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

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

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

#rockbox log for 2012-05-01

00:00:18 Quit Keripo (Quit: Leaving.)
00:00:52 Join [Saint] [0] (~Saint]@unaffiliated/saint/x-8516940)
00:01:58 Quit factor (Read error: Connection reset by peer)
00:03:58 Quit Rower85 (Quit: Hmmm...)
00:04:06 Quit leavittx (Ping timeout: 240 seconds)
00:04:19 Join factor [0] (
00:10:03funmanit shows read access to uninitialized memory, and invalid write access (not stack, not heap)
00:11:50 Quit ender` (Quit: sweater, n a garment worn by children when their mother is cold)
00:13:03 Quit matsl (Ping timeout: 244 seconds)
00:19:38 Join MethoS- [0] (~clemens@
00:24:14 Quit Strife89 (Ping timeout: 272 seconds)
00:25:40TheLemonManpamaury: seems to be something else beside power, the pc recognizes it but cant get the descriptor
00:28:32 Quit bertrik (Remote host closed the connection)
00:30:00 Join matsl [0] (
00:44:11 Part Arsin ("Once you know what it is you want to be true, instinct is a very useful device for enabling you to know that it is")
00:45:21 Join jdgord [0] (~jdgord@rockbox/developer/JdGordon)
00:47:14 Quit jdgord (Client Quit)
00:52:33 Join Scromple [0] (~Simon@
01:22:25 Quit matsl (Read error: Connection reset by peer)
01:27:37***Saving seen data "./dancer.seen"
01:49:00 Quit kadoban (Ping timeout: 246 seconds)
01:51:25 Quit TheLemonMan (Quit: WeeChat 0.3.7)
01:56:13 Quit pamaury (Remote host closed the connection)
01:58:35 Quit bitcraft (Remote host closed the connection)
02:00:22 Join bitcraft [0] (
02:05:03 Quit lebellium (Quit: ChatZilla [Firefox 13.0/20120425123149])
02:14:33 Quit [Saint] (Remote host closed the connection)
02:20:45 Quit remlap (Ping timeout: 272 seconds)
02:26:00 Join Keripo [0] (
02:39:34 Quit XavierGr ()
02:45:59 Join [Saint] [0] (~Saint]@unaffiliated/saint/x-8516940)
03:25:48funmansaratoga: did you run the sim in valgrind too?
03:26:05saratogano i don't know anything about valgrind
03:26:13saratogado you want me to help with something?
03:26:28funmandeciphering skin_parser.c perhaps :)
03:26:39saratogaor maybe kugel
03:26:48funmanJdGordon said he wont have time till last week
03:27:11funmankugel: can you help with fs#12639 ?
03:27:12fs-bluebot lebellium Samsung-like theme for Clip Zip: WPS doesn't load properly and USB not working (bugs, unconfirmed)
03:27:19funmans/last/next/ week
03:27:40***Saving seen data "./dancer.seen"
03:28:02funmanusing valgrind is simple: just add valgrind in front of your cmd line and valgrind will show a backtrace when the error happens
03:28:18funmanwith −−db-attach=yes, it will propose if you want to attach a debugger at the moment when an error happens
03:28:40JdGordonfunman: skin_parser.c is entirely me :)
03:29:04saratogacan you figure out where it is in the code when it dies?
03:29:06funmani am not sure what get_param_text is doing
03:29:30funmansure, i pasted an entire backtrace + gdb output on FS
03:30:41[Saint]That certainly is an awesome bug.
03:30:50saratogaso its trying to allocate memory for a bmp and it dies?
03:31:23funmanno i think it's related to the font
03:31:31funman#1 0x000000000043c322 in parse_font_load (element=0x6edd50, token=0x6eddc8, wps_data=0x77fb68) at /home/fun/rockbox/apps/gui/skin_engine/skin_parser.c:487
03:31:56saratogathat is the second trace right?
03:32:06funmanright, that s the one i am looking now
03:32:17funmanthe first one only happens when changing theme
03:32:23[Saint]I never considered switching out the font(s) when trying to debug this.
03:32:40funman[Saint]: it happens when unplugging USB
03:33:02funman(gdb) print element->params
03:33:02funman$11 = 1892
03:33:21funmanso i believe there is a skin_tag_parameter struct at offset 1892 of skin_buffer?
03:33:24funmanJdGordon: right?
03:34:07saratogaah i see, so get_param_text basically gives something that is not a pointer
03:34:26JdGordonthe skin engine uses offsets everywhere
03:34:40funman(gdb) print *(struct skin_tag_parameter*)((void*)&skin_buffer[1892])
03:34:40funman$13 = {type = 4294967295, data = {number = 65536, text = 8624294395904, code = 8624294395904}, type_code = 0 '\000'}
03:34:44saratogawow i already regret grepping for what get_param_text does
03:34:55[Saint]I found it odd that the theme fails on the first attempt to load, but it loads quite happily on the second attempt.
03:35:04[Saint]That was what interested me.
03:35:14funmansaratoga: indeed
03:35:36[Saint]...and the fact a theme borked USB, that was/is just...weird.
03:35:55funmanno it doesnt break USB
03:36:30funmanit breaks when the theme is reloaded, after USB was unplugged
03:36:42[Saint]Hum, I must be thinking of another issue.
03:37:05[Saint]He managed to make a theme that prevented USB from mounting.
03:37:21[Saint]I thought that was in the same bug report, my mistake.
03:37:45saratogatheres also themes that just data abort on load, so its probably a more general memory allocation problem hiding somewhere
03:37:54JdGordonfunman: well those values are obviously wrong :)
03:37:56saratogausb has to allocate stuff as well IIRC, so it can probably happen there
03:38:15funman(gdb) print *((struct skin_tag_parameter*)((void*)&skin_buffer[1892])+1)
03:38:15funman$20 = {type = INTEGER, data = {number = 0, text = 7318349394477056, code = 7318349394477056}, type_code = 4 '\004'}
03:38:19funmanJdGordon: yeah
03:38:40JdGordonwel no, that could work
03:38:45JdGordondata.number == 0 is fine
03:38:55funman(gdb) print /x (*((struct skin_tag_parameter*)((void*)&skin_buffer[1892])+1)).data.text
03:38:58funman$25 = 0x1a000000000000
03:39:08funmanwe use get_param_text though
03:39:21funmanparam_number == 1
03:39:35funmanJdGordon: please check the backtrace on
03:41:16funman(gdb) frame
03:41:16funman#1 0x000000000043c322 in parse_font_load (element=0x6edd50, token=0x6eddc8,
03:41:16funman wps_data=0x77fb68) at /home/fun/rockbox/apps/gui/skin_engine/skin_parser.c:487
03:41:17JdGordonis it the wps,fms or sbs that is broken?
03:41:19funman487 ptr = strchr(filename, '.');
03:41:21funman(gdb) print *element
03:41:24funman$30 = {next = -1, children = 21328189116645376, data = -2357432, tag = 0x4bc5e0,
03:41:27funman params = 1892, children_count = 0, line = 20, type = TAG, params_count = 2 '\002'}
03:41:30funmanhow can i tell ?
03:41:40[Saint]JdGordon: sbs iirc
03:41:41 Join kadoban [0] (
03:41:44JdGordonprint *wps_data
03:41:48funmandoesnt the backtrace show the wps ?
03:42:00funman$31 = {buflib_handle = -1, tree = -2359388, images = -1, font_ids = -1, font_count = 0, backdrop_id = -1, use_extra_framebuffer = false, albumart = -2358644, playback_aa_slot = 1, peak_meter_enabled = false, wps_sb_tag = true, show_sb_on_wps = false, wps_loaded = false}
03:42:12JdGordonerr no, thats not helpful :)
03:43:10funmansysfont is always in memory, right? (to show HID parameters for example)
03:43:10JdGordonand the sbs looks correct, and font_load() is correct
03:43:34JdGordonI'm intrigued but cant work on this now :)
03:43:56funmanyes you can!
03:44:10[Saint]JdGordon: I looked at the theme itself, its fine as far as I'm aware.
03:44:46funmanhm we dont have a XinitThreads() call in rockbox
03:44:46[Saint]Its some in-core weirdness, and that's where my ability to debug it stops.
03:44:55 Nick perrikwp_ is now known as perrikwp (
03:45:47funmankugel: do we need to call XInitThreads() in SDL code ? I don't know if more that one thread accesses X, but a sim running in valgrind complained about it
03:46:24JdGordonthe scroll thread will talk to the lcd
03:50:13 Quit factor (Read error: Connection reset by peer)
03:52:18 Join kisak [0] (
03:53:36kisakI'm looking at and it looks like there's no rockbox-related ebuilds in the Rion overlay, there is an old ebuild for rbutil in the sunrise overlay
03:54:34 Join enthdegree [0] (~enthdegre@wikimedia/enthdegree)
03:56:09kisakI was wondering if I should adjust that or leave it alone
04:04:15funmankisak: adjusting is welcome i think
04:05:29 Quit amiconn (Disconnected by services)
04:05:29 Join amiconn_ [0] (amiconn@rockbox/developer/amiconn)
04:05:35 Quit pixelma (Disconnected by services)
04:05:36 Join pixelma_ [0] (pixelma@rockbox/staff/pixelma)
04:05:39 Nick pixelma_ is now known as pixelma (pixelma@rockbox/staff/pixelma)
04:05:51 Nick amiconn_ is now known as amiconn (amiconn@rockbox/developer/amiconn)
04:12:18 Quit bieber (Ping timeout: 260 seconds)
04:13:36 Join bieber [0] (~quassel@
04:13:43saratogajhMikeS: on repeat is anything forcefully reset in the codecs, or are they responsible for reiniting everything themselves
04:13:52saratogabasically, is repeat a lot like just seeking to the start of the track?
04:16:23 Quit [Saint] (Read error: Connection reset by peer)
04:24:42saratoganoisetable_exp[i] = noisetable_exp[i]<< 1;
04:24:58saratogayeah thats probably not going to end well if repeat doesn't reinit noisetable_exp automatically
04:33:35 Quit TheSeven (Disconnected by services)
04:33:44 Join [7] [0] (~quassel@rockbox/developer/TheSeven)
04:37:41 Join icarusfactor [0] (
04:38:47 Join [Saint] [0] (~Saint]@unaffiliated/saint/x-8516940)
04:46:52jhMikeSsaratoga: repeat = seek without removing pcm data from buffer
04:47:08saratogaok then that makes a lot more sense
04:47:33saratogaso basically, on init the codec takes a lookup table and doubles it if a certain feature is enabled
04:47:45saratogaon repeat, it will therefore double that table each time it loops
04:48:04saratogawhat is the best way to fix this? make the first value magic and just check to see if its doubled?
04:48:10jhMikeShaha, well, maybe they meant to do that?
04:48:21jhMikeSwhere is this?
04:48:41saratogadifferent codec features need a noise table with different average power
04:49:53saratogano this is my bug i added it 4.5 years ago
04:50:47jhMikeSwhy wouldn't general seeking be a problem or is it?
04:51:07saratogait doesn't reinit the table on seek
04:51:38saratogaoh huh i todes
04:51:43saratogalet me test that
04:52:25jhMikeSrepeat and seek should be doing exactly the same thing, only difference is calling audio_pre_ff_rew first
04:54:24jhMikeSyou know, what you destroy the original table don't you?
04:54:47 Quit [Saint] (Read error: Connection reset by peer)
04:55:34jhMikeSso some new track might need the original or modded original but now it's gone?
04:57:16 Quit anewuser ()
04:57:42saratogai don't know, when does the codec get reloaded?
04:59:48 Quit n17ikh (Ping timeout: 256 seconds)
05:00:17saratogahaha yeah it does mess up subsequent tracks
05:00:41 Join [Saint] [0] (~Saint]@unaffiliated/saint/x-8516940)
05:00:49saratogaits just that the error is quite subtle until you've looped through quite a few times, and its extremely uncommon to find a wma file with this feature enabled
05:01:05saratogaand loading a vorbis file then back to wma fixes it
05:02:24saratogawell its easy enough to check if the table is already doubled or not, then just undouble it if needed
05:04:25saratogaalso, i wonder how long a noise table really has to be for it to sound random, the current 8k table seems a little excessive
05:06:08 Nick icarusfactor is now known as factor (
05:08:30 Join n17ikh [0] (
05:12:03 Join n17ikh_ [0] (
05:13:44jhMikeScodec gets reloaded if another is loaded and it needs it again otherwise never
05:16:35jhMikeSnoise table repeat with 21.5 hz?
05:17:43saratoganoise coding is only used below 11khz IIRC
05:17:50saratogaso really its repeats with about 1 hz
05:18:01saratogaalthough i don't even think it needs one value per sample, so probably even less
05:19:24saratogasince these values are used in the frequency domain, i tend to think even a fairly high repeat rate would be fine, so long as the table length wasn't integer divisible by the number of random numbers needed per frame
05:19:44saratogasince you probably wouldn't notice if the noise repeated at a different frequency each time
05:25:59jhMikeScouldn't use the same thing as the dither output?
05:27:44***Saving seen data "./dancer.seen"
05:31:51saratogajhMikeS: is this too ugly you think?
05:32:08saratogai didn't know about dither when i wrote that, let me check
05:33:55saratogai'm actually not sure what the PDF of the noise needs to be, i guess i could look plot the table and figure it out
05:35:06jhMikeSthe random is just uniform PDF, the triangle comes from x[x] - x[n-1]
05:35:44 Quit bitcraft (Remote host closed the connection)
05:35:57saratogaint32_t random = dither->random*0x0019660dL + 0x3c6ef35fL;
05:36:00saratogathats all you use?
05:37:36jhMikeScheap and not bad results
05:38:07jhMikeSthe next line makes it purple noise
05:38:58saratogai'll look into this later, i don't have time right now to dive into another project
05:39:08saratogafor now i think i'll jsut push my less awesome fix
05:42:40CIA-44Commit 87d3dde in rockbox by Michael Giacomelli: Fix corrupt when repeatidly playing very low bitrate WMA files.
05:43:32saratoga" Fix corrupt when repeatidly playing very low bitrate WMA files. "
05:43:46saratogawow its like i didn't read anything i typed
05:44:37saratogajust substitute for "Fix corruption when repeatedly playing very low bitrate WMA files."
05:45:24CIA-4487d3dde build result: All green
05:46:02[Saint]*playing very low bitrate WMA files repeatedly?
05:46:32saratogalook i don't know
05:46:40saratogai've been writing all day my brain isn't working
05:47:56[Saint]I wasn't trying to criticize, that's just how I would have put it and I wondered if I would have been wrong in doing so.
05:48:11saratogayeah that works too
05:48:15jhMikeSgerrit seems to acting insane
05:48:33 Quit MethoS- (Ping timeout: 252 seconds)
05:48:40saratogahow so?
05:48:46saratogathat was the first time i'd used it, seems nice
05:48:59[Saint]Yeah, its pretty cool.
05:49:03jhMikeSkept giving me errors
05:49:14jhMikeSnow I"m getting 404ed
05:49:24saratogaDNS maybe?
05:49:50[Saint]that's a feature, jhMikeS protection :)
05:49:50*jhMikeS tries browser restart
05:50:04funmansaratoga: svn propedit log.. whoops we can t do that anymore
05:50:20saratogais there a way to edit git commit messages?
05:50:37funmannope, the hash depends on the message
05:51:13[Saint]Oh...yeah, derp. Not after its been pushed to master.
05:51:21funmanin vlc we use git push −−force to replace the commit with a new one with a better message though
05:51:34jhMikeShrm, that seemed to help
05:51:39funmanit needs to be made rapidly before too much people pull from the tree though
05:51:53saratogaso just commit a new message and force a push?
05:52:24funmanyeah, that will give a new hash
05:52:56saratogaso "git push −−force origin HEAD:master"
05:53:14saratogai guess it'll lose the gerrit bits
05:53:26[Saint]You can do −−append iirc, but I'm pretty sure that only works locally, before the push.
05:53:48saratogawell should i try this or will i regret it :)
05:53:58saratogamy grasp of git is weak
05:54:11[Saint]Only one way to find out: )
05:54:30[Saint]...well, only one fun way.
05:55:05saratogano that bounces, "! [remote rejected] HEAD -> master (non-fast forward)"
05:55:34saratogaperhaps it didn't like me pushing the appended commit?
05:57:16[Saint]I'm not too sure how to deal with this wrt gerrit
05:57:48[Saint]Taking gerrit out of the picture, its easy.
05:58:05saratogawhy does gerrit matter?
06:01:52[Saint]Well, it doesn't really...its pushing anywhere that makes it more difficult.
06:02:04 Join n1s [0] (~n1s@rockbox/developer/n1s)
06:03:33[Saint]"git commit −−ammend -m "new commit message"" should do it, unless I'm mistaken.
06:06:02[Saint]Assuming its the most recent commit, that is.
06:06:22[Saint]Its rather different if its not so.
06:09:39*jhMikeS gets into detached head state repeatedly
06:10:30*[Saint] resists
06:10:56*jhMikeS was waiting for that
06:11:09 Join T44 [0] (
06:13:38[Saint]Google seems to suggest to me that "git commit −−append -m "edited commit message" && git push −−force" should 'just work'.
06:13:59 Quit Topy44 (Ping timeout: 246 seconds)
06:14:16 Quit enthdegree (Quit: HydraIRC -> <- Go on, try it!)
06:16:55saratogayeah it rejects that
06:16:57saratogano idea why
06:17:36*[Saint] prods Torne
06:17:56saratogagoogle says its not a good idea, and the directions were complicated enough i don't feel like figuring it out at this hour
06:18:01saratogaso i'm just going to leave it
06:18:49[Saint]Fair enough.
06:19:35saratogaseems like it causes problems for other people
06:28:07 Quit saratoga (Quit: Page closed)
06:33:08 Quit n1s (Read error: Connection timed out)
06:39:53 Join bitcraft [0] (
06:40:06 Quit prof_wolfff (Ping timeout: 276 seconds)
06:40:12funmanyeah IMO Just leave it as is
07:20:26 Join wodz [0] (
07:21:30wodzsaratoga: I think 87d3dde is incorrect. 0x05 doubled is 0x0a NOT 0x10
07:27:48***Saving seen data "./dancer.seen"
07:28:35CIA-44Commit b4eec0d in rockbox by Michael Sevakis: Make INITDATA_ATTR work on everything that has INIT_ATTR enabled for code.
07:31:37CIA-44b4eec0d build result: 0 errors, 3 warnings (Michael Sevakis committed)
07:32:03 Quit mc2739 (Quit: leaving)
07:34:49wodzjhMikeS: I have a question regarding dsp code. Would it be possible to play higher samplerate files without resampling if 1) dsp is not used at all 2) codec hardware can be clocked with higher clocks (I mean I2S clock)?
07:35:14wodzI guess Hifmans people will love us for this
07:35:40jhMikeSI get that question alot
07:36:29 Join mc2739 [0] (~mc2739@rockbox/developer/mc2739)
07:36:57jhMikeSSure, we can do it, but to just switch it all the time would be very problematic. I think it would be best to just select a desired output rate from what the hardware supports.
07:38:25wodzI guess we are talking about 48kHz and 96 maybe. Many codecs can handle such freqs just fine
07:39:17wodzand moreover we advertise in the code what are the codec capabilities
07:39:32jhMikeSSure but I advise agaist switching shared PCM on a per track basis
07:40:11wodzyou mean fade out/in will be f***** up right?
07:40:34jhMikeSnot just crossfade
07:41:30wodzMost people which will 'benefit' from high res files are rather ortodox and disable all effects AFAIK
07:41:34jhMikeSwhich we _could_ avoid between files of different rates. there's also latency between DSP and what you're hearing and there's also the matter of mutiple streams
07:43:08wodzby multiple streams you mean our mixer code (aka beep)?
07:43:48jhMikeSbeep, voice or potentially other stuff
07:45:07wodzSo, how others deal with this problem? Just better resampler?
07:46:50jhMikeSI think just setting the preferred samplerate would be fine. You'd pick what you usually use and just resample to that.
07:48:50wodznot very userfriendly approach
07:49:55 Join n1s [0] (~n1s@rockbox/developer/n1s)
07:49:57jhMikeSI think users that care will figure it out :)
07:52:48wodzHow good/bad is downsampling compared to upsampling? I guess downsampling gives better results so setting freq to say 96 will degrade avarage quality (unless only 96k files are played)
07:53:34jhMikeSupsampling is better really
07:56:17wodzso this all boils down to 'it would be nice to have better resampler' :-)
07:57:24jhMikeSsure it would. even larger upsample ratios sound a bit crunch to me
07:58:40jhMikeSthough depending on the target and the codec, a more expensive algorithm would make certain formats useless
08:00:55CIA-44Commit 0dd200b in rockbox by Michael Giacomelli: Fix audio corruption when sequentially playing low bitrate WMA files
08:02:07 Join saratoga [0] (980329b4@gateway/web/freenode/ip.
08:02:25saratogawodz: good resampling isn't terribly difficult, at least for 16 bit limited audio
08:02:49saratogai think for 96,88.1k audio downsampling makes perfect sense since its pretty easy to do in those case if we had a better resampler
08:03:17saratoga48k to 44.1k is a bit harder, but can still be done with very good quality at reasonable computation cost, we just haven't written the code to do so :)
08:03:47CIA-440dd200b build result: 0 errors, 3 warnings (Michael Giacomelli committed)
08:04:24wodzsaratoga: You mentioned some articles about this. Could you share? I am going to learn something about dsp in general.
08:06:51saratogawodz: are you familiar with the textbook way to do a resampler ?
08:07:18saratoga(as opposed to the less accurate but much more efficient polynomial ones discussed there)
08:07:43wodzsaratoga: no, I know almost nothing about dsp for now
08:07:50saratogaah ok
08:08:01saratogatheres basically two ways you can do it
08:09:39saratogathe DSP way students learn is to upsample to the least common multiple of your source and destination sample rates (LCM(44100,48000)=705600Hz), then low pass filter so that nothing above half your target frequency is left, then throw away the excess samples
08:10:15saratogaunfortunately, that is stupidly slow since it involves doing filtering on an almost 1MHz sample rate
08:10:43saratogaso the smart way is to fit a polynomial to your data, and then read out the points you need to compute by evaluating the fit
08:12:20wodzsaratoga: something like flac compression scheme
08:14:33saratogai guess its not even really so much signal processing as general interpolation math stuff
08:14:56saratogaanyway, i need sleep
08:28:54 Join Dacicus2 [0] (
08:50:09 Quit factor (Read error: Connection reset by peer)
09:03:29 Join factor [0] (
09:03:33Dacicus2Hi, factor.
09:05:00 Join ender` [0] (
09:05:02Dacicus2Hi, ender`.
09:05:49JdGordonDacicus2: if you have a question just ask it
09:05:51Dacicus2How are you doing, JdGordon?
09:06:06Mode"#rockbox +o JdGordon" by ChanServ (ChanServ@services.)
09:06:10*[Saint] points to the topic
09:06:11Mode"#rockbox -o JdGordon" by ChanServ (ChanServ@services.)
09:06:23Mode"#rockbox +o JdGordon" by ChanServ (ChanServ@services.)
09:06:26Kick(#rockbox Dacicus2 ::bye bye bot) by JdGordon!~jonno@rockbox/developer/JdGordon
09:06:28Mode"#rockbox -o JdGordon" by ChanServ (ChanServ@services.)
09:20:15 Quit [Saint] (Remote host closed the connection)
09:20:44 Join [Saint] [0] (~Saint]@unaffiliated/saint/x-8516940)
09:26:59 Quit Scromple (Read error: Connection reset by peer)
09:27:51***Saving seen data "./dancer.seen"
09:46:05 Quit wodz (Quit: Leaving)
09:50:17 Join Rower85 [0] (
09:50:44 Quit Rower85 (Client Quit)
09:54:18 Join stoffel [0] (
09:54:47 Quit bitcraft (Remote host closed the connection)
10:06:42 Join mortalis [0] (~mortalis@
10:29:04fs-bluebotGerrit review #227 at,227 : New crossfeed algorithm for Rockbox: "Meijer" crossfeed by Bertrik Sikken (changes/27/227/1)
10:38:41 Quit stoffel (Ping timeout: 265 seconds)
10:43:16*[Saint] wonders if there was supposed to be more to that or not...
10:43:32[Saint]jhMikeS: ↑?
10:46:52jhMikeSto what?
10:47:27jhMikeSjust "announcing" to those that care
10:47:41jhMikeS..that it was updated
10:48:11[Saint]It sat on flyspray for a year or so...possibly longer.
10:48:48jhMikeSthere was talk of committing it though recently but I said "hold off 'till I'm done with this other stuff"
10:48:57[Saint]I seem to recall Bertrik saying he had no intention of committing it, perhaps I'm mistaken.
10:49:25jhMikeSmaybe someone else did
10:49:36[Saint]I bugged him about it ages ago when he first put it up.
10:49:47[Saint]I'm quite sure, actually.
10:50:06[Saint]Perhaps he flipped on it. Dunno.
10:50:07jhMikeScan't remember exactly, probably in the logs...perhaps something about configurability too
10:50:32CIA-44Commit 92ef7bd in rockbox by Dominik Riebeling: Move bootloader class initialiation into helper.
10:50:32CIA-44Commit 9b363c6 in rockbox by Dominik Riebeling: Bootloader installation: set file filter.
10:50:44jhMikeSI was interesting in trying it anyway
10:50:53[Saint]Its nice.
10:53:03CIA-449b363c6 build result: All green
10:55:06 Quit kadoban (Ping timeout: 256 seconds)
10:55:14n1si'm wondering what to do about g139 I gather it's not ok to drop yielding in dsp_process but it's also a rather ugly thing to have in the lib
10:55:15fs-bluebotGerrit review #139 at,139 : rbcodec refactoring: make dsp_process not yield by Sean Bartell (changes/39/139/2)
10:56:38bluebrotherhmm. Given the number of fixes Rockbox Utility got in the last two month a new release might be in place
10:57:39jhMikeSuse preemptive threads? :)
10:57:46bluebrotherthough I'd really like to figure what to do about linux binaries first ...
10:58:03n1sotoh, codecs yield too so perhaps putting the coop stuff inside some #ifdef is enough?
10:58:25jhMikeSprobably good enough for now
11:00:25jhMikeSDSP can process samples in larger runs now too
11:05:12n1si'm thinking that the dsp should know the samplerate and should be able to calculate the number of samples that are processed in a tick so shouldn't need to read current_tick
11:08:28jhMikeSit processes faster than the samplerate
11:09:23n1syeah but yielding too often as long as it's not excessive isn't too bad, is it?
11:11:39jhMikeSnot too bad. I did want to keep that loop small though and not clutter it.
11:14:25jhMikeScould have inline functions there and define them blank outside of firmware builds
11:14:54n1sdo you prefer the current approach? we could just wrap current_tick in a function call
11:15:54 Join lebellium [0] (
11:21:30n1sit will probably make sense to do the #ifdef stuff after merging the patch that adds an rbcodec config header
11:23:02 Quit [Saint] (Read error: Connection reset by peer)
11:23:18 Join [Saint] [0] (~Saint]@unaffiliated/saint/x-8516940)
11:27:55***Saving seen data "./dancer.seen"
11:28:39CIA-44Commit 4bef502 in rockbox by Nils Wallménius: (Author: Sean Bartell) rbcodec refactoring: autoresumable
11:30:17 Quit amiconn (Remote host closed the connection)
11:30:17 Quit pixelma (Remote host closed the connection)
11:31:55CIA-444bef502 build result: All green
11:32:19 Join pixelma [0] (pixelma@rockbox/staff/pixelma)
11:32:19 Join amiconn [0] (amiconn@rockbox/developer/amiconn)
11:47:39jhMikeSn1s: this satisfies me to not change behavior, it just needs the definitions in the right place under the right conditions:
11:50:07jhMikeSn1s: couple of corrections :
11:52:10jhMikeSargh, stupid typos
11:53:23 Join pamaury [0] (~quassel@rockbox/developer/pamaury)
12:01:32*jhMikeS was transfering an album to his clip v1 and it gave a SWI fault :\
12:01:37 Join mirak [0] (
12:02:42*jhMikeS sees more silly typos in that patch and forgets about it lol
12:10:39n1sshould i apply that (with fixes for typos)?
12:13:16 Quit [Saint] (Remote host closed the connection)
12:16:25 Join TheLemonMan [0] (
12:17:43 Join dfkt|a [0] (~dfkt@unaffiliated/dfkt)
12:28:15jhMikeSn1s: it needs the bits put in the right headers
12:28:55jhMikeSI haven't looked at what exactly is going on with everything warble
12:29:47n1sjhMikeS: yeah, a later patch by wtachi adds rbcodec config headers so i think i'll look into merging that and putting the DSP_PROCESS* stuff in the rockbox specific header
12:30:10jhMikeSsounds good
12:30:10n1sbut that will be later
12:30:46jhMikeSI guess as a bonus the patch gets all the target stuff out
12:31:32jhMikeSjust the header adjustment would be left
12:32:36n1si hadn't looked at that other stuff before but it's nice
12:44:38 Join y4n [0] (~y4n@unaffiliated/y4ndexx)
12:59:51 Quit Keripo (Quit: Leaving.)
13:13:31 Join wodz [0] (
13:19:37 Quit [7] (Ping timeout: 265 seconds)
13:23:59 Quit pamaury (Remote host closed the connection)
13:24:47 Join prof_wolfff [0] (
13:27:59***Saving seen data "./dancer.seen"
13:31:50 Join [7] [0] (~quassel@rockbox/developer/TheSeven)
13:32:10 Quit prof_wolfff (Ping timeout: 246 seconds)
13:38:56 Join prof_wolfff [0] (
13:38:57 Quit [7] (Disconnected by services)
13:39:03 Join TheSeven [0] (~quassel@rockbox/developer/TheSeven)
13:51:41 Quit ender` (Quit: Zero Defects, n. The result of shutting down a production line.)
14:17:51 Join remlap [0] (
14:22:31 Quit TheSeven (Read error: Operation timed out)
14:24:01 Join TheSeven [0] (~quassel@rockbox/developer/TheSeven)
14:29:38 Quit dfkt|a (Quit: AndroIRC - Android IRC Client ( ))
14:58:27 Quit soap (Ping timeout: 246 seconds)
15:01:29 Join lebellium_ [0] (
15:01:43 Join WalkGood [0] (~4@unaffiliated/walkgood)
15:03:10 Quit lebellium (Ping timeout: 246 seconds)
15:03:25 Nick lebellium_ is now known as lebellium (
15:05:46 Join soap [0] (
15:05:46 Quit soap (Changing host)
15:05:46 Join soap [0] (~soap@rockbox/staff/soap)
15:11:08 Quit n17ikh_ (Quit: leaving)
15:13:47 Join stoffel [0] (
15:15:43 Quit n17ikh (Quit: Changing server)
15:15:56 Join n17ikh [0] (
15:28:00***Saving seen data "./dancer.seen"
15:30:56 Quit y4n (Quit: We're fucking 3LN!)
15:36:30 Join y4n [0] (~y4n@unaffiliated/y4ndexx)
15:44:02 Quit bluebrother (Disconnected by services)
15:44:08 Join bluebrother [0] (~dom@rockbox/developer/bluebrother)
15:47:16 Quit fs-bluebot (Ping timeout: 246 seconds)
15:48:23 Join fs-bluebot [0] (
15:53:55 Quit jhMikeS (Ping timeout: 246 seconds)
16:13:06 Join ender` [0] (
16:17:02 Join anewuser [0] (~anewuser@
16:17:02 Quit anewuser (Changing host)
16:17:02 Join anewuser [0] (~anewuser@unaffiliated/anewuser)
16:20:33 Join Guest123 [0] (
16:22:34 Quit Guest123 (Client Quit)
16:35:56 Join webguest198 [0] (
16:38:08webguest198Hello, I am trying to find a way to contact the RockBox crew by email but there are no contact details on the webpage. Where could I send an email?
16:42:53 Quit webguest198 (Quit: CGI:IRC)
16:43:03n1sThe mailing list perhaps?
16:45:28 Quit stoffel (Ping timeout: 244 seconds)
17:05:11bluebrotherwhy contact by email anyway?
17:05:34 Join jlbiasini [0] (
17:05:39 Part jlbiasini
17:14:01 Join kadoban [0] (
17:28:02***Saving seen data "./dancer.seen"
17:42:50 Join LeMiasme [0] (~LeMiasme@2a01:e35:8ab0:54d0:7057:122a:bae:86c8)
17:45:13 Quit LeMiasme (Client Quit)
17:46:15lebelliumhum my theme for Fuze is almost finished and like for the Clip Zip, the USB connection won't work, it says that whenever I connect the USB cable: "*panic*
17:46:16lebelliumUnhandled unmasked IRQ 03: INT_USB (status 0x00000008)"
17:46:43 Quit TheLemonMan (Quit: WeeChat 0.3.7)
17:48:22 Join Strife89 [0] (~Strife89@
18:33:48 Join stoffel [0] (
18:48:06 Join DomasoFan [0] (
18:48:47 Part DomasoFan ("I'm a happy Miranda IM user! Get it here:")
18:53:10n1swhy is root.make adding MEMORYSIZE as a command line define when autoconf #defines it?
18:57:04n1swait a minute
18:58:02n1ssigh, better question is why does warble have an "autoconf.h" in its source?
19:03:04 Join einhirn [0] (
19:11:48CIA-44Commit 45c5b1f in rockbox by Nils Wallménius: Delete autoconf.h file in the warble source, one less headache.
19:15:58 Join [Saint] [0] (~Saint]@unaffiliated/saint/x-8516940)
19:17:52 Quit stoffel (Ping timeout: 244 seconds)
19:19:41funmanlebellium: git pull
19:21:16 Quit Poodlemastah (Ping timeout: 245 seconds)
19:21:38funmanby the way i could reproduce the bug you reported and confirm that your description is 100% correct
19:22:21 Join bertrik [0] (~bertrik@rockbox/developer/bertrik)
19:22:49lebelliumI really don't understand why with my 3 themes (Zip, Fuze+/Gigabeat/R0 and now Fuze) the WPS doesn't load properly at first time. Is there something wrong in my code?!
19:23:48[Saint]When I looked at your cut down versions, they should have been fine.
19:24:43 Quit WalkGood (Quit: ♪ ♫ ♪ ♫ ♪ ♫ ♪)
19:25:24[Saint]I recall one or two things I probably would've done differently, but when I checked over it it looked ok.
19:26:19lebelliumthus I don't understand why it seems to occur only with my themes (although I did not try all the other themes available of course)
19:26:27wodzsomething weird with skin engine is happening nowadays
19:27:19wodzI guess all this weirdness is with some relation to mem corruption valgrind is showing
19:27:25[Saint]I haven't had a lot of/any time for Rockbox themeing for ages. :-S
19:28:05***Saving seen data "./dancer.seen"
19:28:27[Saint]I need to finish what I started with my RaaA themes, but time/real life/etc. hasn't been on my side.
19:32:27 Quit einhirn (Ping timeout: 252 seconds)
19:33:17funmanyes the skin code is broken
19:34:08funmanlebellium: you need d7ead706a
19:37:05lebelliumto fix what issue? Because the WPS never load properly at first since October...
19:39:10 Join Rower85 [0] (
19:40:42funmanlook at the commit log it is fully explained
19:42:00lebelliumok for the USB connection
19:43:28bertrikfunman, oh wow, /me wonders how USB worked at all with the wrong interrupt handler
19:43:42funmanit did not
19:44:31 Join Keripo [0] (
19:44:39bertrikoh, it was broken for just 1 day I see
19:45:34 Join pamaury [0] (
19:45:35 Quit pamaury (Changing host)
19:45:35 Join pamaury [0] (~quassel@rockbox/developer/pamaury)
19:46:29n1sgrr why does gcc warn about redefinition even after #undefing ?
19:46:44funmaninternal define ?
19:48:46n1sthis is DEBUGF which i don't think is one of those as it should then warn everywhere
19:49:04 Join jfc [0] (
19:51:49n1sthis is too messy with redefinitions everywhere :/
19:59:26 Quit mortalis (Quit: KVIrc 4.1.3 Equilibrium
20:04:28 Join petur [0] (~petur@rockbox/developer/petur)
20:11:24peturfocus fail
20:13:00wodzI guess it is using rk27generic as base.
20:13:59 Join saratoga_ [0] (98032941@gateway/web/freenode/ip.
20:14:11saratoga_do we have a good time domain low pass filter in rockbox somewhere?
20:14:22saratoga_i guess the EQ low frequency adjustment code?
20:14:29saratoga_(something simple would be nice)
20:19:27CIA-44Commit 6637e24 in rockbox by Dominik Riebeling: (Author: Jean-Louis Biasini) Update Fuze+ manual (main files) FS #12492
20:21:23bluebrotherfinally got around reviewing this ...
20:22:27CIA-446637e24 build result: All green
20:23:41 Quit wodz (Quit: Leaving)
20:24:32 Join mortalis [0] (~mortalis@
20:25:29bertrikwodz what is a ritmix?
20:25:59saratoga_wodz: (for the logs) the guy who wrote that resampling paper also put up a beginners guide to the topic which might be useful if you want to learn more
20:27:51bertriksaratoga, we have a generic biquad I think, this should be able to do a 2nd order lowpass
20:39:10 Join Transformer [0] (
20:39:48 Quit bertrik (Remote host closed the connection)
20:40:25 Quit Transformer (Excess Flood)
20:43:33 Join bertrik [0] (
20:43:34 Quit bertrik (Changing host)
20:43:34 Join bertrik [0] (~bertrik@rockbox/developer/bertrik)
20:43:48 Quit mortalis (Quit: KVIrc 4.1.3 Equilibrium
20:47:35 Quit factor (Read error: Connection reset by peer)
20:48:52 Quit funman (Ping timeout: 248 seconds)
20:50:08 Quit Strife89 (Quit: Battery running low.)
20:54:12 Quit GeekShadow (Ping timeout: 248 seconds)
20:55:04 Quit Keripo (Quit: Leaving.)
21:00:56 Join factor [0] (
21:01:07 Join GeekShadow [0] (
21:02:00 Join Poodlemastah [0] (
21:03:14 Quit pamaury (Remote host closed the connection)
21:08:25 Join funman [0] (~fun@rockbox/developer/funman)
21:08:38 Quit GeekShadow (Ping timeout: 252 seconds)
21:08:52n1sgrr, now the sim decided to include a system header instead of one from our libc, wth?
21:10:29 Join wodz [0] (
21:11:18 Quit Unhelpful (Ping timeout: 260 seconds)
21:16:02 Quit y4n (Quit: only amiga makes it possible)
21:21:29 Join evilnick [0] (~evilnick@rockbox/staff/evilnick)
21:24:08 Quit bertrik (Read error: Connection reset by peer)
21:24:19 Quit remlap (Ping timeout: 246 seconds)
21:24:30 Join remlap [0] (
21:27:17 Join pamaury [0] (
21:27:17 Quit pamaury (Changing host)
21:27:17 Join pamaury [0] (~quassel@rockbox/developer/pamaury)
21:28:06***Saving seen data "./dancer.seen"
21:29:33 Join domonoky [0] (~Domonoky@rockbox/developer/domonoky)
21:31:59 Join GeekShadow [0] (
21:33:32 Quit Rower85 (Quit: Hmmm...)
21:34:18saratogaso i guess one relatively easy way to make our resampling a lot better would be to use the EQ's lowshelf filter to upsample 2 or 4x, and then do linear interpolation at the higher sampling rate
21:36:36saratogaso for 48>44.1 you'd turn it into 192>44.1 which is only a little slower but much more accurate since the sampling rates are so close
21:40:07saratogathat paper suggests that if we wanted to further improve the results, we could get another 30dB of stop band rejection by going from linear to a 4 point spline fit, at the cost of just an extra 8 multiplies per sample
21:40:31saratogaso theres a nice incremental series of steps we could take
21:41:17saratogaand of course reuse the oversampling code
21:45:13 Quit remlap (Quit: Leaving.)
21:47:03 Quit perrikwp (Read error: Connection reset by peer)
21:47:21 Join bertrik [0] (
21:47:22 Quit bertrik (Changing host)
21:47:22 Join bertrik [0] (~bertrik@rockbox/developer/bertrik)
21:48:17 Join perrikwp [0] (
21:52:00 Join Unhelpful [0] (~quassel@rockbox/developer/Unhelpful)
22:00:22 Quit factor (Read error: Connection reset by peer)
22:08:27 Join factor [0] (
22:17:40 Join fml [0] (
22:19:06 Join remlap [0] (
22:20:13fmlIsn't a closing parenthesis misplaced in the last patch to the manual (IMX233_install.tex: ... \opt{fuzeplus}{(``''} above) ...)?
22:20:46fmlI can't check it now for real, but just from reading the patch I'd say it's misplaced.
22:21:44wodzlooks odd
22:22:06marazindeed it does
22:23:30fmlI can't tell what's wrong exactly. Maybe it's just the } that should be moved to the right. Someone who knows how it works should judge.
22:23:44 Quit petur (Quit: here today, gone tomorrow)
22:27:02 Join Strife89 [0] (~Strife89@
22:31:43 Quit fml (Quit: ChatZilla [Firefox 12.0/20120420145725])
22:36:46 Quit n1s (Quit: Ex-Chat)
22:46:39 Join SeaWeed [0] (
22:47:06 Part SeaWeed
22:49:24 Quit GeekShadow (Ping timeout: 248 seconds)
22:50:00 Join stoffel [0] (
22:57:41 Quit Strife89 (Ping timeout: 276 seconds)
22:58:49 Quit pamaury (Remote host closed the connection)
22:59:04preglowi really wouldn't use any of the eq filters for resampling
22:59:16preglownone of them are straight lowpass filters
22:59:26preglowfir filtering is the way to go for high quality anyway
23:01:06preglowi'd still recommend having a look at the resampling code that comes with speex and adapting that
23:02:05 Quit evilnick (Read error: Connection reset by peer)
23:02:09 Join evilnick| [0] (~evilnick@rockbox/staff/evilnick)
23:03:45 Quit evilnick| (Read error: Connection reset by peer)
23:03:49 Join evilnick [0] (~evilnick@rockbox/staff/evilnick)
23:13:05 Join GeekShadow [0] (
23:15:23 Join Keripo [0] (
23:16:43 Quit stoffel (Remote host closed the connection)
23:22:12 Quit wodz (Ping timeout: 245 seconds)
23:25:34 Quit MCrase (Quit: MCrase)
23:28:08***Saving seen data "./dancer.seen"
23:36:28 Quit evilnick (Remote host closed the connection)
23:36:32saratogapreglow: FIR is slow though
23:37:03saratogathe idea was to use a combination of oversampling and polynomial interpolation since its very fast and still has relatively good output, at least for 16 bit audio
23:48:51 Quit Keripo (Quit: Leaving.)
23:58:46 Join MethoS- [0] (~clemens@

Previous day | Next day