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

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

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

#rockbox log for 2010-04-06

00:00:45kugeltoo long!
00:00:58kugelah the abstract only
00:01:35gevaertskugel: if I understand the documentation correctly, it shouldn't be public. I'm not entirely sure though
00:02:42kwbrgevaerts: works great. thanks
00:03:03kwbrgevaerts: will add the tool to my music sync-script
00:03:26gevaertskwbr: you can also do it using sg3-utils and hand-assembling the SCSI packet :)
00:03:38saratogaamiconn: how do you figure?
00:03:41 Quit RadicalR (Read error: Connection reset by peer)
00:03:42*S_a_i_n_t was fairly certain thata restore with itunes synced the time on an iPod.
00:04:01gevaertsS_a_i_n_t: most people try to avoid restoring too often
00:04:12S_a_i_n_tWell, I've only ever set the timezone on mine, and the time is always right.
00:04:28gevaertsS_a_i_n_t: itunes syncs the clock every time
00:04:32amiconnsaratoga: ldm reduces code size -> better caching, and if the code memory is uncached, same thing - fetching fewer instructions
00:04:36S_a_i_n_tAs do I, but what would cause it to reset the time?
00:04:55saratogaamiconn: but a loop will be entirely cached regardless, its only 1-2 cachelines long
00:05:00gevaertsS_a_i_n_t: what?
00:05:21amiconnSure, but there's not *only* that loop, there's other code around it
00:05:26*kugel shorts the abstract a bit, the full version is still in the proposal
00:05:34kwbrgevaerts: Ha! I'll check to code...
00:05:39amiconn(well, unless you're timing just that loop of course...
00:05:53saratogayeah we're just timing that loop, its a memory benchmark after all
00:05:58gevaertskwbr: has the full details. ipod-time-sync was based on that
00:05:59 Join Doleo [0] (
00:06:00S_a_i_n_tErrr, bad wording. I mean, why would the time be incorrect in the first place?
00:06:11S_a_i_n_tIf the battery ran completely dead...?
00:06:32gevaertsever heard of clock drift?
00:06:58*S_a_i_n_t googles.
00:07:29 Quit Doleo (Client Quit)
00:07:45 Quit captainkewllllll (Quit: Page closed)
00:08:27 Quit bertrik (Ping timeout: 265 seconds)
00:08:31 Join Doleo [0] (
00:09:51 Quit Doleo (Client Quit)
00:12:52amiconnsaratoga: There's also the possibility that the cache isn't completely single cycle, or not properly initialized somehow
00:13:37*amiconn would in fact be very surprised if the PPs would be the only SoCs with bugs like this
00:13:41 Join anewuser [0] (anewuser@
00:13:48 Quit anewuser (Changing host)
00:13:48 Join anewuser [0] (anewuser@unaffiliated/anewuser)
00:13:50saratogaperformance is really good so I think cache is working
00:14:10saratogathough it could be some bug specific to this loop, but i think me making a mistake is more likely :)
00:14:45saratogaprobably cache bugs are less likely on ARM9 and higher since the cache is designed by arm and theres onboard hardware for things like TCM
00:14:49saratogaso less to screw up
00:15:10kugeltcm is only on arm9e
00:15:36 Join Doleo [0] (
00:18:36 Quit Schmogel (Ping timeout: 265 seconds)
00:18:42 Quit shai (Quit: Leaving)
00:21:16 Quit GeekShadow (Quit: The cake is a lie !)
00:24:23 Quit xiainx (Ping timeout: 252 seconds)
00:30:29 Quit kwbr (Quit: Leaving)
00:32:54stripwax_saratoga - mm, my tremor timings seem unchanged so it seems an accurate measurement −− i.e. old rockbox mdct_arm.S negligable benefit on whatever the nslu2 is, versus about 9% faster with FasterMdct
00:34:50 Join xiainx [0] (xiainx@wpa062091.Wireless.McGill.CA)
00:36:43saratogastripwax_: well a lot of the improvement was probably just the ldm/stm stuff, so maybe thats reasonable
00:36:43 Quit antil33t (Read error: Connection reset by peer)
00:37:04saratogai don't know if we ever benched it on ARM9, back in those days we only had the GBF
00:37:32 Quit ender` (Quit: I went to a restaurant that serves "breakfast at any time." So I ordered French Toast during the Renaissance.)
00:37:50 Quit slck (Ping timeout: 240 seconds)
00:43:02 Join antil33t [0] (
00:45:11 Quit JohannesSM64 (Quit: WeeChat 0.3.2-dev)
00:45:43 Quit domonoky (Read error: Connection reset by peer)
00:47:59 Quit TheSeven (Read error: Connection reset by peer)
00:49:30 Quit Battousai (Ping timeout: 240 seconds)
00:49:44 Join slck [0] (
00:49:54 Quit stripwax_ (Quit:
00:49:57 Join Battousai [0] (~bryan@gentoo/developer/battousai)
00:50:41 Join LambdaCalculus37 [0] (~rmenes@rockbox/staff/LambdaCalculus37)
00:52:31 Join JohannesSM64 [0] (
01:03:00***Saving seen data "./dancer.seen"
01:06:54 Join RadicalR [0] (
01:09:41 Join TillW [0] (~Till@
01:12:47 Join fejfighter [0] (
01:13:35 Quit leavittx (Ping timeout: 240 seconds)
01:13:42 Quit leavittx_ (Ping timeout: 265 seconds)
01:13:42 Quit n17ikh (Ping timeout: 245 seconds)
01:16:02 Quit jacekowski (Killed (idoru (Spam is off topic on freenode.)))
01:19:26 Quit anewuser ()
01:19:26 Join n17ikh [0] (
01:23:11 Join leavittx [0] (~leavittx@
01:24:42 Quit Doleo (Quit: ChatZilla 0.9.86 [Firefox 3.5.6/20091201220228])
01:25:16*flyback taking a nap, bbl
01:26:38 Nick fxb is now known as fxb__ (
01:26:38 Join leavittx_ [0] (~leavittx@
01:26:41 Join leavittx__ [0] (~leavittx@
01:27:44 Join jacekowski [0] (
01:28:44 Nick fxb__ is now known as fxb (
01:28:52 Quit dfkt (Quit: -= SysReset 2.53=- Ph'nglui mglw'nafh Cthulhu R'lyeh wgah'nagl fhtagn.)
01:28:58 Quit arbingordon (Ping timeout: 258 seconds)
01:32:00 Quit komputes (Quit: I haven't slept for ten days, because that would be too long.)
01:48:21 Quit TillW (Quit: This now concludes our broatcast day.)
01:48:36 Join TillW [0] (
01:56:19 Join planetbeing_ [0] (
01:56:55 Nick planetbeing_ is now known as planetbeing (
01:57:12 Join arbingordon [0] (~w@unaffiliated/arbingordon)
02:02:55 Quit xiainx (Ping timeout: 268 seconds)
02:12:57 Quit toffe82 (Quit: ChatZilla 0.9.86 [Firefox 3.6.3/20100401080539])
02:15:53 Join robin0800 [0] (
02:18:43 Join cjcopi [0] (
02:27:16 Quit TillW (Ping timeout: 246 seconds)
02:27:43 Join komputes [0] (~komputes@ubuntu/member/komputes)
02:38:47 Nick fxb is now known as fxb__ (
02:42:32 Join TillW [0] (
02:48:07 Quit robin0800 (Read error: Connection reset by peer)
02:54:04 Join robin0800 [0] (
02:58:41 Quit moos (Ping timeout: 276 seconds)
03:00:57 Quit kugel (Remote host closed the connection)
03:03:03***Saving seen data "./dancer.seen"
03:08:06 Join xiainx [0] (
03:16:38 Join Blue_Dude [0] (
03:20:00Blue_DudeHah! Try this out for size... Modified bookmark.c to write optional tokens to bookmarks and at the same time remain backward compatible with existing bookmarks and bookmarks with other optional tokens. Plus it adds pitch and timestretch tokens by default.
03:21:50Strife89Make a patch for Flyspray. :)
03:22:56 Quit robin0800 (Remote host closed the connection)
03:24:01 Join robin0800 [0] (
03:30:15 Join blairb [0] (
03:33:53Blue_DudeAnd it's at FS #11178. Enjoy!
03:34:40*flyback wishes there was test code for c140/c150
03:39:00 Part froggyman
03:42:26 Join shaggy-h [0] (
03:43:27 Quit shaggy-h (Client Quit)
03:45:03 Quit adnyxo (Quit: Leaving)
03:46:01 Quit planetbeing (Read error: Connection reset by peer)
03:46:18 Join planetbeing [0] (
03:46:55 Join shaggy-h [0] (
03:49:34 Join adnyxo [0] (
03:49:40 Quit adnyxo (Remote host closed the connection)
03:50:02 Quit TillW (Quit: This now concludes our broatcast day.)
03:53:47 Join robin0800_ [0] (
03:53:49 Quit robin0800_ (Remote host closed the connection)
03:54:04 Quit robin0800 (Ping timeout: 240 seconds)
03:57:40 Join robin0800 [0] (
03:58:18 Join BHSPitMonkey [0] (~stephen@unaffiliated/bhspitmonkey)
04:03:29 Quit Chronon (Ping timeout: 265 seconds)
04:08:41 Quit LambdaCalculus37 (Quit: Fwump)
04:08:49 Quit Strife89 (Quit: Going to bed.)
04:11:07 Join phanboy4 [0] (
04:11:55 Join TillW [0] (
04:22:50 Quit Blue_Dude (Quit: ChatZilla 0.9.86 [Firefox 3.6.3/20100401080539])
04:25:12 Join bieber [0] (~bieber@
04:27:49 Join Rob2223 [0] (
04:29:31 Join webguest62 [0] (
04:31:30 Quit Rob2222 (Ping timeout: 265 seconds)
04:33:26 Quit webguest62 (Client Quit)
04:41:44 Quit fejfighter (Ping timeout: 248 seconds)
05:00:13 Quit Barahir (Ping timeout: 260 seconds)
05:00:29 Quit bieber (Quit: Leaving)
05:01:53 Join Barahir [0] (
05:03:07***Saving seen data "./dancer.seen"
05:04:24 Join CGL [0] (~CGL@
05:06:39 Quit komputes (Ping timeout: 246 seconds)
05:14:16 Quit mikroflops (Ping timeout: 248 seconds)
05:15:40 Join mikroflops [0] (
05:31:05 Join froggyman [0] (~me@unaffiliated/froggyman)
05:34:47 Part flyback ("Leaving")
05:35:02 Quit froggyman (Read error: Connection reset by peer)
05:35:18 Quit Horscht (Quit: Verlassend)
05:35:42 Join Rob2222 [0] (
05:37:00 Join froggyman [0] (~me@unaffiliated/froggyman)
05:38:49 Quit evilnick (Ping timeout: 246 seconds)
05:38:50 Quit Rob2223 (Ping timeout: 245 seconds)
05:39:10 Quit mikroflops (Ping timeout: 246 seconds)
05:40:24 Join mikroflops [0] (
05:46:40 Quit planetbeing (Read error: Connection reset by peer)
05:47:31 Join planetbeing [0] (
05:49:59 Quit robin0800 (Ping timeout: 268 seconds)
05:51:08 Join blairb_ [0] (
05:52:30 Quit mikroflops (Ping timeout: 246 seconds)
05:52:55 Join mikroflops [0] (
05:53:27CGLHi everyone
05:54:45 Quit blairb (Ping timeout: 258 seconds)
06:02:26 Part froggyman
06:09:05 Join robin0800 [0] (
06:22:21 Quit xiainx (Quit: Good Bye)
06:25:47 Quit mikroflops (Ping timeout: 264 seconds)
06:26:46 Join mikroflops [0] (
06:36:51 Join evilnick [0] (
06:42:16 Quit phanboy4 (Ping timeout: 258 seconds)
06:44:11 Join phanboy4 [0] (
06:45:43 Join phanboy_iv [0] (
06:49:27 Quit phanboy4 (Ping timeout: 276 seconds)
06:52:08 Quit robin0800 (Remote host closed the connection)
06:57:13 Quit saratoga (Quit: Page closed)
06:58:32 Quit mikroflops (Ping timeout: 264 seconds)
06:58:43 Join S_a_i_n_t_ [0] (S_a_i_n_t@
06:58:57 Join mikroflops [0] (
06:59:56 Quit S_a_i_n_t (Ping timeout: 252 seconds)
07:02:51 Quit phanboy_iv (Quit: Leaving)
07:03:10***Saving seen data "./dancer.seen"
07:05:04 Quit CGL (Quit: Saliendo)
07:07:48 Join shai [0] (
07:10:18 Quit panni__ (Quit: ( :: NoNameScript 3.81 :: ))
07:29:11 Quit mikroflops (Ping timeout: 248 seconds)
07:29:36 Join mikroflops [0] (
07:46:29 Quit planetbeing (Read error: Connection reset by peer)
07:46:50 Join planetbeing [0] (
08:09:21 Join Zagor [0] (~bjst@rockbox/developer/Zagor)
08:13:42 Join r2k000 [0] (~r2000@
08:14:08 Join Boldfilter [0] (
08:22:31 Quit Boldfilter (Quit: Boldfilter)
08:35:50 Join ender` [0] (
08:38:04 Join Luca_S [0] (
08:38:19 Quit amiconn (Disconnected by services)
08:38:22 Join amiconn_ [0] (quassel@rockbox/developer/amiconn)
08:38:41 Join mitk [0] (~mitk@
08:38:43 Nick amiconn_ is now known as amiconn (quassel@rockbox/developer/amiconn)
08:39:24Luca_SFuzeV2, latest SVN. I'm experiencing random screen corruptions - boot loader showed corrupted colors in the RB logo (only once), WPS seems to "move to the left" part of the screen (as if it was missing some pixels)
08:39:34 Join pixelma_ [0] (quassel@rockbox/staff/pixelma)
08:39:34 Quit pixelma (Disconnected by services)
08:39:53 Nick pixelma_ is now known as pixelma (quassel@rockbox/staff/pixelma)
08:40:10Luca_Snow I had the same corrupted colors problem in the WPS too
08:48:06 Join flydutch [0] (
08:50:23 Join bertrik [0] (~bertrik@rockbox/developer/bertrik)
08:53:49 Quit mikroflops (Ping timeout: 245 seconds)
08:55:04 Quit drostie (Remote host closed the connection)
08:55:04 Nick S_a_i_n_t_ is now known as S_a_i_n_t (S_a_i_n_t@
08:56:52 Join stripwax [0] (
08:59:26S_a_i_n_tBlargh! FS #11101 stops sim-builds from compiling....I *really* like the behaviour the patch offers (though, the menu/wording for it is *awful*), but it's just not done as well as it could be by someone who actually knows what they're doing. ie. Not me.
08:59:36S_a_i_n_tGood idea, bad implementation.
09:00:16 Quit JohannesSM64 (Quit: WeeChat 0.3.2-dev)
09:00:28 Join mikroflops [0] (
09:00:55Llorean S_a_i_n_t: Is there a comment on the task about that problem?
09:03:13***Saving seen data "./dancer.seen"
09:03:50S_a_i_n_tThere are several about the crappy wording, But I'm just going to add a comment to it now alog the lines of "should use 'charging only mode' and 'mass storage device mode" (or similar) instead of a rather sonfusing 'yess/no' selection, and btw this patch also makes the bootloader complain and messes up sim-builds"
09:04:05S_a_i_n_tIt's a nice idea, just not very well done.
09:04:37S_a_i_n_tsonfusing? "Ahem....*con*fusing"
09:07:45 Quit arbingordon (Quit: `)
09:11:48 Quit freddyb (Ping timeout: 265 seconds)
09:18:09 Join LinusN [0] (~linus@rockbox/developer/LinusN)
09:23:09 Join Eugenpaul [0] (
09:28:25 Quit bertrik (Ping timeout: 252 seconds)
09:28:44 Join lpereira [0] (
09:29:44 Join pamaury [0] (~pamaury@rockbox/developer/pamaury)
09:32:08 Quit mikroflops (Ping timeout: 264 seconds)
09:42:29 Join Bagder [0] (~daniel@rockbox/developer/bagder)
09:46:47 Quit planetbeing (Read error: Connection reset by peer)
09:46:55 Join planetbeing [0] (
09:53:36 Quit pamaury (Read error: Operation timed out)
09:53:44 Join GeekShadow [0] (~Antoine@reactos/tester/GeekShadow)
09:53:50 Join pamaury [0] (
09:53:52 Quit pamaury (Changing host)
09:53:52 Join pamaury [0] (~pamaury@rockbox/developer/pamaury)
09:55:29 Join arbingordon [0] (~w@unaffiliated/arbingordon)
10:13:37 Quit Luca_S (Quit: CGI:IRC (EOF))
10:23:32 Join kugel [0] (~kugel@rockbox/developer/kugel)
10:27:45mitkIt looks like fuze v1 was broken by r25491. It frozes during startup. Anyone tried fuze v1 with build higher than 25490?
10:30:19 Join einhirn [0] (
10:30:47 Join m3dlg [0] (~m3dlg@
10:34:09kugelmitk: I'm not sure how but I'll have a look
10:35:47 Quit BHSPitMonkey (Quit: Ex-Chat)
10:37:07mitkkugel: my fuze lasts with yellow rockbox logo, sbs status bar loaded and wheel led on. Clean config and default theme. 25490 is working
10:40:03 Quit m3dlg (Quit: RAGE QUIT)
10:41:47kugelre: RaaA. I've looked into gnu pth a bit (a cooperative thread library) and it looks pretty straight forward to replace our own implementation with it
10:45:27kugellinux and android only offer pthread, android also java threads, neither is guaranteed to be cooperative, so we're basically forced to a) port over our threads or b) use a third party library which offers cooperative threads explicitely
10:47:49linuxstbOr c) fix the apps/ code to work with pre-emptive threading
10:48:25gevaertsfix *all* code
10:48:43gevaertsIt's not just apps
10:48:49linuxstbHow much firmware/ code will be left in a true RaaA?
10:49:12linuxstbOr how much firmware code where threading is an issue?
10:50:11gevaertsI don't think it's a good idea to switch threading models *only* for RaaA
10:52:18gevaertsand I'd assume that *all* code has issues. Everything that uses global state needs locking, which in rockbox basically means that everything needs locking
10:52:49linuxstbI'm not saying it's the best solution, but should we definitely rule it out?
10:53:11gevaertsI think within the scope of a RaaA project, yes
10:53:13linuxstbIsn't it only things which can be changed by multiple threads?
10:53:33 Join JdGordon [0] (~jonno@rockbox/developer/JdGordon)
10:53:51gevaertsisn't that basically everything?
10:54:41*gevaerts points JdGordon to FS #11175 to see if he has ideas
10:54:50*JdGordon just oened that email
10:55:03pamaurychanging the threading model would be an extremely tedious task
10:55:20BagderI'm not sure it would be _that_ tedious
10:55:27linuxstbI don't know. But that's my point - is it _definitely_ a huge undertaking?
10:55:37Bagderwe already have locking on several places due to system calls' own yieldings
10:56:07JdGordongevaerts: the theme causes the wrong track to start?! thats bloody retarted
10:56:46kugelthe problem is that debugging thread problems is hard, even more with preemptive threads. I do think it would be a very tedious task
10:56:50gevaertsJdGordon: yes, I know, but it's even worse. It causes the wrong track to start, but with the correct resume info so if you restart playback it changes...
10:57:37pamauryB4gder: don't underestimate it, debugging threading is extremely tedious, even when you right the code with preemptive model in mind
10:57:42JdGordonI probably wont get a chance to look at it tonight, but we'll see...
10:58:00Bagderto me, it's about where Rockbox is heading, where the future lies
10:58:06linuxstbSoC isn't meant to be fun, it's a job... ;)
10:58:24kugelI don't want to rule it out completely, but I think it might be a bit too much work for a single gsoc summer.
10:58:52pamaurywhy preemptive threading would be the future ?
10:58:55Bagderyes, it's a huge undertaking to consider it for gsoc
10:58:57kugelbut if it's required I shall give it a try, I guess?
10:59:17linuxstbThis is also why I've been saying that the SoC project shouldn't be focused on a target device - it should be laying the foundations.
10:59:20Bagderpamaury: because I see raaa as the future and thus getting out of the way for such "basic" things such as threading models
11:00:28 Join bluebrother [0] (~dom@rockbox/developer/bluebrother)
11:00:45gevaertsPreemptive threading might of course also make multi-cpu support *slightly* easier
11:00:51linuxstbBagder: I don't agree RaaA is _the_ future. Rockbox as a firmware will still be relevant for a long time, as long as manufacturers keep making dedicated DAPs. But I agree it's _part_ of the future.
11:01:31 Join n1s [0] (~n1s@rockbox/developer/n1s)
11:01:41kugellinuxstb: do you think I should focus on making preemptive threads work over porting to a device?
11:01:49Bagderyes, I'm not suggesting we'd drop anything of what we do now, but I do see a future where users such as myself will not be using dedicated DAPs anymore
11:02:23pamauryDon't get it wrong, raaa is fundamentally different, first of all, nearly all the code in firmware/ becomes useless and you can thus rely on the OS features, which are thread-safe. Then, the relevant code in apps/ can probably be 'fixed' for preemptive threading
11:02:37kugelRockbox works very well with cooperative threads, so I don't see a need to make preeptive threads work. they have also drawbacks
11:03:09Bagderkugel: but so does cooperative... :-)
11:03:14***Saving seen data "./dancer.seen"
11:03:20JdGordon"change is bad"..... :< silly argument
11:03:36Zagorthe proper argument is, rather, "multithreading is hard"
11:03:47linuxstbkugel: I'm just trying to keep an open mind. i.e. there are lots of things to discuss and consider. I just added "use pre-emptive threading" as a third option to your list...
11:03:55Zagorbut it certainly looks like we need to bite this bullet
11:04:10 Quit bluebroth3r (Ping timeout: 264 seconds)
11:04:21Bagdermultithreading _is_ hard
11:04:37JdGordonif we are talking RAAAaaaa wouldnt it be better to talk about splitting apps into logical modules and make them not talk to eachother how they do now?
11:04:55JdGordonpull the playback engine out into a seperate library, ditto drawing, etc
11:05:05Bagderthat'd make sense, yes
11:05:30Bagdera bit like librbtag has been started
11:05:31kugelI'd consider the project more successful if we have an installable app at the end (for a platform we didn't target before), not so much if we work with preemptive threads
11:06:02gevaertskugel: you could get an installable app in a week...
11:06:54kugelwith sdl and without fixing it to work with the target tree, yes
11:07:04Bagderfor the project, the foundation is more important than a single gsoc project though
11:07:25gevaertsof course, but those are also just foundation things, just like threading :)
11:07:54JdGordonalso, for Raaaa, going for a real target for gsoc is stupid.... linux+sdl is the obvious porper first step, then maybe replace sdl with qt or something
11:09:21kugelwell, I followed the project idea on SummerOfCode2010 which mentions porting to an actual device and no threading. but if we now decide that being independant on the underlying thread library is more important I can change my proposal to reflect this
11:09:28*linuxstb wants linux+ncurses, keeping charcell alive ;)
11:10:20*gevaerts wants to separate the GUI from the rest better, so keeping charcell alive becomes a non-issue :)
11:10:27pamaurygevaerts: I just found a library for usb analyzing, it didn't use it yet but it seems interesting:
11:11:03kugelso what has the higher priority?
11:11:07linuxstbgevaerts: I agree. The apps/ code definitely needs changing in the long-term, but it's definitely something to do after the lower-level stuff.
11:11:26gevaertskugel: find a healthy mix :)
11:13:27linuxstbI think it will depend a lot on the choices made, and the problems that are found along the way. So any timeline needs to be flexible. As a minimum, a "desktop" linux+sdl target could be a successful outcome, assuming the right amount of work has been done to get there.
11:19:38 Join TheSeven [0] (~theseven@rockbox/developer/TheSeven)
11:21:05n1sre: raaa, but not exactly what you guys are discussing, i've been out of the loop for a while and just read Bagder's blog post about it yesterday, and while i agree that raaa is *a* future of rockbox i wondered a bit about the part about relying on the device's OF/OS for stuff like networking. How do we feel about more advanced players like the zune for example, which has wifi, are we against rockbox gaining its own network stack? if so do we port linux
11:21:06n1sf.e. and run raaa on top?
11:22:15Zagormy personal feeling is that linux is proably the way to go for more complex code such as networking
11:22:22BagderI agree
11:22:22n1sbtw, i am not saying replacing the firmware entirely is feasible on phones and the like but on the more featured daps and pmps which lack an open os
11:25:10kugelok, I'll add "explore possiblities to run under preemptive multithreading", I propose I have a try but if it turns out too hard I move the focus onto later items on the list. I think it's not really possible to give an estimate on the work needed for preemptive threads
11:26:16n1skugel: i see no reason why you have to deal with all the problems raa brings up at once, i's suggest focusing on a few and getting them right
11:26:39kugelI fear that if I concentrate on the thread issue too much then, no matter of the outcome, porting will fail
11:26:46gevaertskugel: I'd mainly add a note that lots of details and steps could change after discussion
11:27:29*n1s is curious to see how the raaa gui will turn out
11:27:52kugelnot much different from the current one
11:28:14kugelwhich is why a desktop is actually a suboptimal target
11:28:30n1si'd like to run raa on a desktop/laptop
11:29:00n1ss/to run/to be able to run comfortably/
11:29:21n1sof course nothing prevents different UI's to be available
11:30:11kugelsure, but a complete ui overhaul is too much for gsoc imo
11:31:08kugelmaybe if it was *only* the ui overhaul but there are a lot of prerequisites for that
11:33:39n1syeah, i'm not saying you or anyone else should do it
11:37:08kugelbut pthread isn't really more native than pth, so if I'd have to choose one I would pick the one I know rockbox will work with
11:39:41kugelI'd even say pth is more portable
11:41:06gevaertsWell, we'd keep our current abstraction layer for threads anyway I think, so in the end different targets can have their own thread.c
11:43:13kugelI'm thinking that even if we manage to run under preemptive threads, it doesn't have a real advantage
11:45:14linuxstbIsn't the advantage that Rockbox is doing what the underlying OS expects applications to do? So it should be more efficient, and in a way, simpler.
11:46:12kugeland using plain ansi c library calls (make/swapcontext or set/longjmp) is not what the OS expects?
11:46:15linuxstbBut yes, "explore possiblities to run under preemptive multithreading" seems a sensible thing to do first. The decision can be made after that.
11:49:08topikwith rockbox r25494 i get 'undefined instruction at 3077C450' on my fuze v1
11:49:30topikafter the rockbox logo, before entering the main menu
11:50:10n1show many threads will raaa use anyway?, UI, playback, codec, buffering? Does multithreading even bring a benefit?
11:51:20kugellinuxstb: pth doesn't use an underlying thread library. it's a self-contained library, so basically the same as pthread (just cooperative). so I don't think it's less efficient than pthread
11:52:23kugelbesides, pthread isn't completely implemented on android
11:54:50topiksame problem mitk describes i suppose. did you have a chance to have that look, kugel?
11:55:14 Join liar [0] (
11:57:19 Quit kugel (Ping timeout: 264 seconds)
11:58:27 Join JdGordon_ [0] (~jd@rockbox/developer/JdGordon)
12:01:41 Join kugel [0] (~kugel@rockbox/developer/kugel)
12:08:30linuxstbkugel: Linux isn't the only kernel RaaA could run on though. There's WinCE, Symbian, OSX, ... I don't know what threading models they use.
12:09:02linuxstbMaybe targetting something non-Linux would be beneficial, to introduce diversity. Or at least thinking about them.
12:10:01kugelsure, but I wanted to focus on unix like devices
12:10:32kugelbut as I mentioned in the proposal, the final decision on the target device hasn't been made yet, I suggested a decision deadline for july, so I'm still open
12:12:08kugelI would prefer android or maemo because that's the devices I'm interested in. as the gsoc project most probably involves buying one I would rather buy one which I'm interested in outside of gsoc too
12:15:03linuxstbI'm just saying that it would be useful to at least read about a diverse selection of targets, as the aim (IMO) is to lay the groundwork for new ports.
12:15:56Zagorpth looks interesting
12:18:05 Nick fxb__ is now known as fxb (
12:20:01kugeltopik: I think I can imagine the problem
12:20:36topikkugel: just posted a bug report on flyspray. i didn't select the right options though, so feel free to correct or remove it
12:22:00CIA-5New commit by funman (r25495): Revert unrelated part of r25491 : fuzev1 init code works again
12:24:37 Quit Eugenpaul (Ping timeout: 248 seconds)
12:25:56 Join Luca_S [0] (
12:30:02topikthanks funman. works like a charm
12:30:40*JdGordon 's ipod isnt turning on after doing a batt bench :(
12:31:01 Part LinusN
12:31:36Luca_Sfuzev2 screen corruption is much more frequent when boosted - i quite expected the opposite
12:33:14JdGordon18h sitting in the "hw info" debug screen (so no skining running at all)... so either its all bad, or the theme engine doesnt affect runtime at all (anymore)
12:34:48Luca_Sshouldn't panic messages turn on backlight?
12:37:50kugelfunman, it wasn't really an unrelated change
12:39:40 Quit JdGordon_ (Quit: Bye)
12:44:09topikthank you too kugel, for looking
12:44:48kugelI edited my proposal to reflect the comments made and our thread discussion
12:46:21topikis there any possible RaaA target device you already have your eye on, kugel?
12:46:40kugelhtc desire seems nice :)
12:46:44JdGordonRaaa by definition sholdnt be targete to a specific device!
12:47:06topikno, but i read he plans to buy one as a test target
12:47:19topikso i was wondering if there was a leading candidate
12:48:43kugelJdGordon: the point is of course laying the groundwork application port that is easy to port to a specific device (or rather, a specific host OS)
12:49:02linuxstbJdGordon: It has to target specific devices at some point...
12:49:11 Join watto [0] (~watto@
12:49:56topikhtc desire is a shiny device. the raw power will be unusual for rockbox to experience
12:53:10kugelthe port to a specific platform is rather to prove and ensure the resulting application is really portable straight-forward. I think the port doesn't desperatively need to run flawless but it should show the groundwork that has been laid is useful
12:54:28linuxstbkugel: That's something else to think about - how to target different devices running the same OS. e.g. do we want to create many Android ports, or a univeral "run-anywhere" binary.
12:57:37topikother android music devices could be inspiration for an answer to that question
12:57:42topikmusic apps even
12:58:20linuxstbThe problem is that Rockbox currently expects to know everything about the target device at compile-time.
12:58:38linuxstb(or most things...)
12:58:49linuxstbExceptions are some optional hardware, such as FM tuners.
12:59:34kugelmost importantly, display resolution
12:59:38linuxstbBut I guess with Android we don't even know exactly what CPU we'll be running on...
13:01:05JdGordonkugel: "and the skin engine   where I was the first to use it outside of the WPS with the base skin." <- no you wasnt :)
13:02:07kugelI meant "in svn", I didn't count the various dirty hacks before that
13:02:53 Join mikroflops_ [0] (
13:03:18***Saving seen data "./dancer.seen"
13:09:19 Quit liar (Quit: Verlassend)
13:09:31 Quit planetbeing (Quit: planetbeing)
13:09:33 Join liar [0] (
13:12:50 Part watto
13:15:49pamaurygevaerts: did you ever think that there might be something wrong with usb string descriptors ? I was sniffing usb packets for usb audio under windows when I noticed it: even in Linux with svn head, most string descriptor response are ill-formed, I'm not sure if it's a
13:16:01pamauryproblem with wireshark or a real problem
13:21:16 Quit blairb_ (Quit: Leaving)
13:21:40JdGordongevaerts: re that broken theme... if its being caused by the playlist viewer then its a dupe (reported int he forums though i tihnk)... I cant imagine how that causes things to break, but ill have a looky anyway
13:27:32 Quit mikroflops_ (Ping timeout: 245 seconds)
13:28:47 Join Eugenpaul [0] (
13:37:26 Quit lpereira (Read error: Connection reset by peer)
13:37:34 Join lpereira [0] (
13:40:40 Join mikroflops [0] (
13:40:42Eugenpaulhello! I'm a one more student wishing to participate in gsoc
13:41:17EugenpaulI think I'll be able to optimize mp3 codec
13:43:27EugenpaulCan someone give me some advices? Like where can i find current rockbox decoder algorithm
13:44:02gevaertsYou'll need one of our codec people then
13:44:06*gevaerts looks around
13:44:32gevaertslinuxstb, stripwax: ping
13:44:57Bagderthe "algorithm" would be in the source
13:46:06JdGordonkugel: re SBS updating more often. according to my batt benchs the theme engine wastes no additional cpu, so I dont see any reason to not drop that update delay to something like HZ/10 or faster even
13:46:26JdGordonI understand its wasteful updating every button press (pretty much only when scrolling anyway)
13:46:28stripwaxgevaerts - pong, but I'm heading off ...
13:46:40 Quit mikroflops (Ping timeout: 246 seconds)
13:46:49gevaertsstripwax: ok, you can't help then. You know about codecs, right? :)
13:47:02stripwaxEugenpaul - in the rockbox svn code, take a look at the apps/codecs tree
13:47:04linuxstbEugenpaul: All our codecs are in the apps/codecs/ directory in the Rockbox source.
13:47:15TorneJdGordon: why does the SBS vanish while certain things are happening?
13:47:23stripwaxmp3 is implemented using libmad, in apps/codecs/libmad
13:47:24TorneJdGordon: i can understand it not updating but disappearing is weird
13:47:35JdGordonTorne: specific example?
13:47:39TorneJdGordon: splashes
13:47:39 Join mikroflops [0] (
13:47:52 Quit stripwax (Read error: Connection reset by peer)
13:47:55linuxstbEugenpaul: Do you have experience with audio codecs, or embedded programming?
13:47:58Eugenpaulwell, i know)
13:48:09JdGordonwhat is probably happening is the theme is disabled which causes a full screen clear, then it is reneabled which shuold cause an update
13:48:33TorneJdGordon: committing db does it over and over again eveyr second or so, i presume because it's looping and doing splashes repeatedly
13:48:44Tornemakes the status bar flash on and off crazily the whole time
13:48:46JdGordonIIRC splashes always disable the theme
13:49:01JdGordonwhich might be stupid...
13:49:02 Quit TillW (Quit: This now concludes our broatcast day.)
13:49:03Torneit just looks weird :)
13:49:30JdGordonah, yeah, we had the problem where if we didnt disable it we need to somehow clear the splashed area after it shuold be removed
13:49:33JdGordonwhich isnt so easy
13:50:00Tornefair enough. as long as there's a good reason :)
13:51:01kugelJdGordon: as I said, we can consider lowering it to HZ/10
13:58:29 Join hebz0rl [0] (
13:58:41Eugenpaullinuxstb: sorry for delay, i don't have real experience in this, but i know basics of mp3 decoding,
13:59:39 Join robin0800 [0] (
13:59:51BagderEugenpaul: so without experience in this field, why or how do you think you can optimize the mp3 codec?
14:00:14 Join froggyman [0] (~me@unaffiliated/froggyman)
14:00:39BagderI guess that's what your application would need to include :-)
14:01:42 Quit mikroflops (Ping timeout: 245 seconds)
14:02:01Eugenpauli have experience in c, assembler, know about signal processing, love rockbox) and really want to do it
14:02:13linuxstbEugenpaul: If you don't have any experience, then your best choice would be to start work on it now - i.e. download the Rockbox source, and try and identify areas for improvement. Also, just working on the mp3 codec alone may not be an entire summer's work.
14:02:20 Join mikroflops [0] (
14:04:01*amiconn still thinks that RaaA can't be the future of rockbox
14:04:35Bagdercan't be?
14:05:00amiconnImo being a firmware in its own right is what makes rockbox rockbox
14:05:35amiconnRaaA wouldn't be rockbox anymore, but just one among dozens of music player programs
14:05:38linuxstbBagder: Do you see it as the _only_ future for Rockbox? I can't imagine us dropping support for running Rockbox natively...
14:05:39n1sJdGordon: couldn't the update freq be um, smart, so it updates as fast as it needs, for example using the smallest time of a set max time (like 1/5s and smallest alternating subline delay and the delay of any scrolling text) or would that be too complicated
14:06:00BagderI don't see as the _only_ future, no.
14:06:15n1samiconn: i see it as *a* future for rockbox
14:06:25Bagderwe still even support charcells, I don't think we need to drop anything
14:06:30JdGordonn1s: lines are only redrawn if they are dynamic lines. the update freq isthe time between checking if lines should be updated
14:06:34n1sI'd really like a usable raaa on my netbook
14:07:10amiconnNo, as long as devices running rockbox as a firmware still exist, rockbox probably will as well
14:07:13 Join Zagor_ [0] (~bjst@2a00:801:f5:fff8:216:6fff:fe3b:5ef5)
14:07:46Bagderbut to me personally, I won't be using Rockbox unless it can continue as an app...
14:07:50n1sJdGordon: but wasn't the updatefreq discussion triggered by a user with an animation that used alternating sublines with shorter delay than 1/5s?
14:07:56linuxstbSo we all agree that the future is Rockbox running as both an application and as a native firmware?
14:08:17*amiconn doesn't see a point in RaaA
14:08:18n1slinuxstb: i do at least
14:08:45JdGordonn1s: well yes, but the sublines times change depending on which are enabled at any time... I dont tihnk adding those smarts is a good idea
14:09:06 Join watto [0] (~watto@
14:10:10JdGordonwell, at least not untill we have a single timeout for all skin-enalbed screens
14:10:21JdGordonall button loops use mostly rnadom timout values
14:10:26 Join jgarvey [0] (
14:11:47amiconnThe devices which potentially *could* benefit from RaaA can't run it, and the devices which could can also run another music player of choice
14:11:47 Join evilnick_B [0] (~0c140464@rockbox/staff/evilnick)
14:12:48linuxstbWhich devices can't run it?
14:13:11linuxstbBut I think the point is that no other music player of choice has Rockbox's features.
14:13:20 Join TillW [0] (
14:13:47amiconnfoobar2000? VLC? Just to name two
14:14:06 Quit mikroflops (Ping timeout: 264 seconds)
14:14:42linuxstbAnd they run on the devices we're talking about?
14:14:46amiconnAnd the devices which can't run it I was thinking of are those mobile phones only capable of running java stuff
14:15:13amiconnBut then those don't have much storage either
14:16:02pamauryI personally see no point in RaaA as I hate mobile-phones-that-do-everything but my opinion can change, and I understand people have an interest in that
14:16:11Bagderandroid phones have no equivalent music player, just to mention one platform
14:17:07linuxstbpamaury: I agree, I have no desire to play music on my phone. But lots of people just have (or want) one device...
14:17:22n1samiconn: foobar is windows only and i personally don't like vlc as a music player (altough i can't give any concrete arguments)
14:18:04*evilnick_B would dearly love Rockbox on the iPhone
14:18:29n1sanyway, the specialized DAP's will not go away i think, there will always be people that don't have smartphones or want to have massive storage for lossless
14:19:00pamauryFurthermore, mobile phones will never reach the battery autonomy of a DAP ! And mobile phones are huge, heavy, not very practical in some situations
14:19:20amiconnSure, foobar is windows only. That's because I just named two, and I'm more of a windows user than a linux user
14:19:52linuxstbamiconn: I haven't found any music player on Linux I like as much as Rockbox...
14:19:56amiconnVLC is cross platform though, and it's quite good
14:20:33pamaurymobile phones will soon be so powerful that you will be able to run foobar with wine with a linux emulator on your iPhone...
14:20:49 Join mikroflops [0] (
14:20:55*linuxstb wonders how Rockbox would cope with other processes adding/removing files during playback...
14:21:13gevaertslinuxstb: someone will find out :)
14:21:15JdGordonshould work fine
14:21:22JdGordonplayback just skips missing tracks
14:24:54 Join wodz [0] (
14:25:11n1scould anyone who knows coldfire take a quick look at this, looks like yet another gcc bug to me
14:26:46wodzn1s: whats the problem with that?
14:28:09n1swodz, isn't the upper half of a0 any junk that was there before after the move.w?
14:28:28wodza0 is not cleared so this may have side effects in my opinion
14:28:35n1sso then doing cmp.l on that will um, be bad=
14:29:50wodzyes You may have junk in upper word of a0 and doing cmp.l is not good idea here
14:30:06n1sok, so gcc is buggy, yay
14:30:50n1sof course, i should test a newer version
14:30:57pamauryn1s: I don't know coldfire assembly but if the move.w doesn't clear upper part, this is buggy
14:31:10n1spamaury: it doesn't
14:31:30wodzn1s: maybe gcc knows that a0 will be 0 when SATURATE() is called
14:32:05 Join froggymana [0] (
14:32:08n1swodz: sounds unlikely, since i compiled the function in its own file
14:32:24n1smaybe i should try a better testcase
14:32:37pamauryHow would gcc know ? Try to compile it with "extern" so gcc can't make any assumption, I doubt zeroing a0 is part of the calling convention
14:32:46 Nick fxb is now known as fxb__ (
14:33:11 Quit Bagder (Quit: It is time to say moo)
14:33:42 Quit S_a_i_n_t (Ping timeout: 260 seconds)
14:35:07 Quit mikroflops (Ping timeout: 265 seconds)
14:37:21 Join mikroflops [0] (
14:44:23 Join adnyxo [0] (
14:44:56 Quit n17ikh (Read error: Connection reset by peer)
14:45:18 Join n17ikh [0] (
14:47:54 Join dfkt [0] (dfkt@unaffiliated/dfkt)
14:49:40 Join m3dlg [0] (~m3dlg@
14:54:55n1sseems to be doing it in cases where it definitely gets called when a0 != 0
14:55:21*n1s puts "tet bug on new gcc on todo list"
14:56:12n1ss/gcc on todo list"/gcc" on todo list
14:59:08 Quit TheSeven (Read error: Connection reset by peer)
15:03:22***Saving seen data "./dancer.seen"
15:04:32 Quit antil33t (Read error: Connection reset by peer)
15:04:38 Join antil33t [0] (
15:17:11 Quit m3dlg (Ping timeout: 260 seconds)
15:20:19 Join grndslm [0] (
15:39:40 Join Barahir_ [0] (
15:43:52 Quit Barahir (Ping timeout: 258 seconds)
15:44:35 Quit FlynDice (Remote host closed the connection)
15:48:26 Quit Adubb (Ping timeout: 252 seconds)
15:49:20 Join Schmogel [0] (
15:50:18 Join FlynDice [0] (
15:50:51 Quit mikroflops (Ping timeout: 276 seconds)
15:51:07 Quit froggymana (Quit: CGI:IRC)
15:51:51 Join mikroflops [0] (
15:52:09 Join Adubb [0] (~aldubuc@
15:53:19 Quit pamaury (Quit: Quitte)
15:54:18 Join pamaury [0] (~pamaury@rockbox/developer/pamaury)
15:54:34 Join saratoga [0] (~9803c6dd@gateway/web/freenode/x-sjqlvdmdwtxrdoip)
15:54:41saratogaEugenpaul: ping
15:54:55 Join freddyb [0] (
15:57:03saratogaEugenpaul: i have to step out for a few minutes, but we should discuss the mp3 project, I have a lot of ideas about it
15:57:08saratogawill you be online later?
15:57:50 Quit mitk (Quit: Leaving)
15:58:40Eugenpaulsaratoga: yes?
16:00:00Eugenpaulsaratoga: i will be online now for few minutes and about 2 hours later once again
16:03:40 Quit freddyb (Remote host closed the connection)
16:05:25 Join Lixun [0] (
16:07:01 Join anewuser [0] (anewuser@unaffiliated/anewuser)
16:11:59LixunHi, I'm a student interested in participating in GSOC. How is the workload rockbox expecting me to commit?
16:12:25Lixunis it like a full-time internship or part-time?
16:12:40linuxstbIt's supposed to be full-time.
16:12:51linuxstb(I think that's google's advice)
16:13:32 Quit wodz (Quit: Leaving)
16:14:03 Join xiainx [0] (
16:14:09 Quit n1s (Ping timeout: 258 seconds)
16:14:28Lixunthank you. because i have another part-time job during vacation so I'm wondering whether I can balance two or not
16:16:09linuxstbI think it would depend on the hours you were working in the part-time job, and what kind of work it was (and of course, how many hours a day you are personably able to work effectively each day...)
16:16:56 Quit Eugenpaul (Remote host closed the connection)
16:17:53Lixunthe job is like 20hrs per week
16:17:58 Join n1s [0] (~n1s@rockbox/developer/n1s)
16:18:38Lixunit's about C# coding
16:19:35Lixunfor rockbox, I think I should be able to guarantee 30hrs at least per week
16:21:40 Join CGL [0] (~CGL@
16:23:40 Quit anewuser ()
16:24:27linuxstbLixun: Feel free to apply, but all other things being equal, I think we are probably more likely to choose an applicant who doesn't have other commitments. A full-time job is about 35-40 hours per week.
16:28:04LixunI see. Thanks.
16:28:10 Nick YPSY is now known as Ypsy (
16:29:15 Quit saratoga (Changing host)
16:29:15 Join saratoga [0] (~9803c6dd@rockbox/developer/saratoga)
16:29:33gevaertsOf course, depending on your application, all other things might not be equal
16:32:06 Join DataGhost [0] (~dataghost@unaffiliated/dataghost)
16:32:26 Quit pamaury (Quit: Quitte)
16:34:13 Quit mikroflops (Ping timeout: 276 seconds)
16:34:52 Quit JdGordon (Quit: Leaving.)
16:34:53 Join mikroflops [0] (
16:35:14Luca_Show strange. FuzeV2. I was sitting in the debug > buffering info screen while playing a song, when I moved the scrollwheel: playback skipped to the next song o_O is this the intended behavior?
16:35:38saratogaI think it is, so that you can watch what happens to the buffer on track change
16:35:43saratogaits a debug screen afterall
16:36:15 Quit TillW (Remote host closed the connection)
16:36:42Torneyes, it is intentional; the buttons that normally scroll lists are mapped to next/prev track in the buffering screen
16:36:52Luca_Sinteresting - indeed it makes sense. i was surprised because it was using the scrollwheel and not the next/prev buttons, but the prev button is used to exit the screen so it might be reasonable
16:37:09TorneYeah, it's reusing the list context iirc
16:37:18Torneso it has the same button mappings as other lists/menus
16:38:02Luca_Sthanks for the explanations
16:39:12 Join anewuser [0] (anewuser@unaffiliated/anewuser)
16:43:06niekieIs there a way to get the database viewer translated?
16:43:52niekieThings like Artist, Album Artist, Album, Genre, Composer, Track, Year, User Rating . . . etc are not translated currently.
16:44:32niekieSeems the language files aren't used for it, but tagnavi.config
16:46:44 Quit n1s (Ping timeout: 258 seconds)
16:48:18 Join toffe82 [0] (~chatzilla@
16:48:36 Join fyrestorm [0] (
16:50:57 Quit Lixun (Quit: CGI:IRC (EOF))
16:54:10 Join archivator [0] (~archivato@
16:54:56 Nick fxb__ is now known as fxb (
16:59:51 Quit Zagor_ (Quit: Leaving)
17:01:56 Quit kugel (Ping timeout: 248 seconds)
17:02:25 Join Xerion [0] (
17:03:25***Saving seen data "./dancer.seen"
17:05:46 Join TillW [0] (
17:07:07archivatorIs there any interest in a self-sustained minimalistic maintenance app for Rockbox? Rbutil Lite - something that can be put on your DAP Autorun-style and do basic maintenance every time your player connects to a pc - purge the database of expired entries (last time I checked, rockbox only marks them as expired), check for newer versions and run a file system check? I was thinking of something sub-1MB that just works. I remember distinctly a project to
17:07:07archivator merge elf and pex executables into one file, that could be pretty sweet.
17:07:18 Join Blue_Dude [0] (~chatzilla@
17:07:53 Join pamaury [0] (~c2c7a50a@gateway/web/freenode/x-qtznnlcxrpbjnmuk)
17:08:15 Quit pamaury (Changing host)
17:08:15 Join pamaury [0] (~c2c7a50a@rockbox/developer/pamaury)
17:09:56Blue_DudePlease check out FS #11178 - bookmark update. It needs a little tweaking to remove an unused token but it's more or less ready to go.
17:10:49 Join m3dlg [0] (~m3dlg@
17:17:30 Quit toffe82 (Quit: ChatZilla 0.9.86 [Firefox 3.6.3/20100401080539])
17:21:18 Quit Blue_Dude (Quit: ChatZilla 0.9.86 [Firefox 3.6.3/20100401080539])
17:23:49 Join togetic [0] (
17:23:50 Quit togetic (Changing host)
17:23:50 Join togetic [0] (~togetic@unaffiliated/ibuffy)
17:24:06 Quit mikroflops (Ping timeout: 260 seconds)
17:24:47 Join mikroflops [0] (
17:24:59 Join webguest88 [0] (
17:25:46 Join TopyMobile [0] (
17:25:47 Quit robin0800 (Remote host closed the connection)
17:26:52 Quit m3dlg (Ping timeout: 252 seconds)
17:27:59 Join webguest48 [0] (
17:27:59 Quit webguest88 (Client Quit)
17:28:06 Quit lpereira (Quit: Leaving.)
17:29:00 Quit webguest48 (Client Quit)
17:30:39 Join funman [0] (~fun@rockbox/developer/funman)
17:31:00 Join toffe82 [0] (~chatzilla@
17:31:25 Quit mikroflops (Ping timeout: 276 seconds)
17:31:50GodEaterwhilst RaaA on the iphone is probably interesting for a niche crowd, I'm still against using EmbeddedOSX as a target given the problems we'd face ever getting it accepted into the app store. It would only ever run on jailbroken devices and we cannot guarantee that JBing will always be possible.
17:32:16GodEaterAndroid desperately needs a decent music player - the one on the N1 is barely adequate
17:32:23saratogayou're concerned about official support for a rockbox port?
17:32:31saratogawhen has that even been an issue for us in the past?
17:32:39GodEaterit hasn't of course
17:33:07GodEaterbut the audience of iphone users that JB is much smaller than the audience of Android users who wouldn't have to
17:33:28linuxstbGodEater: I don't think anyone thinks embedded OS/X is the priority target (or one we should ask a student to do). Or do they?
17:33:34 Join mikroflops [0] (
17:33:51GodEaterlinuxstb: I was just commenting on evilnick_B's statement earlier
17:34:07saratogawell its very unixy and has very nice dev tools and wide library support
17:34:20saratogaso its probably at least as good a target from a technical stand point as Android
17:34:20GodEaterI was also about to say I agree with your statement about lack of decent media players on linux in general
17:34:27GodEaterI'd love RaaA as a native desktop app
17:34:49Tornesaratoga: apart from in the GUI..
17:34:52GodEatersaratoga: the dev environment is nearly all Objective-C isn't it?
17:34:57 Quit fyrestorm (Quit: Ur skills' fireproof like a wooden panel -- U got feds talking leet on your IRC channel!)
17:34:57Tornewhere you are more or less required to use Cocoa Touch
17:35:06saratogaverses android and java :)
17:35:20Torneporting the UI of an existing app to iphoneos is pretty hard
17:35:27saratogaimo objective c using gcc is a bit closer to what we want then java
17:35:29*linuxstb also isn't keen on Android, but we have to choose something...
17:35:52GodEaterlinuxstb: you already said you're not interested in running RaaA on a mobile at all ;)
17:36:02saratogaI think OSX is probably the easiest target followed closely by pure linux phones and then android
17:36:09 Quit xiainx (Ping timeout: 260 seconds)
17:36:12linuxstbGodEater: Oh yeah... I'll shut up then. ;)
17:36:17saratogamostly because theres so much information available and it uses gcc
17:36:29linuxstbWouldn't maemo be easier?
17:36:38archivatorAndroid has a native development kit, right?
17:36:39GodEaterthe other big issue with the iPhone as a target means you ideally need OSX as a dev platform
17:36:49GodEatersetting up the iPhone tool chain on linux, whilst possible, is not fun
17:36:57GodEater(speaking as a man who's done it)
17:37:03saratogai think the linux tool chain is adequate though
17:37:15GodEatersaratoga: did you ever set it up?
17:37:24GodEaterarchivator: yep, there's the NDK for android
17:37:42saratogaand probably not much worse then setting up embedded tools for most linux phones which typically involved pirating various libraries . . .
17:37:59saratogaGodEater: no but I've read about people using it in the distant past
17:38:00GodEateryep, you definitely need to do that
17:38:15GodEaterNico_P and I both set it up, and it's really not fun
17:38:16 Quit mikroflops (Ping timeout: 265 seconds)
17:38:21archivatorWell, that's pure C with a way to communicate with the dalvik layer, isn't it? Sounds reasonable to me, given that most of the stuff can be kept outside of dalvik.
17:38:37 Join mikroflops [0] (
17:38:55 Join geertvdijk [0] (
17:39:33GodEaterthe toughest bit is extracting some of the libs from the the X-Code image files on linux
17:40:03saratogai don't think you're pirating anything, IIRC the SDK is freely available from Apple, you're just misusing it a bit
17:40:15GodEatersure - pirating is a strong term
17:40:18 Quit r2k000 (Quit: r2k000)
17:40:19GodEateryou're violating the EULA
17:40:48 Join killan [0] (
17:41:02 Join hd [0] (
17:41:02 Quit hd (Changing host)
17:41:02 Join hd [0] (~jd@Wikipedia/HellDragon)
17:41:32saratogafor the moto smartphones I think you actually have to BT stuff they never officially released to the public, i.e. actual piracy not just ignoring a EULA clause
17:41:38 Join ender [0] (
17:41:52 Quit Schmogel (*.net *.split)
17:41:52 Quit Barahir_ (*.net *.split)
17:41:52 Quit ender` (*.net *.split)
17:41:52 Quit jd (*.net *.split)
17:41:52 Quit parafin (*.net *.split)
17:41:52 Quit killan_ (*.net *.split)
17:41:52 Quit Topy44 (*.net *.split)
17:42:03 Join Kitr88 [0] (~Kitr88@
17:42:33 Join xiainx [0] (
17:42:41 Quit TopyMobile (Ping timeout: 258 seconds)
17:42:45 Join Schmogel [0] (
17:42:45 Join Barahir_ [0] (
17:42:45 Join ender` [0] (
17:42:45 Join jd [0] (~jd@Wikipedia/HellDragon)
17:42:45 Join parafin [0] (
17:42:45 Join Topy44 [0] (
17:42:58 Quit jd (Max SendQ exceeded)
17:44:35 Quit Schmogel (Ping timeout: 270 seconds)
17:44:35 Quit ender` (Ping timeout: 270 seconds)
17:45:02 Quit Kitar|st (Ping timeout: 246 seconds)
17:46:24 Join n1s [0] (~n1s@rockbox/developer/n1s)
17:47:18 Quit Kitr88 (Ping timeout: 276 seconds)
17:48:34GodEaterat least with android the SDK is very easy to acquire
17:48:42GodEatercan't speak for maemo
17:50:06archivatorregarding gsoc, is it a problem if I work at a slow pace till the end of June? I have exams (quite a few, and all difficult at that) and would hate to choose between gsoc and actual academics :)
17:50:56geertvdijkGodEater: Maemo is supposed to be easy to develop for, although I don't think there's an off-the-shelf sdk such as you might find with android. I think it's just the editor/IDE you prefer + Qt libs
17:51:15geertvdijkor GTK+
17:51:23 Join Kitar|st [0] (
17:51:38funmanthere's "hildon" although i don't know if it's just a set of gtk+ widgets or something else
17:52:10archivatorgeertvdijk: there's this:
17:52:29geertvdijkfunman: Hildon I believe is indeed just a set of GTK+ widgets designed for mobile
17:52:32 Nick ender is now known as ender` (
17:52:39geertvdijkarchivator: didn't know that, thanks!
18:04:25 Join kugel [0] (~kugel@rockbox/developer/kugel)
18:04:30 Join TopyMobile [0] (
18:05:03 Join komputes [0] (~komputes@ubuntu/member/komputes)
18:08:35TorneMaemo is currently GTK+ with the Hildon widgets if you want them, yeah. Qt is not until the next version of the OS
18:10:59gevaertsYou can do Qt right now if you want though
18:13:15Torneit only integrates nicely fairly recently, though, i believe
18:13:21Tornethe migration path is only just getting going
18:15:10gevaertsI don't know the details very well
18:15:14 Join domonoky [0] (~Domonoky@rockbox/developer/domonoky)
18:15:20funmanpressing buttons have no effect on Clip+ when running test_boost
18:15:33funmansame with unless i mark action as volatile
18:15:58funmanand then going by the disassembly (of the .rock) the only difference is r0 (return of get_action) is stored on the stack then read back immediately after
18:16:47kugelmaemo would also be an interesting target imo
18:16:48 Quit mikroflops (Ping timeout: 258 seconds)
18:17:13kugelplus, I'd actually like to learn how I can make it run on my mini2440 ;)
18:18:04funmanah nevermind now it works
18:18:14 Join mikroflops [0] (
18:18:15 Join stoffel [0] (
18:18:34*funman adds that to the "random bug" list
18:19:06kugelis that the same one as the mind blow list? :)
18:19:39funmanprobably, perhaps with enough tests (power off+power on) we would see the good value
18:19:56saratogado we know why test_codec doesn't work if the codec is in RAM on the clipv2/+?
18:20:43funmani didn't know it didn't work anymore let me look
18:21:02kugelsaratoga: it should be in iram now
18:21:08funmankugel: btw why did you make this change in (and do you know why it didn't work on fuzev1?)
18:22:05funmanwith your change the init code would just be in plugin ram instead of codec ram
18:22:12kugelI somehow thought putting the init code in the iram (where codec is now) clashes with the rockbox' portion of the iram
18:22:42kugeldon't ask me why I thought that, and don't ask me why it didnt work (actually I was just about to ask you the same)
18:22:58kugelah :\
18:23:02kugelright, I remember why
18:23:12 Join bertrik [0] (~bertrik@rockbox/developer/bertrik)
18:23:32kugelinit calls loads skins at some point, and skin loading uses the plugin buffer for caching the .wps/.sbs files
18:23:44saratogametadata parsing fails according to the error message
18:24:01funmansaratoga: ah i had the same problem but didn't find the reason
18:24:08funmanmetadata parsing works fine in the wps
18:24:22kugelsaratoga: did you do a clean recompilation (nuking the build dir)? that often helps if linkage-related stuff changed
18:24:46funmanthat happens with clean build, i had seen that when i tried to put the codec buffer in iram
18:25:30funmanwell not the 'failed' message but metadata length was 0
18:25:35 Join Eugenpaul [0] (
18:25:59funmani'll try md5sum to see if file reading is really reliable
18:26:46Luca_Smaybe that could explain the WPS backdrop sometimes disappearing on fuzev2? bitmap failing to load?
18:26:47archivatordomonoky: I have updated FS #11160 as per our discussion. I can't think of anything else to fix/do. The rest is making SAPI and Carbon thread-safe, really.
18:27:05funmanLuca_S: come on, that's the only problems you can think of ? there's more :)
18:28:17Luca_Snot as many as it may seem! today I reorganized all my music in a single folder so I could play it in rockbox as fine as in the OF :)
18:28:23 Join TheSeven [0] (~theseven@rockbox/developer/TheSeven)
18:35:47*bertrik wonders how many jpeg implementations we have
18:36:17funman4 runs boosted, 4 runs unboosted, md5sum OK for the same 4.2 Mb file
18:36:33 Quit anewuser (Quit: Like renoise + like music? 3 days to submit your entry!)
18:38:08domonokyarchivator: looks good. 3 comments: 1. what is this strange comment in encodeList ? (the list is surely used later). 2. The skipping of empty toSpeak values seems to have got lost (or is it not needed anymore ?) 3. There are TABS instead of spaces in this patch.
18:38:44funmanah but the plugin boosts
18:40:19funmanmd5sum still OK on 2 runs
18:41:27 Join bmbl [0] (~Miranda@unaffiliated/bmbl)
18:42:54funmanwell playing with test_boost doesn't make the Clip+ crash faster (it is boosting/unboosting several times already when it uses a codec)
18:43:32Luca_Stoday I noticed that when boosting the fuzev2 has frequent screen corruptions (missing pixels, color shift)
18:43:50Luca_Swhen unboosted they're still there, but less frequent
18:43:55funmani see corruptions too
18:45:45bertrikare the display data pins still shared with other functions on the fuze v2?
18:46:01archivatordomonoky: 1) Yes, it is, but as far as I can tell, we can safely delete the entries we're not going to process from it. Duplicates would point to the same file as another entry, so they can go. Unvoiced entries never created a file, so they can go as well. That's all I meant by "not used" - that the cleanup routines won't mind us deleting a few things from it (especially *these* things)
18:46:58archivatordomonoky: 2) It's right there: line 154 in the patched file - "if (!entry.voiced && !entry.toSpeak.isEmpty())" 3) damn, I thought I had gotten vim to behave! Will fix and reupload.
18:47:21 Join Lear [0] (chatzilla@rockbox/developer/lear)
18:48:00 Quit Luca_S (Quit: CGI:IRC)
18:48:23 Join panni_ [0] (
18:49:55 Quit scorche (Ping timeout: 252 seconds)
18:50:39 Join scorche [0] (~scorche@rockbox/administrator/scorche)
18:51:24funmanbertrik: you mean buttons?
18:51:46bertrikfunman, yes, for example
18:51:51funmandoesn't seem so
18:52:37domonokyarchivator: about 1) the list is used later by the filecopiing from talkfile.cpp, but it probably doesnt need the invalid entrys.
18:52:53pamauryfunman: in svn there is a file called usb-drv-as3525.c, it contains some init code for a usb driver. What is this usb driver ? Of which device(s) I mean.
18:53:56archivatordomonoky: well, yes, exactly - we basically make the list have distinct entries filename-wise. I'll change the comment to reflect that, I can see how it can be confusing.
18:54:09domonokyarchivator: removing the dublicates might get you problems with talkfiles.
18:54:11funmanpamaury: i tested it on Clipv1 or Fuzev1 (don't remember which one) but it doesn't work at all
18:54:22 Quit Lear (Quit: ChatZilla 0.9.86 [Firefox 3.6.3/20100401080539])
18:54:48domonokypamaury: there might be dublicate strings to talk, but they could exist in different directorys, so the filecopiing needs all entrys.
18:54:49funmani have some more code here which i tested on Fuzev1 but it works randomly, some times i have an USB interrupt, some times I see that Linux saw the device in dmesg, sometimes nothing happens
18:55:22ranmabertrik: Re dbop, you said you had an idea what might cause the button readout noise on C200v2?
18:55:31domonokyups, archivator instead of pamaury :-)
18:55:43bertrikranma, yes, it goes away if we disable the tristate between accesses
18:56:00 Join Lear [0] (chatzilla@rockbox/developer/lear)
18:56:45domonokyarchivator: removing entrys where voicing failed might be fine, but the dublicates are a problem.
18:57:35 Join Slim [0] (
18:57:41Slimafternoon ..
18:58:38archivatordomonoky: I'm not sure I follow. The talkfiles are duplicated, not the encoded filenames. I haven't looked at the actual values but I always assumed they held the whole path. I'm not sure I understand how having two identical entries contributes anything to the process.
18:58:58 Quit scorche (Ping timeout: 258 seconds)
18:59:11 Join scorche [0] (~scorche@rockbox/administrator/scorche)
18:59:17archivatorOh, sorry.
18:59:26bertrikfunman, I'm checking if those buttons are actually not sharing pins with DBOP
18:59:33archivatordomonoky: I was confusing wav and talk, I see now.
18:59:54funmanbertrik: you have a fuzev2 ?
19:00:06SlimCan anyone tell me how to determine if a DB initialisation is complete? I have an 80G video iPod running the latest build, and 540 albums. Just don't want to reboot it until it's finished.
19:00:19bertrikI suppose this part is the same as on the as3525
19:00:38archivatorYeah, that makes sense, but how would I skip them (assuming they've been voiced successfully)? I would need yet another flag to check..
19:00:58ranmabertrik: Ok, I'll try that.
19:01:31funmankugel: itcm/dtcm registers read as 0 on Clip+
19:03:28***Saving seen data "./dancer.seen"
19:03:53domonokyarchivator: couldnt you just set the status of the dublicate entry to encoded=true ? then the encoderstep can skip it.
19:04:28 Quit antil33t (Read error: Connection reset by peer)
19:04:34 Join antil33t [0] (
19:04:57domonokyjust check the encoded flag before encoding, similar to how its done for voicing.
19:05:38bertrikfunman, on as3525, GPIO port C is shared with the DBOP data pins, which also are connected to most of the buttons on the fuze v2. So if some interrupt routine is switching these to input to read them, it might disturb display writing and explain the screen corruptions
19:05:38archivatordomonoky: yeah, that would work. Will fix it tonight.
19:05:45domonokyarchivator: good :-)
19:06:06bertrikof course I'm not sure if this is the same for the SoC in the fuze v2, since I don't know the pin map of that
19:06:11ranmabertrik: I changed dbop_read_input to not tri-state, but I'm still seeing the noise.
19:06:12 Quit einhirn (Quit: Miranda IM! Smaller, Faster, Easier.
19:06:32bertrikweird, that seemed to fix it for me
19:06:55Eugenpaulsaratoga: ping
19:07:29funmanbertrik: ok, I don't know if kugel tried to read the buttons from dbop
19:07:54 Join mitk [0] (
19:08:29Eugenpaulsaratoga: it's me again with mp3 project
19:09:36ranmaFWIW: Baseline (music or no music, no display updates): almost zero noise, about 20% during display updates (HW info), abou 27% during display updates (HW info) while playing music.
19:10:25ranma(Used this patch to check for noise and show percentage in HW info: ranma/rockbox-c200v2/rockbox-c2x0v2-backlightpatch-20100407-patchset/dbop_noise_check.patch">
19:10:26funmandid you try to force boosting?
19:12:25Slimok, thanks anyway - I guess it's a hard one.
19:12:33 Quit Slim (Quit: CGI:IRC)
19:13:10 Quit Adubb (Ping timeout: 240 seconds)
19:13:53funmanranma: btw about the backlight, any problem with setting both pins?
19:15:42ranma20% while playing with forced boost.
19:15:53ranmafunman: Yes, A5 is buttonlight on mine.
19:16:40FlynDicefunman: (or any other Samsa folks)My head is exploding with v2 #ifdef's, do you have a problem with splitting set_cpu_frequency() into a v1 and v2 like this:
19:17:11funmanranma: and D7 is nothing?
19:17:35funmanFlynDice: no problem
19:17:35ranmaD7 is the second buttonlight (menu text below power button)
19:17:49funmanthere are 2 button lights? Oo
19:17:55ranmaBut A5 is the main one (with 4 leds aparently)
19:18:07funmanhm ok
19:18:15funmanI don't even know if both are enabled by D7, bertrik ?
19:18:22funman=> on the other model
19:18:44 Join Adubb [0] (~aldubuc@
19:19:19bertrikboth on c200v1 and c200v2, they are both enabled
19:19:21 Join mikroflops_ [0] (
19:20:51funmanranma: did you find how the OF decides which one to use?
19:21:39ranmaYes, it's switching A7 to input and reads the value to decide which one to use.
19:21:42 Quit mikroflops (Ping timeout: 265 seconds)
19:22:38bertrikdidn't I post that information here already on IRC some time ago?
19:22:47CIA-5New commit by FlynDice (r25496): SansaAMS: Only use INT_MCIO with as3525v1, it is unused in as3525v2.
19:22:53CIA-5New commit by FlynDice (r25497): AS3525v2: Set XPD to SD-MMC interface in sd_init() for HAVE_MULTIDRIVE. ...
19:22:56funmanok, what do we need for it to be in svn ?
19:22:58CIA-5New commit by FlynDice (r25498): Sansa AMS: Split set_cpu_frequency() into 2 separate functions for as3525v1/v2 as the code is quite different for each model.
19:24:02 Quit mikroflops_ (Ping timeout: 246 seconds)
19:24:23funmanranma: looks good, D7 just needs to be set unconditionally of the variant?
19:24:53funmando we ask bertrik to test, or commit and see if someone reports problems in the forum?
19:26:06ranmabertrik: Did you test the rockbox binary I sent you?
19:26:25 Join Horscht [0] (~Horscht2@xbmc/user/horscht)
19:26:41ranmaIf not yet, please do rather test ranma/rockbox-c200v2/">
19:28:11ranmafunman: I'd vote for committing ;)
19:28:25FlynDiceI see the red on r25498 but its not my doing.....
19:28:26ranmaBTW the OF does the check in init_hardware at 0x8000
19:29:35ranmaDoes sd access generate events that cause _buttonlight_on() to be called?
19:29:40*funman slaps FlynDice with an archos player
19:29:45funmanhm that's heavy :/
19:30:11ranmaIf not, sd access is so noisy that it defeats my noise check in dbop_read_input()...
19:30:21*FlynDice rubs eye and swears he didn't touch nuthin over there.....
19:30:23funmanranma: no but since sd accesses yield() there could be another thread doing it
19:31:26ranmaBut another thread should only be calling it if there is some button action, right?
19:33:01bertrikranma, your zipfile works, I can control the backlight brightness now
19:33:34ranmaI've half-added the buttonlight to backlight pwm in that patch so that the buttonlight brightness is increasing a bit when _buttonlight_on() is called and during initial buffering I see heavy buttonlight flicker. :)
19:33:50funmanranma: grep only shows drivers/button.c and plugin api
19:34:01 Join mt [0] (~chatzilla@
19:34:09ranmabertrik: The working part is where it detects the variant correctly ;)
19:34:58ranmabertrik: It should show 'C200v2 variant 0' in HW info
19:35:31ranma(And if it would missdetect you should have a dark backlight anyway)
19:35:54bertrikyes it does say variant 0
19:36:23bertrikI can see the display flicker a bit if I move the player around
19:36:37bertrikI'll check if this is also present in the OF
19:36:44ranmabertrik: Yeah, that's because of the software pwm. Looks the same in OF
19:36:57ranmaIf you increase brightness to max it should be constant on.
19:37:49bertrikthat's why I set it to exactly half brightness :)
19:38:16funmanranma: did you run battery_bench already ?
19:38:38ranmaOk, so I suppose I'll commit the backlight variant detection real soon now :)
19:38:55ranmafunman: I did one once
19:39:04funmanyou remember the runtime?
19:39:40ranmaShould be in the forum thread, IIRC about 8-9 hours.
19:39:48bertrikdo you think the backlight has its own variant, or is it the whole player that has two variants?
19:39:52 Join Luca_S [0] (
19:40:11bertrikI can confirm the OF flickers too, but it seems a bit less obvious than rockbox
19:40:54ranmaHmm, make that 4:17
19:41:08funmanranma: hm did you run it after my fix for FM radio?
19:41:20ranmaI was about to say that was before your FM radio fix :)
19:42:26ranmabertrik: It's most likely a higher frequency. The current one uses the 100Hz timer and doesn't increase frequency for PWM.
19:46:02 Join mikroflops [0] (
19:46:47 Quit TillW (Remote host closed the connection)
19:53:08 Quit kugel (Ping timeout: 240 seconds)
19:53:17 Join Strife89 [0] (~michael@
19:57:37 Quit leavittx__ (Read error: Operation timed out)
19:57:40 Quit leavittx (Read error: Operation timed out)
19:58:50 Quit Lear (Quit: ChatZilla 0.9.86 [Firefox 3.6.3/20100401080539])
19:59:55 Quit leavittx_ (Ping timeout: 258 seconds)
20:00:42 Join freddyb [0] (
20:01:09 Quit Luca_S (Quit: CGI:IRC (EOF))
20:01:19 Quit hd (Read error: Connection reset by peer)
20:01:46 Join jd [0] (
20:01:47 Quit jd (Changing host)
20:01:47 Join jd [0] (~jd@Wikipedia/HellDragon)
20:02:04 Quit funman (Quit: free(random());)
20:03:36 Join Luca_S [0] (
20:04:02 Join kugel [0] (~kugel@rockbox/developer/kugel)
20:05:19 Join saratoga_lab [0] (~9803c20d@gateway/web/freenode/x-hhgpfavtipetxiki)
20:05:28saratoga_labEugenpaul: ping
20:05:40 Join leavittx_ [0] (~leavittx@
20:06:35 Join TillW [0] (
20:06:36Eugenpaulsaratoga_lab: i'm here
20:08:40saratoga_labEugenpaul: i suggested the mp3 project
20:08:57saratoga_labthe idea was to look at libmad and other open source decoders and figure out how fast they could be made
20:09:08saratoga_labthen make improvements in libmad until it was as fast as possible
20:09:26saratoga_labat present we have ideas about how to improve libmad, but no solid idea how good the codec itself is compared to other decoders
20:09:57 Quit stoffel (Remote host closed the connection)
20:10:13saratoga_labthe project would involve learning about mp3 decoding and going piece by piece through the codec comparing to other implementations
20:11:55 Join leavittx [0] (~leavittx@
20:12:48saratoga_labits a difficult project in that we don't have a good idea how much improvement is possible
20:12:54 Join leavittx__ [0] (~leavittx@
20:13:13saratoga_labyou said you had done some DSP before, but never any codec stuff right?
20:13:59 Join wodz [0] (
20:14:07Eugenpaulyes, i know about dsp from university
20:16:39saratoga_labwhat stuff have you learned about?
20:18:51 Join _zic [0] (
20:19:50 Quit n1s (Ping timeout: 258 seconds)
20:19:58freddybDid some testing on Fuze v2 recording and the FIFO runs out in the do_sw_pwm function. I tried changing the INT_TIMER2 so that it stacks do_sw_pwm in front of whatever was executing on return so that do_sw_pwm would be interruptible but I don't know ARM well enough. (The linker called me some very bad names.)
20:21:59kugelfreddyb: can you try ? it removes the scrollwheel portion of the button read
20:24:14Eugenpaulsaratoga_lab: i'm trying to remember... it was something like fourie transform, multiplexing...
20:24:45saratoga_labEugenpaul: can you be more specific, DSP is something you can take a class in or a doctorate, knowing where you are along that line is helpful
20:25:06 Quit _zic (Quit: Leaving)
20:25:33kugelfreddyb: is the fifo so small? I find it strange that it doesn't surive the tick tasks
20:26:30kugelwe could do it like the of, it blocks button reading if the display is busy but I would rather avoid that
20:28:27freddybkugel: i'm compiling your scrollwheel patch now
20:29:15Eugenpaulsaratoga_lab: it was classes
20:29:40freddybFIFO is 32 words
20:29:54 Quit TillW (Ping timeout: 248 seconds)
20:30:44 Quit mitk (Quit: Leaving)
20:31:20wodzI found out that to play mp3 on my MPIO hd200 it runs at ~100MHz while other coldfire based DAPs needs ~30MHz. Could it be that difference comes from generic ata code vs. asm optimized reads?
20:31:32kugelis there some fifo almost empty interrupt? maybe it's possible to run that in the fiq so it's possible to refill it during the timer isr
20:31:35gevaertsunlikely I think
20:31:56saratoga_labwodz: i would check that you have IRAM properly defined and that you are enablign all the coldfire ASM code in the codecs
20:32:11saratoga_labmissing either of those would result in a huge decrease in performance on coldfire
20:32:26kugelwodz: you probably need to go through the codec files and look for IRAM defines which are not yet enabled for *your* cf target. coldfire's ram is slow
20:32:49CIA-5New commit by ranma (r25499): Detect C200v2 variant by reading A7, use A5 or A7 to control backlight and buttonlight depending on the result.
20:32:51 Quit flydutch (Quit: /* empty */)
20:32:54saratoga_labI think IRAM should be all or nothing, very few defines are target dependent
20:33:14wodzsaratoga_lab: whats strange ogg decoding is realtime ~120MHz
20:33:16saratoga_labcheck the .map file for libmad and see if it links to IRAM
20:34:02freddybKugel: We interrupt on half full, almost full, and full. I tried adding almost empty yesterday without any difference.
20:34:03saratoga_labor vorbis if you prefer
20:34:22 Quit TopyMobile (Ping timeout: 246 seconds)
20:34:41kugelfreddyb: we'd need the fiq in order to interrupt irqs
20:34:52Eugenpaulsaratoga_lab: maybe i understand you incorrect? what should i say? my knowledge is not deep enough, but i think i have basics and i am ready to learn
20:35:13saratoga_labEugenpaul: no I just want to know what you've already learned about DSP
20:35:39kugeli believe it's currently used for the pcm callback, but audio playing (with pcm involved) and recording at the same time is not possible so it should be possible to use it for the recording callback
20:36:58freddybKugel: I don't think it would take much to use fiq instead of irq but the way we do irq's the fiq will not trump it. I think the whole do_sw_pwm is in uninterruptible irq mode. I tried some changes based on the sample code in PL190 datasheet but didn't have any luck.
20:37:37kugelthe fiq is a special interrupt on arm which can interrupt irqs
20:41:00freddybKugel: I like the patch, can the same thing be done for the buttons? It seems stable for recording at 44kHz but still crashes on 96kHz.
20:41:16 Nick fxb is now known as fxb__ (
20:41:46kugelI dont think it can, the button pins don't change without setting some other pins before
20:42:13wodzsaratoga_lab: from part is linked to dram part to iram
20:42:31saratoga_labwodz: that should be fine then
20:42:55 Join TillW [0] (
20:43:27saratoga_labcan you also verify that it defines "CPU_COLDFIRE" in the codecs?
20:43:51saratoga_labthat enables EMAC instructions in libmad, which are very important for performance
20:44:03saratoga_labif that is defined, I would also check that ICACHE is properly enabled
20:44:30saratoga_labalso that you've clocked the CPU as high as you think you have :)
20:44:41bertrikgevaerts, does the IO priority commit also apply to flash targets?
20:44:43kugelfreddyb: how do you like the scrollwheel behavior with it?
20:44:58freddybVery much.
20:45:05gevaertsbertrik: it's enabled for flash. Whether or not it helps much is a different matter
20:45:15bertrikdid it get slower or faster?
20:45:19wodzsaratoga_lab: hmm where should be CPU_COLDFIRE defined?
20:45:26kugelfreddyb: better than before?
20:45:56kugelwodz: config.h probably
20:46:42freddybThe scroll speed seems the same but much more responsive.
20:47:44wodzok I definitely have CPU_COLDFIRE defined than
20:47:58 Join TopyMobile [0] (
20:48:03*archivator just found about Cepstral LLC and their fabulous voices
20:48:14archivatorPerhaps we should add their engine to rbutil?
20:48:38wodzCC apps/codecs/libmad/imdct_mcf5249.S makes me think it takes the correct one
20:49:21 Join fml [0] (
20:50:12fmlWouldn't the sansa C200v2 variant checking code an ideal candidate for the init func (recently introduced)?
20:50:52kugelI think so
20:51:09kugelranma: I'm not sure we want " /* vim:set ts=4 sw=4 et: */" in our files
20:52:51gevaertsI'm pretty sure we don't want it
20:54:49fmlkugel: I have no SVN access now so if you want you can introduce the attribute
20:55:08bertrikwhat init func?
20:55:51fmlbertrik: function that's only run once during the initialisation and then overwritten
20:56:06bertrikah ok
20:56:50bertrikgevaerts, I'm not noticing any obvious delays or speedups on my e200
20:57:36gevaertsbertrik: well, I don't think it can cause delays at all, and speedups will be limited due to flash already being fast
21:01:28archivatorI'd like to repeat a question from earlier today: is there any interest in a small cross-platform maintenance utility to be run Autorun-style? I'm thinking database purging, file system checks and updating as part of the functionality. Something small that runs everytime you connect the player to a PC..
21:02:27gevaertsarchivator: if you want to do that, go ahead, but personally I think it's not a wonderful idea
21:03:31***Saving seen data "./dancer.seen"
21:03:51archivatorWell, if I'll be doing it, I'd like it to be more than a long script. I see it as an rbutil extra, updated as part of the tree.
21:05:26gevaertsI don't really understand what you mean by database purging, filesystem checks can't really work if the tool to run them is on the same filesystem (and I think automated filesystem checks on plugin are a bad idea anyway), and I think automatic updates are a bit dangerous
21:06:17 Quit fml (Quit: CGI:IRC)
21:08:18archivatorgevaerts: last I checked, rockbox doesn't really delete entries from the database, it just sets a flag to skip them. That's what I mean - actually removing the deleted entries. A filesystem check can check for errors (and, on Windows, fix some of them). A warning along the lines of "Your filesystem is corrupted, fix it!" can be quite helpful. Automatic updates can be dangerous but they can also be something the user wants.
21:13:41gevaertsWouldn't a host-side thing be better? i.e. a set of udev scripts, a windows service, that sort of thing?
21:16:17archivatorCould be, I'm just brainstorming at this point. A windows service is a clumsy thing, though. I doubt I can get it to be notified on device insertion (I think you need a window for that). Plus, it would mean maintaining 3 projects instead of one cross-platform one (Qt-based, for example).
21:16:45bluebrotherarchivator: I'm not too fond of such an idea.
21:17:43bluebrotheryou can already install Rockbox Utility on the player. Why not include a new "cleanup" task group that can do such stuff?
21:18:29 Join Sergio [0] (
21:18:33bluebrotheralso, adding command line parameters to Rockbox Utility is something that's on my todo list since ages. Unfortunately the current design makes such a thing rather hard to implement (one of the issues I hope to address when reworking the UI)
21:19:10archivatorbluebrother: RockboxUtility is not exactly small. Plus, it would need a way to do things on startup (which it lacks). Also, I'm not sure how it handles being on the player but will changing mount points affect it?
21:19:47 Quit freddyb (Remote host closed the connection)
21:19:52bluebrotherarchivator: yes, but I'm planning to change this (i.e. check if we have a portable installation and if yes automatically adjust the mountpoint)
21:20:02bluebrotherit seems portable installation is rarely used, though.
21:20:07 Nick adnyxo is now known as needhelp (
21:20:51bluebrotherthe problem with the size of Rockbox Utility is Qt −− linking it statically simply makes it big. So I don't think it's possible to create such a maintainance utility with less than 1MB if it's Qt based.
21:21:22bluebrotherhowever, the ~4MB it has on Windows is acceptable IMO
21:21:45 Nick needhelp is now known as adnyxo (
21:22:07archivatorbluebrother: in any case, a new utility would be far smaller than Rockbox Utility (which, imho, contains far too many things to be considered adequate for the task). Plus, you can use UPX on Windows.
21:22:48 Quit Sergio (Ping timeout: 240 seconds)
21:23:06bluebrotherwe already upx Rockbox Utility ;-)
21:24:32bluebrotherbut what about maintainance for such a new utility? Given how many people actually work on Rockbox Utiltiy I don't think there will be a lot of people working on such a tool either.
21:26:05archivatorWell, there won't be much to maintain. It would only need to reflect changes in download locations and database structure.
21:27:20archivatorBoth seem pretty stable and can be made incredibly easy to change (actually, the database bit can be taken from core rockbox - similar to how the database utility works)
21:27:53bluebrotherhmm. Well, feel free to work on such a thing. I'm not really convinced though.
21:29:49 Part watto
21:32:04 Quit RadicalR (Ping timeout: 245 seconds)
21:32:53 Join Sergio [0] (
21:37:43CIA-5New commit by bertrik (r25500): Make array static const in apps/recorder/jpeg_load.c
21:41:51 Join kugel_ [0] (
21:42:07 Quit kugel (Disconnected by services)
21:42:11 Nick kugel_ is now known as kugel (
21:42:15 Quit kugel (Changing host)
21:42:15 Join kugel [0] (~kugel@rockbox/developer/kugel)
21:45:46bluebrotherdomonoky: any thoughts on a new rbutil release? I think we have a nice bunch of fixes now.
21:45:59archivatordomonoky: I uploaded a new version of FS #11160
21:46:08bluebrotherunfortunately there was less response on my call for translators than I hoped :/
21:48:49domonokybluebrother: no problem with a new release. should we commit this tts/encoder paralellising now, or after the release ? :-)
21:49:35bluebrotherIMO doing it after a release would be better −− it gives us more time to try and fix possible issues :)
21:50:29domonokyhm, noone uses svn builds but developers, so putting it into the release will probably give faster reports of broken things :-)
21:50:47archivatorI think it should be left for after release if only so we can get the other engines parallelized as well.
21:50:50*bluebrother hopes for moos to look after the french translation soon
21:51:13bluebrotherdomonoky: well, I do provide svn binaries for non-developers.
21:51:43bluebrothermaybe I should implement something to count downloads of those files. Too bad I don't have access to the server logs
21:53:05pixelmain case it needs testing on MacOS 10.4...
21:53:39domonokypixelma: not currently, TTSCarbon isnt currently paralelizied.
21:54:16domonokybut general rbutil testing on Mac 10.4 before release would be surely good :-)
21:54:58archivatorSAPI would be tricky, though. I can't think of a way to parallelize it without serializing access to the script's stdin (which would effectively serialize the entire process, more or less)
21:56:08pixelmawell, I helped testing the voice fixes and remember having installed a theme to test back then and a build (I believe). I did not try a bootloader install though - neither separately nor wirh a "complete installation"
21:56:27pixelmaor with
21:58:01 Quit xiainx (Ping timeout: 246 seconds)
21:58:34pixelmathe theme was a 160x128 grey one for my M5 which reminds me of a specific feature request/bug report for RbUtil and another idea we discussed a while ago
21:59:29*bluebrother just pushed the weblinguist code to github, in case someone is interested. It isn't really nice code right now, though.
22:00:36pixelmahmmno - the feature request/bug report would actually be for the theme site
22:00:51Eugenpaulsaratoga_lab: I'm about my dsp experience. So, this is what i can remember from lectures (and what i think is useful in project): modulation, discretization, quantization, multiplexing by everything, coding (includes and Huffman coding), masking. Is it still too abstract?
22:01:19saratoga_labEugenpaul: no thats a fine description
22:02:15Eugenpaulsaratoga_lab: and should i know anything else?
22:02:54saratoga_labwell the application should be specific and convince us that you're qualified for the task
22:02:57 Quit Xerion (Ping timeout: 252 seconds)
22:03:11 Join JohannesSM64 [0] (
22:03:15bertrikIs there anything special I have to think of when committing a language patch?
22:03:32saratoga_labmaybe having secondary goals would be a good idea (working on aac decoding or some other codec) in case the mp3 stuff goes quickly
22:03:35 Quit hebz0rl (Quit: Ex-Chat)
22:04:42Eugenpaulok, thank you
22:05:10pixelmabertrik: a new language or an update?
22:05:10 Join robin0800 [0] (~robin0800@
22:05:21 Quit bmbl (Quit: Bye!)
22:05:26bertrikan update, from flyspray
22:06:05 Nick shaggy-h is now known as chrism (
22:07:23pixelmaa language which is spoken natively by a committer? Other than that... check if the translator is in CREDITS and also translators are usually mentioned in the lang file heade
22:10:49 Quit pamaury (Quit: Page closed)
22:11:19 Quit kugel (Ping timeout: 265 seconds)
22:11:39 Quit Strife89 (Ping timeout: 268 seconds)
22:15:54 Quit ender` (Read error: Connection reset by peer)
22:16:28 Join ender` [0] (
22:20:49CIA-5New commit by bertrik (r25501): Slovak language update ...
22:22:52archivatordomonoky: I'm starting to think the parallelization should be user-controlled. At least there should be an option to turn it off. I have reason to believe festival splits the work into two processes already and with things like multisyn voices we might be slowing down the process rather than speeding it up.
22:25:52 Join kugel [0] (~kugel@rockbox/developer/kugel)
22:27:03 Quit wodz (Quit: Leaving)
22:29:04 Join notlistening [0] (
22:29:26 Quit Zarggg (Quit: Zarggg)
22:29:29notlisteningHi just tried the latest build on the clip+ oh dear ;)
22:31:36FlynDicenotlistening: Is that a "This is all screwed up" oh dear or a "this is fantastic" oh dear?
22:32:57 Quit kugel (Remote host closed the connection)
22:33:19 Join Zarggg [0] (
22:35:43notlisteningit's a wow this is screwed up oh dear and now it is a can not get it to screw up again oh dear just trying to reproduce the problem
22:36:20notlisteningthe inability to write to disk is a bummer
22:37:18 Join bieber [0] (~bieber@
22:38:03bieberAre Summer of Code applications publicly visible as they're submitted, or do they just get shown after decisions are made?
22:38:49gevaertsThe abstracts for ones that are selected will be shown after the decisions
22:41:38saratoga_labI'm thinking about the show settings and the "show music" option specifically
22:41:52saratoga_labperhaps it would make sense to change it to "show media" so that mpg files are included as well
22:42:29saratoga_labit seems like more people would rather see both then see just one, and it can be quite confusing to new users
22:42:30 Quit scorche (Ping timeout: 264 seconds)
22:43:12saratoga_labat least my understanding is that the option is mostly to keep you from seeing playlist files and album art, so it seems like seeing mpg files would be useful as well
22:43:28gevaertsthat makes sense I guess
22:43:39 Quit mikroflops (Ping timeout: 260 seconds)
22:44:16notlisteningmakes sense to me to
22:44:22saratoga_labthe use case that would be impacted negatively by this would be people who have mpg files mixed in with their music and do not want to see the mpg files
22:45:04 Part froggyman
22:45:11 Join mikroflops [0] (
22:45:29LloreanI'd say that "Show Media" shouldn't be an option until we have videos in the playlist system, and then it should be an additional option rather than replacing "Show Music"
22:45:50bieberPerhaps "Show Music," "Show Videos," and "Show Media" settings would be best?
22:45:50bieberWith the third being the union of the first two, of course
22:46:00LloreanI can't imagine why "Show Music" is confusing to new users, though
22:46:08LloreanIt's a quite explicit name, and it's not the default.
22:46:35saratoga_labpeople seem to ask about it
22:46:38 Join BdN3504 [0] (
22:46:42gevaertsMedia might actually be confusing. Does that include images? Text?
22:46:44saratoga_labi think that option tends to get toggled without people thinking about it
22:46:51Lloreansaratoga_lab: People have honestly asked "I've set it to show music and can't find my videos?"
22:46:59saratoga_labin the forums just now
22:47:11LloreanHe knew he'd set it to show music?
22:47:18LloreanOr it was set to show music without him knowing (something we can never solve)
22:47:26saratoga_labwell if he did he didn't mention it, but he set it to music at some point
22:47:35LloreanThat's my point - as long as there's *any* setting that can hide their music (such as "Show Playlists") then it can't be solved.
22:47:36saratoga_labpresumably because he didn't realize that would exclude video
22:47:51BdN3504if i restore an ipod on a pc using the "software version 1.3" for apple ipod gen 5, will i be able to rockbox it? or do i need to manually choose a different version of the apple software?
22:48:02Lloreansaratoga_lab: I'd bet he accidentally set it due to the quickscreen's bad controls
22:48:10LloreanIt's hard to leave the quickscreen without changing the "show files" option
22:48:14saratoga_labhmm well that should be fixed then
22:48:22saratoga_labmaybe the show files option should be a little harder to find
22:48:26LloreanIt's not the setting that's the problem, but the fact that you can set that setting without even knowing it.
22:48:27 Quit robin0800 (Read error: Connection reset by peer)
22:48:35 Join robin0800 [0] (~robin0800@
22:48:38*Llorean has said "show files" shouldn't be on the quickscreen for a while.
22:48:51saratoga_labi still think its odd that show music doesn't show mpg files, which are quite off music videos and such
22:49:04LloreanBut anything on the vertical axis on the quickscreen is going to trigger accidental changes. You hold "Menu" on iPods to get there, but can't tap "Menu" to leave it (like you can many other things)
22:49:11saratoga_labi do agree with it not being on the quickscreen FWIW
22:49:24bluebrotherBdN3504: I'm not aware of any specific version required when restoring an Ipod with Itunes.
22:49:24Lloreansaratoga_lab: Change "Show Music" to "Audio Files"
22:49:55BdN3504bluebrother: i am more concerned about rb compatibility
22:50:00notlisteningbluebrother did you manage with the wine detection bit for rbutil?
22:50:02saratoga_labit still seems more natural to me to group video and audio together since mpg contains audio
22:50:20LloreanGames contain audio too...
22:50:26bluebrotherone thing I really *hate* about the 4th option on the quickscreen: you can't leave the Quickscreen with the same button you entered it (something that worked before, and it's *really* annoying on an ipod)
22:50:37Lloreanbluebrother: I just mentioned that. :)
22:50:57Lloreansaratoga_lab: There should definitely just be a setting for "formats that can be played while in the WPS" whatever it's called.
22:51:16LloreanOr "formats that can be playlisted"
22:51:26bluebrotherLlorean: ok, I'm obviously too slow then :)
22:51:29saratoga_labwell i think in the long term those limitations on video are kind of annoying and should be removed
22:51:40saratoga_labso i don't really see that as an argument for the setting now
22:51:59bluebrotherIMO we should really consider dropping the quickscreen completely and replace it with a user customizable menu (now the Quickscreen is customizable)
22:52:00 Join scorche [0] (~scorche@rockbox/administrator/scorche)
22:52:05Lloreansaratoga_lab: Until this limitations are removed, though, the setting can clean up a directory view for people only wanting to see things they can insert in their playlist.
22:52:07 Quit Sergio (Ping timeout: 240 seconds)
22:52:22LloreanIt'd be silly to make a mess of things now on the *hope* a feature will exist in the future that we've been saying should exist for years now.
22:52:32bluebrothernotlistening: yes, see r25471
22:52:52saratoga_labbluebrother: i also like that idea
22:52:58Lloreanbluebrother: Personally, I think the quickscreen should show the "load .cfg file" screen, in which the user can have as many .cfgs of their favorite settings and combinations of settings as they like
22:53:01bluebrotheri.e. just replace the quickscreen with a list a user can put any option in he likes.
22:53:08LloreanNo complication creating custom menus (or a UI to create them) but all the power of it, and more.
22:53:21bluebrotherlike "my menu" various devices have these days
22:53:33saratoga_labi don't know about cfg files, but i do dislike hwo the current quick screen works
22:53:41saratoga_labi tend to get into it by accident and then change settings trying to get out of it
22:53:48bluebrotherLlorean: well, IMO a simple list should be rather straight forward
22:53:50saratoga_labits just ackward, at least on the players i've used
22:54:00bluebrotherthe quickscreen was nice. Until it got changed
22:54:11Lloreanbluebrother: I just think .cfg files would actually be quicker to change (more like the current quickscreen)
22:54:39LloreanBut since the quickscreen is not likely to ever support the old Archos Rockbox quickscreen functionality, I'd say it should be dumped for some sort of list-based menu now.
22:54:56bluebrotherLlorean: probably, but it requires more work to set it up (and that might be quite problematic for average users)
22:55:43Lloreanbluebrother: Add to the context menu of each setting "add current value to .cfg" like inserting things into the playlists in playlist catalog?
22:55:46bluebrotheryep, that's my point. If we get the old way back (didn't that work on h100 at some point?) there's no much point in keeping it as such a special case it is
22:56:00LloreanBut a single customizable menu in place of the quickscreen also isn't bad.
22:56:18LloreanI'm pretty sure the H100 is the reason it stopped working - button combinations necessary for it to work weren't available.
22:56:42LloreanIt was some reason like that, for some player at or shortly after.
22:56:50bluebrotherLlorean: that of course makes it easier. However, you still have to enter a filename which seems awkward for creating a customizable menu.
22:57:25bluebrotheriirc Mode + Joystick can get recognized on h100. Unless my memory serves me wrong.
22:58:05bluebrotherfrom my understanding the reason is that it broke at some time, and few developers knew (and know) how that feature was intended to work in the first place
22:58:14pixelmawasn't it Play which you could use with other buttons on the H100ß
22:58:43 Join Strife89 [0] (
22:58:48 Quit Eugenpaul (Quit: Eugenpaul)
22:58:51bluebrotherhmm. Probably play, yes.
22:58:55pixelmathere are no (usable) button combos on the Iaudios though
22:59:54bluebrotherspeaking of buttons: why does the jpegviewer plugin not invoke the menu on long Select on Ipods? Pressing a combo is awkward when plain Select does nothing
22:59:57BdN3504Rockbox works fine, if an ipod 5.5g gets restored to the apple ipod software version 1.3
23:00:39bluebrotherBdN3504: there is no dependency between Rockbox and the apple firmware version. If the bootloader runs Rockbox will run.
23:02:51 Join planetbeing_ [0] (
23:03:29 Quit scorche (Ping timeout: 264 seconds)
23:03:33***Saving seen data "./dancer.seen"
23:04:10 Join scorche [0] (~scorche@rockbox/administrator/scorche)
23:04:19 Quit Zagor (Quit: Clint excited)
23:06:03 Nick Ypsy is now known as YPSY (
23:06:38 Quit notlistening (Remote host closed the connection)
23:07:44 Quit BdN3504 (Quit: CGI:IRC (EOF))
23:11:19 Nick fxb__ is now known as fxb (
23:13:06 Quit chrism (Ping timeout: 240 seconds)
23:14:17 Quit dockimble (Quit: WeeChat 0.2.1)
23:14:48 Join dockimble [0] (~dockimble@
23:19:04 Quit liar (Quit: bye boomshanka)
23:27:32 Join Surge97 [0] (
23:28:13 Quit goffa (Ping timeout: 276 seconds)
23:32:41 Join goffa [0] (~goffa@
23:33:02 Join froggyman [0] (~me@unaffiliated/froggyman)
23:40:57 Join stripwax [0] (
23:43:36 Quit bieber (Quit: Leaving)
23:47:48 Quit planetbeing_ (Quit: planetbeing_)
23:48:07 Quit bertrik (Quit: Leaving)
23:48:13 Join Blue_Dude [0] (
23:48:20 Quit GeekShadow (Quit: The cake is a lie !)
23:48:28 Quit evilnick_B (Quit: Page closed)
23:49:35CIA-5New commit by Blue_Dude (r25502): Bookmark.c cleanup, still no functional changes... yet
23:54:52 Quit archivator (Quit: sleep time!)
23:56:52 Quit Blue_Dude (Quit: ChatZilla 0.9.86 [Firefox 3.6.3/20100401080539])

Previous day | Next day