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).

Notice: Only Gecko based browsers prior to FF4 support the multipart/mixed "server push" method used by this log reader to auto-update. Since you do not appear to use such a browser, this page will simply show the current log, and not automatically update.

#rockbox log for 2010-07-07

00:03:23bieberI'll have to look into deb packaging, see if I can maybe get a PPA for it, now that the editor's starting to be useful
00:05:05CIA-6New commit by 03bluebrother (r27325): Add Theme Editor application icon for Windows
00:05:31bluebrotherbieber: the icon is now visible on all platforms :)
00:05:38bieberAwesome :D
00:06:09bieberDo you have a big version for OSX? I've got an XCF in the source tree that's significantly bigger than the .png being used for window icons, if you need it
00:06:26 Quit ender` (Quit: Perfection is reached, not when there is no longer anything to add, but when there is no longer anything to take away.-- Antoine de Saint-Exupery)
00:07:09 Join togetic [0] (~togetic@unaffiliated/ibuffy)
00:07:13CIA-6r27325 build result: All green
00:08:00 Quit jgarvey (Quit: Leaving)
00:09:19 Quit domonoky (Read error: Connection reset by peer)
00:09:23pixelmabluebrother: if you give me the link, I'll try tomorrow... errr.. later today ;)
00:10:00bluebrotherpixelma: usual location is −− all updates will go there (except the official stuff of course ;-)
00:10:48bluebrotherbieber: yes, I've started with the xcf. OSX can go up to as high as 512x512px. I've stopped at the resolution the xcf has.
00:11:08 Join wincent [0] (
00:11:08bluebrotherI was also thinking about creating a svg of the icon. Can be useful on Linux.
00:12:00 Quit wincent_balin (Ping timeout: 245 seconds)
00:18:51 Join zack [0] (~chatzilla@
00:19:14 Nick zack is now known as ZGruk (~chatzilla@
00:19:33 Quit ZGruk (Client Quit)
00:25:29 Join JdGordon [0] (~jonno@rockbox/developer/JdGordon)
00:30:56 Quit evilnick_B (Quit: Page closed)
00:31:58 Quit kugel (Remote host closed the connection)
00:33:18 Join halmi [0] (
00:35:45Buschelstripwax: if find some time to debug -> this patch experiments with replacing libfaad's fft with the codeclib's. the main three changes are encapsulated by "//AB: use codeclib's fft" (1st declaration of FFTComplex variable, 2nd filling this variable and calling codeclib's fft, 3rd fprint-debugging).
00:36:05pixelmabluebrother: thanks, will try it
00:36:35Buschelstripwax: I do not understand why the results of libfaad's and codeclib's fft are different (even researched bitreversing and conjugated complex input).
00:37:09Buschelstripwax: would be great if you could help as I assume there might be a large speed difference (~10 MHz)
00:38:59Buschelsaratoga/mt/lear/n1s: any other codec dev might be interested in this as well...
00:40:18Buschelhave to quit now, will be happy to read any hints in the log
00:40:22 Quit Buschel ()
00:40:34stripwaxBuschel - quick q - your diff calls both ff_fft_calc_c and fft_dif, is that right?
00:41:15 Join t0rc [0] (~t0rc@unaffiliated/t0rc/x-5233201)
00:42:04stripwaxso at a guess, libfaad uses regular bitreverse ordering , which differs from our codeclib (which uses the blended split-radix digitreverse ordering)
00:42:28stripwaxbut i know nothing about libfaad. but if the diff is assuming that our codeclib uses regular bitreverse ordering, then that could explain why you are getting diffs
00:42:31stripwax^ all for the log
00:43:52stripwaxfortunately we have a digitreverse table in codelib too, called revtab I think. just shift it right to leave the remaining significant bits. e.g. index = (revtab[i])>>(12-nbits)
00:46:38***Saving seen data "./dancer.seen"
00:46:42 Quit halmi (Read error: Connection reset by peer)
00:47:43 Join BHSPitMonkey [0] (~stephen@unaffiliated/bhspitmonkey)
00:51:48 Quit JdGordon (Ping timeout: 240 seconds)
00:59:44 Join anewuser [0] (anewuser@unaffiliated/anewuser)
01:16:40 Quit bertrik (Quit: :tiuQ)
01:21:33 Quit stripwax (Quit:
01:22:29 Quit stoffel (Ping timeout: 265 seconds)
01:24:54 Quit jennifur (Ping timeout: 265 seconds)
01:26:35 Join Horschti [0] (~Horscht2@xbmc/user/horscht)
01:28:36 Quit Horscht (Ping timeout: 240 seconds)
01:36:40 Join JdGordon [0] (3a601b9e@gateway/web/freenode/ip.
01:37:13 Join jennifur [0] (
01:37:39 Quit jennifur (Read error: Connection reset by peer)
01:45:44 Join jennyf [0] (
01:47:11 Quit jennyf (Client Quit)
01:59:45 Quit robin0800 (Remote host closed the connection)
02:05:08 Join evilnick_ [0] (
02:20:41 Quit evilnick_ (Ping timeout: 265 seconds)
02:27:18 Join DerPapst1 [0] (
02:29:22 Quit DerPapst (Ping timeout: 260 seconds)
02:32:57 Quit t0rc (Quit: Leaving)
02:39:47 Quit Xerion (Read error: Connection reset by peer)
02:40:31 Join Xerion [0] (
02:46:40***Saving seen data "./dancer.seen"
02:48:30 Quit JdGordon (Quit: Page closed)
02:50:53 Quit BSOD90 (Ping timeout: 260 seconds)
02:57:34 Join BSOD90 [0] (
03:00:21 Join JdGordon [0] (3a601b9e@gateway/web/freenode/ip.
03:04:19 Quit merbanan (Ping timeout: 240 seconds)
03:08:50 Quit Xerion (Read error: Connection reset by peer)
03:09:34 Join Xerion [0] (
03:19:46 Quit togetic (Ping timeout: 551 seconds)
03:25:40saratogaBuschel: you're probably already aware of this since i keep talking about it, but rather then optimize the dct4_kernel function, you can package all the data into an MDCT and use our mdct
03:26:09saratogasince the MDCT is still a DCT, with rotation you can get the same result as the dct4_kernel with a LOT less code
03:26:35saratogathe ffmpeg version is 19 lines long :)
03:27:45saratogathey basically just multiply times the SBR window coefficients, reshuffle to get into the right order for an MDCT, and then call the MDCT
03:29:59saratogahmm though looking more carefully at dct4_kernel I guess its not tremendously different
03:30:41saratogasince it basically just rewindows and repackages for an FFT of size N/2
03:34:40saratogai guess its pretty much the same, just with the mdct you have to write some fixed point code, and with dct4kernel you have to optimize some code :)
03:43:53 Quit Xerion (Read error: Connection reset by peer)
03:45:56 Quit S_a_i_n_t (Ping timeout: 240 seconds)
03:47:18 Join S_a_i_n_t [0] (S_a_i_n_t@
03:48:04 Join Xerion [0] (
03:48:46 Join fdinel [0] (
03:51:26 Join jennifur [0] (
03:53:18 Quit S_a_i_n_t (Read error: Connection reset by peer)
03:56:17 Join togetic [0] (~togetic@unaffiliated/ibuffy)
03:57:31 Join S_a_i_n_t [0] (S_a_i_n_t@
04:06:31 Join Simran [0] (4673bb7c@gateway/web/freenode/ip.
04:07:44jennifurhi Simran
04:08:34 Join t0rc [0] (~t0rc@unaffiliated/t0rc/x-5233201)
04:09:21 Quit S_a_i_n_t (Ping timeout: 265 seconds)
04:11:06 Join S_a_i_n_t [0] (S_a_i_n_t@
04:16:35 Quit Strife89 (Quit: Bed - before midnight, for once. XD)
04:16:49 Quit pixelma (Disconnected by services)
04:16:50 Join pixelma_ [0] (quassel@rockbox/staff/pixelma)
04:16:51 Quit amiconn (Disconnected by services)
04:16:54 Join amiconn_ [0] (quassel@rockbox/developer/amiconn)
04:17:05 Nick pixelma_ is now known as pixelma (quassel@rockbox/staff/pixelma)
04:17:13 Nick amiconn_ is now known as amiconn (quassel@rockbox/developer/amiconn)
04:17:18 Quit Simran (Quit: Page closed)
04:25:47 Quit Barahir_ (Ping timeout: 240 seconds)
04:27:59 Join Barahir [0] (
04:34:51 Quit TheSeven (Ping timeout: 245 seconds)
04:38:56 Join TheSeven [0] (~TheSeven@rockbox/developer/TheSeven)
04:39:18 Join DerPapst [0] (
04:41:31 Quit DerPapst1 (Ping timeout: 245 seconds)
04:46:42***Saving seen data "./dancer.seen"
05:09:23 Join kramer3d [0] (~kramer@unaffiliated/kramer3d)
05:11:16 Quit DerPapst (Quit: Leaving.)
05:30:10 Quit panni_ (Quit: ( :: NoNameScript 3.81 :: ))
05:39:18 Join clochette [0] (
05:39:37 Quit Horschti (Quit: Verlassend)
05:40:41 Quit clochette (Client Quit)
05:53:01 Quit Zarggg (Read error: Connection reset by peer)
05:53:08 Join Zarggg [0] (
05:53:56 Quit Zarggg (Read error: Connection reset by peer)
06:00:33 Join n17ikh [0] (
06:09:07 Join bzed_ [0] (
06:09:35 Nick bzed_ is now known as Guest65862 (
06:09:37 Quit t0rc (Quit: Leaving)
06:10:03 Quit bzed (Read error: Operation timed out)
06:16:52 Quit kramer3d (Quit: Leaving)
06:30:35 Quit jennifur (Ping timeout: 260 seconds)
06:33:16 Quit fdinel (Quit: Miranda IM! Smaller, Faster, Easier.
06:35:53 Join jennifur [0] (
06:35:53 Quit jennifur (Client Quit)
06:39:48 Quit scorche` (Ping timeout: 240 seconds)
06:40:25 Join sandra_f [0] (
06:42:06 Quit sandra_f (Client Quit)
06:45:35 Quit anewuser (Quit: Sixth edition of 25 samples + 48 hours + detroit party (july 31) (maybe streamed online!))
06:46:45***Saving seen data "./dancer.seen"
06:52:40 Join scorche [0] (~scorche@rockbox/administrator/scorche)
07:11:21 Join froggyman [0] (~me@unaffiliated/froggyman)
07:14:58 Join bieber_ [0] (
07:19:32 Join stoffel [0] (
07:32:12 Join esperegu [0] (~quassel@
07:38:05 Join Buschel [0] (
07:41:58 Quit togetic (Ping timeout: 260 seconds)
07:50:33 Quit liar (Ping timeout: 240 seconds)
07:52:19 Join liar [0] (
07:55:25 Join togetic [0] (~togetic@unaffiliated/ibuffy)
07:57:52 Join einhirn [0] (
08:00:04 Join Zarggg [0] (
08:11:19 Join bmbl [0] (~Miranda@unaffiliated/bmbl)
08:14:33 Quit binaryhermit (Read error: Connection reset by peer)
08:14:53 Join binaryhermit [0] (
08:17:05CIA-6New commit by 03bieber (r27326): Theme Editor: Moved volume options in device configuration panel
08:18:28 Join mitk [0] (~mitk@
08:18:49CIA-6r27326 build result: All green
08:22:44 Join lauanana [0] (
08:24:01 Quit krazykit (Ping timeout: 260 seconds)
08:24:11 Quit lauanana (Client Quit)
08:25:53 Join krazykit [0] (~kkit@
08:26:50 Join ender` [0] (
08:33:19Buschelsaratoga/stripwax: thanks for the hints you gave last night. just updated FS #11461. codeclib's fft is used now. further optimization to follow. decoding speed af aac-he is ~107 MHz now (svn 130 MHz) −− on PP5022
08:37:49 Quit Buschel (Ping timeout: 264 seconds)
08:39:40 Quit linuxguy3 (Ping timeout: 276 seconds)
08:43:44 Quit BHSPitMonkey (Remote host closed the connection)
08:46:48***Saving seen data "./dancer.seen"
08:50:32CIA-6New commit by 03bieber (r27327): Theme Editor: Made all lines of text render as a single graphic, viewport size limits now enforced on text width
08:50:55 Quit bieber_ (Remote host closed the connection)
08:52:03CIA-6r27327 build result: All green
08:53:58S_a_i_n_tbieber: I'm not sure I understand the "viewport size limits now enforced on text width"
08:54:49 Quit binaryhermit (Ping timeout: 265 seconds)
08:58:55bieberS_a_i_n_t: Up until now, if text exceeded the width of the viewport it would just keep on spilling out
08:59:06bieberNow it cuts off like it's supposed to, I just need to implement scrolling
08:59:46S_a_i_n_tAh right, the way I read it was that "text can't exceed the width of the viewport", just checking ;)
08:59:46 Join binaryhermit [0] (~binaryher@
09:00:51 Join Zagor [0] (bjst@rockbox/developer/Zagor)
09:09:54CIA-6New commit by 03wodz (r27328): HD200 - fix misleading comment in system-hd200.c
09:10:04 Quit bmbl (Read error: Connection reset by peer)
09:11:32CIA-6r27328 build result: All green
09:13:23 Join flydutch [0] (
09:13:34 Join wodz [0] (
09:15:16wodzbattery_bench on MPIO looks quite good - old result with cutoff at 3500 mV was 4:14:10, new result with cutoff at 3650 mV is 7:10:59
09:18:03bieberMeh, it's getting slow on large themes
09:18:41wodzI think I'll do some benches with CPUFREQ_NORMAL set to 33.8688 MHz instead of 45.1584. The most popular codecs should run still unboosted
09:19:34S_a_i_n_tbieber; Large as in "lots of text in the WPS/SBS/FMS etc. files" or, large as in "lots of skinbuffer usage"?
09:19:52bieberI'm sure it's text
09:20:13bieberIt looks like I'm not going to be able to get away with my grossly inefficient text rendering ;)
09:20:50S_a_i_n_thehehe, some of my WPS are in the region of 3~400 lines ;)
09:20:55S_a_i_n_tShould be interesting :P
09:21:55bieberNo worries, I'm fixing this nonsense before I move on to anything else
09:29:42 Quit scorche (Disconnected by services)
09:29:53 Join scorche` [0] (~scorche@rockbox/administrator/scorche)
09:33:37simonrvnyou can imagin how slow it got on an old athlon and radeon 9000 ;)
09:34:46 Join swilde [0] (
09:35:30S_a_i_n_tThe GFX card really shouldn't have anything to do with that...
09:36:07JdGordonit would if gnu would hurry up and get gcc going on the gpu :)
09:36:09S_a_i_n_tI've run it on an AMD 1.10Ghz with an nVidia Vanta16 ;)
09:36:14 Join petur [0] (d408b802@rockbox/developer/petur)
09:36:46bieberNo, it's the fact that every time the scene is rendered (i.e. every time you change the text ) it's loading every font from disk and rendering every line of text once for each character in the line ;)
09:37:28bieberSooo, I need to do some caching, and find a way to delay rendering text until I know how much is going to be on the line
09:38:24JdGordondont bother rendering unless the time between key presses is large maybe?
09:38:29JdGordonand keep fonts in memory
09:38:39CIA-6New commit by 03bieber (r27329): Theme Editor: WorkingWorking on speeding up the renderer, replaced for-loops with faster method call
09:39:15bieberJdGordon: The problem with delaying rendering is that I want to be able to have a timer running rendering at a reasonable framerate, so you can see animations and such
09:39:28 Quit rado1 (Read error: Operation timed out)
09:39:47bieberI'm hoping that caching fonts and just rendering each line once will significantly speed everything up
09:40:14CIA-6r27329 build result: All green
09:40:34JdGordonjust disable that timer when the user is typing?
09:41:04JdGordonyou sholdnt need to hit the disk more than once per object
09:41:42bieberI mean when the timer is running, I want to have it rendering significantly faster than the user could change the text anyways, so just delaying things isn't really an option
09:42:13bieberBesides, I have no doubt that constantly reading the font from disk has to be the problem. It was handling just as zippy as could be before I started doing that
09:43:39JdGordonit would be pretty funny if a PC couldnt keep up with the ipod drawing skins :)
09:43:48 Join Buschel [0] (
09:46:29 Join bertrik [0] (~bertrik@rockbox/developer/bertrik)
09:46:36 Join rado1 [0] (
09:48:16bieberOkay, got text rendering delayed until the end of the line without any significant speedup, it's gotta be disk access
09:49:02CIA-6New commit by 03bieber (r27330): Theme Editor: Delaying rendering of text lines until newline reached
09:50:31CIA-6r27330 build result: All green
09:50:36wodzload fonts to memory PC has plenty of
09:50:52wodzdisk is sloooow
09:51:45 Join Rob2222 [0] (
09:51:59wodzand moving pointers is muuuuuch faster than seeks
09:54:52 Quit Rob2223 (Ping timeout: 245 seconds)
09:55:33 Quit JdGordon (Quit: Page closed)
09:56:38simonrvnyeh, and i've got slow disks
09:57:00simonrvni imagine for anyone in a similar position it would be slow too
09:58:28bieberDisks in general are just really, really slow compared to memory
09:59:09bieberI should have known better than to think I could get away with loading KB of data from disk multiple times a second
10:00:14 Quit Buschel (Ping timeout: 260 seconds)
10:03:23wodzmemmap() should do
10:04:16 Join pamaury [0] (~quassel@rockbox/developer/pamaury)
10:05:38bieberwodz: memmap?
10:06:24bertrikDon't worry about performance too much unless it really gets in the way
10:06:50bieberOh, interesting. It's apparently POSIX only, though, which would kill my Windows compatibility
10:07:08bieberI just whipped up a font cache, and it's introduced a new bug without making rendering _that_ much better :/
10:07:26wodzbieber: I think there is similar call on windows
10:09:14wodzbieber: This may be interesting for You also:
10:12:16wodzbieber: And finally windows equivalent
10:12:57bieberIt's interesting stuff, but I don't want to resort to platform-dependant ifdefing for simple caching
10:13:53wodzfor few hundreds of kB it is not worth the effort probably
10:17:51gevaertsIsn't there something like Qmmap?
10:21:33bieberThere doesn't seem to be
10:21:59bieberI just put together a simple class to implement a global font cache, but it's not helping as much as I thought it would
10:23:48bieberOh wait, that's probably because I'm having the font class load the info from the cache, and then load it from the file all over again
10:24:29 Quit binaryhermit (Ping timeout: 240 seconds)
10:25:28bieberOh, thanks
10:25:50gevaertsSeems a bit simplistic in that I don't see anything that wraps madvise(), so you're relying on OS heuristics, but I'd expect OS heuristics to be just fine for smallish files (also for plain file reading actually...)
10:27:24mitkbieber: Just fyi. Open Document dialog doesn't show directories starting with . (dot). Also doesn't show files with no extension even when "All files" option is selected.
10:27:42biebermitk: Thanks, I'll have to have a look at that
10:28:20bieberI think you can right-click in Qt's file dialog to show hidden files, and the lack of no-extension files should be an easy fix
10:31:16 Join hamish_ [0] (~hamish@
10:31:46S_a_i_n_tIIRC Qt does have the "right click - show hidden files" thing, yes.
10:31:54mitkbieber: Yeah it works for hidden files.
10:32:06 Quit hamish_ (Client Quit)
10:33:20bieberAny recommendations for a Linux code profiler?
10:33:41mitkMaybe it is good to show .rockbox directory without right clicking. Especially on simdisk.
10:34:07 Join DerPapst [0] (
10:36:34bertrikamiconn, for the extra-keypress-required-after-exiting-a-plugin-problem, maybe we should just remove the button_clear_queue call (called just after returning from a plugin) and see what happens.
10:41:24wodzbieber: Valgrind may also be interesting
10:41:39CIA-6New commit by 03bieber (r27331): Theme Editor: Fixed rendering bug that caused text in sublines not to appear, implemented a global font cache
10:43:23CIA-6r27331 build result: All green
10:43:59biebergprof was insanely helpful
10:45:07bieberApparently it's actually rendering the text that's taking up most of the program's time, not font loading
10:45:17bieberSo it's the images I should be caching
10:46:49***Saving seen data "./dancer.seen"
10:48:12 Join JdGord [0] (~jd@
10:54:42 Quit rvvs89 (Changing host)
10:54:42 Join rvvs89 [0] (robotnik@pdpc/supporter/base/rvvs89)
11:06:10 Join dfkt [0] (dfkt@unaffiliated/dfkt)
11:08:01wodzwhere from codec gets master clock signal on ipod mini? Is it generated by PP SoC (like on coldfires) or is it independent oscillator circuit?
11:09:13 Quit esperegu (Remote host closed the connection)
11:09:56 Join esperegu [0] (~quassel@
11:11:36 Quit JdGord (Quit: Bye)
11:16:59bieberMeh, even with fonts and images cached it's hanging a little on rapid text edit :/
11:19:04 Join evilnick_ [0] (
11:23:51 Quit evilnick_ (Remote host closed the connection)
11:24:51*S_a_i_n_t wonders how rapidly people are going to be editing said text?
11:25:25S_a_i_n_tI don't imagine anyone writing WPS syntax at 60wpm ;)
11:29:02bieberEhh, the problem comes when you hold down the backspace key and it hangs
11:29:22bieberFor now I'm just putting a limit on the time between renders
11:29:49 Join Luca_S [0] (
11:32:14 Quit wodz (Quit: Leaving)
11:32:24Luca_Sfunman (for the logs): I have my FuzeV2 again. I'm starting a series of accelerated battery benches with backlight always on and 1) plain svn 2) dynamic cpu freq (latest patch) 3) audio on pllb. 1) is already done (no crashes), 2) just started.
11:33:49CIA-6New commit by 03bieber (r27332): Theme Editor: Implemented caching for rendered text, added profiling info to debug build, added a 500msec delay when rendering after code changes to ...
11:35:22CIA-6r27332 build result: All green
11:41:54 Join dfkt_ [0] (dfkt@unaffiliated/dfkt)
11:44:09 Quit Luca_S (Quit: CGI:IRC)
11:45:02 Quit dfkt (Ping timeout: 264 seconds)
11:45:33 Quit esperegu (Read error: Connection reset by peer)
11:49:13bertrikhm, button_clear_queue() is all over the place, so removing one call from plugin.c is not going to solve a lot
11:49:17 Join esperegu [0] (~quassel@
11:54:01 Quit esperegu (Remote host closed the connection)
11:56:41 Join esperegu [0] (~quassel@
12:09:08 Nick dfkt_ is now known as dfkt (dfkt@unaffiliated/dfkt)
12:12:21 Join watto [0] (~watto@
12:14:36 Join wodz [0] (
12:40:03 Join kramer3d [0] (~kramer@unaffiliated/kramer3d)
12:43:46 Join ucchan [0] (
12:45:46 Quit Galois (Ping timeout: 240 seconds)
12:46:53***Saving seen data "./dancer.seen"
12:51:51 Quit kramer3d (Quit: This computer has gone to sleep)
13:06:24 Quit Unhelpful (*.net *.split)
13:06:24 Quit antil33t (*.net *.split)
13:06:24 Quit jhMikeS (*.net *.split)
13:06:24 Quit literal (*.net *.split)
13:06:24 Quit advcomp2019_ (*.net *.split)
13:06:25 Quit fxb__ (*.net *.split)
13:06:25 Quit rasher (*.net *.split)
13:06:25 Quit lostlogic (*.net *.split)
13:09:38 Join Unhelpful [0] (~quassel@rockbox/developer/Unhelpful)
13:09:38 Join antil33t [0] (
13:09:38 Join jhMikeS [0] (~jethead71@rockbox/developer/jhMikeS)
13:09:38 Join literal [0] (
13:09:38 Join advcomp2019_ [0] (~advcomp20@unaffiliated/advcomp2019)
13:09:38 Join fxb__ [0] (
13:09:38 Join rasher [0] (~rasher@rockbox/developer/rasher)
13:09:38 Join lostlogic [0] (~lostlogic@rockbox/developer/lostlogic)
13:25:33pixelmabieber: for what it's worth, a quick test run of the theme editor (r27323 built by bluebrother) went fine on MacOS 10.4
13:35:32CIA-6New commit by 03uchida (r27333): plugin api: add the following functions. ...
13:37:37CIA-6r27333 build result: All green
13:42:56wodzWaiting for battery_bench to finish I started disassembling of MPIO HD300 firmware (bigger brother of my HD200). Firmware for both players is build from the same source base for sure.
13:46:08 Join Dannyboy [0] (
13:46:34Dannyboycould someone look at this
13:46:41Dannyboyand explain to me how to use it
13:46:45Dannyboyi dont understand the instructions
13:46:50Dannyboyi want to create a modified firmware
13:51:47 Join t0rc [0] (~t0rc@unaffiliated/t0rc/x-5233201)
13:52:39mitkDannyboy: Did you tried this ->
13:56:29wodzmitk: maybe I misunderstood but he don't want to install rockbox but to modify firmware
13:57:47mitkMaybe you, may be me :) Only Dannyboy knows it.
14:00:00Dannyboyi want to modify the firmware
14:00:07Dannyboyas i installed rockbox onto ram
14:00:19Dannyboyand i have to use modified original firmware to restore it
14:00:50 Join panni_ [0] (
14:01:00wodzyou mean rom right?
14:01:55Dannyboyif someone could make me a bin file using the 1.66 us firmware
14:04:29wodzHonestly I don't get it. If you have rockbox in rom and you want to restore player you need to flash original firmware. (unless you want to have rb still installed but not in rom)
14:05:38Dannyboyin order to flash original firmware dont you need to boot from disk
14:05:50Dannyboywhen i try to flash original untouched firmware i get an error message
14:06:16Dannyboysaying not an original firmware file
14:06:54wodzIt shouldn't matter. Rockbox copies itself into ram and runs from there irrespective if it is loaded from disk or from flash
14:07:26CIA-6New commit by 03uchida (r27334): text viewer: fixed the following bugs. ...
14:07:26Dannyboyso all i need to do is get firmware file which is .hex
14:07:28Dannyboyand rename to .bin?
14:09:13CIA-6r27334 build result: All green
14:09:22wodzI would wait and ask somebody skilled in irivers particulary.
14:09:58Dannyboyyes, ive been asking this question a few days
14:10:14 Quit mitk (Quit: Leaving)
14:10:35 Join merbanan [0] (
14:11:49wodzSo maybe ask this question on ML? Some devs do not follow IRC discussions
14:15:46amiconnwodz: Not entirely true. On some targets it is possible to have rockbox directly running from ROM. Code and .rodata, that is - it still has to copy .data and iram stuff, and use dram for .bss
14:17:09wodzamiconn: I see - isn't it painfully slow runing this way?
14:17:20wodzDannyboy: Mailing List
14:17:26Dannyboyoh ok
14:19:11Tornewodz: why should it be? on olde rplatforms, NOR flash is not significantly slower than SDRAM to read
14:20:26 Quit rado1 (Ping timeout: 264 seconds)
14:20:51wodzTorne: maybe You know what is delivering MCLK signal to the codec in ipod mini? Is it PP SoC or is it external oscilator circuit?
14:20:58Torneno idea.
14:23:32amiconnDepends. On SH1 it's about twice as slow, as the EEPROM is 8 bit and the DRAM is 16 bit
14:23:58amiconnOn coldfire you probably don't see much difference for code. or data I don't know
14:38:35 Join rado1 [0] (
14:40:09AlexPI've never actually done a test, but running my H140 out of ROM, or using the RAM image doesn't feel any different
14:42:17CIA-6New commit by 03uchida (r27335): plugin api: new functions set the end of the structure. thanks to Frank Gevaerts.
14:44:17CIA-6r27335 build result: All green
14:45:50 Join Galois [0] (
14:46:55***Saving seen data "./dancer.seen"
14:48:13 Quit sneak (Ping timeout: 245 seconds)
14:49:37 Join JdGordon [0] (~jonno@rockbox/developer/JdGordon)
14:52:56 Quit t0rc (Remote host closed the connection)
14:54:17 Join sneak [0] (debian-tor@gateway/tor-sasl/sneak)
15:00:41 Quit ucchan (Quit: Leaving...)
15:00:41 Quit antil33t (Read error: Connection reset by peer)
15:00:47 Join antil33t [0] (
15:08:45 Join teru [0] (
15:13:29 Join BSOD90_ [0] (
15:15:44 Quit JdGordon (Ping timeout: 245 seconds)
15:15:45 Quit BSOD90 (*.net *.split)
15:15:45 Quit w1ll14m (*.net *.split)
15:15:45 Quit bluebrother (*.net *.split)
15:15:45 Quit mk429_ (*.net *.split)
15:15:45 Quit simabeis (*.net *.split)
15:17:05 Join bluebrother [0] (~dom@rockbox/developer/bluebrother)
15:17:55 Join simabeis [0] (
15:19:06 Join linuxguy3 [0] (
15:21:32 Join w1ll14m [0] (
15:22:35 Join mk429 [0] (
15:27:24 Join komputes [0] (~komputes@ubuntu/member/komputes)
15:29:44 Join jgarvey [0] (
15:34:04 Join tchan [0] (~tchan@lunar-linux/developer/tchan)
15:36:50 Quit komputes (Ping timeout: 260 seconds)
15:37:20 Join peschmae [0] (
15:42:22 Part froggyman
15:42:51 Quit peschmae (Quit: CGI:IRC (Ping timeout))
15:43:02 Quit r0b- (Ping timeout: 276 seconds)
15:43:33 Join Kitr88 [0] (
15:43:46 Join r0b- [0] (
15:43:49 Quit Kitar|st (Read error: Connection reset by peer)
15:48:03 Quit Kitr88 (Ping timeout: 264 seconds)
15:50:13 Join komputes [0] (~komputes@ubuntu/member/komputes)
15:53:29 Join Kitar|st [0] (
15:57:52 Join freddyb [0] (
16:09:06 Quit einhirn (Quit: Miranda IM! Smaller, Faster, Easier.
16:09:15 Join t0rc [0] (~t0rc@unaffiliated/t0rc/x-5233201)
16:17:25 Quit Dannyboy (Quit: CGI:IRC (Ping timeout))
16:30:06 Quit bluebrother (Disconnected by services)
16:30:08 Join bluebroth3r [0] (~dom@rockbox/developer/bluebrother)
16:32:58 Join ansuz [0] (
16:37:21 Join bor_ka [0] (
16:44:39 Join kramer3d [0] (~kramer@unaffiliated/kramer3d)
16:46:58***Saving seen data "./dancer.seen"
16:49:29 Join Strife89 [0] (
16:55:03 Quit kramer3d (Quit: Leaving)
17:00:09S_a_i_n_tAnyone want to close out "FS #11449" as "Not a bug, RTFM"?
17:02:53S_a_i_n_tFS #11426 can also be closed now.
17:03:03S_a_i_n_tFixed by JdGordon yesterday.
17:03:13TorneS_a_i_n_t: i would've expected a wps to work from anywhere, tbh ;)
17:03:45Tornei mean, it's not particularly critical, i guess.
17:03:46S_a_i_n_tIt needs to be in the WPS dir, pretty sure the manual states this.
17:04:00S_a_i_n_tIt needs to be in the same dir as the wps folder also
17:04:08S_a_i_n_t(image folder for the wps file)
17:05:17 Join chrisb [0] (
17:05:28chrisbdo you know if the line-out works for the iPods which can take rockbox?
17:05:50Torneyes, all the ipods with a dock connector have line out on the dock
17:05:58Tornethere's a setting to enable/disable it in rockbox (because it takes some power)
17:06:08chrisbTorne: thanks
17:06:18Torneyou may also need to turn on the accessory power depending what you are connecting it to
17:06:30chrisbTorne: understood
17:06:46chrisbi have a Belkin dock with external power, so i can do that
17:07:11TorneYah. It might still need accessory power enabled, some docks seem to use that to detect the ipod being present
17:07:28Tornewe have an option because again, it takes some power even when nothing is connected
17:07:47chrisbnow i just have to find a way to replace the ipod Classic 6 Gen with something that will fit in that dock and take rockbox
17:08:03 Quit teru (Quit: Quit)
17:08:27chrisbthe tigerdirect description never said 6th Gen, just 2G, so i thought i would be OK
17:08:35chrisbfor rockbo
17:09:04S_a_i_n_t2G as in "GB"?
17:09:14S_a_i_n_tOr, as in, "totally mistaken"?
17:09:29Tornethe second gen ipods are from 2002, so if you thought you were buying one that wasn't second hand you were worng ;)
17:09:47Tornealso, the 1g/2g are not particularly great, they're huge and bricklike and can't shut off.
17:10:20S_a_i_n_tAnd have weird controls
17:10:20 Join Jaykay [0] (
17:10:51Tornenone of the ipods currently available to buy new work with rockbox, anyway
17:11:30 Part Zagor
17:11:58chrisbS_a_i_n_t: the description said 2G as in "2nd Gen" and had that written on one of the photos
17:12:24chrisbS_a_i_n_t: but they shipped a brand new 6th Generation or 7th Generation
17:12:42S_a_i_n_tthere's a 7th now?
17:12:52chrisbsee for yourself
17:13:29S_a_i_n_tIt really pays to investigate the generation of the iPod yourself (assuming the images are correct)
17:13:44S_a_i_n_tApple has a nice thing on their site for identifying iPod models
17:13:51chrisbbut both labels 2nd Generation and Gen 7 iPod Classic are on the box i received
17:14:20TorneYes, that's clearly described as an ipod classic
17:14:20S_a_i_n_tIFAIK, there it *no* 7th Gen
17:14:27S_a_i_n_tAnyone feel free to correct me...
17:14:37TorneS_a_i_n_t: some people seem to think the later classics are different enough to warrant being called 7th gen
17:14:44chrisbi hate apple. i dropped my guard, and this is the first apple product i ever bought
17:14:53Tornei don't think we've found any hardware differences yet that are interesting, but i also don't think we have looked very hard
17:14:59S_a_i_n_tI go by Apple's description of Generation
17:15:01Tornesince we can't run on the original classics yet
17:15:06S_a_i_n_tAFAIK it stops at 6 now
17:15:08chrisband i hate it, i hate the appleness of it and i hate its hostility to gnuness and hackability
17:15:09TorneApple don't *have* generatoins for most of them :)
17:15:20TorneUsing only apple's terminology gets very verbose very fast
17:15:47 Join AzureWurk [0] (
17:15:55S_a_i_n_tchrisb: Hate to say it dude, but yes, the site you linked very clearly states it is a "classic"
17:16:08S_a_i_n_tie. 6Th Gen
17:16:20Torne"classic 2nd gen" would be what other people call "7th"
17:16:22chrisbS_a_i_n_t: i can take my lumps, i know i screwed up
17:16:34Tornelike "mini 2g" or "nano 2g" is not the same as 2g
17:17:38 Join MethoS- [0] (~clemens@
17:18:18*S_a_i_n_t loves the "A closer look" section of that page that shows the same image, but flipped horizontally ;)
17:18:23chrisbin my defense, it did not say 6th Gen, it said 2G at the time I ordered, and it has a 2nd Gen label on the box, with a Gen 7 label slapped above it
17:18:50S_a_i_n_tNone of which are correct, nice ;)
17:19:06S_a_i_n_tGotta love misleading information... :/
17:19:26TorneS_a_i_n_t: well, it's all consistent with the terminology commonly used..
17:20:03Tornethe problem is that now apple call the new one the classic, people assume that all non-mini/nano/touch ipods are ipod classics
17:20:09S_a_i_n_tHahaha...that site is awesome, lets show the same image, but flip it horizontally to give the impression that they are actually left/right angle views.
17:20:12Torneand thus that the classic is in some way related to the older disk based models
17:20:17Tornewhich it's not, other than looking similar ;)
17:20:55chrisbTorne: bingo, that's the haze i'm emerging from
17:21:39S_a_i_n_tIf all else fails, look at the picture and head to Apples "identify your iPod" page
17:22:00Torneapple's terms only give generation unmbers to a few models
17:22:01S_a_i_n_t(assuming the correct image has been posted)
17:22:12Torneand describes all the rest by.. descriptive stuff and/or release date
17:22:58 Quit panni_ (Read error: Connection reset by peer)
17:24:23 Quit antil33t (Read error: Connection reset by peer)
17:24:30 Join antil33t [0] (
17:25:41chrisbi'll check the pawn shop for a 5G or 5.5G...
17:26:39chrisbthe only lossless format iPod will play is .wav, right? no flac....
17:27:29Torneapple lossless, ALAC
17:27:47Tornebut no flac, no
17:27:50chrisbTorne: ok, thanks
17:28:17chrisbis .wav free or proprietary?
17:29:05TorneWAV is a container, it doesn't really belong to anyone but it's not a published public standard either
17:29:41Tornethe normal format for wav, linear pcm audio, is too trivial for anyone to own i would hope ;)
17:35:05*chrisb looks at and thinks maybe 5G doesn't take rockbox
17:35:43Tornewhat gives you that impression?
17:35:47TorneApple: iPod 1g through 5.5g, iPod Mini and iPod Nano 1g
17:36:31freddybThere's a patch (FS #10386) for adjusting the thumb wheel speed on iPods that's been dormant for a while. I have it working. Is anyone interested in this?
17:36:39chrisbi looked at and did not see one for 5G or 5.5G
17:37:12Tornechrisb: ipod video
17:37:20Tornefreddyb: we have, in the past, explicitly said we don't want that, i think
17:38:14chrisbTorne: ok. no wonder i ended up with a 2nd Gen relabeled as a a Gen 7 iPod
17:38:20chrisbi'm in apple hell
17:38:39Tornechrisb: <- this is the horse's mouth on ipod models
17:38:47Torneyou'll notice that almos tnone of them have a numeric generation :)
17:39:04S_a_i_n_tfreddyb: If it weren't specific to the Video...*I* may be interested in this, but I doubt it has a future in SVN personally
17:39:16Tornealso that when they sold lots of those models, they sold them under names other htan the one listed on that page
17:39:19Tornesooo, yeah
17:39:26S_a_i_n_t(by interested, I mean for my personal builds)
17:39:57 Join bmbl [0] (~Miranda@unaffiliated/bmbl)
17:40:01Tornescroll speed/acceleration shouldn't really be something you have to adjust
17:40:27S_a_i_n_tno, it should "just work", but it's nice to be able to turn it down selectively
17:40:46S_a_i_n_tI had a two stage sensitivity thing going for a while, but I lost the patch
17:40:57S_a_i_n_tSome plugins seem more sensitive than others.
17:41:03TorneThey aren;t.
17:41:11Torneand we've had working acceleration for a long time now
17:41:16S_a_i_n_tnote: *seem*
17:41:40Torneseriously: if it doesn't seem right, then adjust the speed/acceleration params and offer that as a patch :)
17:41:47 Quit merbanan (Ping timeout: 240 seconds)
17:42:07Torne*shrug* i don't personally see the need for it to be user-settable; ther'es an incredibly narrow range that's even useful
17:42:29S_a_i_n_tToning down scroll sensitivity for the virtualy keyboard is what made me persue it in the first place
17:42:46Tornethe virtual keyboard is just worthless on a scrollwheel anyway
17:42:53Tornea much better fix is to replace the virtual keyboard on scrollwheel targets
17:43:02Tornewhich there are several ideas, and one or two patches, for
17:43:29Tornewith a different kind of virtual keyboard, written with scrollwheels in mind
17:43:43Tornethe current one assumes you have up/down/left/right, i.e. a 4-way control of some kind
17:44:16Tornei've got a couple on my list, hang on a sec :)
17:44:48freddybTorne: the reason I went looking for this patch was because the ipod wheel is too fast for my virtual keyboard.
17:45:00S_a_i_n_tAwesome..I'm genuinely interested. I use the v-keyboard a lot
17:45:06TorneS_a_i_n_t: *how*?
17:45:21Tornergardless of scroll speed i find the virtual keyboard totally unusable on ipod
17:45:31S_a_i_n_t"with great difficulty" ;)
17:45:31Tornetyping more than two or three chars makes me want to throw it out the window :)
17:45:44Torneit's why testing frotz was a pain ;)
17:45:51S_a_i_n_tI take it you often hit "menu" and exit the thing?
17:45:57S_a_i_n_t(I used to do that a LOT)
17:46:03Torne"close trapdoor" "take key" was as far as I ever could stand to get into Curses
17:46:09TorneNo, I know the controls
17:46:28Tornebut using scrolling for one axis and the scrollwheel's buttons for the other axis is just impossible ;)
17:46:38S_a_i_n_tI do too...I still make the up/down/left/right mistake very often though
17:46:47Torne <- there are two implementatoins here that are much better
17:46:56Torneone using bank-safe-like use of the wheel
17:46:57S_a_i_n_tAwesome, thanks
17:47:01Torneand one just making the keyboard 1-d and having scrolling only
17:47:10freddybTorne: Thanks!
17:47:25Tornethere are some other ideas around as well, on the forums and as (now-closed) feature requests
17:47:32Tornebut none of those ever actually got implemented
17:47:40Torneit's on my list of "things to look at one day"
17:48:04Tornesince i did port frotz, which uses about a million times as much text input as anything else on rockbox ;)
17:48:41Tornefreddyb: seriously, though: the virtual keyboard moves up and down at exactly the same rate as any list view does
17:49:01Torneso the notion that it's too fast for the keyboard seems bizarre to me, that should mean it's too fast *everywhere*
17:49:16Tornescrolling up or down by one list item should be exactly the same movement as up or down one row on the vkbd.
17:49:45freddybYea, but in the virtual keyboard going too far is a problem, other places it's no big deal.
17:50:23freddybI wrote it on a Fuze, which worked great, but it was unbearable on iPod Video
17:51:58Torneer, it is?
17:52:01S_a_i_n_tshit, sorry
17:52:09TorneHow is it any more of a problem in the virtual keyboard than other contexts?
17:52:18Torne(serious question)
17:52:51S_a_i_n_tI think it annoys me because I *want* the vkeyboard to feel precise
17:52:57S_a_i_n_tI don't feel in control of it
17:53:30freddybMy keyboard patch uses left and right scrolling to move on y and x axis of the grid. Going too far means scrolling all the way back around.
17:53:31S_a_i_n_ton a 176x132 screen its *really* hard to get a specific glyph
17:54:03TorneOh, yes, you are the author of the patches i just linked to ;)
17:54:13S_a_i_n_tand, I have made my own .kbd file which helps a LOT
17:54:17Tornei forgot
17:54:27S_a_i_n_tthe default vkeyboard with userfont is just...impossible
17:54:53S_a_i_n_t(on 176x132)
17:54:57Tornetbh i think the vkeyboard needs to be rethought entirely, not just on scrollwheel targets
17:55:06Tornethere are a *lot* of cases where it's barely usable
17:55:06S_a_i_n_tit's so damn crowded with glyphs I never use./
17:55:15Tornenot just the layout, but the entire implementation
17:55:36S_a_i_n_tlayout is fine, especially sine a .kbd file uses the userfont
17:55:51S_a_i_n_tit needs some creative thinking to make a nice layout
17:56:09S_a_i_n_tbut, I now have qwerty and dvorak layouts for my vkeyboard ;)
17:56:13S_a_i_n_t(pointless I know)
17:57:24freddybWhat don't you like about it on non-scrollwheel targets?
17:57:32Tornefreddyb: still really slow
17:57:38Tornenot as much, but hey
17:57:59Tornewhat can be done better is kinda limited, though, with the tiny number of controls we have.
17:58:22Tornethere are clever schemes on various handheld games devices but they generally have more buttons or at least the buttons laid out better for use by separate fingers
17:58:54freddybI thought about a 2d bisection input for models with 4way directional control but never wrote it.
17:59:40Torneit's not necessarily easy to do diagonals, is the problem with that kind of thing
18:00:10Torneclever layouts are usually 8-direction-based
18:00:33Tornesince that's enough to map a lot of symbols with only two presses per symbol
18:00:55Tornewithout easily-usable diagonals you need more than two
18:01:11simonrvnpie menu? or is that too lame?
18:01:55Tornesimonrvn: well, that's more or less what i'm talking about: it's kinda problematic to select 40-50+ symbols from a pie menu that's only split into four
18:02:28Torneand there are quite a few targets where actually pressing the diagonals would be a pain
18:02:37Torneor maybe even wouldn't work at all, depends on the keypad matrix
18:03:47S_a_i_n_tperhaps using "tap" style navigation would be a tad better for the iPods? (think rockboy)
18:03:55S_a_i_n_tthen, diagonals could be used.
18:04:12TorneWell, that will only work on touch models
18:04:20Tornethe ones with a physical wheel (which is not just ipods) can't do that
18:04:35TorneAlso, if you're going to use the touchpad that way, you can do gestures :)
18:04:39Torneinstead of taps
18:04:44S_a_i_n_twell, yeah...but screw them. We're talking touch-wheels here :P
18:04:54Tornewell, i was talking about scrollwheels in general
18:05:00Tornebut yes, someone suggested gestures
18:05:08Tornedivide the wheel into 8 sectors
18:05:17S_a_i_n_tfor the ones that can use it, it would be awesome.
18:05:18Tornedrag finger across it as if it were a touchpad
18:05:27Tornekinda like, er, clock hands?
18:05:29Torneif that makes sense
18:05:38S_a_i_n_tit does (to me at least)
18:06:22 Quit petur (Quit: shopping)
18:06:28S_a_i_n_tthe only problem with using "taps" would be to stop it from repeating, ie. one "tap" == one movement. Repeating can be DAMN annoying in rockboy
18:07:34S_a_i_n_tI imagine I would quite like to use the vkeyboard in this way, but haven't the faintest idea how to implement it.
18:10:02freddybI don't think the virtual keyboard is ever going to be much faster than texting on a phone...
18:10:30Tornesome people can text really, really fast ;)
18:10:32*Torne shrugs
18:10:37Tornei'm just mentioning some of the ideas i've seen around
18:10:39S_a_i_n_tsure, but DAPs can't send who cares :P
18:11:00freddybWe don't have an app for that?
18:11:26S_a_i_n_tsure, you just need the iPod wifi/3G dock connector ;)
18:15:21 Join domonoky [0] (~Domonoky@rockbox/developer/domonoky)
18:16:02bluebroth3rpixelma: nice to hear :)
18:16:22S_a_i_n_tfreddyb: You seem to know the vkeyboard quite well, how would you feel about looking into the "rockboy-ish" 'tap-to-input' suggestion? I could certainly try help where I can, but my guess is that you know your way around the code much better than I. Even if diagonals were ignored, and it was simple "tap up/down/left/right/ to move one glyph in that direction" I think would feel quite nice.
18:16:44S_a_i_n_tmaking "keystrokes" not repeat would be a major plus IMO
18:18:14freddybI don't know rockboy. Why tap and then direction? Just to get around menu/play?
18:19:27S_a_i_n_tthe iPods don't have enough buttons for rockboy, so the scrollwheel is split into 8 "buttons" that you "tap"
18:19:41S_a_i_n_tTry rockboy on your Video to get a feel for the idea
18:20:04S_a_i_n_ts/ipods/touchwheel iPods/
18:20:50S_a_i_n_tIn theory, this would also make diagonal navigation possible
18:21:19 Quit rado1 (Ping timeout: 265 seconds)
18:22:40TorneS_a_i_n_t: if you entirely disabled repeating i think that would laso get quite tedious :)
18:23:10freddybI see what you mean now.
18:23:14S_a_i_n_tperhaps, but...I'd like to try it.
18:24:03TorneS_a_i_n_t: touching repeatedly without actually pressing the button, to move multiple steps, feels like quite an unnatural movement
18:24:30S_a_i_n_tTorne: So, you don't play rockboy then? ;)
18:24:37pixelmaS_a_i_n_t: FS #11449 is a bug, at least it once worked as he described. WPSs should load when chosen but they won't be remembered (so at next boot it'll load something from .rockbox/wps) - and you'll *probably* find this explanation in the manual...
18:24:39TorneS_a_i_n_t: no, of course not :)
18:24:53Torneand repeating isn't optional, surely?
18:24:59Torneyou need to be able to hold buttons down to play most games :)
18:25:49 Join rado1 [0] (
18:26:11pixelmaTorne: e.g. on the Iaudios you can'T do diagonals, no button combos except with the Power button which is also not desired as holding Power for like 10 seconds is a hard power-off
18:26:29S_a_i_n_tperhaps not optional, but surely it could be ignored for X period of time?
18:26:46TorneS_a_i_n_t: well, it's not *repeating* on rockboy, surely.. but rather holding htem down
18:26:54Torneif you stay touching "up" then it needs to emulate holding down up
18:26:58Tornewithout letting go
18:27:02Torneotherwise many games wouldn't work :)
18:27:18Torneand no, i haven't ever actually used rockboy, maybe it's broken :)
18:27:42mthmm .. libm linking problem again
18:27:48mt(in the sim)
18:27:48S_a_i_n_tIt probably does work as you say, its just incredibly sensitive I guess.
18:28:13Tornewell, i would suspect it can't really help that
18:28:29Tornecapacitive touch sensors *are* really sensitive
18:28:31Torneby definition :)
18:29:33S_a_i_n_tI noticed a while ago that on some days I don;t actually even need to touch the wheel to navigate, but rather hover my finger over it, and others it simply won't work this way.
18:29:53S_a_i_n_tNo idea why it would change, but experimentation suggests it indeed does.
18:30:09S_a_i_n_tI guess "it" doesn't, but *I* do.
18:31:52TorneYup, probably depends how sweaty your finger is ;)
18:33:40 Join funman [0] (~fun@rockbox/developer/funman)
18:37:19 Join joeyg [0] (
18:37:25 Quit joeyg (Client Quit)
18:38:22 Quit pamaury (Remote host closed the connection)
18:39:42 Quit swilde (Quit: ERC Version 5.3 (IRC client for Emacs))
18:39:45 Quit togetic (Ping timeout: 240 seconds)
18:46:53mtAnyone knows why the sim is failing to link libm ? I can't see an obvious commit that would have affected that since r27015
18:47:01***Saving seen data "./dancer.seen"
18:47:39funmanmt: it works fine for me, what is your OS ?
18:48:19mtubuntu 8.04, but I didn't have this problem in my older revision
18:48:34funmancan you paste the log?
18:48:36mtfunman: could you try building the sim with libwmapro ?
18:48:59 Quit rado1 (Ping timeout: 240 seconds)
18:49:20funmanshould i just add wmapro.c to apps/codecs/SOURCES ?
18:49:37 Join Luca_S [0] (
18:50:00 Join rado1 [0] (
18:50:11mtfunman : not just that .. but you could cd to libwmapro and apply wmapro_mainbuild.patch and it will do the changes.
18:52:13mtbtw here's the output :
18:54:16funmanyeah i can reproduce it
18:54:47 Quit S_a_i_n_t ()
18:54:59 Quit freddyb (Ping timeout: 240 seconds)
18:55:06 Join S_a_i_n_t [0] (S_a_i_n_t@
18:55:18mtseems that codecs aren't linking libm
18:55:42funmanno codec needs it so far
18:55:47 Join freddyb [0] (
18:56:52mtit could be needed for the early phases of porting, but I didn't have this problem in r27015 anyway
18:57:12 Join Jerom [0] (~jerome@
18:57:21funmani added -z defs to the linker to make linking fail if a library is missing
18:57:55funmanbefore that the linker relied on rockboxui (which loads the .codec) to be linked with libm too
19:03:52mtfunman: Thanks :)
19:06:03funmanmt: you want me to commit it ?
19:07:33mtThat would be great.
19:09:23wodzcan someone tell me what does it mean?
19:10:17 Quit DerPapst (Quit: Leaving.)
19:10:28CIA-6New commit by 03funman (r27336): wmapro codec sim build needs to link with libm ...
19:11:03funmanwodz: what exactly?
19:11:30wodzthe place where it is time inconsistent
19:11:40funman00:-1 ?
19:12:02wodzafter 06:53:40 there is 05:49:39
19:12:24CIA-6r27336 build result: All green
19:12:42 Part rado1
19:12:42wodzline 372
19:12:55funmanwodz: hm the voltage also is higher at this point
19:14:04wodzI have a copy of the first bench and it is the same with the begining of this file up to the 372 line
19:15:11funmanwodz: copy/paste or file system error ?
19:15:37funmanline 372-378 are the same than 297-313
19:17:18wodzfunman: time is different
19:18:38wodzfunman: by copy of first bench I mean I made a copy of battery_bench file after first run
19:18:43funmansorry, line 307-313
19:20:10CIA-6New commit by 03alle (r27337): Describe what 'Update on stop' does
19:20:45wodzthe question is what is the real time of the second bench
19:20:48 Join Buschel [0] (
19:21:04bertrikI see that sometime the battery bench has ended at the expected voltage but does not have a final line saying why the battery bench was ended
19:21:53CIA-6r27337 build result: All green
19:22:55wodzwaiting over 6 hours to see something like this is quite irritating :-/
19:24:08wodzanother question is why first bench starts at 01:03:22
19:24:29 Join fml [0] (
19:24:30bertrikand there's still this funny thing that the battery benchmark file contains a warning to not stop the battery benchmark by loading another plugin, but basically the only way to be able to read that is to open the file with the text viewer plugin.
19:24:57wodzdid it really not logged for 1 hour?
19:25:30pixelmawodz: did you start your player 1 hour before actually starting the battery bench?
19:25:46funmanwodz: bench start with player's uptime
19:25:55wodzpixelma: could be - it was charging before bench
19:25:59funmanwhich is IMO is confusing
19:26:43fmlDo we always have the name of the file being played in the struct mp3entry?
19:26:55wodzso it run for 6:59:40 - 1:03:22 in the first bench right?
19:27:00*amiconn thi9nks it's logical
19:27:29pixelmame too with the exception of "charger connected"
19:28:57funmanamiconn: why?
19:30:55CIA-6New commit by 03alle (r27338): Even more readable code
19:30:56pixelmafunman: it's already using power for this amount of time
19:31:05 Join OverLOAD [0] (
19:31:56funmanwe could print the uptime once at the start and then print time relative to plugin start
19:32:41CIA-6r27338 build result: All green
19:32:50 Join voRia [0] (
19:32:57wodzpixelma: but it may charge (and probably is) before bench start
19:33:14 Nick OverLOAD is now known as OverLOAD_x (
19:33:24 Quit OverLOAD_x (Client Quit)
19:34:27freddybS_a_i_n_t: I tried the rockboy thing and I think it would be too easy to hit accept or cancel by accidentally clicking instead of tapping. It could be really frustrating.
19:35:07 Quit fml (Quit: ChatZilla 0.9.86 [Firefox 3.6.3/20100401080539])
19:35:15pixelmawodz: that's what I meant with my "exception" statement
19:35:44pixelmafunman: starting the bench with the uptime number is the same amount of information
19:37:03funmanpixelma: it requires more work to know how much time charging/discharging took, or if you want to plot the data
19:37:50pixelmaI can't see how
19:38:12wodzfunman: do you use some script for plotting battery_bench results?
19:38:13funmanin a graph the origin is 0 (battery bench started)
19:38:50pixelmaorigin 0 would be player switched on
19:38:53pixelmato me
19:39:04funmanwodz: (gnuplot)
19:39:29funmanpixelma: why? you don't have information between when the player was switched on and when bench started
19:39:59funmani charge my players to 100% before starting a bench, i don't care how much time the player was on before i started the bench
19:40:57 Join Hillshum [0] (4ba5f499@gateway/web/freenode/ip.
19:42:49pixelmaI want to know how long the battery lasts, not how long the bench ran. As I said, maybe it should not count the uptime as long as the charger is connected but everything after that makes perfect sense to me (if I decide to only start the bench one hour after using the player then I still want to know that it already ran 1 hour, even if I don't have data for it)
19:42:54 Quit Buschel (Ping timeout: 276 seconds)
19:45:03pixelmaand if you have the charger connected, you should get a mark in the log, so you can also figure it out (if this isn't broken or I misremember)
19:45:34 Quit voRia (Quit: Leaving.)
19:46:03funmanthe 'A' represents charger connected
19:49:45 Join binaryhermit [0] (
19:49:45 Join Buschel [0] (
19:49:45 Join DerPapst [0] (
19:49:45 Quit bor_ka (Quit: CGI:IRC (Ping timeout))
19:54:34 Join bor_ka [0] (
19:55:31bor_kafunman, if you remember my problem with glitches and skips on fuzev1, I believe I have traced it to the incorrect sector reads
19:56:10bor_kaOr is it more ranma's code, if it can be said so?
19:56:26funmanwhat do you mean?
19:57:26bor_kaI mean who is the best person to describe it to ;)
19:57:39funmanthe best person is #rockbox
19:57:57wodzsetting normal frequency to 33.8xxx MHz instead of 45 gives +30min on MPIO in ideal conditions
19:58:05bor_kawell... tried it several times already, will try again
19:59:02bor_kaWhen playing mp3-320 on fuze v1 I get strange glitches, parts of other songs to be more precise, from 1/10 sec to several seconds
19:59:52bor_kawhat have I done: I added "watermarks" to the sectors of the "problematic" files, that is changed last 4 bytes of every file's sector to the sector number
20:00:22bor_kathen I have added test, if the sector being read contains this watermark
20:00:23Luca_SFuzeV2: - battery bench with backlight on, plain svn
20:00:42Luca_SFuzeV2: - battery bench with backlight on, plain svn + dynamic cpu frequency
20:01:11funmanLuca_S: the last patch (v11) which sets PLLA to 248MHz ?
20:01:16bor_kain fat_read_sectors - since there I could test for the file first cluster and check only "watermarked" ones
20:01:19Luca_Syes, v11
20:01:53funmanLuca_S: please post the results (runtime) on FS
20:02:15wodzdo you drain battery down to 3.2V ?
20:02:36bor_kathen I got panic - sector didn't contains watermark in the buffer, but did have it when read directly from the disk
20:02:37bieberpixelma: Thanks :)
20:02:43wodz3.3 actually
20:02:52Luca_Safter recharging, is there any other test that could be useful? i'm planning to test the PLLB patch
20:03:25bor_kaso, the result is - sometimes transfer in fat.c does not read correct sector - and it directly calls sd_read_sectors
20:03:51wodzdraining down to such low voltage will shorten battery life significantly
20:04:13bor_kafunman, I believe I can add check to the sd-as3525.c - but what should I look to?
20:05:03bor_kaI mean, what variables or something should I trace in the panic?
20:05:14funmanno idea, i'm not sure to understand your description
20:05:28funmanfat_read_sectors() does something wrong ?
20:05:50bor_kano, I believe sd_read_sectors does something wrong
20:06:38amiconnImo it is logical because starting the plugin after starting playback takes time, and that time would be missed otherwise
20:07:08wodz44 seconds for example?
20:07:26funmankugel: ping
20:07:29amiconnAlso, battery_bench is not only usable for benching discharging, but also for benching charging
20:08:02bor_kafunman, I have added check, and storage_read_sectors did read sector incorrectly
20:08:04amiconnFurthermore it gives you one more info than if you start at 0:00
20:08:26funmanbor_ka: did you bissect the problem to a specific commit ?
20:08:40wodzbut it makes comparision of different benches harder
20:08:55funmanamiconn: see the patch i posted
20:09:14funmani'm testing it and i'll commit
20:09:24funmanuptime information is still there but individual entries start at 0
20:09:28amiconnwodz: how/ why?
20:09:43 Quit MethoS- (Remote host closed the connection)
20:10:04bor_kafunman, I believe it is almost impossible, because for some revisions there were filesystem corruption, and tests with this corruption will not show anything
20:10:15wodzamiconn: you have to normalize data to plot it with common timescale
20:10:44funmanbor_ka: you can disable writing: make sd_write_sectors() return -1;
20:11:02bor_kafunman, hmmm
20:11:49wodzmaybe as a compromise we can introduce one more column with relative time
20:12:22funmanwodz: amiconn:
20:13:04amiconnwodz: Not if you (re)boot fresh before doing the bench (which I always do)
20:13:30amiconnThen it makes things easier, because the timescale is automatically normalized, even if I forget to start the plugin immediately
20:14:04funmanamiconn: with this patch no need to reboot?
20:15:02amiconnI'd do it even if not needed
20:15:23amiconnIn fact on many targets it is *required* because they don't charge properly in rockbox (yet)
20:16:48 Join Horscht [0] (~Horscht2@xbmc/user/horscht)
20:16:50wodzamiconn: from the other hand if you testing if charging in rb has impact on runtime I prefer not to reboot
20:16:51bor_kafunman, but wouldn't rockbox panic if it can't write?
20:17:22funmanbor_ka: no
20:17:39CIA-6New commit by 03funman (r27339): battery_bench: make log time start at 0 (since plugin started) ...
20:19:29CIA-6r27339 build result: All green
20:24:32 Quit binaryhermit (Ping timeout: 240 seconds)
20:25:11BuschelCan anyone with a coldfire target (both MCF5250 and MCF5249) test FS #11461?
20:27:58 Quit bor_ka (Quit: CGI:IRC (Ping timeout))
20:28:56wodzBuschel: I can test this on MCF5249 a bit later
20:30:19CIA-6New commit by 03funman (r27340): fix r27310 and r27317 ...
20:31:55CIA-6r27340 build result: All green
20:34:16Buschelwodz: great :)
20:41:41 Quit Luca_S (Quit: CGI:IRC)
20:47:04***Saving seen data "./dancer.seen"
20:49:03 Quit t0rc (Remote host closed the connection)
20:49:30 Join bor_ka [0] (
20:51:07bor_kafunman, without the sd_write it can't access playlist control file, and I don't know how to start playback
20:51:45funmanopen a file?
20:52:02funmanthis message is just a warning, ignore it
20:54:19bor_kaOk, I have deleted it from the PC - it started playing :) Thanks
21:09:12funmani have used a bit my fuzev1 but didn't hear glitches so far
21:09:26 Part watto
21:09:40 Join kramer3d_ [0] (~kramer@unaffiliated/kramer3d)
21:09:56 Nick kramer3d_ is now known as d3remark (~kramer@unaffiliated/kramer3d)
21:10:16 Join wombatman [0] (
21:11:11 Join merbanan [0] (
21:11:57 Quit esperegu (Ping timeout: 276 seconds)
21:14:06 Quit Hillshum (Quit: Page closed)
21:17:57bor_kafunman, it happens on some folders only
21:18:26bor_kaand I believe it can be a timing issue - so, the memory chip can matter
21:18:37bor_kaand they can differ in different batches
21:18:56 Quit d3remark (Quit: Leaving)
21:19:13bor_kathe point is - I have never heard glitches from the uSD, it is twice the internal memory size...
21:27:15 Quit freddyb (Quit: ChatZilla 0.9.86 [Firefox 3.5.9/20100401213457])
21:31:58 Join petur [0] (~petur@rockbox/developer/petur)
21:36:50 Quit Jerom (Quit: Leaving.)
21:37:23 Part AzureWurk
21:37:44 Join Jerom [0] (~heidi@
21:40:26 Quit bor_ka (Quit: CGI:IRC (EOF))
21:41:14 Join roolku [0] (
21:42:55roolkufunman: r27340 breaks buildzip (on cygwin at least): 0/.rockbox: No such file or directory at /home/robert/roolku/tools/ line 478
21:43:34roolkufunman: I suspect you still need the check for $install!=0
21:44:25funmanroolku: oh i forgot that sorry
21:44:46funmanadding "&& ($install != 0)" would do the trick?
21:45:09wodzBuschel: on MCF5249 64kaache.m4a 367.95MHz svn, 289.47MHz aac-he_v8.patch
21:46:24roolkufunman: not a perl person, so no clue about the operators. I would leave the if statement for clarity
21:49:29funmanroolku: works ? (not a perl person myself)
21:54:02roolkufunman: what was wrong with lear's approach?
21:54:55funmannothing, but having install be both an integer and a string is wrong to me
21:55:54saratogaI would just do unless($install)
21:56:13saratogarather then if (not equal to null)
21:56:17 Join captainewkl [0] (2669ecc2@gateway/web/freenode/ip.
21:56:36funmansaratoga: unless($install) checks if install is initialized or not ?
21:56:53saratogaoh i guess you want the opposite
21:56:57saratogaso if($install)
21:57:22saratogayeah it checks that
21:57:39funmanand can we access $install anyway ?
21:57:46roolkuthat is what lear did
21:58:32saratogastrings in perl are false if they are blank, not inited or "0"
21:59:03funmanso "XXX $install YYY" should give an error ?
21:59:35roolkuif you change $install to "", abs_path("$install/$rbdir") will be /.rockbox instead of 0/.rockbox which doesn't seem right either
21:59:55funmanhm right
22:00:14roolkuif you use the if it will never be evalated
22:00:32roolku(for $install not set)
22:00:34saratogawhat was wrong with the original?
22:00:49funmansaratoga: which original ? ;)
22:00:55roolkuoriginal = lear ?
22:00:56saratogathe one the diff is against
22:01:33funmanthe error that roolku pasted, which r27317 was supposed to fix
22:02:04funmanbut i broke it again
22:06:31 Quit captainewkl (Quit: Page closed)
22:07:09 Join jennifur [0] (
22:08:22funmanroolku: works for me
22:09:54funman(same url just edited): works too and is simpler
22:10:21 Quit jennifur (Read error: Connection reset by peer)
22:10:41funmancan you try make PREFIX=a install on cygwin ? i don't know how '/' works
22:10:47funman.. in cygwin
22:12:46funmanfucking perl
22:13:22roolkuyour unedited patch worked without a prefix installed...trying the other now
22:13:45funmanmake install without a prefix shouldn't work
22:13:58funmanunless in the sim
22:14:15 Join jennifur [0] (
22:14:26funmanand i have yet another thing to fix
22:14:46roolkuI did a make zip (that was not working before)
22:14:58funmanPREFIX is only for installing
22:15:15funmaneverything is installed in ./.rockbox
22:15:32funmanthen .rockbox is zipped, or copied to $PREFIX/ (if we're installing)
22:15:52funmanif we're installing .. to another place than '.'
22:17:18roolkuyes, but your commit still broke make zip :)
22:17:36funmanit broke make zip, on cygwin
22:20:05funmanroolku: can you try make PREFIX="$PWD/sdqfqsfqsd" install ? (just make sure the qsqsqsdf part doesn't exist)
22:20:41roolkuusing which patch?
22:21:53funmani don't understand the abs_path error
22:22:38funmanperhaps it's because of concatenating itneger with string, or perhaps abs_path() doesn't like non existing paths. But why is it an error only in cygwin.. ?
22:24:03roolkufunman: /home/robert/clean/Ngigabeatfx/sdqfqsfqsd/.rockbox: No such file or directory
22:24:39funmanok the cygwin's perl clearly sucks
22:25:31roolkumy quick google seems to suggest different abs_path implementations; cygwin seems to take into account the actual file systen, rather than just do string manipulation (which I suppose is better/more correct)
22:25:50funmanstring manipulation is more correct yes
22:26:08Buschelwodz: thanks for testing. good result :)
22:31:37CIA-6New commit by 03funman (r27341): more fixes ...
22:33:04 Quit bmbl (Quit: Bye!)
22:33:24CIA-6r27341 build result: All green
22:35:15 Quit funman (Quit: free(random());)
22:39:19 Join MethoS- [0] (~clemens@
22:47:06***Saving seen data "./dancer.seen"
22:50:12roolkufunman: (for logs) make zip is working now, but make install still barfs even if target dir exists because of missing .rockbox subdir
22:56:05roolkufunman: you could take $rbdir out of both sides of the equation perhaps?
22:56:18CIA-6New commit by 03bieber (r27342): Theme Editor: Made auto-expand/highlight of parse tree optional (through preferences dialog), added Simulation Time variable to device config panel, ...
22:58:04CIA-6r27342 build result: All green
22:59:31 Quit w1ll14m ()
23:00:04 Quit Jaykay (Ping timeout: 264 seconds)
23:01:06 Join fml [0] (
23:03:16bluebroth3rI'm considering to move rbutil/tools to utils/something. Any ideas what's the best for something? common? Or still name it tools? utils/tools might be a bit strange
23:03:26fmlHello. I'd like to change the way bookmarks are created. I'd like to introduce a new setting "Unique per Track". If set to "yes", only one bookmark per track will be retained −− both for normal and recent bookamrks. What do you think?
23:03:58fmlbluebroth3r: what do the tools do?
23:04:05 Join togetic [0] (~togetic@unaffiliated/ibuffy)
23:04:37bluebroth3rfml: currently it's only bin2c. I'd like to eventually merge the rbutil and utils folder −− there isn't much point in having both IMO
23:04:59bluebroth3rbin2c is used by the patchers, but also by the (already in the utils folder located) beastpatcher.
23:05:28fmlbluebroth3r: then a merge is the right way IMO, i.e. no subdirs
23:05:49bluebroth3rsome of the patchers even keep their own copies of bin2c (with small modifications). Merging those mostly-identical copes is another thing that's on my list
23:06:24 Quit roolku ()
23:06:26 Quit jennifur (Read error: Connection reset by peer)
23:06:35bluebroth3rfml: well, I still need to place bin2c into some folder. Putting it into utils/ doesn't seem sensible −− all tools are in some subfolder of utils/
23:06:51 Join jennifur [0] (
23:07:56fmlbluebroth3r: I mean no intermediary subdir "tools"
23:09:18bluebroth3rfml: well, what folder would be appropriate then? common? Btw, I'm also planning to generalize the deploy script used for Rockbox Utility and place it into that folder too.
23:09:50bluebroth3rto have it do the deployment stuff for both the Theme Editor and Rockbox Utility
23:10:53bluebroth3rsure, bin2c and the deployment script could go directly into utils/. Not sure if that's a good way though.
23:11:10bluebroth3rbtw, isn't that newparser a bit misplaced in utils/?
23:15:15 Quit wombatman (Quit: ChatZilla 0.9.86 [Firefox 3.6.3/20100423140709])
23:15:20 Quit fml (Quit: ChatZilla 0.9.86 [Firefox 3.6.3/20100401080539])
23:17:07*domonoky thinks utils/common/ sounds good.
23:21:26 Join Hillshum [0] (4ba5f499@gateway/web/freenode/ip.
23:21:36 Quit Hillshum (Client Quit)
23:25:51 Quit stoffel (Remote host closed the connection)
23:29:58 Nick jennifur is now known as bunnyboi (
23:39:18 Quit flydutch (Quit: /* empty */)
23:40:24 Quit Jerom (Read error: Connection reset by peer)
23:42:05CIA-6New commit by 03alle (r27343): More consistent formatting of setting names
23:43:43CIA-6r27343 build result: All green
23:45:45 Quit wodz (Ping timeout: 245 seconds)
23:47:16 Quit jgarvey (Quit: Leaving)
23:51:40 Join wodz [0] (
23:53:36 Quit komputes (Quit: I haven't slept for ten days, because that would be too long.)

Previous day | Next day