46:16ArmandiuxGSWhere I can get help for this issue?
02:46:21saratogaor just play a mono file
02:46:47saratogaArmandiuxGS: if you want help you should document the problem and file a bug report
02:47:25ArmandiuxGSI dont know how I can make it.
02:47:27 Join planetbeing__ [0] (
02:47:39saratogatheres a bug link on the left side of the front page
02:48:01saratogai wonder if the xiph people will be receptive of my proposal to drop the stupid low accuracy mode in tremor
02:48:54 Quit planetbeing (Ping timeout: 252 seconds)
02:48:54 Nick planetbeing__ is now known as planetbeing (
02:48:54 Nick planetbeing is now known as 30BAACDCE (
02:48:54 Nick planetbeing_ is now known as planetbeing (
02:49:30 Join Lss [0] (
02:50:58ArmandiuxGSI do not know what kind of bug is.
02:51:23saratogai think you said it was a problem with plugins
02:51:54saratogaif not pick the one that looks closest to whatever it was that didn't work, someone will fix it if you're completely wrong
02:56:34ArmandiuxGSThank you saratoga
02:56:52ArmandiuxGSI will try to make a task.
02:57:03 Quit planetbeing (Ping timeout: 276 seconds)
02:57:57ArmandiuxGSWhat would be the appropriate title?
02:59:29ArmandiuxGS"Undefined instruction"?
03:00:09pamaurysaratoga: hum, with 0x1234 one channel and 0x5678 I get the same thing on output except that there are a few glitches here and there and that I can hear them...
03:01:29pamaurySo perhaps the problem is due to n
03:01:44pamauryunreliable iso code, I will invistage further tomorrow
03:03:31 Quit pamaury (Quit: Page closed)
03:03:52 Part ArmandiuxGS ("Saliendo")
03:09:37 Quit Barahir (Ping timeout: 258 seconds)
03:10:26 Join Barahir [0] (
03:10:49 Join planetbeing [0] (
03:29:34 Quit 30BAACDCE (Quit: 30BAACDCE)
03:35:13S_a_i_n_tJust for the record, Nano2g bootloader (r25139) is broken in some way...hangs on the "Apple" scrren indefinitely.
03:35:43S_a_i_n_tThough crypt_firmware works again. Yay! which gave me a false sense of security.
03:36:51S_a_i_n_tIts still good to be able to pass a broken bootloader through a working encryptor :P
03:39:49 Quit Adubb (Read error: Connection reset by peer)
03:39:59 Join Adubb [0] (
03:47:14 Join planetbeing_ [0] (
03:57:08 Quit planetbeing_ (Quit: planetbeing_)
04:06:41 Quit perfectdrug_ (Read error: Connection reset by peer)
04:16:30 Quit Barahir (Ping timeout: 318 seconds)
04:17:00 Quit Adubb (Read error: Connection reset by peer)
04:18:48 Join Adubb [0] (
04:19:46 Quit YPSY (Ping timeout: 316 seconds)
04:20:02 Join Ypsy [0] (
04:20:17 Nick Ypsy is now known as YPSY (
04:21:56 Quit Adubb (Read error: Connection reset by peer)
04:26:37***Saving seen data "./dancer.seen"
04:30:06 Join Adubb [0] (
04:34:20 Quit Adubb (Read error: Connection reset by peer)
04:37:23 Quit TheSeven (Disconnected by services)
04:37:37 Join The_Seven [0] (~theseven@rockbox/developer/TheSeven)
04:37:50 Nick The_Seven is now known as TheSeven (~theseven@rockbox/developer/TheSeven)
04:40:36 Join Adubb [0] (
04:45:23 Join planetbeing_ [0] (
05:22:12 Quit Llorean (Quit: Leaving.)
05:26:54 Join Zarggg [0] (
05:34:17 Quit Zarggg (Quit: Zarggg)
05:36:56 Join Zarggg [0] (~zarggg@2001:0:4137:9e74:0:fbe6:beb1:ba3d)
05:39:59 Quit saratoga (Quit: Page closed)
05:40:28 Quit anewuser ()
05:42:53 Quit Zarggg (Quit: Zarggg)
05:45:32 Join Zarggg [0] (
05:46:57 Quit planetbeing (Read error: Connection reset by peer)
05:46:58 Join planetbeing__ [0] (
05:46:58 Nick planetbeing__ is now known as planetbeing (
05:47:03 Join planetbeing__ [0] (
05:48:24 Quit JdGordon_ (Ping timeout: 265 seconds)
05:48:37 Quit JdGordon (Ping timeout: 268 seconds)
05:50:09 Quit planetbeing_ (Ping timeout: 264 seconds)
05:52:57 Join Rob2223 [0] (
05:53:15 Join froggyman [0] (
05:56:22 Quit Rob2222 (Ping timeout: 265 seconds)
06:02:49 Join Llorean [0] (
06:02:49 Quit Llorean (Changing host)
06:02:49 Join Llorean [0] (~DarkkOne@rockbox/user/Llorean)
06:10:33 Join Barahir [0] (
06:15:34 Quit GeekShadow (Ping timeout: 240 seconds)
06:25:28 Part froggyman
06:26:05 Join froggyman [0] (
06:26:39***Saving seen data "./dancer.seen"
06:26:59S_a_i_n_tAnyone around with an accurate understanding of the %Sx tag?
06:27:15S_a_i_n_t*(it's "translate X string)
06:38:52 Join JdGordon [0] (~jonno@
06:41:05CIA-5New commit by uchida (r25161): wave metadata parser reduces binsize.
06:41:32S_a_i_n_tI'm wondering if two strings can be translated in the same conditional, ie. would %?ia<%ia|%Sx|Artist|: |%Sx|Unknown> display "Artist: Unknown" (assuming there are no id3 tags) in the appropriate language if the strings Artist and Unknown are both present?
06:42:39S_a_i_n_terrr... %?ia<%ia|%Sx|Artist|: |%Sx|Unknown|> rather.
06:45:55 Join karashata [0] (
06:47:01 Join planetbeing_ [0] (
06:47:01 Quit planetbeing (Read error: Connection reset by peer)
06:47:01 Nick planetbeing_ is now known as planetbeing (
06:47:07 Join planetbeing_ [0] (
06:51:40S_a_i_n_tOr would it be: %?ia<%ia|%Sx|Artist|: %Sx|Unknown|> ? I *think* the syntax is correct, assuming both Album and Unknown are translated strings...but I'd really like to talk to someone who has used the tag before, or preferably the author of it ( JdGordon? ).
06:51:40S_a_i_n_tI've yet to see an example that translates two strings at I wonder if it's even possible.
06:51:40 Quit planetbeing__ (Ping timeout: 264 seconds)
06:51:40JdGordonWhy wouldn't that work?
06:51:40S_a_i_n_tI'm not I guess that means "it *should* work?". I've only ever seen one string translated per I thought there may have been a reason for that.
06:53:24JdGordonSx is an ordinary tag.
06:54:33S_a_i_n_tsweet, it's just that the added |'s make the syntax look really crazy.
07:02:30 Join JdGordon1 [0] (
07:03:09 Quit karashata (Quit: The fluffy dragon has left completely!)
07:05:12 Join karashata [0] (
07:15:48 Join Horschti [0] (~Horscht2@xbmc/user/horscht)
07:17:19 Quit Horscht (Ping timeout: 265 seconds)
07:20:18 Join Farthen [0] (
07:47:27 Join planetbeing__ [0] (
07:50:39 Quit planetbeing (Ping timeout: 260 seconds)
07:50:39 Nick planetbeing__ is now known as planetbeing (
07:51:35 Join HatchlingByHeart [0] (~Hatchling@
07:53:35HatchlingByHeartcan anyone tell me if there's any plans for supporting the 6th gen iPod?
07:53:47HatchlingByHeartI couldn't find anything in the FAQs
07:53:51JdGordon1we dont make plans for future targets
07:53:55S_a_i_n_tplans...yes, probaly.
07:54:01JdGordon1if it happens, it happens
07:54:02S_a_i_n_twill it happen any time soon?
07:54:35HatchlingByHeartdamn. I just bought a 6th gen 160GB iPod and I like the look of Rockbox
07:54:52HatchlingByHeartthe utility told me mine is unsupported T_T
07:55:28S_a_i_n_tthe main page would have told you that also.
07:56:06HatchlingByHeartoh... 5.5g - 1g... how'd I miss that? o.o
07:57:01HatchlingByHeartwell, thanks for your help guys
07:57:11 Part HatchlingByHeart
07:57:24 Quit karashata (Quit: The fluffy dragon has left completely!)
07:58:12 Join CGL [0] (~CGL@
08:10:58 Quit arbingordon (Quit: `)
08:13:55 Join arbingordon [0] (
08:15:33 Join r0b- [0] (
08:15:44r0b-are the e200v2's now FULLY supported?
08:15:59 Quit froggyman (Quit: froggyman)
08:17:12 Quit JdGordon (Quit: Bye)
08:19:10 Quit panni_ (Read error: Connection reset by peer)
08:26:40***Saving seen data "./dancer.seen"
08:45:27 Quit JdGordon1 (Read error: Connection reset by peer)
08:45:57S_a_i_n_tr0b-: the main page lists them as stable...if that's what you mean.
08:46:56*S_a_i_n_t has forgotten how to build the fontpack again... make "insertwhatnow?"
08:47:37 Quit planetbeing (Quit: planetbeing)
08:47:37 Nick planetbeing_ is now known as planetbeing (
09:00:29 Join stoffel [0] (
09:07:59 Quit CGL (Quit: Saliendo)
09:08:04 Join JdGordon [0] (~jonno@rockbox/developer/JdGordon)
09:33:01 Join Highlander [0] (
09:41:39 Join JdGordon_ [0] (
09:41:45 Quit JdGordon_ (Changing host)
09:41:45 Join JdGordon_ [0] (~user@rockbox/developer/JdGordon)
09:50:30 Quit CaptainKewl (Remote host closed the connection)
09:55:19 Quit JdGordon_ (Quit: JdGordon_)
09:57:38 Quit kaniini (Ping timeout: 245 seconds)
10:03:00 Join flyasky [0] (
10:04:02flyaskyI have VoiceRecorder Samsung YV-150 - will rockbox work on it?
10:05:37S_a_i_n_tflyasky: Is it on the Roclbox main page?
10:06:44flyaskyno (
10:07:31S_a_i_n_tright, sorry that was a bit blunt. But it answers your question.
10:08:04S_a_i_n_tThere's a "New Ports" page on the Wiki however used to get a new project started.
10:08:20S_a_i_n_tOr to try to gain interest in starting a new port rather.
10:08:25 Join ender` [0] (
10:08:41flyaskyare they compatible in core? I want try to port to YV-150
10:11:20pixelmayou need to find out a lot about your device (hardware components, how to upload your own firmware etc.) if you want to port Rockbox and it's impossible without all this info to say how easy a port would be
10:11:36S_a_i_n_tI know *nothing* about the target in question...but I can offer you this:
10:12:37pixelmait's always a lot of work, but a few ports were easier because code to use certain components was already around
10:18:28 Quit r0b- (Ping timeout: 245 seconds)
10:19:26 Join r0b- [0] (
10:19:45 Quit moos (Quit: ChatZilla 0.9.86 [Firefox 3.6/20100115144158])
10:22:32 Quit JdGordon (Read error: Connection timed out)
10:23:18 Join JdGordon [0] (~jonno@rockbox/developer/JdGordon)
10:26:16 Join kaniini [0] (
10:26:42***Saving seen data "./dancer.seen"
10:32:49 Join bmbl [0] (~Miranda@unaffiliated/bmbl)
10:33:01 Quit tmzt_ (Read error: Connection reset by peer)
10:33:05 Join tmzt_ [0] (
10:35:33 Quit S_a_i_n_t ()
10:35:50 Join S_a_i_n_t [0] (S_a_i_n_t@
10:49:49 Join yuanxu [0] (yuanxu@
10:52:14 Quit S_a_i_n_t ()
10:52:36 Join S_a_i_n_t [0] (S_a_i_n_t@
10:52:48 Quit yuanxu ()
11:05:40 Join Casainho [0] (
11:12:41 Quit S_a_i_n_t ()
11:14:02 Join JdGordon_ [0] (
11:14:02 Quit JdGordon_ (Changing host)
11:14:02 Join JdGordon_ [0] (~user@rockbox/developer/JdGordon)
11:14:34JdGordonpixelma: with the fm patch, does the buttons in the wps all work fine?
11:15:58 Quit flyasky (Ping timeout: 245 seconds)
11:16:23 Join bertrik [0] (
11:16:51 Quit planetbeing (Quit: Poof.)
11:19:23 Join JdGordon1 [0] (
11:20:14pixelmawell, the buttons work fine. Everything still hangs though when I want to exit the radio screen, with yesterday's patch (which I think is the "latest" :\ ). AlexP was able to reproduce a hanging sim when entering the radio screen with a simple test.fms (containing %Tn or %Tf), set through the menu
11:20:54pixelmamy test was on my M5
11:22:35 Quit JdGordon (Ping timeout: 265 seconds)
11:25:54 Quit JdGordon_ (Read error: Connection reset by peer)
11:29:58 Join Lear [0] (chatzilla@rockbox/developer/lear)
11:33:08 Join Barahir_ [0] (
11:36:30 Quit Barahir (Ping timeout: 258 seconds)
11:36:44JdGordon1pixelma: was it you that asked why cabbiv2.fms was being loaded?
11:38:19pixelmano, AlexP was
11:38:43JdGordon1or whoever asked... its because thats the default one
11:40:11 Join stooo [0] (
11:41:06 Part stooo
11:47:42JdGordon1haha, I found the hang AlexP hit
11:49:28 Quit JdGordon1 (Read error: Connection timed out)
11:50:01 Join stooo [0] (
11:50:23 Join JdGordon [0] (~jonno@rockbox/developer/JdGordon)
11:53:09 Join flyasky [0] (
11:54:25pixelmaI'm running out of ideas what to test for and which could help finding out what happens, but I get the effect on all my three targets
11:56:17flyaskythank you! I'll read first - when ask again )
12:02:33 Quit stooo (Quit: Leaving.)
12:04:36 Quit bzed (Remote host closed the connection)
12:04:53 Join bzed [0] (
12:06:39 Quit chaos (Ping timeout: 248 seconds)
12:06:39 Quit ps-auxw (Ping timeout: 248 seconds)
12:07:44 Join chaos [0] (~chaos@gentoo/user/ch4os)
12:09:02 Join ps-auxw [0] (~arneb@2001:470:c807:0:1532:4e5f:2ad3:4123)
12:12:08 Join S_a_i_n_t [0] (S_a_i_n_t@
12:23:10pixelmaJdGordon: looks like it has something to do with th %Tn tag alone in the test fms (not using the "latest" patch though). I now put it in a %?tt<%Tn|> and it doesn't hang anymore
12:24:41pixelmabut %Tn shows up very seldom, even though it sounds like the station is tuned (I'm currently in preset mode and the fmr is correct)
12:25:26pixelmastill testing with the M5
12:25:51AlexPpixelma: I thought about that, but also got the m5 sim hang with %Tf alone, which really should be possible - people might just want a big frequency and nothing else
12:26:33pixelmaAlexP: JdGordon put a new patch u which supposedly fixes that
12:26:41AlexPAh, OK
12:26:42pixelmaup too
12:26:46***Saving seen data "./dancer.seen"
12:28:15pixelmaTf also has to do with a preset, tf is the current frequency, or did that change?
12:29:24pixelmaI think you can use it for a "next/previous station list" IIUC
12:29:36pixelmaTf that is
12:30:05AlexPAh right
12:30:15AlexPI can't rememeber off the top of my head
12:30:40 Quit JdGordon (Read error: Connection timed out)
12:32:05CIA-5New commit by jdgordon (r25162): fiddle with the skin debug output so the load lines arnt shown unless debugwps is enabled
12:32:17 Join JdGordon [0] (~jonno@rockbox/developer/JdGordon)
12:33:31 Join CFP [0] (
12:34:23 Quit CFP (Client Quit)
12:35:10pixelmaI'm sure the station name shows up more often with SVN (and goes away when not tuned, e.g. in scan mode)
12:36:21pixelmaI'll need an SVN build for comparison
12:39:05pixelmaand I'd really appreciate version numbers for the radio skin patches, more accuracy when you talk about it and easier to keep old versions around
12:39:24AlexPThe station name shows up when it should in the patch for H100/S
12:39:35AlexPactually, make that I think
12:39:44pixelmainside an %?tt ?
12:39:47AlexPIn fact, forget that, I'd need to check :)
12:41:56pixelmawell, it shows up fine without the %tt bit but then I get the freezed
12:42:07pixelmaor freezes
12:42:18 Quit ps-auxw (Quit: leaving)
12:42:23AlexPSo what is the remaining issue? A fms with %Tn \n %tf seems to work fine in the m5 sim
12:43:02 Join ps-auxw [0] (~arneb@2001:470:c807:0:1532:4e5f:2ad3:4123)
12:44:04JdGordonit needs a bigger buffer first, so probably the resigin patch, and the hwcodec issues
12:44:11pixelmaeven if it works now, my gripe would still be that it can't show the recording info on hwcodec targets which record radio from the radio screen as SVN does
12:44:26JdGordonit has the tags to do it
12:44:45JdGordonits even setup forr the hard coded to do it
12:44:49AlexPJdGordon: What is the status of the resizing patch?
12:45:05JdGordonkugel said he is having some wierdness with it
12:45:09pixelmaJdGordon: no, you can't show the peakmeter conditionally
12:45:59JdGordonI'd be happy to just increase the buffer size to get this in untill the resize patch is finished
12:46:27pixelmaand I'm not sure - does the prerecording time count up as the buffer fills? I imagine with the patch it would just show the set value (I could be wrong though)
12:46:27JdGordonpixelma: that is a seperate issue
12:46:50JdGordonit should display everythign exactly as svn apaprt from the bars
12:46:53pixelmanot, if you say "it's the same as SVN"
12:47:08pixelmaminus the comma
12:47:47JdGordonmaking the peak meters work conditionally is not a small change
12:48:17 Join liar [0] (
12:48:17JdGordonit would pretty much involve rewriting the pm handling, which isnt necessarily a bad thing
12:48:46AlexPJdGordon: I'd be fine with resizing the buffer a bit for now - on greyscale it is currently tiny
12:50:54pixelmayou'll not see the prerecording time info with the hardcoded fms now anyway due to the pm issue - it does not look the same as SVN currently anyways (and I remember finding one difference annoying, but have to look at the hardcoded one once more)
12:51:04 Join JdGordon1 [0] (
12:51:15 Quit JdGordon (Read error: No route to host)
12:52:54JdGordon1anyway, I think that should happen, and it sholdnt be required for fm skin once the rest is working fine
12:55:00soapscorche, is there a plugin or module for SMF which prevents user editing of forum posts after it as been replied to?
12:55:35soapIt's only an occasional problem on the forums, but introduced said feature a few months ago and it has been nothing but good there.
12:56:39JdGordon1there are legit reasons to edit a replied to post thoguh
12:57:01JdGordon1like "this whole post was wrong, removing to not make more people stupid"
12:57:02 Join pamaury [0] (~c2c7a50a@rockbox/developer/pamaury)
12:57:09S_a_i_n_tI agree, but not really after its been replied to.
12:57:18S_a_i_n_tcouldn't you just remove it in that case?
12:57:20linuxstbHow do you know a post was replied to? All you know is that there was another post in the same thread.
12:57:22AlexPI often edit my own after it has been replied to to correct spelling :)
12:57:37S_a_i_n_tAlexP: Spellcheck :P
12:57:38pixelmaJdGordon1: like editing the original bug report after a part has been fixed? :\
12:57:50AlexPS_a_i_n_t: Doesn't get everything
12:57:56pixelmawhen it actually wasnt
12:58:09soapRemoval of replied-to posts seems like a perfect action for the staff to take care of. Retractment of statements seems to be a bad idea unless you're trashing the whole thread.
12:58:25JdGordon1if your being subtle about a not fixed bug then either out with it or dont bother... because i have no idea what your talking about
12:58:55AlexPI don't actually mind either way, but it isn't really a problem on our problems at the moment I don't think
12:58:57soapbug reports go in the tracker, regardless, so that point is moot.
12:59:58soapAlexP, I agree it is not a large problem by any sense of the imagination - but the "fix" appeared to have no negative side effects on HA, so why not do it if it is just as seamless on SMF?
12:59:59pixelmaJdGordon1: no, it's fixed now and some time ago. But I remember you editing the original report and removing unfixed things
13:00:24pixelmaI could never understand why you edited the original post
13:00:26AlexPsoap: Sure - I'm not objecting at all - I just can't be bothered to investigate it myself :)
13:00:55AlexPAs long as it only affects the users and I can still edit mine afterwards :)
13:10:01 Join dfkt [0] (dfkt@unaffiliated/dfkt)
13:22:19JdGordon1pixelma: have you tried putting %pm inside a conditional viewport?
13:23:48pixelmathat would probably work but then you get font size "dependencies"
13:25:51pixelmamaybe the most flexible you could do there is putting these viewports at the bottom and work with negative positions (never tried yet)
13:27:15JdGordon1I have no idea how to put %pb inside a conditional in a way that makes sense
13:28:11JdGordon1from the code POV... i can see plenty of reasons why being able to do that would be cool
13:30:41TheSevenliar: did you have a chance to test my latest commit?
13:31:27pixelmaJdGordon1: I thought we were talking about the peakmeters (%pm), but IIRC %pb needs to be on its own line too. But then - how does the current radio screen do it?
13:32:11JdGordon1yes, sorry, pm
13:33:50liarTheSeven: r25162 works fine
13:34:12liaroh sorry
13:34:13liarit doesnt
13:34:29liarTheSeven: FTL: scheduling bank ...
13:35:34TheSevenand the version before (r25156) is fine?
13:36:29liariirc r25154 is the last good one, ill try r25156
13:41:54liarTheSeven: r25156 works fine
13:43:28CIA-5New commit by theseven (r25163): More detailed panic message for Nano2G VFL fast read problems
13:48:18 Quit Adubb (Read error: Connection reset by peer)
13:48:30 Join Adubb [0] (
13:50:05 Join Buschel [0] (
13:50:18BuschelLear: you there?
13:50:36Buschelperfect :o)
13:50:57liarTheSeven: VFL fast read failed, RC 0000, bank 1, RET 100, base 903180
13:51:22BuschelSaw your bug report reagrding mpc resume. The nug eas not introduced with the sv8 library...
13:51:44BuschelAnd it seems aac and alac use it the same way...
13:52:04LearThey do?
13:52:31LearI'm pretty sure AAC doesn't...
13:52:55liarTheSeven: VFL fast read failed, RC 0000, bank 2, RET 100, base 1600520 for a different trak
13:53:16BuschelSorry, it saves the position via set_offset(elapsed_time)
13:53:25LearHm, elapsed time for set_offset looks wrong...
13:53:25Buschelwhich is not correct either
13:53:51LearIt seemed to work fine when I implemented the raw seek though.
13:54:21 Join m3dlg [0] (
13:55:07LearAnd the use of "sound_samples_done" in the initial seek is just a question of re-using another variable. Probably not worth the confusion it can cause. :)
13:56:13BuschelLear: can you review / test this mpc patch? ->
13:56:18Buschelworks for me
14:01:04LearLooks reasonable... But file offset is known, so that would be better to save, wouldn't it?
14:07:11S_a_i_n_tJdGordon1: Is there a way the "playlist viewer" can have the offset set as "current track +1"? or *could* there be?
14:07:32S_a_i_n_tI figure then you could have "next AA" displayed
14:09:03JdGordon1the offset is the offset
14:09:10JdGordon11 will always be the next track
14:09:18JdGordon12 is always the track after next
14:09:29JdGordon1you cant display the next AA currently
14:09:51S_a_i_n_tah..shit..there's that plan.
14:10:01BuschelLear: just saving ci->curpos?
14:10:03S_a_i_n_tOh well :D
14:11:40S_a_i_n_tNext AA is pretty big on my list, as well as a few other peoples I imagine. You ticked off number 1 on that list for me (multifont), so I guess I should be more patient :P
14:12:48*JdGordon1 is good at ticking people off :p
14:13:09S_a_i_n_tNumber 2 would probably be userfont splashes...or bitmap scrollbars.
14:13:38 Quit m3dlg (Ping timeout: 268 seconds)
14:13:43*S_a_i_n_t can't decide between those two...err, whichever's easier to implement :P
14:13:49 Quit Tomis (Quit: Tomis)
14:14:41LearBuschel: I don't think you really need to save that. E.g., the mp3 codec doesn't do that, since it doesn't do any internal buffering (that I know of).
14:14:59LearVorbis probably does, so it calls raw_tell.
14:15:26BuschelLear: If I remove the call of set_offset it does not resume at all...
14:15:51pamaurygevaerts: I will soon create FS task for my usbaudio work. Do you think I should also put the test code I use for iso transfers or should I only put the usb audio one ? Or both put in different patches (I don't like this option because I'll do the manual splitting). The test drivers allows to send lots of OUT transfers and check for CRC, do some debug and statistics (there is rockbox code+linux code that uses libusb)
14:17:03gevaertspamaury: test code is useful I guess
14:18:43 Join Tomis [0] (~Tomis@
14:18:53BuschelLear: seems like this is because mpc does not use/call ci->advance_buffer as all this low level stuff is done within the decoder itself...
14:19:08pamaurywhat upset me is that I can send 256K/s of OUT transfers without any errors with my test code (I do manual CRC check even though the hardware should do it also) but yesterday I record the usb audio output and it was good except there were some samples zeroed here and there and they sound weird. I don't know where is the problem
14:19:26BuschelLear: calling ci->set_offset(ci->curpos) seems just right
14:19:42 Quit jd (Ping timeout: 265 seconds)
14:19:53LearBuschel: Ah, yes, that's why.
14:20:41gevaertspamaury: did you test the test code while playing audio (the usual way)?
14:21:23BuschelLear: so, for aac ci->set_offset(elapsed_time) should just be removed. aac uses ->advance_buffer()
14:23:23pamaurygevaerts: I can't, my test code is a special usb interface with special requests do to some iso transfers. But this afternoon, I'll do some more test witht the usb audio code, using usbmon to see if there is a usb transfer error or if I'm messing up things somewhere
14:23:24LearSeems like it, yes. Still seems to work as it is... :)
14:24:09gevaertspamaury: I mean, try just playing some audio files in rockbox while using the test interface
14:26:33pamaurygevaerts: ah ok. but the problem I encountered yesterday (well today at 3am) is that if I play a file with left channel=0x1234 (say) and right=0x5678, if the usb audio record it's output to a file. Then the file contains the expected audio but with some errors. So the problem is not (only) playing audio.
14:26:47***Saving seen data "./dancer.seen"
14:28:59 Quit anti[v4] (Remote host closed the connection)
14:32:18CIA-5New commit by Buschel (r25164): Fix FS #11103. Resuming musepack files was handled wrong since ages. This change converts the decoders exact sample position to an estimated byte ...
14:35:36 Join GeekShadow [0] (~Antoine@reactos/tester/GeekShadow)
14:37:43CIA-5New commit by learman (r25165): Make resume handling in the AAC codec less confusing. No functional change.
14:39:26 Join tvelocity [0] (
14:39:50BuschelLear: Saving ci->curpos keeps the exact file position. But in combination with mpc's variable bitrate the calculated sample index is just an estimation. When saving a (virtual) file position that is calculated by the reversed formula of the resuming formula, the calculated sample index will be precise.
14:41:49LearBuschel: But if you check set_offset, you'll see that it subtracts the pcmbuf latency. So it won't be an exact reverse, I think...
14:42:10 Quit JdGordon1 (Ping timeout: 265 seconds)
14:42:17BuschelLear: yes. But that's the same for all the other codecs as well.
14:43:50BuschelLear: what I wanted to state -> the committed solution is more precise for variable bitrate files
14:47:55 Join robin0800 [0] (
14:56:27pixelmaJdGordon (for the logs) - your latest patch also fixes the freezes I get with a simple %Tn fms (even though presets were loaded and displayed before it hung on exit)
14:56:35 Quit S_a_i_n_t (Ping timeout: 240 seconds)
14:56:36 Join S_a_i_n_t_ [0] (S_a_i_n_t@
14:56:54pixelmamaybe they were invalidated on exit?
15:01:30 Quit S_a_i_n_t_ (Ping timeout: 264 seconds)
15:07:01 Join flydutch [0] (
15:11:29 Quit Casainho (Ping timeout: 240 seconds)
15:22:51 Join Kitr88 [0] (
15:26:08 Quit Kitar|st (Ping timeout: 264 seconds)
15:26:54 Quit robin0800 (Quit: No Ping reply in 180 seconds.)
15:27:11 Quit Kitr88 (Ping timeout: 245 seconds)
15:27:19 Join robin0800 [0] (
15:28:08Buschelhmm, why ci->id3->offset has strange results for alac format when entering the codec? e.g. the correct byte position seems to be saved (e.g. 1.500.000), but when re-entering the codec it is 2.020
15:29:16 Quit kaniini (Ping timeout: 245 seconds)
15:29:45 Join S_a_i_n_t [0] (S_a_i_n_t@
15:32:03linuxstbBuschel: What do you mean by "re-entering the codec" ?
15:32:19S_a_i_n_tTheSeven: you there?
15:33:00 Join Kitar|st [0] (Kitr88@
15:33:50Buschellinuxstb: entering codec_main of atrac.c
15:34:37linuxstbAre you talking about alac or atrac?
15:34:45*Buschel needs to slow down a bit...
15:34:48S_a_i_n_tTheSeven: your latest commits seem to make CygWin compile "dirty" (not sure how else to put it...)
15:35:03Buschelsorry, my mistake. of course I am talking of alac.c
15:35:08S_a_i_n_tIt seems to compile...but isn't happy about something.
15:38:15 Join Casainho [0] (
15:41:15LearBuschel: ALAC doesn't support resuming at all, at the moment, it seems. Copying the aac.c code ought to work though.
15:42:34BuschelLear: I know, I wanted to implemement it right now. But id3->offset's value seems odd (printf debugging in pcsim).
15:43:07 Join findus [0] (
15:45:02findushi folks, just had a quick question about future developement....(please don't hate me)
15:45:20LearBuschel: Added a case for it in audio_finish_load_track (playback.c). Shouldn't make any difference, but my overlap patch expects it at the moment. :)
15:45:38findusI just purchased a sansa clip plus player, is there a rockbox port in the works now, or coming in the foreseeable future. Thanks for any replies
15:46:30robin0800findus: is it on the front page?
15:47:07findusyes, but it does not mention if something is in the work :)
15:48:01TheSevenS_a_i_n_t: that's just some annoying warning triggered by some code i added to help debugging. this will be gone again soon.
15:48:58 Quit Highlander (Quit: Quitte)
15:50:47S_a_i_n_tAh...right, I was just testing out the build. So, it doesn't actually *mean* anything, errrr...not to me at least?
15:52:07 Join kaniini [0] (
15:55:14AlexPfindus: The clip v1 works more or less
15:55:26AlexPthe clip plus + v2 may or may not
15:55:34 Join DerPapst [0] (
15:55:35AlexPDepends on when someone does the work
15:55:42AlexPThere are never any plans as such
15:56:20findusthank you AlexP and robin0800
15:58:01AlexPI think some people are looking at the clip+, but there are never any ETAs. Spare time volunteer project, and you never know whether something is going to be done/possible until it is
15:59:04liarTheSeven: have you seen my messages above? (vfl fast read panic)
15:59:10FlynDicefindus: take a look here:
16:01:29 Join kugel [0] (~kugel@rockbox/developer/kugel)
16:02:07TheSevenliar: yes, but that only confused my even more ;-)
16:02:41S_a_i_n_tliar: It *Panic*'s when you...?
16:02:48TheSeventhere is a fast read failure somewhere, so it resets the bank and tries a slow read, which also fails, because of a spare ecc error...
16:02:52S_a_i_n_tI can't seem to get that build to fail.
16:02:54TheSeventhis *might* actually be a failing page
16:03:01findusYeah, I know the deal with opensoftware and appreciate that it's all based on volunteer work."Release date? - Thursday"
16:03:03TheSeven(because it was not properly written?)
16:03:18TheSevenS_a_i_n_t: which flash type do you have?
16:03:24TheSevensee debug => display hw info
16:03:25liarTheSeven: but it works with the previous revision?
16:03:51TheSeventhe previous revision wasn't using fast reads at all, so it actually tried slow reads twice
16:04:02TheSevenmaybe that was a little more fault-tolerant
16:04:23liardiskmode does not remap i either
16:04:40findusthanks FlynDice, bookmarked and will check back on that one later!
16:06:11S_a_i_n_t^ TheSeven
16:06:56 Join CGL [0] (~CGL@
16:07:13BuschelLear: I do not understand what is happending here :/ will do a rebuild now
16:07:17S_a_i_n_tAlso...(sorry, I know it'll probably be the *least* of your worries), the Nano2g SVN bootloader is hanging on the "Apple"
16:07:58S_a_i_n_t...but at least crypt_firmware works again :D
16:08:46TheSevenoh? who fixed it?
16:08:50TheSevenit fixed itself by magic?
16:09:16S_a_i_n_tFunny thing, I used to have that HW info patch for the 2g nano in my "essential patches" folder, glad you committed it :D
16:09:32S_a_i_n_tmagic I guess...if you didn't touch it.
16:10:08S_a_i_n_tas I had the stack at 4000 myself anyway, and it failed that can't of been it.
16:10:17S_a_i_n_t(which was my first guess)
16:12:10CIA-5New commit by theseven (r25166): Nano 2G VFL: try slow read twice if fast read failed
16:12:33TheSevenS_a_i_n_t: aha, so the cached flash type also supports interleaved reads, that's nice to know
16:12:54liarTheSeven: i just wanted to suggest that :-) (slow read 2 times)
16:13:23S_a_i_n_tyou're ( TheSeven ) the only one that touches the nano2g code I suspect you fixed crypt_firmware by accident.
16:13:28*bluebrother found a problem with rbspeex.c
16:13:37S_a_i_n_tI didn't see a commit from anyone else regarding it.
16:13:58bluebrothernow lets see if that is the cause for rbspeex not working correctly on ppc
16:14:13TheSevenliar: i doubt that this fixes your problems though
16:16:46 Join jd [0] (
16:16:47 Quit jd (Changing host)
16:16:47 Join jd [0] (~jd@Wikipedia/HellDragon)
16:18:34S_a_i_n_tthe "screendump causes *panic*" issue with nano2g is interesting...the first time I tried, it worked fine, but every time since I've had a "iTunes white screen of death" fom it.
16:18:54S_a_i_n_t*from also.
16:19:29S_a_i_n_tI'm too scared to try it a fifth time.
16:19:55S_a_i_n_ts/scared/can't be assed restoring the thing/
16:21:42liarTheSeven: yep that fails too
16:24:23*S_a_i_n_t congratulates Apple for getting to OFW to run on each of the different NAND's properly, but suspects if TheSeven had their doccumentation things would be different for RB
16:25:42 Quit r0b- (Ping timeout: 256 seconds)
16:26:05 Join r0b- [0] (
16:26:51***Saving seen data "./dancer.seen"
16:27:01CIA-5New commit by theseven (r25167): Nano2G VFL: reset the bank again before the second slow read try
16:28:03TheSevenliar: I don't think this will help either, but it is resembling the old behavior even closer now when the fast try fails
16:29:19TheSevenyour chip was an xx55xxEC type? this is interesting, as mine is also one of those... (samsung interleaved)
16:31:19 Quit robin0800 (Quit: No Ping reply in 180 seconds.)
16:31:44 Join robin0800 [0] (
16:31:50pixelmabluebrother: sounds nice and thanks for looking into this. I'll try as soon as possible (tomorrow evening, I guess) :)
16:32:27BuschelLear: found it. Calling qtmovie_read changes ci->id3->offset :/
16:34:55LearYep, that's why aac.c copies the value... :)
16:35:51BuschelAn application note would have been helpful... Well, found at least
16:36:31 Quit Lear (Quit: ChatZilla 0.9.86 [Firefox 3.6/20100115144158])
16:36:49liarTheSeven: 2555D3EC yes
16:37:53TheSevenliar: does the current build seem to work?
16:39:15liarTheSeven: different panic message
16:39:30liar(RC 0080 instead of RC 0000)
16:39:31TheSevenoh? which one?
16:39:38TheSevenoh damn
16:39:44TheSevenso 2 banks failed at the same time now
16:40:22TheSevendamn, that's almost the same chip that i'm using!
16:40:39TheSeveni just read back 3.2 GB of data using that build without a single problem
16:41:28 Join adnyxo [0] (
16:41:42CIA-5New commit by theseven (r25168): Don't parallelize Nano2G NAND reads, just do pipelined ECC. (10% performance loss :-/ )
16:41:54TheSevenliar: ok, now does this one fix it?
16:42:17TheSeventhis should behave *exactly* the same, besides maybe different timings and an additional retry, from the nand point of view
16:42:34TheSevenit's just doing the ECC operations after starting the next page read, not before
16:43:50 Join korlen [0] (~4436d135@gateway/web/freenode/x-iumuilofryvknqub)
16:44:06korlenis anyone here?
16:44:14TheSeventhere are 129 people here.
16:44:46korlenwell im wonder if anyone can answer my question
16:45:12korlen can i use rockbox on sansa fuze v02 yet?
16:45:41TheSevenit it listed on the front page?
16:46:19korlenit says sansa fuze only v01, im just wonder.. idk how offen rockbox updates their website
16:46:22*bluebrother points to the front page
16:46:53*findus can answer!
16:46:59bluebrotherkorlen: to figure that I'd suggest checking if that "Subversion" stuff is recent ...
16:47:09 Join dfkt_ [0] (dfkt@unaffiliated/dfkt)
16:48:00bluebrotheroh, and check the _last_ line of the front page :)
16:49:51liarTheSeven: *panic*
16:50:43 Quit dfkt (Ping timeout: 265 seconds)
16:52:50TheSevenliar: same oneß
16:53:02*TheSeven is confused
16:53:07liarTheSeven: same one
16:53:14liar(with RC 0080
16:53:37korlenoh ok thanks
16:55:14TheSevenliar: sorry, I'm out of ideas.
16:55:41 Quit korlen (Quit: Page closed)
16:56:58 Quit adnyxo (Quit: Ex-Chat)
17:05:14CIA-5New commit by Buschel (r25169): Implement resume for alac codec.
17:05:38 Join Darkknight512 [0] (
17:06:56CIA-5New commit by theseven (r25170): Revert Nano2G fast VFL reads for now. They just won't work on some hardware.
17:10:54CIA-5New commit by theseven (r25171): Nano2G NAND: Detect if the chip supports interleaved and/or cached writes.
17:12:31CIA-5New commit by theseven (r25172): Nano2G VFL/FTL: Use fast writes
17:13:17CIA-5New commit by theseven (r25173): Nano2G FTL: Enlarge FTL buffers for faster write operations
17:13:48TheSevenliar: can you check the latest one again?
17:14:14TheSeventhis time, there should be trouble when writing, if anything.
17:14:34Darkknight512I can test if you need
17:14:44liarTheSeven: i dont want to reformat my partition again :-(
17:15:13TheSevenDarkknight512: which chip type do you have?
17:15:17Darkknight512i havent even put my music back on since yesterdays iloader prob, so reformatting wont be a prob
17:15:24Darkknight512umm, can i check from within RB?
17:15:32TheSevenyes, system => debug => view hw info
17:15:42BuschelAny thoughts about FS #10016 (reading id3v1, if id3v2 contains few data)?
17:16:59Darkknight512NAND: Banks4 ID: 2555D3EC
17:17:37TheSevenok, so the same chip type as liar
17:17:53TheSevenyou could also do the test then
17:18:13 Join saratoga [0] (~463f90ed@gateway/web/freenode/x-vavuyirumqlajxuv)
17:18:40Darkknight512alright what u need?
17:18:50kugelBuschel: I'm not a fan
17:20:09liarTheSeven: remember, it did not happen with every track being played, so it might not happen with every write access
17:20:16TheSevenDarkknight512: just download r25173 when it finishes building and try copying some data to/from your ipod, and check if it works
17:20:30Buschelkugel: can you please explain?
17:21:13TheSevenliar: the writes did fail very consistently on some chips, but i know the cause of that, and circumvented it. this fix seemed to work the last time i committed it, so i'm confident that it will also work now. i just want to make sure.
17:22:18pixelmadoesn't reading ID3v1 need an additional seek to the end of the file (which possibly won't help if the ID3v1 doesn't contain more data than the v2.x tag)
17:23:27pixelmaI'd rather encourage people to tag their files correctly (and consistently)
17:23:31liarah so thats not the same bug/problem?
17:23:38Buschelyes, id3v1 needs an additional seek. But if the id3v2 tag does not contain artis/title/album I am of the opinion that such a seek is valuable.
17:23:58 Join dfkt [0] (dfkt@unaffiliated/dfkt)
17:24:21Buschelpixelma: The "learning factor" is best reason so far :o)
17:24:27liar*as the vfl fast read problems?
17:24:35Darkknight512TheSeve: You need me to test writes correct?
17:24:48pixelmaBuschel: such a seek could be useless (as I said, you don't know whether you'll get more data out of the ID3v1 beforehand)
17:24:54TheSevenliar: no, your chips supports interleaved writes, while some others don't, which caused problems on those of course. i'm detecting this now, and only enable interleaved writes if they are supported, so this should be sage.
17:25:05TheSevenDarkknight512: yes
17:25:53kugelBuschel: I don't think we should support mixed tags
17:26:38kugeland I agree with pixelma
17:26:45Darkknight512dosent appear to be a problem
17:26:58 Quit dfkt_ (Ping timeout: 265 seconds)
17:27:41pixelmaI remember there was a discussion with rasher when he filed the bug report last year
17:28:05 Quit DerPapst (Quit: Leaving.)
17:29:26Buschelpixelma: a pitty it wasn't added to the fs entry. I will ad some remakr to the fs entry
17:29:43rasherI just think requiring people to have perfect tags is a losing battle
17:30:22Darkknight512TheSeven:Writes and reads appear to be ok
17:30:24rasherIt's *easy* for us to check the v1 tag if the v2 tag is missing one or more of (artist, title, album)
17:30:40rasherIt's hard for the user to track down which of his songs have this problem
17:30:55 Join polobric1lo [0] (
17:30:56rasherAnd for the user with perfect tags, such a seek would never happen
17:31:16rasherSo I don't see exactly who we're inconveniencing terribly
17:32:30CIA-5New commit by theseven (r25174): Implement Nano2G fast NAND read API as a wrapper around the slow one.
17:33:19 Join panni_ [0] (
17:33:43saratogaI agree with rasher here
17:33:47pixelma started the 12th it seems
17:33:58CIA-5New commit by theseven (r25175): Re-enable fast reads in the Nano2G VFL/FTL. (Using the wrapper API in the NAND driver)
17:35:13TheSevenliar, Darkknight512: can you check reads again?
17:36:09 Quit polobric1lo (Ping timeout: 265 seconds)
17:38:22 Quit FlynDice (Remote host closed the connection)
17:40:04liarTheSeven: FTL: Scheduling bank ...
17:40:15liar(the same bank & block as before)
17:40:25TheSevenah-ha. a highlevel bug? that only happens with some chips? weird.
17:40:27rasherSo it's (so far) Buschel, saratoga, gevaerts, bluebrother and me on one side and pixelma, kugel, JdGordon and iirc amiconn on the other. I'm not sure how we're going to reach any conclusion here
17:40:48TheSevenso we were digging around the wrong place all the time :-P
17:41:06saratogawhats the argument against? that it might cause people to inadvertantly reduce their battery life if they have screwy tags?
17:41:18rashersaratoga: That seems to be it, yes
17:41:40saratogameh if you have screwed up tags, and a player where the disk actually matters, you get what you deserve
17:42:04Darkknight512TheSeven: I just copied half a gb of stuff from the Nano to my computer, no errors
17:42:05TheSevenhow likely is it to even cause additional power consumption? it won't if the whole file is being buffered at a time, right?
17:42:22TheSevenDarkknight512: did you *ever* get FTL panics before?
17:42:27saratogaprobably only matters for really big files that can't be all in the buffer at the same time
17:42:33kugelthe metadata is fetched before the file is buffered
17:42:42Darkknight512wait, nvm, i tried copy again, now it died
17:42:42saratogaoh yeah
17:42:59liarTheSeven: the panic screen looks very strange :-)
17:43:07Darkknight512FTL: Scheduling bank 2 block 492 remap!
17:43:16TheSevenwhen reading or writing?
17:43:24TheSevenliar: in terms of what?
17:43:47liarshould i have mentioned that before? wait ill make a photo
17:44:13pixelmanot that I've seen many complaints of users who say why doesn't XYZ show in the WPS even though my files are correctly tagged (except rasher)
17:44:43*kugel hasn't seen a single complaint
17:44:57rasherSo because no one has complained, we don't fix things?
17:45:22pixelmaI wouldn't call it a fix
17:45:27rasherI would.
17:45:37kugelI wouldn't
17:45:48pixelmayay, kindergarten
17:46:51kugelin my eyes it's trying to fix a problem which doesn't exist
17:47:03saratogaif it doesn't exist theres no harm in it
17:47:11saratogasince it will never effect anyone
17:47:15pixelmareading last year's coversation again... where do you draw the line after which Rockbox will not try to read the ID3v1 after getting some info from ID3v2?
17:47:59rasherIf artist and title is present
17:48:00pixelmasaratoga: if it adds code complication it'll affect everyone
17:48:18rasherpixelma: that's so miniscule it's not worth considering. Honestly
17:48:31rasher"Oh no, it'll shave off 2ms of battery life"
17:48:31saratogai don't think an if() parse_id3() block really adds much complication
17:48:47saratogawhy not just battery bench this
17:49:05pixelmarasher: ah, and what do you tell someone who asks about "and my album/year/whatever tag"?
17:49:09saratogai really doubt the difference is measurable even in the worst case, but its easy enough to test if you've got a hard disk placer
17:49:42rasherpixelma: I tell them that we have to draw the line somewhere, and an id3v2 tag that has artist and title is "good enough"
17:50:10pixelmaand what about songs where you deliberately leave the tag empty (because you don't know it or else)?
17:50:18rasherWhat about them?
17:50:38pixelmathen Rockbox will seek and won't even find something in the ID3v1
17:50:58rasherI thought the answer to anything bad happens is "fix your tags"?
17:51:11saratogaa sequential seek 10MB forward is maybe a 20 ms operation at most
17:51:18rasherAlso, if that happens enough that your battery life is significantly affected, I'll buy you a beer
17:51:29rasherThat goes for all Rockbox users.
17:52:01Buschelrasher: I would only slightly move the line to "good enough = has album, title, artist". The added code will only perform the seek/id3-reading, if this is not fulfilled. I cannot imagine a battery bench would show any effect.
17:52:04rasherI'd rather have an occasional seek in vain than an occasional missing info
17:52:15 Join froggyman [0] (
17:52:23pixelmathere it starts
17:52:42kugelrasher: fix your tags then :)
17:52:50*pixelma doesn't have a HD target anymore
17:53:27rasherkugel: You can't expect 100% of users to have 100% perfect tags
17:53:28kugelI'm not seeing the whole point in id3v1 tags anyway, id3v2 is clearly superior
17:53:31rasherIt's just not reasonable
17:53:43BuschelThis discussion is less technical and more about notion.
17:53:57kugelI don't expect, I just don't care about people with bad tags
17:54:08rasherThat's not really a great attitude to have at your users
17:54:16pixelmarasher: then you should also start supporting APE tags in MP3 files or in ogg vorbis etc. ... :/
17:54:18kugelso what
17:54:20TheSevenwhat about just someone actually *doing* a battery bench with that fix applied, and checking if there is a difference of more than a minute?
17:54:35rasherkugel: please never talk to our users again.
17:54:42rasherYour attitude is terrible
17:54:43BuschelCan't we just make a voting? ;o)
17:54:43AlexPTheSeven: Don't be silly, that'd end a good argument :)
17:55:06kugelstop with that. we make rockbox for us, not "my users"
17:55:32TheSevenyou could still argue while the bench is running...
17:55:38rasherAh, the nice old cop-out
17:55:49TheSevenerm, WTF?
17:55:55saratogaif you need 300mA to keep the disk spinning, and each id3v1 check takes 30ms, and you listen to 15 songs an hour, then I get that you would increase current consumption by 40 microamps
17:55:59kugelTheSeven: a bench will show nothing, I don't think it'll have any impact on battery life
17:56:04Darkknight512liar: Lol
17:56:14rasherkugel: so what's the harm?
17:56:29rasherkugel: do you just want to be able to point and laugh at users with slightly weird tags?
17:56:47TheSevenshouldn't panicf() disable interrupts rather early? at least *before* writing to the LCD?
17:57:10saratogayou'd probably have to strech each id3v1 access out to a couple hundred milliseconds before we could even measure it, maybe more
17:57:17Darkknight512TheSeven: should i give it a try and see if it heppens to mine?
17:57:27kugelthere's no harm, but I rather teach people to have good tags instead of messing with mixing them with an old and obsolete standard
17:57:42liarTheSeven: once i even saw the title bar
17:57:43soapHow common, really, is incomplete v2 tags AND well-formed v1 tags? I've seen plenty of truncated v1 tags paired with proper v2 tags (most common scenario most likely), I've seen v2 OR v1 only. But incomplete v2 and complete v1? I can't imagine the situation outside intention which would create that.
17:57:44rasherOh ffs. So you're arguing this because you want people to suffer?
17:57:46kugelyes, I want to laugh at them...
17:57:47saratogathe problem is theres little feedback that the tag format is the problem . . .
17:57:52TheSevenkugel: then just drop id3v1 support completely.
17:58:03kugelTheSeven: I can live with that :)
17:58:07rasherkugel: please stay out of this actual rational discussion then
17:58:10rasherchristing fuck
17:58:11AlexPIf it makes no battery difference, then why not just help people out?
17:58:16pixelmarasher: APE tags?
17:58:23rasherAlexP: Because we hate users, apparently
17:58:26saratogai still want ape tags
17:58:40rasherI'm not opposed to reading ape tags
17:59:03Darkknight512rasher: developersdevelopersdevelopers
17:59:08rasherObviously after v1 and v2
17:59:21saratogai've actually found a few albums with better apev2 tags then id3v2 tags in my collection
17:59:24pixelmaoh cool, more complication
17:59:29kugelrasher: please keep the discussion at a reasonable level
17:59:36AlexPDo ape tags come in the middle and require another seek? :)
17:59:36saratogaprobably some old foobar version did that
17:59:40rasherkugel: which is why I want you to stop being in it
17:59:48saratogaapev2 is next to id3v1
17:59:52saratogaso its the same case as we have now
17:59:57*Buschel was distracted when first using rockbox, which forced him to use APE tags on mpc files (which is non sense, as each audio format might use each tag format)
18:00:11TheSevenliar: you just have a weird ipod ;-)
18:00:16saratogai thought ape was the standard for mpc?
18:00:18rasherkugel: Sorry, I just have a low tolerance for being actively hostile towards other users of the software
18:00:23Buschelnevertheless I searched for a tool and replaced the tags on _all_ of my files...
18:00:28liarDarkknight512: is that happening for you too?
18:00:34Darkknight512im trying
18:00:34AlexPI really really fail to see the issue here - if there are no measurable negative side effects, then why not support the widest range we can?
18:00:39Darkknight512it hasent died yet
18:01:02soapand IF APE is supported, shouldn't the priority go v2 -> APE -> v1, rasher?
18:01:20kugeland I have a low tolerance for your attitude towards me, right now
18:01:23Darkknight512liar: were you in WPS or just diskmode screen when it did that?
18:01:26*TheSeven votes for ID3v2 and nothing else
18:01:37rashersoap: I'd say v2 > v1 > ape, but shrug
18:01:42TheSevenDarkknight512: apparently wps
18:02:04Darkknight512nope, mine is normal, well as normal as a crash is =D
18:02:06soaprasher, except for the fact if there was APE you'd expect them to be more complete than v1.
18:02:12*TheSeven has never seen ape tags being used by anything but mp3gain
18:02:27saratogaape and id3v1 could be done configurably since they're always sequential laid out back to back!
18:02:37liarDarkknight512: it does not happen always
18:02:44saratogaTheSeven: foobar used them for a long time
18:02:45soapkugel, except for the fact there are plenty of good reasons to use v1 tags. And since v1 tags are universally supported, yet limited there is plenty of reason for a file to carry two sets of tags.
18:02:51TheSevenhaha, read both and compare the id3v1 strings to ape, and use ape if they were truncated, but v1 if they are just different :-D
18:02:57saratogathey were very popular with HA's crowd
18:03:12Darkknight512liar: ok then, try #4
18:03:24 Part rasher
18:04:12soapyea, the foobar defaulting to APE for "higher complexity" MP3 tags really created the mess.
18:04:47Darkknight512No my crashes are normal
18:04:47kugelsoap: can you name anything that doesn't read id3v2? it's around for many many many years now
18:05:00soapkugel, my DVD player, my car radio.
18:05:34soapand even the hardware which DOES do v2 likely wants v2.3 or 2.2, not 2.4, and VERY likely doesn't want UTF-8.
18:05:38saratogawell the idea made sense circa 2003 when half the hardware out there couldn't read id3v2 properly but id3v1 still worked fine
18:06:10saratogaso you had id3v1 for hardware devices, and apev2 for software/archiving
18:06:12soapv1 tags are next to free, and give you near universal hardware support. Not much of a downside.
18:06:33 Quit r0b- (Ping timeout: 256 seconds)
18:06:44saratogado we still lock threads asking about unusable builds like we did when there were supported and unsupported builds?
18:06:55 Join r0b- [0] (
18:07:32soapI don't. I've laxed up on locking those since the signal-to-noise ratio on the forums is well under control.
18:07:35pixelmasoap: and if it doesn't like UTF, how does ID3v1 help there?
18:07:47soappixelma, it reads the v1 tags just fine.
18:08:16pixelmathose won't be UTF-8 either, no?
18:08:17 Quit r0b- (Client Quit)
18:08:36soapnope, but some tags > no tags.
18:08:52TheSevensaratoga: such threads (depending on the kind of question) *might* even help the developers of that port sometimes
18:09:41kugelwe have the development threads for the helpful stuff :p
18:12:47 Quit Topy44 (Ping timeout: 256 seconds)
18:13:00saratogahuh we don't even build the HDD63x0
18:13:33saratogamaybe that should be added to the build system, lowlight actually favored marking it as unstable
18:14:15 Join r0b- [0] (
18:15:26saratogaoh wait that was the hd16x0
18:16:00saratogahaha chrome thinks is in "irish" and offers to translate to english for me
18:16:52 Join u42p [0] (
18:17:07u42pi installed rockbox on my cowon d2 some weeks ago. how to i completely remove it again?
18:19:06saratogahmm though the manual instructions sound a bit weird
18:20:15 Join dfkt_ [0] (dfkt@unaffiliated/dfkt)
18:22:42 Join mt [0] (~mtee@rockbox/developer/mt)
18:23:00 Join Topy44 [0] (
18:23:46 Quit dfkt (Ping timeout: 265 seconds)
18:24:13 Join dfkt [0] (dfkt@unaffiliated/dfkt)
18:24:41 Quit r0b- (Ping timeout: 256 seconds)
18:25:19 Quit dfkt_ (Ping timeout: 260 seconds)
18:26:08CIA-5New commit by theseven (r25176): Nano2G: switch over to slow writes at VFL level
18:26:25TheSevenliar, Darkknight512: this build should be fine again
18:26:43TheSevenerm, it should say reads, not writes -.-
18:26:52***Saving seen data "./dancer.seen"
18:27:58u42psaratoga: yeah, it seems incomplete. i still have the bootloader on it now
18:28:08 Join Conformist [0] (
18:28:56 Join perfectdrug [0] (
18:31:35 Join r0b- [0] (
18:32:05saratogai think the manual uninstall is actually a different section with the wrong header
18:33:40 Quit saratoga (Quit: Page closed)
18:34:25Darkknight512umm i gtg TheSeven, cant test the read for you
18:36:01u42pi simply reinstalled cowon firmware, seems to have worked
18:38:51 Quit Topy44 (Ping timeout: 256 seconds)
18:41:27 Join Lear [0] (chatzilla@rockbox/developer/lear)
18:48:41liarTheSeven: works fine
18:49:42TheSevenliar: can you have a look at the diffs between the last 3 revisions? do you spot anything suspicious? the only change that i can see is that it will retry failed reads immediately with the working builds, and only after the other ones are completed with the failing builds
18:50:01 Join CaptainKewl [0] (~jason@
18:51:46 Join Topy44 [0] (
18:51:50 Join polobric1lo [0] (
18:51:57 Join planetbeing [0] (
18:53:34 Quit Casainho (Quit: ChatZilla 0.9.86 [Firefox 3.6/20100115133306])
18:56:25 Quit Topy44 (Ping timeout: 256 seconds)
19:01:45 Quit robin0800 (Quit: No Ping reply in 180 seconds.)
19:02:09 Join robin0800 [0] (
19:03:42 Join m3dlg [0] (
19:05:44 Quit robin0800 (Client Quit)
19:06:02 Join robin0800 [0] (
19:07:13 Join DerPapst [0] (
19:10:51 Join planetbeing_ [0] (
19:20:22CIA-5New commit by bluebrother (r25177): Fix rbspeex on big endian hosts. ...
19:20:47 Quit Curtman (Ping timeout: 240 seconds)
19:22:27 Join Curtman [0] (
19:22:32 Quit robin0800 (Quit: No Ping reply in 180 seconds.)
19:22:58 Join robin0800 [0] (
19:25:45CIA-5New commit by Buschel (r25178): Minor clean up in playback.c
19:26:03 Join Highlander [0] (
19:27:48 Join T44 [0] (
19:28:35flyaskyhi again! ) i want make wiki-page for Samsung YV-150/ how can I?
19:28:49 Join Topy [0] (
19:29:34pixelmabluebrother: are you up for a LaTeX question?
19:30:03pixelmaflyasky: are you registered at our wiki yet?
19:31:55pamaurygevaerts: I have a possible explaination for the usb audio problem. I just read the spec once again and to my disappointement, I noticed that the spec wants the bInterval value of the iso endpoint to be 1. It seems that the linux usb audio driver allows other values but only honors them if there are not to high (might be 1,2,3 and even 4). This might definitely be the problem. I need to do more tests before I can confirm this though
19:32:33 Quit T44 (Ping timeout: 240 seconds)
19:32:34flyaskypixelma: no
19:32:54flyaskyflyasky: i will
19:33:49 Join Topy44 [0] (
19:34:09 Quit Rob2223 (Quit: Rob2223)
19:37:08flyaskypixelma: now I registered in wiki
19:38:25 Quit bmbl (Ping timeout: 245 seconds)
19:42:00 Join bmbl [0] (~Miranda@unaffiliated/bmbl)
19:43:08 Quit stoffel (Remote host closed the connection)
19:47:20pixelmaflyasky: you should be able to edit the wiki now and create new pages
19:47:38flyaskypixelma: thank you
19:52:13 Quit tmzt (Ping timeout: 276 seconds)
19:53:35 Join tmzt [0] (
19:54:02 Quit tvelocity (Quit: Αποχώρησε)
19:58:03 Quit robin0800 (Read error: Connection reset by peer)
19:58:30 Join Omlet [0] (
20:00:30 Quit crazed (Read error: Operation timed out)
20:03:10 Quit pamaury (Ping timeout: 252 seconds)
20:08:13 Join Rob2222 [0] (
20:12:54ranmachanIf I want to test recording on my C200v2, do I need to change anything esides HAVE_RECORDING in firmware/export/config/sansav200v2.h?
20:13:31ranmachanI tried that and got a broken main menu (texts don't match the menu items anymore)
20:13:51ranmachanmake clean didn't help, only undefining HAVE_RECORDING...
20:23:14kugelare you sure you unzipped the complete
20:23:35kugelranmachan: what you're experiencing sounds like you loaded an outdated lang file
20:24:02ranmachanAh, I only updated rockbox.sansa
20:24:10ranmachanThat would explain it :)
20:26:55***Saving seen data "./dancer.seen"
20:28:14 Quit flydutch (Quit: /* empty */)
20:42:02CIA-5New commit by alle (r25179): Polish character set for 08-Atadore font (FS #11082 by Tomasz Kowalczyk), part 1
20:45:13CIA-5New commit by alle (r25180): Rename some glyphs in 08-Atadore font to standard unicode names (part 2 of FS #11082)
20:49:14 Part u42p ("Leaving")
20:49:39 Quit findus (Ping timeout: 245 seconds)
20:53:01 Join planetbeing__ [0] (~planetbei@
20:56:33 Quit planetbeing (Ping timeout: 276 seconds)
20:56:33 Nick planetbeing__ is now known as planetbeing (~planetbei@
20:56:34 Nick planetbeing is now known as 17SAALP2R (~planetbei@
20:56:34 Nick planetbeing_ is now known as 77CAAC9VV (
20:57:25 Quit Curtman (Quit: Leaving)
21:10:18 Quit amiconn (Disconnected by services)
21:10:19 Join amiconn_ [0] (quassel@rockbox/developer/amiconn)
21:10:20 Join pixelma_ [0] (quassel@rockbox/staff/pixelma)
21:10:20 Quit pixelma (Disconnected by services)
21:10:37 Nick pixelma_ is now known as pixelma (quassel@rockbox/staff/pixelma)
21:10:46 Nick amiconn_ is now known as amiconn (quassel@rockbox/developer/amiconn)
21:14:04 Quit GeekShadow (Read error: Connection reset by peer)
21:17:08 Join domonoky [0] (~Domonoky@rockbox/developer/domonoky)
21:22:43bluebrotherpixelma: sorry, had a phone call.
21:25:59CIA-5New commit by theseven (r25181): Finally fast Nano2G NAND reading, also works on remapped blocks this time.
21:26:42TheSevenliar, Darkknight512: please test :-)
21:26:48tmzt_TheSeven: sorry, is that for onenand devices?
21:26:58TheSevenno, plain old nand
21:27:03tmzt_okay, thanks
21:27:08 Quit dfkt (Read error: Connection reset by peer)
21:27:24tmzt_issues with linux mtd
21:27:51liarTheSeven: so it was a problem with remapped blocks?
21:28:06tmzt_just wanted to see what your solution was, but I don't think it's applicable
21:29:05TheSevenyes, it always used the VFL data for bank 0
21:29:29pixelmabluebrother: I can \opt the last line in a table, so that for some targets the last line is atually above that. If I use \nopt for this, I'll get the following error when compiling a manual for the excluded targets: . Is this unavoidable due to \nopt being a hack, or is it fixable?
21:29:35TheSevenand your flash seems to be one of the few around that has enough bad blocks that you actually hit one during testing
21:31:50bluebrotherpixelma: have you ended the \nopt line with a % so the input line continues?
21:34:21liarTheSeven: it has pretty much bad blocks yes...
21:35:25 Join dfkt [0] (dfkt@unaffiliated/dfkt)
21:35:25pixelmabluebrother: the nopt list fits on one line (which is why I prefer it (opt takes two lines) and I saw it before. Last time I rearranged stuff so that the last line in the table wasn't (n)opted but didn't draw the connection to \nopt - only now I saw it working with \opt
21:35:47liarTheSeven: works
21:37:35TheSevenanybody with some knowledge about the build client around?
21:38:05gevaertsa bit
21:38:54bluebrotherpixelma: can you post the code somewhere?
21:42:04pixelmaI'm unsure which part - this is from a fully reworked button table in goban tex for better readability. I have the non-working \nopt still in there but commented out, below is the working \opt
21:42:18gevaertsTheSeven: do you have curl, zip and perl available?
21:42:59gevaertshm, those are checked explicitely, they won't be the problem anyway
21:43:04TheSeveni installed curl, zip was already there, and as the client is running, perl just must be.
21:43:11gevaertssvn and make?
21:43:37TheSevenand the toolchain is in the path
21:43:42TheSeven(only arm for now)
21:43:48pixelmabluebrother: what would you need - the new goban.tex, a diff, the new button table?
21:44:28TheSevenbtw, do we already know which archlist name eabi will use?
21:46:02 Join Schmogel [0] (
21:46:10gevaerts"Command not found" is apparently sent if the build log has "not found" anywhere in it
21:46:25gevaertsHave you tried building in that tree?
21:47:04TheSevennot yet
21:47:10TheSeveni'm installing the other toolchains right now
21:47:25TheSeventhis is going to take ages on that machine
21:48:30 Join bluebroth3r [0] (
21:48:30 Quit bluebroth3r (Changing host)
21:48:30 Join bluebroth3r [0] (~dom@rockbox/developer/bluebrother)
21:48:30 Quit bluebrother (Disconnected by services)
21:51:45 Join Curtman [0] (
21:51:48 Nick 77CAAC9VV is now known as planetbeing (
21:55:44CIA-5New commit by theseven (r25182): Nano2G NAND: Don't continue reading on that bank if starting the read failed.
21:57:39 Quit dfkt (Quit: -= SysReset 2.53=- Ph'nglui mglw'nafh Cthulhu R'lyeh wgah'nagl fhtagn.)
22:12:47bluebroth3rpixelma: played around with it a bit but without success. I can reproduce the issue though. Sorry :(
22:13:04bluebroth3rreplacing all this opt stuff with a real preprocessor would be a really good thing.
22:16:31 Join perfectdrug_ [0] (
22:17:32 Quit Llorean (Quit: Leaving.)
22:18:37 Join Llorean [0] (~DarkkOne@rockbox/user/Llorean)
22:18:37 Quit perfectdrug (Ping timeout: 265 seconds)
22:22:44 Quit CGL (Ping timeout: 258 seconds)
22:23:42 Quit m3dlg (Ping timeout: 265 seconds)
22:24:10 Join m3dlg [0] (
22:26:58***Saving seen data "./dancer.seen"
22:28:11 Join RadicalR [0] (
22:38:32CIA-5New commit by funman (r25183): sd-as3525v2: delay a bit before reading the command response ...
22:38:38CIA-5New commit by funman (r25184): sd-as3525v2: do not reverse 2 times long responses, read them directly in the needed order
22:44:07 Quit Schmogel (Ping timeout: 245 seconds)
22:46:26CIA-5New commit by funman (r25185): sd-as3525: do not reverse 2 times long responses, read them directly in the needed order
22:48:43 Join planetbeing_ [0] (
22:50:32 Quit planetbeing (Ping timeout: 240 seconds)
22:50:32 Nick planetbeing_ is now known as planetbeing (
22:51:08 Join funman [0] (~fun@rockbox/developer/funman)
22:57:19funmanhum i shouted victory too early on the forum : reading from Clip+ works fine but not write
23:00:33 Join rasher [0] (~rasher@rockbox/developer/rasher)
23:04:20funmandeleting a directory is stuck on "deleting..."
23:05:28 Join saratoga [0] (~9803c6dd@gateway/web/freenode/x-jbawgofqaczrqwsh)
23:06:33saratogathe xiph people seem interested in our new mdct
23:11:14funmansaratoga: 58.36MHz needed for mp3@192kbps on clip+
23:11:28funmansounds a lot
23:11:53funmanbut then i'm not sure of fclk & pclk
23:12:30saratogai'm guessing pretty slow :)
23:12:37saratogahow much IRAM does the clip+ have?
23:13:22pixelmaI'm not sure if deleting from Rockbox is working correctly even on stable port targets. I had some file system errors on my M5 - a chkdsk /F seemed to fix everything but left a FOUND000 folder on the device which I deleted from within Rockbox - repeating chkdsk showed an error in this folder which shouldn't exist anymore (but was fixable again)
23:13:44 Quit kugel (Ping timeout: 265 seconds)
23:13:59pixelmabut I didn't try to reproduce yet
23:14:08funmansaratoga: same than clip afaik (didn't try to see if it was larger)
23:15:00gevaertsZagor: I had another one of those "Fatal build error: Missing log file" things, which seems related to the weirdness at"> . The database results for "select * from log where revision=25184 and value like '%zenvisionm30sim%';" also look strange to me
23:15:06funmanpixelma: seems to work on clipv1
23:15:56gevaertsI got the error at 22:44:23
23:16:24pixelmafunman: well, deleting seemed to work too according to Rockbox on the device
23:17:10amiconnrasher, saratoga: It's not only "screwed up" tags. Sometimes tags deliberately don't contain some information (I often leave out the album tag for single songs)
23:17:31 Quit Farthen (Read error: Connection reset by peer)
23:17:47rasheramiconn: And the response to that seems to be "fix your tags"
23:17:48amiconnI would hate it if rockbox does a useless seek in that case (completely useless, since my tracks don't have id3v1 tags at all)
23:18:06amiconnrasher: Since it isn't broken there is nothing to fix
23:18:16saratogathe seek won't have a measurable impact on battery life so i don't see why anyone should care
23:18:37rasheramiconn: Most songs come from an album.
23:18:48gevaertsnot all audio is songs
23:18:58gevaertsIsn't the current proposal to try v1 if *none* of album, artist or title are there?
23:18:59rasherAnyway, mostly what saratoga said
23:19:08amiconnThen take the year as another example. There are albums where the year simply isn't given
23:19:36rasherI'd say if none of artist or title is set
23:19:54 Quit Highlander (Quit: Quitte)
23:19:57saratogai'm not really seeing an argument against always checking all tags to be honest
23:19:58amiconn(e.g. Machinae Supremacy has no years given for their downloadable albums)
23:19:58rasherI still don't understand why we're even having this discussion, since no one will ever notice
23:20:09saratogawe're talking a few milliseconds her per song
23:20:29amiconnThe seek is useless, and id3v1 is inferior
23:20:43saratogaits also irrelevent
23:20:51*gevaerts isn't against trying v1 if none of the important tags is there for v2
23:20:52amiconnIt isn't irrelevanrt
23:21:30rasherWe do lots of "useless" things to check if various conditions are true
23:21:36saratogawe could parse the id3v1 tag 3 times over on every song load and no one would ever be able to measure the difference
23:21:55amiconnWe used to have a setting which id3 version to use, and that was okay
23:21:56gevaertsI don't care too much if the important things are "title and artist" or "title, artist and album" or some other combination
23:22:10 Join karashata [0] (
23:22:39amiconnsaratoga: I would expect to get a measurable difference
23:22:41rasheramiconn: it still wouldn't solve this problem
23:22:48CIA-5New commit by theseven (r25186): ftl-nano2g.c: s/\(\*([^)]+)\)\./\1->/
23:22:57rasheramiconn: it still wouldn't solve this problem (if we brought back that setting)
23:23:09saratogaamiconn: how long do you think it takes to read an id3v1 tag then?
23:23:27 Join CGL [0] (~CGL@
23:23:27amiconnI would guess around 100ms. Perhaps even more for very long tracks
23:23:33saratogai would expect much less
23:23:50saratogathe latency here would just be the rotational latency since the tag is some number of MB later
23:24:00amiconnOn flash targets, yes. But there are those hdd targets
23:24:12saratogasure but the disk is already spinning and you don't have to seek the head
23:24:23 Join Farthen [0] (
23:24:23saratogajust wait for the current track to advance 5-10MB
23:24:35amiconnNope. For a seek, you have to walk the fat chain, then access the data area
23:24:45amiconnIt's way more than just rotational latency
23:25:01 Join kugel [0] (~kugel@rockbox/developer/kugel)
23:25:11saratogathe chain for the file isn't already in memory?
23:25:28saratogabah thats annoying
23:25:42amiconnThat is, one (or a few) fat sectors are in memory
23:25:44TheSevenno fat buffering?
23:25:48 Quit Lss (Read error: Connection reset by peer)
23:25:53amiconnBut that doesn't help if the file is fragmented
23:26:26rasherAnyway, this extra talk won't happen for every song, or even most songs. Just very occasionally.
23:26:40saratogaso 100ms @ say 300mA and 4 minutes a song means 160 uA extra current
23:26:47TheSevenaha. fragmented files with incomplete id3v2, complete id3v1 on a hard disk target. now *that's* what i call a corner case.
23:27:00saratogaassuming we do an id3 look up every song
23:27:13amiconnFragmented files aren't a corner case. FAT fragments quite a lot
23:28:03rasherTheSeven: the "complete id3v1" part isn't part of the corner case to be fair
23:28:40*gevaerts would say that if none of the "standard-v1" set of tags is there in v2, reading v1 isn't a bad idea
23:28:50saratogamy sansa is at 4% fragmentation after 3 years and probably a billion rockbox updates, so I think assuming fragmentation is common isn't so good
23:28:57 Quit Farthen (Ping timeout: 256 seconds)
23:29:24saratogajust 31 fragmented songs!
23:29:53amiconnOne fat sector (512 bytes, i.e. 128 entries on fat32), references 4MB of data (at the maximum cluster size of 32KB). That means even a completely unfragmented, typical song at 192kbps (or similar vbr) spans more than one fat sector
23:30:17amiconnYour sansa doesn't hold as much data as a typical hdd target
23:30:29saratogameaning I swap out music more often?
23:31:04*amiconn checks his recorder
23:31:39saratogaUSB transfers tend not to fragment since UMS performance goes off a cliff if you try to copy two files at the same time
23:31:45saratogaat least on evcery device ive tried
23:32:26saratogaand if you only ever copy one file at a time onto a blank disk it'll rarely if ever fragment
23:32:54saratogamy fragmented files seem to be ones where i went through and edited the id3v2 tags
23:34:04 Quit Rob2222 (Quit: Rob2222)
23:34:06rasheramiconn: Okay, so even if we allow half a second to check the id3v1 tag. That'll be what, 10 seconds of extra work during a day? At most?
23:34:11CIA-5New commit by Buschel (r25187): Correction of musepack SV8 replaygain. The album/title peak is saved in a logarithmic representation and needs to be converted to linear fixed point ...
23:34:23rasherConsidering few files will fall into this trap
23:34:46saratogalets just do a battery bench and find out
23:34:56saratogaone of you must still have a hard disk target
23:35:26rasherIf we're going to have a "no feature must ever affect me negatively no matter how little", I could use the same argument against supporting multiple codecs, since I only use vorbis.
23:36:04saratogaspeaking of codecs, I think aac is the only one that still needs a gigantic 1MB codec buffer, we should really see about shrinking that
23:36:25 Quit m3dlg (Ping timeout: 246 seconds)
23:37:22 Join m3dlg [0] (
23:39:09amiconnrasher: Features which can have negative effects should be disableable
23:39:52 Join Rob2222 [0] (
23:40:04amiconnFeatures implemented that way are e.g. logging, database in ram, variable speed, ...
23:40:07rasheramiconn: Then it never ends
23:40:18 Quit m3dlg (Client Quit)
23:40:45rasherWe have to draw the line somewhere. We really can't have every damn feature be disablable because someone might not use it
23:40:47amiconnOh, voice too, of course
23:41:04rasherNo one would get anything done like that
23:41:43 Quit liar (Ping timeout: 245 seconds)
23:41:47saratogai don't really think it makes sense to have an option to disable a feature whose worse case cost is a hundred microamps
23:42:19saratogathere comes a point where small numbers become indistinguishable from 0
23:43:11BuschelWhy is this issue discussed that much? It is all about being more tolerant to imperfect tagging. When files are perfectly tagged, there is no additional seek/load or whatever.
23:44:22S_a_i_n_thow often are files perfectly tagged though?
23:44:44S_a_i_n_tmine are, your may be...but I've seen some libraries in some *right* states.
23:44:45BuschelFurthermore the v1 tag is only read, if the the v2 tag did not contain main data like artist/title/album (f which album could be dropped as well)
23:44:53 Quit Battousai (Read error: Operation timed out)
23:45:58BuschelWe are losing nothing (except the learning factor for users with baldy tagged files), and winning better usability. Just my 2 cents.
23:46:47 Join Battousai [0] (~bryan@gentoo/developer/battousai)
23:47:33 Quit Buschel ()
23:52:43kugelplease, no setting for this one
23:53:31 Quit domonoky (Read error: Connection reset by peer)
23:54:11 Quit Lear (Quit: ChatZilla 0.9.86 [Firefox 3.6/20100115144158])
23:55:28rasherIt just confuses me that people who would quite literally never ever notice this if someone had just done it (and they didn't happen to see the commit)
23:55:34rasher... are opposing this.
23:55:35funman+1, or at least add a setting to show/hide this setting
23:58:05 Quit Omlet (Quit: ( :: NoNameScript 4.22 :: ))
23:58:11amiconnI mean, it's okay to try id3v1 if id3v2 isn't there, or completely emtpy (because that's very likely an invalid tag)
23:58:45amiconnBut if id3v2 contains only partial information, how likely is it that the id3v1 tag contains the missing pieces?
23:58:46rasheramiconn: But why not if artist+title are missing? What do we lose by doing this?
23:59:00rasherWhy *not* do it?
23:59:00gevaertsamiconn: in Buschel's current patch, v1 would be tried if *all* of artist, title and album are missing

Previous day | Next day