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 2009-05-25

00:00:16Unhelpfulgevaerts: obviously we can't "really" operate on the filesystem while running as a USB block device, but perhaps there could be a middle ground, of treating the filesystem as read-only, and updating any cached FS structures that are written to by the USB host
00:00:49gevaertsUnhelpful: that would provide a pretty efficient cache, but is it really worth it?
00:00:51Unhelpfulthe benefit being that we might be able to identify file reads and pre-cache blocks before they're requested.
00:01:24Unhelpfulie, "host read the directory, then the first block of this file, and we know where the rest of the file is stored"
00:01:41gevaertsIf filesystems are reasonably defragmented, just getting the next chunk in advance will be good enough in 95% of cases
00:02:13Unhelpfula fat filesystem that's reasonably defragmented? ;)
00:02:28gevaertssurely those exist :)
00:03:15*Unhelpful is certain they are purely theoretical ;)
00:03:50gevaertsSeriously though, we're working with 64k blocks. I'd expect even a horribly fragmented fat filesystem to gain a lot from reading 64k in advance, Not as much as by knowing the FAT, but good enough
00:04:54gevaertsactually, in the FAT case we can just use the FAT itself and ignore directories altogether. That will tell us where the next block is
00:04:57Unhelpfulfair enough then. there is also the side effect that the dircache, if loaded into RAM, could be kept synced.
00:05:18 Quit kushalone (Client Quit)
00:05:19Unhelpfuli thought the FAT itself was only used for files in the root
00:05:43gevaertsisn't the FAT this huge linked list of blocks?
00:06:30gevaertsI'm a bit afraid to make a horribly complicated thing that turns out to improve read speed (which almost nobody notices anyway) by 2%
00:07:46 Quit miepchen^schla ()
00:08:00gevaertsAlso, I expect that pre-reading the wrong block from disk will actually make things slower
00:08:37 Quit amiconn (Nick collision from services.)
00:08:40 Join amiconn_ [50] (n=jens@rockbox/developer/amiconn)
00:08:49 Quit pixelma (Nick collision from services.)
00:08:49 Join pixelma_ [50] (n=pixelma@rockbox/staff/pixelma)
00:09:00 Nick amiconn_ is now known as amiconn (n=jens@rockbox/developer/amiconn)
00:09:03 Quit HellDragon (Client Quit)
00:09:07 Nick pixelma_ is now known as pixelma (n=pixelma@rockbox/staff/pixelma)
00:11:44 Join kushalone [0] (n=kushal@
00:13:32 Join MarcGuay [0] (
00:13:42kugelJdGordon1: I think I found something accaptable for the splash issue
00:14:25 Join HellDragon [0] (
00:14:35saratogakugel: I was wondering about the clock issues being discussed in the New Ports thread, are they important to getting the MMU working?
00:14:52kugelsaratoga: I don't know
00:15:25saratogait seems like battery benchs aren't too important without MMU
00:15:29kugelbut probably, as we seem to be too fast after enabling mmu&caches. That won't get better with optimized frequencies
00:15:47saratogai expect the relative importance of some of the bus clocks to change quite radically once the caches are on
00:17:26kugelif FlynDice's find helds true and we can do mp3 @ 90MHz without caches or mmu, I'm wondering how urgent getting the caches to work is then
00:18:53saratogathats on the clip?
00:19:09kugele200v2 iirc
00:19:41saratogamp3 is almost entirely out of IRAM, so what hes probably doing is optimizing the IRAM access times
00:20:09kugelhe maxed the pclk
00:20:13saratogathat said I dislike the idea of running without the CPU caches, you're not supposed to do that
00:22:21kugelwhich is the clock for peripherals (which includes the as3514 codec) IIUC
00:22:22 Join miepchen^schla [0] (
00:22:43kugelsaratoga: sure, I didn't want to say we shouldn't
00:23:38 Quit SirFunk (Read error: 110 (Connection timed out))
00:24:00 Quit pixelma (Nick collision from services.)
00:24:00 Join pixelma_ [50] (n=pixelma@rockbox/staff/pixelma)
00:24:07 Join Llorean1 [0] (
00:24:18 Nick pixelma_ is now known as pixelma (n=pixelma@rockbox/staff/pixelma)
00:25:10 Join faemir [0] (
00:26:28 Quit archivator ()
00:28:43 Quit MarcGuay_ (Read error: 110 (Connection timed out))
00:30:35 Quit kushalone (Client Quit)
00:40:31 Quit Llorean (Read error: 110 (Connection timed out))
00:55:03 Quit ender (" It's a good thing money can't buy happiness. We couldn't stand the commercials.")
00:56:41 Nick Llorean1 is now known as Llorean (
00:56:46 Quit bertrik (Read error: 113 (No route to host))
00:57:03 Quit DrMoos ("Rockbox rules the DAP world")
01:11:25 Join fyrestorm [0] (
01:12:14 Quit fyrestorm (Read error: 104 (Connection reset by peer))
01:22:00saratogasvn is really crawling tonight
01:25:23 Quit barrywardell (Read error: 104 (Connection reset by peer))
01:25:54 Join barrywardell [0] (n=barry@rockbox/developer/barrywardell)
01:28:05saratogais doing % 2 on an int32 going to use a divide or a shift ?
01:28:09 Join Thundercloud [0] (
01:35:13***Saving seen data "./dancer.seen"
01:40:22linuxstbsaratoga: It should just use an AND
01:41:34saratogaof course, thanks!
01:42:36CIA-38New commit by saratoga (r21071): Commit FS #10234 - Spacerocks respawn invulnerability by Eric Clayton. Gives you a couple seconds of invulnerability after respawn.
01:43:33 Quit barrywardell (Remote closed the connection)
01:51:52Unhelpfulx % 1, but gcc should be doing that for you
02:04:10kugelyou mean & ?
02:08:29Unhelpful... of course. this appears to be a bad day for me. i'll go back to editing the scaler ;)
02:11:12 Quit HellDragon (Client Quit)
02:13:52 Quit jordan` (Read error: 104 (Connection reset by peer))
02:14:02 Nick fxb is now known as fxb__ (
02:14:59 Join jordan` [0] (i=gromit@
02:15:58 Quit jordan` (Read error: 104 (Connection reset by peer))
02:17:18 Join jordan` [0] (i=gromit@
02:17:36 Quit jordan` (Read error: 54 (Connection reset by peer))
02:18:49 Join jordan` [0] (i=gromit@
02:24:07 Quit jordan` (Read error: 104 (Connection reset by peer))
02:30:37 Join HellDragon [0] (
02:32:52 Join jordan` [0] (i=gromit@
02:36:08 Quit Llorean ("Leaving.")
02:37:50 Quit efyx_ (Remote closed the connection)
02:40:04 Join cool_walking_ [0] (i=cb3b81c3@gateway/web/ajax/
02:42:39 Quit HellDragon (Client Quit)
02:46:42CIA-38New commit by saratoga (r21072): Forgot to add Eric Clayton to the CREDITS file for FS #10234.
03:11:37 Quit gevaerts (Nick collision from services.)
03:11:49 Join gevaerts [0] (n=fg@rockbox/developer/gevaerts)
03:19:13 Join HellDragon [0] (
03:25:44kugelJdGordon1: so, what was the reason to deactivate statusbars before loading a plugin?
03:30:19 Quit Thundercloud (Remote closed the connection)
03:35:17***Saving seen data "./dancer.seen"
03:38:23 Join perrikwp [0] (i=4aa794a0@gateway/web/ajax/
03:50:02saratogakugel: do we clock off of the internal 24MHz oscillator on AMS?
03:50:36kugelI don't know
03:58:48 Join toffe82 [0] (
04:02:14 Quit kugel ("exit(0);")
04:17:54 Quit saratoga (" ajax IRC Client")
04:30:15 Join tmzt [0] (
04:34:38 Join secleinteer [0] (n=bend@
04:35:20secleinteerhi, i found this ( thread on the ml, and i was wondering if any progres has been made since on running rockbox on android
04:35:25secleinteeranyone know?
04:40:31tmzthow would that even work? android is java-based, rockbox is c even the sdl version
04:40:32 Part toffe82
04:43:28Unhelpfuluntil there's a native-code C API for android, it's unlikely that there will be a proper rockbox port
04:43:38 Quit miepchen^schla (Read error: 101 (Network is unreachable))
04:44:12secleinteertmzt: well that message is from almost a year ago...maybe there have been some developments
04:44:14martian67since android is not a single cpu
04:44:18martian67i doubt that will happen
04:44:44martian67or even a single archetchure
04:44:55martian67its made to run with java, they dont care much about C apis methinks
04:46:28Unhelpfulmartian67: as long as there is a C compiler, a fair amount can be done without worrying much about CPU-specific details. gcc has libraries for supporting almost all C math/logic operations on CPUs that don't have instructions for them.
04:47:07tmztis there a sdl pseudo-target for console linux?
04:47:20martian67well of course, but that wouldnt be a port to android
04:47:34martian67it would be a port to one or many of the platforms android runs on really
04:50:02Unhelpfultmzt: there are the SDL-based simulators.
04:50:22tmztyes, but those are currently X or win32 based?
04:50:38Unhelpfulmartian67: i'm just saying, a C API is a real possibility, and it will be needed if CPU-intensive code is ever to be part of an application.
04:50:42tmztI know there was the beginnings of an ezx port, using the simulator code I think
04:50:47Unhelpfultmzt: they are SDL based.
04:51:09tmztis SDL used for audio as well?
04:51:31martian67Unhelpful, mmh of course
04:53:10Unhelpfuli don't know, but i can't imagine it would use anything else when it already depends on SDL for graphics and input...
04:53:59tmztokay, I might try using that for some of the htc-linux projects which has a similar approach to ezx, it would save writing a lot of code for a media application
05:00:14 Quit AndyI (Read error: 110 (Connection timed out))
05:07:21secleinteeri don't know much about java; does it have support for calling c code, like python does?
05:07:27 Join lymeca [0] (
05:07:35secleinteerperhaps that could be used to port rockbox?
05:09:15secleinteermoreover, android is oss, so it should be possible to create a 3rd party api
05:11:45 Join [TiZ] [0] (
05:12:36 Part [TiZ] ("Leaving")
05:25:47 Join [TiZ] [0] (
05:26:47[TiZ]Hi. I'm using Rockbox 3.2 on my 5th generation iPod, and I need disk mode. Built-in disk mode is not working, and the original firmware won't boot. Is there any way at all to activate Rockbox's disk mode in 3.2?
05:27:23cool_walking_No, there isn't because it just isn't even compiled in in 3.2.
05:27:48[TiZ]I see. Bummer. :(
05:28:56Unhelpful[TiZ]: use a current build.
05:29:39[TiZ]I intend to if I manage to get the iPod to connect to my computer somehow.
05:31:03cool_walking_You could try to get an adapter for the HDD, so you can connect it the your PC and install a current build that way.
05:32:17[TiZ]I suppose that is an option, for last resort.
05:34:56cool_walking_I think you may have a bigger problem if disk mode isn't working.
05:35:21***Saving seen data "./dancer.seen"
05:35:41[TiZ]I should hope it's restricted to the OF. Rockbox itself is working perfectly fine.
05:36:56cool_walking_But, (asking more knowledgable people here) isn't the OF separate from disk mode?
05:41:05 Nick JdGordon1 is now known as JdGordon (
05:47:49 Quit Horscht ("Verlassend")
05:55:12 Quit mirak ("Ex-Chat")
05:58:40 Join BHSPitLappy [0] (n=BHSPitLa@unaffiliated/bhspitmonkey)
06:14:25 Join Llorean [0] (
06:27:44*JdGordon has the proof-of-concept background WPS patch working
06:30:49JdGordonI think as long as the WPS doesnt try to update the viewport where the UI is it should work pretty easily...
06:31:07 Quit [TiZ] ("Leaving")
06:34:06JdGordonarg, the e200 cabbie wps doesnt use conditional viewports :?
06:34:23JdGordon /?/////////// :p
06:35:27 Quit fdinel ("Miranda IM! Smaller, Faster, Easier.")
06:43:58JdGordonbah! why does my test .wps not work but cabbie does?!
06:51:45 Join LinusN [0] (n=linus@gateway/web/cgi-irc/
06:51:56 Part LinusN
06:52:03 Join LinusN [0] (n=linus@rockbox/developer/LinusN)
06:59:27JdGordonvery possibly
06:59:49JdGordonno, all the statis wps tags need a bit of work to get working correctly
07:18:34 Quit Rob2223 (Read error: 54 (Connection reset by peer))
07:18:37 Join Rob2222 [0] (
07:19:22 Join renke [0] (n=renke@
07:30:06 Quit advcomp2019 (Read error: 104 (Connection reset by peer))
07:30:28 Join advcomp2019 [0] (n=advcomp2@unaffiliated/advcomp2019)
07:35:24***Saving seen data "./dancer.seen"
07:36:20 Join matsl_ [0] (
07:39:49 Join Lss [0] (
07:40:10 Quit bubsy ("Mrrrrreow!")
07:48:46 Join Mathiasdm [0] (
07:50:16 Join bertrik [0] (
07:57:59 Quit BHSPitLappy ("Ex-Chat")
08:02:28 Join ender` [0] (
08:06:44 Quit Mathiasdm ("Yuuw!")
08:07:59SlasheriBagder: hi, unfortunately i have to stop providing the rbclient service from because that host will be moved to a new virtual machine. But maybe i could allocate a separate virtual machine for the rbclient after that :)
08:15:45 Quit jmillikin (Read error: 110 (Connection timed out))
08:26:33 Quit Tristan (Remote closed the connection)
08:27:39 Quit bertrik (Read error: 60 (Operation timed out))
08:29:55 Quit renke (Read error: 110 (Connection timed out))
08:36:42 Join petur [50] (n=petur@rockbox/developer/petur)
08:36:54 Join fyrestorm [0] (
08:37:23 Join Zagor [242] (n=bjorn@rockbox/developer/Zagor)
08:37:24 Join Rob2223 [0] (
08:54:40 Quit Rob2222 (Read error: 110 (Connection timed out))
09:00:24 Join flydutch [0] (
09:13:56 Quit Lss (Read error: 113 (No route to host))
09:14:19 Join Lss [0] (
09:15:45 Quit itcheg (" ajax IRC Client")
09:34:45 Join petur2 [50] (n=petur@rockbox/developer/petur)
09:35:25***Saving seen data "./dancer.seen"
09:39:41 Quit Rob2223 ()
09:44:57 Join efyx_ [0] (
09:46:05 Join Rob2222 [0] (
09:52:18 Quit petur (Read error: 110 (Connection timed out))
09:56:07 Quit fyrestorm (Read error: 104 (Connection reset by peer))
10:09:32 Join fyrestorm [0] (
10:12:34 Join Sedgewick [0] (n=Sedgewic@
10:26:39markunBagder: well, it *is* :)
10:26:53*Bagder thirds that
10:27:26markunthe backdrop is out of place?
10:27:39BagderI don't know what he means by that
10:28:31ZagorI think he just means it doesn't suit his taste
10:28:48Bagderprobably, yes
11:02:06 Quit cool_walking_ (" ajax IRC Client")
11:11:31 Nick petur2 is now known as petur (n=petur@rockbox/developer/petur)
11:35:28***Saving seen data "./dancer.seen"
11:38:17 Join mcuelenaere [0] (
11:38:39 Join dfkt [0] (i=dfkt@unaffiliated/dfkt)
11:41:44 Join mcuelenaere_ [0] (
11:42:25 Quit mcuelenaere (Read error: 60 (Operation timed out))
11:42:44 Nick mcuelenaere_ is now known as mcuelenaere (n=quassel@rockbox/developer/mcuelenaere)
11:49:59 Nick fxb__ is now known as fxb (
12:05:28mcuelenaereUnhelpful: does something like look good?
12:06:25 Nick fxb is now known as fxb__ (
12:27:53 Join renke [0] (n=renke@
12:33:47 Join Vytenis [0] (
12:34:42 Quit Vytenis (Client Quit)
12:34:59 Join VytenisS [0] (
12:35:19CIA-38New commit by funman (r21073): mkamsboot: really error out if OF model is different from bootloader model ...
12:36:38 Join Thundercloud [0] (
12:37:34 Join mirak [0] (
12:37:44VytenisSHello, I need a little help. When i generate .lng file with genlang ant choose clip as the target the language file wont work on player, when chossen it's not accepted and gets reset to the previous language, but for example if i choose e200 as a target and place the .lng file on my clip it will work, but however wont show language specific symbols
12:37:52 Join funman [0] (n=fun@rockbox/developer/funman)
12:46:54 Join FrankTM [0] (
12:57:20 Join Horscht [0] (i=d53dbb62@gateway/web/ajax/
13:12:30CIA-38New commit by mcuelenaere (r21074): * read_bmp_*(): add FORMAT_RETURN_SIZE ...
13:14:14FrankTM!seen atz
13:17:45 Join einhirn [0] (
13:18:25 Quit einhirn (Read error: 54 (Connection reset by peer))
13:18:49 Join einhirn [0] (
13:29:09amiconnmcuelenaere: lua.rock doesn't link for crosscompiled win32 sims...
13:29:37amiconnTons of "undefined reference to `_errno'"
13:31:37tmztwould it need something like mingw?
13:33:31mcuelenaereamiconn: since last commit or even before?
13:33:53amiconnI have no idea what it did before. Didn't build rockbox for some time...
13:33:59 Join Casainho [0] (
13:35:32***Saving seen data "./dancer.seen"
13:42:48mcuelenaerehmm is up to date?
13:45:09mcuelenaereI get '*** Your compiler (/usr/bin/colorgcc) does not produce Win32 executables!' upon configuring
13:45:17*mcuelenaere guesses colorgcc is messing up things
13:45:59 Quit HellDragon (Read error: 104 (Connection reset by peer))
13:47:32amiconnmcuelenaere: Same problem exists on Cygwin
13:49:51amiconn (linux)
13:55:11mcuelenaereamiconn: I'm able to reproduce them
13:57:50mcuelenaerehmm rockpaint seems to also user errno, but ifdef's them out for simulators and mingw
13:58:04mcuelenaereerr #if !defined(SIMULATOR) || defined(__MINGW32__) || defined(__CYGWIN__)
14:00:15CIA-38New commit by mcuelenaere (r21075): Fix mingw & cygwin builds
14:02:16 Quit Horscht (" ajax IRC Client")
14:02:20funman;topic=14064.0;attach=3694;image : does the very fast battery drop indicate wrong battery levels ?
14:08:03FrankTMdoes anyone know how to determine weather i have a sansa fuze v1 or v2
14:08:28krazykit FrankTM you check the firmware version.
14:09:57FrankTMis it that simple :P
14:10:44krazykitit is. i mean, you could open it up too if you like. either way.
14:10:53krazykitthat's a v1 fuze, then
14:18:17 Join robin0800 [0] (
14:19:28VytenisSAnyone know why clip is'nt showing any symbols that is not in the english alphabet ? For example if i chose russian language that it wont show even a word, just everything black
14:20:15mcuelenaereVytenisS: you probably don't use an unicode font
14:21:19VytenisSyup, you are right
14:22:25 Join kugel [0] (n=kugel@rockbox/developer/kugel)
14:22:42VytenisSOne more question, why clip wont load a .lng file that i generated with genlang with clip as a target, but will load if i choose e200 as a target ?
14:23:54Bagdervery good question
14:23:58Bagderthat sounds... weird
14:24:18funmanVytenisS: did you run a diff between the files ? also with the .lng generated in normal clip build ?
14:24:55VytenisSdiff between the files ? I am quite new at this
14:26:03kugelfunman: hey
14:26:04funmanVytenisS: you can do like that: xxd file1.lng > file1.txt ; xxd file2.lng > file2.txt ; diff -u file1.txt file2.txt
14:26:22funmanhi kugel :)
14:26:35kugelfunman: What's that number being written into CCU_SRC?
14:26:36VytenisSok will try that
14:27:04kugelin system_init()
14:27:23funmankugel: devices to be reset mask : all devices but ROM, VIC, IRAM & DRAM (last 4 bits)
14:27:55kugelCCU_SRC has only 16bits
14:28:51funmanit has 24 for me (datasheet) and is defined as long in as3525.H
14:29:32kugelI have v1.13 of the DS, and it has 16bits
14:29:48funman5882ac9c50e5a0e6b57b12042e56e63e AS3525_Datasheet_v1_13.pdf
14:30:08VytenisSfunman: I can't build .lng file with the normal clip build only with genlang. When compiling it shows: make: *** No rule to make target `/home/vytenis/rockbox/rockbox/lithuanian/apps/lang/Lithuanian.lng', needed by `/home/vytenis/rockbox/rockbox/lithuanian/lang/max_language_size.h'. Stop.
14:30:11kugelah lol
14:30:30kugelthe other 8bits are on the other page
14:30:34funmanah i understand :) description starts at bit 16 on the 2nd page
14:30:49funmanVytenisS: try make clean before ?
14:31:29 Quit funman ("leaving")
14:31:59kugelthe number is still more than 24bit, isn't it?
14:32:25VytenisSshows the same error after make clean
14:34:52kugelhave you added the lang to SOURCES?
14:34:59kugelapps/lang/SOURCES I mean
14:35:04VytenisSYes I did
14:36:13VytenisSWell found mine mistake, i wrote Lithuanian with capital in the SOURCES
14:37:26 Quit tvelocity (Read error: 104 (Connection reset by peer))
14:51:13 Join AndyI [0] (i=AndyI@
14:53:19 Part secleinteer
14:56:18 Quit jfc (Read error: 54 (Connection reset by peer))
15:00:58 Quit renke (Read error: 110 (Connection timed out))
15:03:29 Join Mathiasdm [0] (
15:06:08 Join barrywardell [0] (
15:14:50 Quit perrikwp (" ajax IRC Client")
15:15:00 Join perrikwp [0] (i=4aa794a0@gateway/web/ajax/
15:17:50 Nick fxb__ is now known as fxb (
15:18:52 Quit Sedgewick (Connection timed out)
15:22:20 Quit kugel (Read error: 110 (Connection timed out))
15:22:53 Join Mathiasdm2 [0] (
15:33:00 Join {phoenix} [0] (
15:33:13 Quit Mathiasdm2 ("Yuuw!")
15:34:24 Quit Mathiasdm (Read error: 110 (Connection timed out))
15:35:34***Saving seen data "./dancer.seen"
15:52:19 Join LambdaCalculus37 [0] (n=rmenes@rockbox/staff/LambdaCalculus37)
15:56:34 Join kugel [0] (n=kugel@rockbox/developer/kugel)
15:57:45 Join n1s [0] (n=n1s@rockbox/developer/n1s)
16:00:57 Quit Casainho ("ChatZilla 0.9.84 [Firefox 3.0.10/2009042316]")
16:15:01Unhelpfulamiconn: as it happens, i was reading the wrong programmer's manual, since addi is not an sh-1 instruction. what i was doing was reformulating the 32x32->top32 multiply a bit, so that it includes a rounding add, and based on the assumption that the top 24 bits are always zero and are discarded.
16:15:26 Join Tristan [0] (
16:16:15Unhelpfulwhat i ended up with was this:
16:17:01LambdaCalculus37Just curious... who was working on the majority of the GoGear ports? Was it lowlight?
16:18:49 Quit Thundercloud (Remote closed the connection)
16:19:25 Join funman [0] (n=fun@rockbox/developer/funman)
16:20:48Unhelpfulmcuelenaere: that looks good, i would amend it to this, though:
16:21:20CIA-38New commit by mcuelenaere (r21076): Lua: port viewports + add test_viewports.lua
16:21:26Unhelpfulthen you can get the size needed to load the image at the desired size, do the load, and realloc down when you get the actual image size
16:22:53mcuelenaereproblem is: I'm not sure how to realloc Lua data.. (if that's even possible)
16:24:07 Quit perrikwp (" ajax IRC Client")
16:24:20 Join perrikwp [0] (i=4aa794a0@gateway/web/ajax/
16:25:36 Nick fxb is now known as fxb__ (
16:26:33 Join Horscht [0] (n=Horscht@xbmc/user/horscht)
16:29:01Unhelpfulmcuelenaere: you have a lua_malloc... that basically turns an malloc allocation into a garbage-collected lua one?
16:29:57 Join itcheg [0] (i=62db4c46@gateway/web/ajax/
16:30:09mcuelenaereUnhelpful: sort of, it turns a malloc() allocation into a garbage-collected lua variable of type userdata
16:30:39mcuelenaereI could look in the internals for another lua_malloc() though, there's probably some way to realloc() somewhere
16:30:49Unhelpfulcould your image loader get the size-to-load, malloc that, do the load, realloc down, and then lua_malloc that?
16:30:52 Quit Horscht (Read error: 104 (Connection reset by peer))
16:31:45 Join tchan1 [0] (
16:32:12 Join Horscht [0] (n=Horscht@xbmc/user/horscht)
16:32:54 Part LinusN
16:33:25mcuelenaerehmm there's no API for that, but it's possible
16:33:53 Quit tchan1 (Client Quit)
16:34:25 Join tchan1 [0] (
16:35:11Unhelpfulit looks like your malloc lib has a realloc, so that could certainly work. and the realloc shouldn't be able to fail, since it's always going to be smaller than the original allocation.
16:36:36 Quit tchan (Nick collision from services.)
16:36:41 Nick tchan1 is now known as tchan (
16:36:47 Join Blue_Dude [0] (
16:36:53 Part Blue_Dude
16:37:48mcuelenaereyes the malloc lib has, but Lua wouldn't know the buffer has been resized though
16:38:18mcuelenaereit would be easier to 'wrap' a buffer into a Lua object, but I can't really see how to do that
16:38:34mcuelenaereunless I take care myself of the GC
16:38:46 Quit Horscht (Read error: 104 (Connection reset by peer))
16:38:52 Quit barrywardell (Read error: 104 (Connection reset by peer))
16:39:05Unhelpfulhrm, so the lua_malloc calls malloc itself?
16:39:09 Join Horscht [0] (n=Horscht@xbmc/user/horscht)
16:39:37gevaertsdoes it matter that lua doesn't know it?
16:41:03Unhelpfulhrm, good question - a bitmap should be treated as read-only data anyway.
16:41:57n1swhat if you want to make a paint-like app?
16:41:58gevaertsif all lua does is keep track of it to free() it when it's not used anymore, a realloc() halfway should not be a problem
16:42:18mcuelenaerehmm yes but it also keeps hold of its size
16:43:51 Join saratoga [0] (i=9803c6dd@gateway/web/ajax/
16:44:41 Quit funman ("leaving")
16:44:42 Quit Horscht (Read error: 104 (Connection reset by peer))
16:45:41 Join Horscht [0] (n=Horscht@xbmc/user/horscht)
16:45:46mcuelenaereI'm not sure whether that impacts some sort of memory usage calculation
16:45:57mcuelenaereUnhelpful: currently it isn't..
16:46:17mcuelenaereand I don't see why it should?
16:49:13 Quit Horscht (Client Quit)
16:50:58 Join Horscht [0] (n=Horscht@xbmc/user/horscht)
16:52:35 Quit Horscht (Read error: 104 (Connection reset by peer))
16:53:06 Join Horscht [0] (n=Horscht@xbmc/user/horscht)
16:53:50Unhelpfulhrm, i guess you could treat it as writable. in that case you need to handle the size correctly. i don't really see somebody rewriting rockpaint in lua, though. ;)
16:55:06mcuelenaereUnhelpful: a Lua image is always an object containing height, width & a pointer to the data
16:55:20mcuelenaereusers can never write directly to the data without passing a boundary check
16:55:36Unhelpfula boundary check on the coordinates?
16:57:00*kugel wonders whether lua scripts could be auto-run
16:57:01CIA-38New commit by mcuelenaere (r21077): Fix some typos
16:57:22kugelso I could make me a nice welcome screen :)
16:57:41mcuelenaerekugel: it could yes :)
16:57:42Unhelpfulif the user always has to address the image via coordinates, after a boundary check, and the length is never needed for one, it seems to me like you can safely resize the data after load.
16:58:08mcuelenaereare there C functions for that?
16:58:18*mcuelenaere wouldn't want to resize an image in Lua
16:58:39mcuelenaereyou'd have to do a boundary check for every pixel..
16:59:25Unhelpfulno no, i mean to resize the allocation. the loader resizes the data :)
17:00:26Unhelpfulit seems to me as if the boundary check on the coordinates would protect from writes past the end of the buffer. lua shouldn't need to know the buffer size for anything else.
17:01:08mcuelenaerehmm yes, but lua doesn't really handle the boundary checks; my wrapper does this
17:01:42 Quit Zagor ("Don't panic")
17:02:02mcuelenaereperhaps I should've said 'a Rockbox Lua image'
17:02:35FrankTMthis may be a stupid question :)
17:02:39FrankTMwhat it UMS-mode?
17:02:53kugeluniversal mass storage
17:03:06gevaertskugel: no :)
17:03:14gevaertsusb mass storage. Also known as msc
17:03:30FrankTMah. i'm wondering how to get my fuse into that mode: P
17:03:57kugelgevaerts: but the U in usb stands for universial? :)
17:04:21gevaertskugel: (pedantic mode) No, it doesn't ;)
17:15:36 Quit matsl_ (Read error: 110 (Connection timed out))
17:19:30 Join bmbl [0] (n=Miranda@unaffiliated/bmbl)
17:19:37CIA-38New commit by mcuelenaere (r21078): Fix FORMAT_RETURN_SIZE in read_bmp_*() when scaling (thanks to Andrew Mahone)
17:20:06saratogaUnhelpful: regarding IMDCT, I got a reply from Loren over at ffmpeg saying that the MDCT is symmetric and thus we don't need to compute the whole thing
17:20:21saratogawhich reminded me of this blog entry i'd read ages ago:
17:23:44 Quit kugel (Read error: 110 (Connection timed out))
17:35:35***Saving seen data "./dancer.seen"
17:37:22 Quit Lss (Read error: 104 (Connection reset by peer))
17:37:45 Join Lss [0] (
17:38:56 Quit Lss (Read error: 104 (Connection reset by peer))
17:42:47 Quit Horscht (Read error: 104 (Connection reset by peer))
17:43:38 Join Horscht [0] (n=Horscht@xbmc/user/horscht)
17:46:15 Join salty-horse [0] (
17:46:26 Join renke [0] (
17:46:36salty-horseyo. recompiled rockbox. unzipped to device. getting "incompatible version" error when starting plugins. why?
17:47:01 Join Lss [0] (
17:47:08evilnick_homeThat usually happens if you didn't overwrite the plugin directory
17:49:35salty-horseI did the regular "make unzip and unzip on device" method :)
17:50:06n1smake unzip?
17:50:50n1swell, try to make clean rebuild and make zip and unzip again. Also make sure to safely remove/unmount the dap befor pullign the cable
17:51:49salty-horsen1s, "make zip"
17:52:32 Join domonoky [0] (n=Domonoky@rockbox/developer/domonoky)
17:52:48 Join Ubuntuxer [0] (
17:53:35 Quit preglow (Remote closed the connection)
17:55:12salty-horsewhat's pdbox? a new kind of music browser?
17:57:31saratogai believe its the puredata gsoc project
17:58:54domonokyits kind of a "music instrument"
17:59:30domonokyand wincents GSOC project. :-)
17:59:54salty-horsen1s, I forgot to restart the device. the "should reboot?" prompt is too easy to skip without noticing
18:03:05 Join wincent [0] (
18:03:12salty-horsethe text viewer isn't very intuitive. I thought "reflow lines" was supposed to wrap long lines, but instead it appears to remove newlines
18:03:44salty-horsewincent, hi!. what's pdbox? (please don't use "puredata" to describe it since I don't know what that is from their website)
18:04:23wincent"Sound environment" is imprecise too :-)
18:04:53wincentIt is a program that creates sound.
18:05:18salty-horseyou feed it patterns in various forms and set up channels and play them?
18:06:18wincentYou build synthesizer (or anything which works with sound) in a full-featured PD environment, and then you let pdbox play it on a DAP.
18:06:22 Quit petur ("work->home")
18:06:44n1swincent: I hope you'll write some documentation for it ;)
18:07:17wincentI will! (I will have to :-) , that much I understood already)
18:08:20wincentThe documentation of the building process of Pure Data designs is what I intend to do after I delivered pdbox at the end of GSoC.
18:09:14n1ssounds great!
18:10:30 Join barrywardell [0] (
18:11:29 Join tvelocity [0] (
18:14:00linuxstbgevaerts: Couldn't you simulate bad sectors?
18:14:53 Quit LambdaCalculus37 ("gotta run!")
18:14:55gevaertslinuxstb: I can, up to a point. I'd also like to compare behaviour with the OF
18:15:42linuxstbAh OK.
18:16:20 Quit itcheg (" ajax IRC Client")
18:17:08 Join funman [0] (n=fun@rockbox/developer/funman)
18:20:56 Quit martian67 (Connection timed out)
18:21:19 Join SirFunk [0] (n=Sir@
18:23:27 Join itcheg [0] (i=62db4c46@gateway/web/ajax/
18:25:32 Join Riku [0] (
18:25:46 Nick fxb__ is now known as fxb (
18:28:28 Quit Ubuntuxer ("Leaving.")
18:31:56 Join Rondom [0] (
18:32:12 Quit funman ("leaving")
18:37:59 Join kugel [0] (n=kugel@rockbox/developer/kugel)
18:38:51JdGordonkugel: hey, i took 8799 and made it work with the wps updating in the background... sort of :)
18:39:30kugelah, interesting
18:40:13JdGordonexcept right now only the dynamic tags get updated
18:40:20 Quit SirFunk (Read error: 110 (Connection timed out))
18:40:20JdGordonand actually, my text wps doesnt work at all :p
18:40:44kugelJdGordon: What I planned for the splashes for is to add a one-shot event at each splash(0,...), then the caller must send the cleanup event (the callback will just update the screen [not clear])
18:41:22JdGordonyeah, that could work... i still dont think its really needed
18:41:52kugelI really don't want dead splash parts on the screen
18:43:58 Quit Lss (Read error: 110 (Connection timed out))
18:44:33 Join _Auron_ [0] (
18:45:32JdGordonanyway, the ncie thing about the wps thing was its a 5 line addition to your patch :)
18:45:42JdGordon+ a bit more to actually make it work
18:46:17JdGordonthe question is.. do we want to keep the current statusbar stuff around? we can just about make it in the wps anyway
18:46:41kugelI don't care
18:46:58salty-horseamiconn, here?
18:47:01kugeldid you read my questions about the statusbar stuff in SVN?
18:49:29kugelWhy is the statusbar disabled when entering plugins. also, it seems to get re-enabled twice
18:49:39kugelafter exiting
18:51:34JdGordonbecause of the whole viewport change thing... 99% of plugins expect the have the full screen, the statusbar got in the way
18:52:16kugelI think we should change those 99% then to deactivate it explicitely
18:53:02JdGordonsure, go for it....
18:53:24kugelbut, though it's deactivated, a do_menu() with hide_bars = false will show the statusbar..
18:55:00 Join bertrik [0] (n=bertrik@
18:55:37JdGordonit gets activated there
18:56:05kugelyes, but that seems weird to me
18:56:57 Quit MarcGuay (Read error: 104 (Connection reset by peer))
19:00:04 Quit einhirn ("Miranda IM! Smaller, Faster, Easier.")
19:05:16 Quit dmb (Read error: 104 (Connection reset by peer))
19:08:51 Join dmb [0] (n=dmb@unaffiliated/dmb)
19:10:13 Join Ubuntuxer [0] (
19:11:44 Quit VytenisS ("Quit")
19:19:14 Join fyre^OS [0] (
19:19:54Unhelpfulsaratoga: quite a bit of that is over my head :/
19:22:03 Join miepchen^schla [0] (
19:25:13Mikachugevaerts: i just tried with a friends ipod video, and that works fine as a hid device, but my nano doesn't
19:27:20saratogaUnhelpful: I'm pretty sure our IMDCT already exploits that symmetry, but we can probably improve it by removing the symmetric coefficient calculation to the windowing code, which would save several loads and stores per sample
19:27:26saratogaand of course a good deal of memory
19:28:35Unhelpfulok... i see that the MDCT doesn't seem to have a DC value (ie a constant basis function)?
19:30:25saratogaUnhelpful: yes because of that factor of 1/2 added to the frequency
19:31:02saratogai assume thats to prevent the transform from being even about 0 but i don't know
19:33:03Unhelpfulthere's an optimization specific to fixed-point implementations that involves the DC coefficient. i saw it in the 16-point expanded IDCT, and copied it to the other IDCT sizes for core jpeg. plugin jpeg still doesn't have it, and i'm trying to figure out if the mpeg plugin does, but that has an entirely different IDCT implementation
19:34:12Unhelpfuland obviously it doesn't apply to MDCT...
19:34:14saratogafiguring out what different DCT related transforms do compared to one another is astonishingly complicated
19:34:36saratogayou would think understanding the transform would make understanding the implementation easier but that has not been the case in my experience
19:35:29kugelsaratoga: I have mp3 @ 80Mhz
19:35:29 Join BryanJacobs [0] (
19:35:31kugelon the fuze
19:35:36***Saving seen data "./dancer.seen"
19:36:06 Quit fyrestorm (Read error: 110 (Connection timed out))
19:36:09kugelthe datasheet says that asynchronous clocking creates some sync overhead
19:36:24saratogayeah you'd have to have at least one extra cycle of delay to sync the clocks
19:36:51kugelI don't know whether this is going to be different with caches, but as of now it seems the 60 extra MHz cannot even it out
19:37:09saratogayou save 60MHz just by syncing the buses?
19:37:39Unhelpfulsaratoga: there's *some* value added to the DC in our MPEG IDCT before any other math is done, but it doesn't seem to be a correct rounding factor.
19:37:41kugelmp3 is @ 160MHz or so with 248MHz cpu and async
19:37:43 Join dp45349 [0] (n=621cc6c1@gateway/web/cgi-irc/
19:38:01kugel248/31 cpu
19:38:04saratogait maybe that theres some additional overhead when accessing the IRAM when async
19:38:13kugelwith cpu @ 192/64, mp3 is at 80MHz
19:38:34kugelI'm fairly sure other codecs are faster too
19:38:38saratogaif i remember correctly the iram had some rather complicated clocking needed since its actually 1T DRAM, so maybe it doesn't like async
19:39:12kugelthe key seems to be the PCLK too, having it at 64MHz all the time gives much better performance overall
19:39:13bertriknice to see some benchmarks in the forum thread :)
19:39:18dp45349quick question why hasn't case game been committed? is there something i can help with?
19:39:25saratogawe really need MMU working so we can see which of these bus changes actually matter :(
19:39:25Unhelpfulthe optimization in question adds the 1<<(shift-1) rounding factor to the DC coefficient before including it in the output, rather than adding a rounding factor to each output value before shifting them down to the output size
19:40:05kugelsaratoga: if it turns out that 248MHz but with async are never worth it, we can cap it to 192 and reduce the core voltage too
19:40:38bertrikif the codec i2c is really a problem to enable MMU and more advanced clocking, I'm in favour of a workaround.
19:40:39kugelthe next integer multiple is 256, which is slightly overclocked, I'm not sure whether it's possible to use that
19:41:14saratogai'd be ok with almost anything to get MMU working in SVN, then try to fix things from there
19:41:32saratogai remember with PP it was months and years before we had things like boosting and DMA working in SVN
19:42:12 Join petur [0] (n=peter@rockbox/developer/petur)
19:42:30kugelI think we should do another bench to see if PCLK constantly at 64MHz has much impact, and if not change SVN to use the 192/64 combo
19:42:30 Quit dp45349 (Client Quit)
19:42:53saratogaif 192/64 works I would be ok with just using it for now
19:42:58kugelif no bad impact, that is
19:43:00saratogawe can always power optimize things later
19:43:23kugelthe lcd is much faster with that pclk too
19:44:17kugeli was getting 80fps for fullscreen update a few hours ago
19:44:52bertrikthe as3525 datasheet has a set of clocking settings IIRC. Knowing and documenting the exact settings of the OF would be a good thing too IMO.
19:45:14kugeliirc, what we have in SVN is what the OF does
19:45:48kugelbertrik: the ds only lists some fastbus settings
19:46:00kugelwith fastbus we're limited to 65MHz IIRC
19:46:06bertrikhmm true, they are all fastbus
19:46:15saratogawhat is fastbus?
19:46:34kugelFCLK = PCLK, mostly
19:46:54kugelpage 100 describes the 3 settings
19:49:26saratogais that the same as running the bus synchronously?
19:51:08kugelsync is another setting. fastbus means the bus clocks must be the same, sync means the cpu clock can(must actually) be higher than the ahb bus freq but must be a integer multiple, with async it doesn't have to be an integer multiple
19:52:14 Quit robin0800 (Remote closed the connection)
19:52:14saratogaah ok so basically sync runs the cpu clock off a multiplied version of fclk, while fast bus runs the peripheral and fclk together
19:52:17bertrikif only this clocking business wasn't so flexible, the choice would've been much easier :|
19:52:43kugelwe could possible also use a sync setting with cpu 248/62, pclk 62
19:53:34gevaertsMikachu: can you submit a bug report?
19:53:51Mikachugevaerts: sure, do you know if it is working for anyone with a nano?
19:54:25Mikachulucky galeon remembered my password, i haven't logged into flyspray in 3 years or so :)
19:54:59gevaertsno idea, but if it works normally otherwise, I don't see any reason why your nano should be different
19:55:16Mikachumy bootloader is pretty old, but that shouldn't matter right?
19:55:44gevaertsdoes usb work otherwise?
19:55:53Mikachuyeah, i can transfer files fine
19:55:59 Quit mirak (Read error: 113 (No route to host))
19:56:02Mikachuand lsusb shows the hid endpoint
19:56:42gevaertsthen I don't really see how the bootloader could matter, no
19:57:33bertrikkugel, saratoga, I'll have a go at an codec i2c workaround soonish. THe problem seems to be that the i2c 'busy' bit doesn't get set immediately after starting a transfer, making the CPU think it's already done.
19:58:00 Join funman [0] (n=fun@rockbox/developer/funman)
19:58:18saratogabertrik: so it just needs a little delay?
19:58:28funmanbertrik: just adding a delay before busy check is enough
19:59:06*kugel overclocks his fuze
19:59:12 Join mirak [0] (
19:59:23funmanthe real problem is that the devices deadlock. at reproducable occasions though (the same timeframe of a given .mpg for example), especially when using pcm
19:59:34 Quit Ubuntuxer ("Leaving.")
19:59:47bertrikthe OF uses a nice interrupt-driven i2c driver, but when doing a codec read/write, it *polls* until the nicely written interrupt-handled transfer is done.
20:00:01saratogaI saw on ABI that using crossfade apparently gives an immediate freeze as well
20:00:22kugelwhy can I get 80MHz mp3 with 192/64, but not with 256/64?
20:00:28funmanbertrik: you can try to patch the OF (from the bootloader!)
20:01:07kugelah there we go, it just needed some time :>
20:01:59bertrikhow bad would it be to 'idle' at 64 MHz (instead of a lot slower)?
20:02:31saratogafor now its fine, but eventually we want to idle slower
20:02:32Mikachugevaerts: would you say it is "Drivers"?
20:02:53kugelI don't think 62 or 31 makes much difference
20:02:54saratogaideally unboosted should be ~25MHz on ARM9
20:03:08saratogabut for now losing a little battery life is hardly a pressing concern
20:03:37kugelalso, we don't have a idle state. If we run the pclk below 61 we get much slower decoding overall
20:04:02CIA-38New commit by MarcGuay (r21079): Specify that the Battery Capacity setting is used for runtime estimation, not battery percentage remaining.
20:04:05gevaertsMikachu: close enough
20:04:06kugelthe beast shows that running at 250 or 500MHz all the time doesn't have much impact too
20:04:11BryanJacobslinuxstb: ping
20:04:27kugelthough, that's arm11
20:05:02saratogai think more advanced arm cores can internally adjust some of their clocks automatically
20:05:47bertrikand it's not really idling, but in a lower power 'wait-for-interrupt' state, right?
20:06:04 Join MarcGuay [0] (
20:06:24 Quit funman ("leaving")
20:06:49kugelI wonder if we could run at 256MHz?
20:07:29kugelI've not noticed a problem so far. And the chip has a anti-overheat mechanism
20:07:49Unhelpfulbertrik: something like that
20:09:03 Join jmillikin [0] (
20:09:15BryanJacobsif anyone wants to do codec development on a PC without using the simulator, contains a command-line utility that will just run your codec on an input file and produce a WAV as output
20:09:28kugeland 256 is just 6MHz over specs (2,4%)
20:09:48BryanJacobsit also ignores sleep() so you can run as quickly as your computer is capable of going - no real-time decoding
20:10:52BagderBryanJacobs: "as quickly as your computer is capable of" in a single core then I guess?
20:10:59 Quit mirak (Read error: 113 (No route to host))
20:11:11BryanJacobsBagder: yes, it pretends to be a single core, no threading right now
20:11:23BryanJacobsis that a really important feature?
20:11:31Bagdernono, I was just curious
20:12:04BryanJacobsI just meant that it won't pause for the timestamp to catch up with decoding progress
20:12:32BryanJacobsI tested with the wavpack codec and the output is correct
20:12:54BryanJacobsnext step: port the correction-file handling to the wavpack codec
20:15:30gevaertsThanks. I'll point tomers to it as soon as I see him
20:15:52Mikachui didn't see you or him in the assignee list
20:15:52 Quit itcheg (" ajax IRC Client")
20:17:04gevaertshm, that list seems out of date
20:17:12Mikachuit lists me :)
20:17:38gevaertsexactly :)
20:18:10 Quit renke ("leaving")
20:18:39amiconnsalty-horse: am now
20:19:09amiconnBagder: What do you think - would it be a good idea if a few build servers would crosscompile win32 sims?
20:19:36amiconnThere was this breakage of win32 sims today which didn't affect linux sims at all...
20:19:39salty-horsehi amiconn. I'm wondering about the reflow feature of the text viewer. (I see that you committed something related to it ages ago)
20:19:53salty-horseshould it really join short lines into one line? is that what's it is supposed to do?
20:20:39amiconnIt is supposed to join adjacent lines into one, and only leave paragraphs if there is an empty line inbetween
20:20:40Unhelpful"reflow" sounds to me like it ought to remove single line breaks, wrap long lines, and treat double line breaks as paragraph breaks
20:21:43kugelreflow sounds like nothing to me
20:21:44salty-horseamiconn, ok
20:21:54salty-horseshouldn't be a separate feature to only wrap long lines?
20:21:56Unhelpfulbasically, converting text for a fixed line length to read nicely on the screen
20:23:06 Join itcheg [0] (i=62db4c46@gateway/web/ajax/
20:23:36kugelbubbles is playable now on the fuze
20:23:52amiconnsalty-horse: Iirc one of the line wrap options does that
20:24:23AlexPBagder: Is it also possible to have a table that check if the manuals build? Right now it is quite hard to spot if they are broken.
20:24:55BagderAlexP: yeah, I'd like that... is "just" for someone to do it...
20:24:57salty-horseamiconn, can you tell me what option is that specifically?
20:25:11AlexPBagder: hehe, as always :)
20:25:42 Join akpup420 [0] (
20:26:09amiconnBagder: Problem is that this would make sim builds somewhat undeterministic
20:26:26Bagderamiconn: why?
20:26:39amiconnBecause of how builds are distributed
20:26:51Bagderwe'd create a new "arch" that does cross-compiled sims
20:27:04amiconnIf some build servers are building linux sims and some are building win32 sim, reds would jump around
20:27:04Bagderas servers would need to be specificly marked to support that
20:27:13amiconnHmm, you mean build every sim twice?
20:27:23Bagderno, I meant cc-sims to be a special build
20:27:31Bagderin its own column
20:27:38amiconnRight now the selection is automatic, based on which sdl-config is found by 'configure'
20:27:43kugelmore columns????
20:27:45 Join SirFunk [0] (
20:27:48Bagdermore columns!
20:28:10akpup420hello, i am looking at, i am thinking about trying it out, but idk if my nano is a first gen, how would i find this out?
20:28:24amiconnIf you have native sdl-config first in the path, it will build linux sims, if you have migw32 sdl-config first, it will crosscompile win32 sims
20:28:43 Quit BryanJacobs ("null")
20:28:43kugelsaratoga: the fuze is incredibly responsive with my current setting
20:28:47amiconnRequirements are mingw32, and a crosscompiled sdl (which is simple)
20:28:48Bagderamiconn: well, I consider that a pretty shake way to do things anyway so I'd rather also change that
20:29:14amiconnHmm, but how?
20:29:22AlexPakpup420: There is a page on the Apple website that should help you - I don't the address sorry. A google should also help
20:29:23amiconnThe command is 'sdl-config' in both cases
20:29:36Bagdersearch for _all_ sdl-configs, then check which ones that seem to be for cross-compiling
20:29:49akpup420thanks alexp
20:29:54Bagderall in the path I mean
20:30:20*amiconn wouldn't know how to do this
20:30:43 Quit flydutch ("/* empty */")
20:30:47Bagderwhy would that be tricky?
20:31:03Bagdersurely the cross-compiled one has some properties we can cehck for?
20:32:54 Join BryanJacobs [0] (
20:33:24kugeltest_codec is weird!
20:33:48kugelhm wait
20:34:20 Join biengo [0] (
20:35:11amiconnBagder: Hmm, sdl-config −−libs lists (among others) -lmingw32
20:35:23amiconnAnd -mwindows
20:35:34Bagderthat looks like some nice strings to scan for
20:35:51akpup420iPod nano (4th generation)
20:36:24amiconnHmm, but how to find all "sdl-config" in the path?
20:36:58AlexPakpup420: Bad luck
20:37:02Bagderfindtool in the configure script already "walks" the path, it should just do something similar
20:37:41pixelmaI wonder if checking crosscompiled sims will help fix those warnings...
20:38:42amiconnpixelma: The -ffunction-sections one? I guess this isn't fixable in a sane way
20:38:55 Quit Xerion (Read error: 104 (Connection reset by peer))
20:39:04pixelmaI meant these
20:42:16 Quit saratoga (" ajax IRC Client")
20:42:38 Part akpup420 ("Leaving")
20:45:49 Join mirak [0] (
20:49:15dionoeamcuelenaere: hi. You could use lua_getfield() to simplify your code (check_tablevalue, instead of lua_pushstring and lua_gettable)
20:49:44mcuelenaerehmm thanks
20:49:51mcuelenaereis there also a shortcut for the lua_settable?
20:49:55 Quit Bombe (Read error: 60 (Operation timed out))
20:50:04dionoeamcuelenaere: which lua_settable?
20:50:38mcuelenaerethe opposite of lua_gettable :p (it sets a value to a key in a table, e.g. tab['test'] = 42)
20:52:07dionoeabtw, if you have the opportunity to borrow it at some library, you should check "Programming in Lua, Second Edition". It's really nice to grasp all the possibilities of lua
20:53:41dionoeayou might also want to have a look at luaL_newmetatable( L, "typename" ) to initialize or fetch the type's meta table when needed
20:54:10 Join msscott [0] (
20:54:39mcuelenaeredionoea: I think I use luaL_newmetatable() for struct rockboxlua_image
20:55:04dionoeaah right, you use it in the init.
20:55:21dionoeaThe way I use it most of the time is, in the object's constructor code:
20:55:44mcuelenaeredionoea: does look wrong? (can't get it working yet)
20:55:47dionoeaif (luaL_newmetatable(L, "type")) { initialize to metatable; } lua_setmetatable(L,-2);
20:56:08dionoeathat way you don't need a specific init function. (luaL_newmetatable() is like a lua_getmetatable() if it already exists)
20:56:26BryanJacobsah-ha, I finally found some info on the Wavpack algorithm itself:
20:56:46mcuelenaerehmm yes, currently being done in rli_new() & rli_alloc()
20:56:55mcuelenaeres/currently/that's &/
20:58:11dionoeamcuelenaere: looks fine. (but then that stack is always tricky)
20:59:05mcuelenaerehmm it seems to work now
21:01:11 Join funman [0] (n=fun@rockbox/developer/funman)
21:01:22mcuelenaeredionoea: hmm if lua_isnoneornil(L, -1) == true, does that mean a none/nil value is on stack?
21:01:50dionoeayes, it means that the top element is none or nil
21:02:02dionoea(if it's none that would mean that the stack is empty)
21:02:16dionoeahi funman
21:03:19dionoeamcuelenaere: is there rockbox specific code in anything besides rocklib.c ?
21:03:36mcuelenaeredionoea: there is
21:03:48mcuelenaeresome code is commented out/removed and a little bit got added
21:04:07 Join p3tur [50] (n=petur@rockbox/developer/petur)
21:04:14mcuelenaerebut all the wrappers are in rocklib.c (currently)
21:04:37mcuelenaere(except the dynamically generated ones, see actions.lua)
21:05:13 Join taylor_ [0] (
21:05:23dionoeadocall() in rocklua.c could be simplified (using -1 instead of lua_gettop())
21:05:32 Part taylor_ ("Leaving")
21:05:56CIA-38New commit by mcuelenaere (r21080): Lua: re-use the viewport pointer; also use the shorter lua_getfield() & lua_setfield() notations (thanks Antoine Cellerier)
21:05:58kugelso, test_codec says "126MHz needed for realtime"
21:06:09dionoeaah no, maybe not.
21:06:15kugelbut the buffering screen shows that it decodes at 80MHz???
21:06:17mcuelenaeredionoea: feel free to improve :)
21:06:42dionoeaOk, I'll commit anything I find then :)
21:07:16 Join saratoga [0] (i=98039dd8@gateway/web/ajax/
21:08:23kugelwho's right there?
21:08:24saratogaBryanJacobs: I sent you that link a week ago:
21:08:46saratogaits from a book on data compression by the wavpack author
21:08:51msscottIs there an install method for unkown player models?
21:09:07BryanJacobssaratoga: must have missed it, sorry, reading now
21:09:17saratogakugel: maybe the timer is wrong on AMS?
21:09:35saratogathat or the defines aren't right in the debug screen, which is probably more likely
21:09:48kugelI don't think the debug screen is wrong
21:10:06kugelit's mostly at 64MHz, only boosts sometimes shortly
21:10:13dionoeamcuelenaere: why aren't the math, io and os libs provided? to save on memory?
21:11:01 Join Hillshum [0] (n=chatzill@unaffiliated/hillshum)
21:11:14dionoea(package would also be nice to share code between lua plugins)
21:11:45HillshumCould someone please delete for me? I didn't mean to double post it
21:12:23kugeleither someone was really fast, or you didn't actually double post
21:12:42HillshumWait, didn't double post
21:13:10saratogayou need it undeleted?
21:13:23 Quit SirFunk (Remote closed the connection)
21:13:31saratogathough i think thats been discussed a lot before
21:13:36*Hillshum just reposted
21:14:18kugelsaratoga: and I think it should be done
21:15:42funmanhi dionoea
21:16:41mcuelenaeredionoea: mostly because they aren't compatible with Rockbox
21:16:49 Join SirFunk [0] (n=Sir@
21:16:52mcuelenaere(ie no FP, not fully POSIX-compatible etc)
21:17:24mcuelenaerefeel free to implement them though :)
21:17:26dionoeamcuelenaere: ok. most of os, io and package should work though. I'll give it a try if I find some time
21:17:30mcuelenaere(I've taken a short look at io)
21:18:06 Quit itcheg (" ajax IRC Client")
21:19:10dionoeaI guess that io.popen() is the only io function that couldn't be supported
21:19:48 Quit Rondom ("Ex-Chat")
21:20:00dionoeaos.execute() and os.setlocale() (setlocale could be changed to a nop)
21:21:07mcuelenaereyeah but io works with FILE*, and RB works with int's
21:22:12kugelbut FILE works too, doesn't it?
21:22:38dionoeaah ok. shouldn't be too complex to replace the FILE functions with fd ones :)
21:22:56mcuelenaerekugel: does Rockbox has support for FILE?
21:24:06mcuelenaereand as this Lua is fixed-point only, all math functions will need fixed-point ones (I've seen there exist some of them in pluginlib)
21:26:30saratogawhat is lua useful for in rockbox? general scripting ?
21:26:52BryanJacobsthe maximum amount of sample history necessary for decoding a WavPack sample is the last 8 samples - just saying it here for history
21:27:31mcuelenaeresaratoga: something like that yes; easing the development of plugins/games I guess
21:28:09saratogaso it can draw to the screen?
21:28:31 Part BryanJacobs
21:28:40mcuelenaereyes it can
21:28:56kugelmcuelenaere: I'm not sure. I seem to remember use of it, but I can't find it anymore
21:28:58mcuelenaereit's supposed to be able to do everything a normal plugin can do :)
21:29:12 Join VytenisS [0] (
21:29:25Mikachucan you write lua scripts in the text_editor plugin and then run them?
21:30:10mcuelenaerekugel: perhaps you mixed up with DIR*?
21:30:29mcuelenaereMikachu: if you really wanted to, you could yes
21:31:05*Hillshum considers text input on RB to be a bit slow for writing code on it
21:32:06Mikachuif i finish 5165 it could be pretty quick :)
21:32:31 Join kugel_ [0] (
21:32:41 Quit kugel (Nick collision from services.)
21:32:47 Nick kugel_ is now known as kugel (
21:33:03*linuxstb has a habit of missing BryanJacobs... ;(
21:33:54Mikachu;a=blob;f=apps/plugins/aaaatextinput.c;h=c28e42c788166 is how far i got btw :)
21:34:47HillshumMikachu: Have you posted it to flyspray?
21:35:37kugelMikachu: what does that do?
21:35:39***Saving seen data "./dancer.seen"
21:36:20MikachuHillshum: i have not
21:36:28Mikachukugel: it does the honeycomb thing mentioned in a comment
21:37:08mcuelenaereMikachu: can't you do drawline() with a helper function to reduce code duplicity?
21:37:20Mikachumcuelenaere: that is super ugly debug code
21:37:29mcuelenaerestill ;)
21:37:31Mikachufrom 3 years ago
21:37:58HillshumIt might get more attention on flyspray
21:38:07Mikachuthan on my hd, yes :)
21:38:22Mikachui only just started looking at my old patch pile yesterday
21:38:43Mikachuamazingly most of it worked without a lot of changes after 3 years
21:39:27Mikachui also like how i put h instead of u in row 5
21:39:34 Join BryanJacobs [0] (
21:40:16kugellinuxstb: ^
21:40:46MikachuBryanJacobs: linuxstb misses you
21:40:58BryanJacobsMikachu: ?
21:41:05Mikachu21:33:02 * linuxstb has a habit of missing BryanJacobs... ;(
21:41:29BryanJacobsah well, we'll catch up eventually
21:41:33linuxstbBryanJacobs: Hi...
21:41:43BryanJacobsmy prediction came true!
21:41:44*Mikachu throws confetti
21:41:54*Mikachu predicts a possible anti-climax
21:42:22BryanJacobslinuxstb: I got the harness working, you can now run whatever plugin you like as a standalone executable
21:42:33BryanJacobsnot plugin
21:42:34linuxstbI was just looking at your "harness", and wondering if you were planning on making it work with the existing files in-place, rather than copying everything into a new directory. I don't know if those files have been changed at all.
21:42:51BryanJacobslinuxstb: the only files not changed were the wavpack ones
21:43:08BryanJacobsand I figured since the codec is actively being developed, it's OK to work on it out-of-tree
21:43:51linuxstbDoesn't that make it less useful in the long-term?
21:44:25BryanJacobsI don't understand - could you explain the use case?
21:45:04Hillshumkugel: would the hardest part be getting the screen to go sideways?
21:45:34kugelrotating fonts, but that's already done in mpegplayer
21:46:21Hillshumthen getting the keymaps
21:47:20linuxstbBryanJacobs: I mean it can be useful for general codec testing/hacking for any codec. It just seems more practical to be able to compile the existing rockbox files in-place, rather than needing to cp them.
21:47:34BryanJacobslinuxstb: give me a minute
21:48:27HillshumMikachu: I don't think it'd be worth it to bother with them
21:49:16 Join itcheg [0] (i=62db4c46@gateway/web/ajax/
21:51:28 Join robin0800 [0] (
21:52:32 Join SirFunk_ [0] (
21:54:01 Join HellDragon [0] (
21:54:58BryanJacobslinuxstb: now I get a "floating point exception" <sigh>
21:55:09BryanJacobsit compiled with the wavpack codec still in-tree though
21:56:25BryanJacobsoops, it's because I didn't set the sample rate
21:58:37BryanJacobsok, all set:
21:58:45 Quit robin0800 (Remote closed the connection)
21:59:46 Join robin0800 [0] (
22:02:56 Quit Hillshum ("ChatZilla 0.9.83 [Firefox 3.0.3/2008092417]")
22:04:34CIA-38New commit by MarcGuay (r21081): Improve the c100 keymap and button names.
22:04:38 Join Nico_P [50] (n=nicolas@rockbox/developer/NicoP)
22:04:40linuxstbBryanJacobs: That looks better. Could you post it to flyspray - I think it would be useful to commit (maybe not exactly in its current form, but something similar)
22:04:53BryanJacobslinuxstb: doing cleanup / code style now
22:06:59 Quit SirFunk (Read error: 110 (Connection timed out))
22:07:09gevaertsHorscht: I think FS #10239 is now reasonably safe if you want to have a go
22:07:41gevaertsOnly test if you have backups though, I can't promise perfection
22:07:44 Quit biengo (Read error: 113 (No route to host))
22:07:46MarcGuaymcuelenaere: See the red?
22:07:55Nico_PHi everyone!
22:08:18MarcGuayMono issues, seems.
22:08:21mcuelenaereMarcGuay: now I do
22:08:50linuxstbBryanJacobs: Also, can you think of a more descriptive name than "harness"?
22:09:20BryanJacobslinuxstb: idk - what should we call it, "thing which lets you run codecs as standalone executables"?
22:09:49BryanJacobsMikachu: it currently doesn't verify the output, but works for me
22:09:53BryanJacobsyour vote, linuxstb?
22:10:22Mikachuor runcodec, testcodec
22:10:48domonokycodec-executer :-)
22:10:58BryanJacobsdomonky: I like that, functional naming is good
22:11:10BryanJacobslet's be british and call it codec-executor
22:14:49kugelfunman: I'm going to do a 256/64 bench, I'm quite interested in the outcome
22:15:05CIA-38New commit by mcuelenaere (r21082): Should fix red
22:15:09funmankugel: with the default&normal frequencies being?
22:16:10funmanfor synchronous cloking you must use a fclk higher than, and an integer multiple of, pclk
22:16:21kugelwhat's the the default frequency for anyway?
22:16:23funmanso 128MHz for 64MHz pclk
22:16:38funmandefault is for more power saving (usb mode?)
22:17:03kugelmore than normal?
22:17:25kugel64MHz fclk unboosted works so far, faster than what's in SVN
22:17:52 Quit itcheg (" ajax IRC Client")
22:18:04kugelswitching to fastbus didn't get me anyway, all frequencies dropped horribly to 24MHz or something
22:18:46funmankugel: 64MHz fclk with 64MHz pclk without fastbus is illegal
22:19:19kugelit's in integer multiple :) 1x
22:19:46kugelI read that it's illegal, but it seems to work. Also, SVN seems to be more illegal?
22:20:07kugelas I said, I failed at switching to fastbus somehow
22:20:20funmanyes, this is why i ran the benchmarks with the nice graphs, to make SVN legal :)
22:20:46funmankugel: I don't understand what "all frequencies dropping 24MHz" means
22:21:33kugelI wanted to do 192/64 with pclk at 64 too, unboosting was just switching to fastbus
22:21:50kugelbut I didn't 64MHz, rather 24MHz (and it was horribly slow)
22:22:21FlynDicefunman: kugel: I've almost got a patch put together to make changing frequency schemes on the ams sansa a bit simpler, should I post it to flyspray or the forum do you think?
22:22:59funmankugel: _you_ didn't 64MHz ?
22:23:01FlynDiceOK let me tie it up should be 30 mins or so
22:23:16kugelfunman: I didn't get to*
22:24:14kugelAHB freq = pclk?
22:24:22funmankugel: here is the patch I used in my benchs
22:24:32funmanAHB freq = APB freq = pclk on the Sansa AMS
22:24:45funman(i learned it from FlynDice :P )
22:28:47kugelfunman: that's the same for the unboosting part as what I did
22:29:46 Join tessarakt [0] (
22:30:55funmani see no performance loss with it
22:32:22saratogaBryanJacobs: that looks very useful
22:32:26 Join Bombe [0] (n=droden@freenet/developer/Bombe)
22:32:33saratogawill it work with any other codecs?
22:33:41BryanJacobssaratoga: should work with any codec
22:34:02BryanJacobsthe biggest problem with using it is avoiding having to pull in half of Rockbox because of metadata parser dependencies
22:34:45BryanJacobsbut you can always just fake the metadata decode by hardcoding in the length and sample rate of the file you're using...
22:34:47 Join martian67 [0] (n=martian6@about/linux/regular/martian67)
22:36:12 Join matsl_ [0] (n=matsl@
22:37:28 Quit funman ("leaving")
22:38:27 Join BHSPitLappy [0] (n=BHSPitLa@unaffiliated/bhspitmonkey)
22:38:29 Join funman [0] (n=fun@rockbox/developer/funman)
22:39:30BryanJacobsguests - bye all
22:40:01 Part BryanJacobs
22:41:22kugelfunman: I think I got it
22:41:39kugelbut it's slower, and playback stopped at the first rebuffer
22:42:50 Quit linuxstb (Remote closed the connection)
22:43:50kugelhm, maybe not slower
22:44:42kugel10% boost rate
22:44:43funmanplayback stops for me with svn as well
22:44:56kugeltotally fine for me
22:45:10kugelI made two battery benches without problems
22:46:16 Quit {phoenix} (Remote closed the connection)
22:46:17 Quit matsl_ (Read error: 104 (Connection reset by peer))
22:46:37funmanfor me too, it seems to stop randomly
22:47:06bertrikit get some stops occasionally too
22:47:10kugelI'm only getting stopping playback on my clip
22:47:22funmanvery much more frequent on the clip too ..
22:47:25kugelnever on my fuze (not on my previous one too)
22:47:42 Quit efyx_ (Read error: 54 (Connection reset by peer))
22:47:52bertrikI had some aac's (IIRC) making some kind of bwwwooooAAAAAH noise sometimes
22:48:28kugelwasn't that an issue with the codec?
22:48:37 Join Thundercloud [0] (
22:49:00 Join efyx_ [0] (n=efyx@
22:49:43kugelok, rebuffering also works now again
22:50:21 Quit mcuelenaere (Remote closed the connection)
22:50:56kugelfunman: I've noticed that going to 62MHz pclk already makes decoding much slower again
22:51:16funmando you mean 64 -> 62 ?
22:51:19kugelI think I needed 110MHz instead of 90Mhz which I got for 64Mhz
22:51:53funmando you have these results with test_codec
22:52:05kugeltest_codec seems unreliable
22:52:12 Part wincent ("Kopete 0.12.7 :")
22:52:35kugelthat one told me that I was at 126MHz for realtime, while the bufferscreen says 80MHz
22:52:59funmani'm more encline to trust test_codec (didn't read bufferscreen code)
22:53:13 Join linuxstb [0] (n=linuxstb@rockbox/developer/linuxstb)
22:53:56kugelwell, the buffering screen just records the frequencies, and calculates avarage
22:54:25kugelbut that includes the cpu time for the frequent lcd updates, of course (but that actually means test_codec should tell me even less that 80Mhz)
22:55:19kugelor that 126MHz mean something different than what the buffering screen shows
22:56:05kugelbut it's definitely hardly boosting, so 80MHz seem correct
22:57:12 Join Hillshum [0] (
22:59:15 Quit efyx_ (Remote closed the connection)
22:59:23saratogakugel: have you changed firmware/export/config-fuze (or clip) ? I think the buffering screen uses the CPU_FREQ define in there rather then the one in the target tree
22:59:26saratogabut i could be mistaken
22:59:27 Join efyx_ [0] (
22:59:47FlynDiceI'm just about to upload this to FS, see what you think
23:00:06saratogaFlynDice: remove apps/plugins/SOURCES?
23:00:22FlynDicesay what?
23:00:39funmansaratoga: they are in clock-target.h
23:00:53saratogaunless its just for testing then never mind
23:01:17kugelsaratoga: it uses CPUFREQ_MAX (which is not in config-fuze.h)
23:01:39funmanFlynDice: and mmu-arm* :)
23:01:53kugelwell, on the other hand, it decods the file in 200% realtime.
23:02:16 Quit p3tur ("Zzzzz")
23:02:24FlynDiceach yes I see it, mixed up with the mmu stuff, I tried to remove most of it
23:02:26kugeldoes that directly translate to the MHz needed for realtime as in the calculation test_codec uses? (it does jsut CPU_FREQMAX / speed)
23:02:34kugelindependantly of CPU_FREQ_NORMAL?
23:04:16 Join matsl_ [0] (
23:04:29kugelbtw, are we using PLLB? if not, we can power it down
23:05:04Hillshumkugel: try running without it?
23:05:18funmanit's powered down already
23:05:25 Quit petur ("Zzzz too :)")
23:05:40 Join salty_horse [0] (
23:06:00 Quit n1s ("Lämnar")
23:06:15funmannot explicitely though - it's powered down at reset
23:06:57kugelare you sure? I apparently missed that part when reading the ds
23:07:50 Join topbloke [0] (
23:08:46topblokecan you turn off the rockbox USB on ipod video?
23:09:12Mikachuyou mean you want to charge it without going to disk mode?
23:09:16Mikachujust hold down menu when you plug it in
23:09:45topblokeit doesn't work on PS3 if you just plug it in when its off
23:10:03Hillshumturn on the iPod first
23:10:31topbloke:) that works
23:10:40topblokebut with the old one you didn't have to!
23:10:48CIA-38New commit by lowlight (r21083): 3 new ports: Samsung YH-820, YH-920, and YH-925. Mostly functional. Audio working on 820 & 925 (untested on the 920). No battery readings. No ...
23:11:00Bagdernow that's a commit
23:11:02 Join Dauron [0] (
23:11:19kugel finally!
23:11:32 Join low_light [0] (i=ad58bb86@gateway/web/ajax/
23:11:38 Join itcheg [0] (i=62db4c46@gateway/web/ajax/
23:11:44kugellow_light: <3
23:11:51Bagderlow_light: nice one there
23:11:54Mikachunice entrance
23:12:12low_lightBagder: 9 new builds to add ;)
23:12:19Bagderhaha indeed
23:13:38*funman builds a yh920 rockbox
23:13:55*kugel builds for yh925(GS)
23:13:57low_lighttest the usb. it works fine for me
23:14:14low_lightwhat's different with the GS version?
23:14:51funmankugel: you can verify that CGU_PLLBSUP == (1<<3)
23:15:21 Quit itcheg (Client Quit)
23:18:47kugellow_light: I'm not sure, but that one is designed for canada
23:18:56kugelthe VID/PID are the same
23:21:15 Quit bmbl ("Woah!")
23:22:10 Quit salty-horse (Read error: 110 (Connection timed out))
23:22:23Bagderthe configure menu is getting impressive
23:22:32BagderI mean the target selection
23:22:45Hillshumand the build table
23:22:50 Join _fml [0] (n=4fd3c951@
23:23:21_fmlHe-he. I see "HAVE_AK4537". Will we also have "HAVE_AK47"?
23:23:33kugelsound works too
23:23:33Hillshumwhat does the build system do with the builds it makes for unsupported targets?
23:23:50kugelthat's a secret :)
23:24:26saratogathey're not added to the download server
23:24:36saratogawell some are (beast) but most aren't
23:24:47kugelams sansa's are too
23:24:58Hillshumams sansa's are?
23:24:59kugelI think all are
23:25:28 Join newbe [0] (n=4e3612d5@gateway/web/cgi-irc/
23:26:49Bagderwell, it depends on what you mean with where they appear
23:26:49 Quit _Auron_ (Read error: 110 (Connection timed out))
23:26:49Horschtgevaerts, thanks for the headsup
23:26:49 Nick Dauron is now known as _Auron_ (
23:27:28HorschtI didn't find the time to test this out today. Depending on wether i have time tomorrow i'll give it a whirl
23:27:29 Quit newbe (Client Quit)
23:27:35Hillshuma quick look at says no
23:27:51kugelthey're sort of hidden
23:27:54Bagderright, the ones in daily/ are explicitly built daily
23:28:08kugelthat too :)
23:28:10Bagderthe zips made in the distributed builds are differenet
23:28:29Bagderthey can all be found if you craft the url a bit
23:29:27 Quit _fml ("CGI:IRC (EOF)")
23:30:31kugellow_light: USB works now :)
23:31:19Hillshuminteresting. Might stop compiling then
23:33:34 Quit Nico_P (Remote closed the connection)
23:34:41 Quit martian67 (Remote closed the connection)
23:34:44 Quit Hillshum ("Ex-Chat")
23:35:11 Join martian67 [0] (n=martian6@about/linux/regular/martian67)
23:35:16Bagderlow_light: yellow...
23:35:37 Quit martian67 (Remote closed the connection)
23:35:41***Saving seen data "./dancer.seen"
23:36:05 Join martian67 [0] (n=martian6@about/linux/regular/martian67)
23:36:33Bagderhow did that warning pop up?
23:37:04funmantools rebuild?
23:37:12Bagderah yes
23:38:34low_lightfunman: The 920 sim uses the 925 image for now (they look almost identical). If you feel the need to change it you can.
23:39:39funmanlow_light: usb works fine now
23:39:45low_lightOn a sad note, the microdrive in my 820 died.
23:39:56kugelCF mod it, then?
23:40:02low_lightI have to see if it's the same as the mrobe100 or hdd1630
23:41:57funmannot sound however .. causes deadlocks
23:42:46CIA-38New commit by bagder (r21084): fix yellow by acknowledging the fread() return code and also allow ...
23:43:02low_lightfunman: that's what happed to the 925 before I found the magic gpio bit.
23:43:20funmanhum ok, i know what to do now ;)
23:43:44funmani suppose the OF bootloader doesn't used sound?
23:44:43low_lightfunman: akcodec-yh82x_yh92x.c line 60.
23:45:07low_lightI looked and that bit was also set in the 920.
23:45:31funman92205cb8e23724a802ceb24482cf6b96 fw_yh920.mi4
23:45:42 Quit BHSPitLappy (Remote closed the connection)
23:46:31low_lightThe OF bootloader will have a sound test section.
23:47:28kugellow_light: the key bindings for dir skipping are in the way :(
23:48:07kugelwe could remove them, you can also skip dirs by pressing, releasing, then holding |<< and >>| respectively
23:51:29CIA-38New commit by bagder (r21085): Added: ...
23:51:30low_lightkugel: "Keymap needs work" ;)
23:51:44kugelhence I'm already looking at it :)
23:51:53Bagderaha, the cia bot also announces www commits
23:53:16Mikachuheh, i have cia for rockbox set up in another channel and it shows commits differently than here
23:53:33BagderMikachu: and later I bet ;-)
23:53:43Mikachuactually the timestamp was identical
23:54:06Bagderthis one runs in the post-commit hook on the svn server
23:54:44Bagderand I believe zagor modified the output
23:57:36kugeland the backlight of the 925 is still overly weird!

Previous day | Next day