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 2010-05-16

00:00:11fmlgevaerts: could you comment on the 0..N-1 question? If I'm correct we could slightly reduce bin size.
00:00:32 Quit robin0800 (Quit: No Ping reply in 600 seconds.)
00:00:57 Join robin0800 [0] (
00:01:44gevaertsfml: it all depends on what exactly you want I guess. What you propose will make sure it's between 0 and N-1, but it could be argued (for some cases) that negatives should become 0 and >=N should become N-1
00:01:53Buscheldownloaded svn build 26070 for iPod Video crashes with "undefined construction at 000A7F4C"
00:02:16Buschelwhere do we have the map files?
00:02:50gevaertsfml: in that case, your code would be smaller while doing the same, yes
00:02:53fmlgevaerts: sure. But what I'm asking about is how to get the value into 0..N-1 by adding or subtracting N as needed.
00:03:04gevaertsunless I'm too sleepy right now :)
00:03:11*Buschel is a bit in a hurry as he will start his 1 week travel tomorrow morning and wants to have usable and up-to-date rockbox on his iPod
00:04:11LloreanBuschel: Any chance you accidentally put the wrong RAM size version on it?
00:04:19pixelmafml: does your fm frequency mean, I'll only get onw decimal now automatically?
00:04:48BuschelLlorean: No. This also happens on my local builds.
00:04:58AlexPpixelma: If you have a region set that changes in 0.1 MHz steps, yes
00:05:12fmlpixelma: yes, if you use European region settings. You'll get two digits if you use Italy (IIRC).
00:05:22gevaertskugel: you know there are still remaining reds?
00:05:31kugelyes, just saw
00:06:03amiconnImo the language/ voice mechanism shouldn't be used for stuff like wps
00:06:36amiconnI mean, for translating/ voicing strings which aren't part of the core (or plugins, once the support for that is done)
00:06:47pixelmadoesn't the tuner still use 0.05 for searching? I remember the c200 tuner even saving presets this way - I know that's wrong but I still would like to see that
00:08:00pixelmaamiconn: there already is a "use lang sting blah" tag and used for the "Next:" etc. in cabbiev2
00:08:01fmlpixelma: you mean when auto scanning?
00:08:27amiconnThat's a different thing - you only reuse a string that already exists
00:08:56amiconnBtw, I'd like to know who messed up voice ids recently...
00:09:15pixelmanope, there were 2 or so added just for this purpose, "Next:" being the example that comes to mind
00:09:24fmlMight be. But to be honest: in 99.99% of the cases you'll have a zero at the end which is not needed.
00:09:30pixelmaonly for cabbiev2 though
00:09:45*Buschel suspects r26051 is the case for the crahes and builds 26050
00:10:20kugelthe database tool needs rework :\
00:11:07fmlpixelma: were you talking to me?
00:11:22pixelmafml: but how can I find out now if the c200 tuner got it all wrong?
00:11:37gevaertskugel: everything that tries to use rockbox code on an OS needs rework...
00:11:49kugelthat's correct :(
00:11:59fmlpixelma: you'll hear it
00:12:00pixelmathe "Next" and cabbiev2 statements were directed to amiconn
00:12:17pixelmafml: I doubt it
00:12:23fmlpixelma: if there is noise then something is wrong and the tuner is not properly tuned
00:12:29kugelI'm not entirely sure why the database tool uses io.c..
00:12:55gevaertskugel: because that made it compile at the time I guess
00:12:56fmlpixelma: it's only displaying. The values is stored as before, and tuning is maded as before
00:13:07pixelmafml: reception isn't great with this tuner and sometimes the x.x5 sounded even better
00:13:11kugelprobably to provide file io because the system's functions were incompatible
00:14:02fmlpixelma: but to make 0.05 changes you have to switch to a region that allows that. And then you'll automatically get two digits after the point.
00:14:08pixelmafml: a preset that is (even falsely) saved as x.x5 - how will it be displayed?
00:14:26fmlpixelma: as x.x
00:14:52 Quit robin0800 (Remote host closed the connection)
00:15:17pixelmarounded up or down - what if it saved 91.55 and 91.60? and I've seen that happen
00:15:36fmlpixelma: I see: it's a trade off. But I think your case is rather an exception while mine is more common.
00:16:05fmlpixelma: the second digit is just cut, i.e. round down
00:16:24 Join Kitr88 [0] (
00:16:44pixelmait somehow feels inaccurate to me :\
00:16:47fmlpixelma: 91.5 and 91.6, respectively
00:17:11pixelmawell, imagine 91.5 and 91.55 then
00:17:47gevaertsIsn't the fundamental issue here that the c200 FM driver even provides those numbers?
00:17:59fmlI can make the formatting so that the trailing 0 (second digit) is cut, but a non-zero not. Would it be better?
00:18:10 Quit dfkt (Quit: -= SysReset 2.53=- Ph'nglui mglw'nafh Cthulhu R'lyeh wgah'nagl fhtagn.)
00:18:18pixelmagevaerts: maybe it is
00:18:39 Quit Kitar|st (Ping timeout: 240 seconds)
00:18:56fmlBut I'll have to leave in a minute. If you have any thoughts please post them on the mailing list.
00:19:10 Part fml ("Bye!")
00:19:14pixelmathe TEA tuner in my other targets is noticably better
00:21:17 Quit Kitr88 (Ping timeout: 268 seconds)
00:21:47 Quit pamaury (Quit: Page closed)
00:21:59 Join pamaury [0] (~c2c7a50a@rockbox/developer/pamaury)
00:22:53 Join Kitar|st [0] (
00:25:08CIA-5New commit by kugel (r26072): Fix last reds, database tool definitely needs rework.
00:25:48pixelmakugel: there was not only red for database tool, also for the OndioFM sim for some reason ;)
00:26:03kugelI saw that
00:26:21saratogaupdated with the rational we discussed today
00:28:23gevaertskugel: *those* reds are fixed :)
00:29:17 Quit TheSeven (Remote host closed the connection)
00:29:31kugelhm that was not a bad red
00:31:07kugelideally button-sdl.h should be named button-target.h so that it's not included in sim builds
00:31:10saratogathat email about running our libcook on solaris big endian hardware
00:31:21saratogaam i right in thinking theres no way such a device exists that doens't have an FPU?
00:31:26CIA-5New commit by kugel (r26073): Correct button_read_device prototype for HAVE_BUTTON_DATA.
00:31:41BuscheljhMikeS: can you please urgently take a look at FS #11277 ?
00:32:52kugelthe build table is pretty damn useful these days
00:33:30 Join TheSeven [0] (~TheSeven@rockbox/developer/TheSeven)
00:33:48gevaertssaratoga: 32 bit solaris isn't too new. It will depend on versions I guess
00:35:17saratogawikipedia says sparc v7 had an FPU in 1986
00:35:22 Quit ender` (Quit: If Java had true garbage collection, most programs would delete themselves upon execution. -- Robert Sewell)
00:35:41saratogathough it looks like our makefile is broken on ubuntu too, so maybe this should be fixed
00:36:59gevaertssaratoga: I'm not entirely sure if *all* sparcs have an FPU, but yes, the fact that it's one that runs solaris makes it more likely
00:42:56Buscheldark times for rockbox... green tables, but: iPod Video Sim does not build on cygwin, iPod Video for target crashes at startup...
00:43:23 Quit pamaury (Quit: Page closed)
00:43:34 Join pamaury [0] (~c2c7a50a@rockbox/developer/pamaury)
00:43:50Buschelso, the build table is not "pretty damn useful", but "pretty damn misleading"...
00:45:11saratogaobviously it can only check for compiler errors
00:46:05*Buschel is just frustrated...
00:46:09 Quit TheSeven (Remote host closed the connection)
00:46:19Buschelpc sim build error ->
00:47:21gevaertsit builds fine with mingw here
00:47:32Buschelin fact, I just wanted to update my iPod before my trip. Now I am stuck since 1 hour or so :(
00:47:56Buschelwill use my good old r26050 and hope that the dust settled until end of next week
00:47:57AlexPBuschel: Can you use a slightly older revision for now?
00:48:11BuschelAlexP: that's what i will do :)
00:48:43AlexPgood plan :)
00:49:06 Join TheSeven [0] (~TheSeven@rockbox/developer/TheSeven)
00:49:36 Quit pamaury (Quit: Page closed)
00:50:04Buschelok guys, have a good night and see you next weekend!
00:50:17 Quit Buschel ()
00:50:57kugelBuschel: try if that helps against the EXIT_SUCCESS error:
00:52:05gevaertskugel: the %zs introduced in r25850 and r25852 still "break" mingw builds
00:54:20kugelyes, that's unfortunate
00:54:45kugelfor some reason that mingw ansi stdio define doesn't help
00:59:55pixelmaI asked before but didn't get an answer - would updating gcc possibly help? I think I'm not at the latest possible under cygwin
01:00:20kugeli doubt it
01:00:32kugelupdating cygwin might be a good idea anyway :P
01:00:45gevaertspixelma: I doubt it. The issue isn't exactly simple unfortunately :\
01:01:16kugelgevaerts: hm, I think it does help actually
01:01:43kugelreplacing the snprintf in menu.c in rockbox with __mingw_printf makes the warning disappear
01:03:06gevaertskugel: we're actually confusing issues here I think. Cygwin and mingw aren't *exactly* the same. If you do a regular build in cygwin, you'll get a binary that's linked against cygwin's libc, not msvcrt
01:03:24gevaertsThere's also the fact that my mingw here doesn't even have __mingw_printf
01:03:46Tornegevaerts: they aren't identical, no, but the difference is only their specfiles
01:03:57Torneand perhaps not the same exact version/patches
01:04:07kugelit seems to have something to do with some plugins re-#defining the functions in questions which maybe breaks the mingw mechanism
01:04:13Tornethey're the same *target*, more or less :)
01:04:18gevaertsTorne: but this isn't actually a gcc issue, it's a libc issue
01:04:47Tornewell, the import libraries and headers for msvcrt should be the same in both, barring again minor version diffs
01:05:34gevaertsTorne: yes, if you build a mingw/msvcrt binary.
01:05:45gevaertsNot if you build a cygwin binary
01:05:55TorneOh, er, of course
01:05:56kugelgahh, __mingw_snprintf still doesn't know %lld!
01:06:24Torneyeah, the spec for -mcygwin has a different include/lib path than -mno-cygwin. *blames brain*
01:06:43 Join solexx_ [0] (
01:10:07 Quit solexx (Ping timeout: 264 seconds)
01:10:27kugelnot sure how to solve the midiutils.c one
01:13:05gevaertshm, I don't see that one
01:13:26kugel/home/kugel/rbdev/rockbox-git/apps/plugins/midi/midiutil.c:137: error: redefinition of ‘printf’
01:13:26kugel/usr/lib/gcc/i586-mingw32msvc/4.4.2/../../../../i586-mingw32msvc/include/stdio.h:252: note: previous definition of ‘printf’ was here
01:14:17pixelmaI got that one too
01:14:19gevaertsI'd say that plugins have no business stealing common names
01:14:49gevaertsJust rename it to my_printf() in that plugin
01:15:15kugelit should still be perfectly valid to override the system's printf, no?
01:16:02gevaertsmaybe, but that doesn't make it a good idea
01:16:06kugelI can only imagine prototype mismatch but it doesn't look like it
01:16:48kugelgevaerts: the %z things can be solved with -D__USE_MINGW_ANSI_STDIO=1 here
01:17:03kugelI don't know about %zd but it works for %z(u)
01:17:27 Quit petur (Quit: Zzzzz)
01:17:36gevaertsyes, but it doesn't work everywhere
01:17:42kugelthe redefinition of rockboy/doom breaks mingw's mechanism to replace it with __mingw_snprintf but that's solvable
01:17:47saratogahaha TTA can have both ID3 and APE tags
01:17:54gevaertsand unless we understand where and why, that makes it unsafe I think
01:18:05kugelwhere does it not work?
01:18:39gevaertswell here
01:19:17kugelthat's very strange
01:19:34gevaertsAnd from that mailing list thread you pointed to, it seems that this is indeed something that doesn't work in all versions
01:20:01kugelcan you post your mingw's stdio.h?
01:22:12kugelit doesn't have "extern int __mingw_stdio_redirect__(fprintf)(FILE*, const char*, ...);" and friends
01:26:16 Join krabador [0] (
01:31:09kugelgevaerts: did you compile mingw yourself?
01:31:45kugelubuntu's package seems to have it correctly
01:37:55gevaertskugel: what
01:38:04gevaertss your version of mingw32-runtime?
01:39:03kugelVersion: 3.15.2-0ubuntu1
01:39:26gevaertsok, I have 3.13
01:41:49gevaertswhich is nearly three years old :\
01:43:59kugeldebian doesn't seem to offer any newer versions
01:44:10gevaertsIf I install ubuntu's mingw32-runtime, it works fine
01:45:37gevaertsWe'll just have to decide if requiring this newer set of libraries is ok or not
01:46:41kugelI'd say yes, but debian is pretty major :/
01:47:18gevaertsit is indeed, but this is a package with no dependencies and no conflicts, so grabbing the package from ubuntu is easy
01:47:27kugelbut even then it's not completely solved. there's %lld used somewhere
01:47:53kugeldoom, d_deh.c
01:49:42gevaertsalso a few in speex.c
01:50:59***Saving seen data "./dancer.seen"
01:51:06gevaertsStill, I prefer a solution that leaves eight cases looking possibly ugly to one that makes nearly every single format string hurt eyes
01:53:15gevaertsare you sure? %lld seems to work here
01:54:31 Quit Boldfilter (Ping timeout: 264 seconds)
01:58:15gevaertsthat's really weird. It works in my test program
02:00:34 Quit TheSeven (Ping timeout: 268 seconds)
02:00:55kugelright, it also works here in a test program
02:01:44 Join TheSeven [0] (~TheSeven@rockbox/developer/TheSeven)
02:02:05kugelthis is all very weird :(
02:03:33gevaertsI also don't get any warning in doom
02:06:31 Quit TheSeven (Ping timeout: 264 seconds)
02:07:21kugelgevaerts: trying the doom cflags for the test app without success :(
02:08:07gevaertskugel: you did add the -D__USE_MINGW_ANSI_STDIO=1 to GCCOPTS?
02:09:17gevaertsand it's there if you run make V=1?
02:09:23gevaerts(for the doom files)
02:10:23 Join TheSeven [0] (~TheSeven@rockbox/developer/TheSeven)
02:16:11gevaertskugel: I get for d_deh.c
02:18:12 Join Rob2222 [0] (
02:18:51kugel is mine
02:21:56 Quit Rob2223 (Ping timeout: 248 seconds)
02:22:14*kugel likes the comment "// All deh values are ints or longs"
02:22:34kugelit means they use int64_t even though it's casted to long at max
02:25:20gevaertsAll I can see is that you end up with -D__USE_MINGW_ANSI_STDIO=1 twice, and that you put it in a different place in the commandline
02:25:40gevaertsApart from that (and a different target) they're identical
02:26:02kugelI added it to EXTRADEFINES as well
02:27:20 Quit DerPapst (Quit: Leaving.)
02:30:07gevaertsfg@zorac:/tmp$ /usr/bin/i586-mingw32msvc-gcc −−version
02:30:07gevaertsi586-mingw32msvc-gcc (GCC) 4.2.1-sjlj (mingw32-2)
02:30:19gevaertsI suspect that one will also be the same though
02:30:50kugelgevaerts: explains why that 3.15 introduced __D_USE_MINGW_ANSI_STDIO
02:30:58kugelwhich is apparently also the cause for the printf error
02:31:14 Join wincent [0] (
02:31:31 Quit wincent (Changing host)
02:31:32 Join wincent [0] (~wincent@rockbox/developer/wincent)
02:33:19 Quit TheSeven (Ping timeout: 245 seconds)
02:33:24gevaertsok, and that one should also be gone with new enough mingw runtime
02:35:58kugelI can't find the file the cvs commit mentions
02:36:25gevaertswhich one?
02:36:38 Join TheSeven [0] (~TheSeven@rockbox/developer/TheSeven)
02:38:09gevaertsAnyway, if compiler-as-packaged-by-distribution-is-not-good-enough is the only problem, I see no serious reason to not just add a mingw toolchain to
02:38:57kugelthat's a good idea
02:40:18kugelah, wrong repo :\
02:42:10kugelhm, the static keyword is still there in the gnuc part of __mingw_stdio_redirect__ definition
02:43:18kugelah, interesting, the stdio functions expand to functions within the source file
02:46:45gevaertshm, I think plugin.h includes too much
02:51:46kugelgevaerts: I wonder if the __mingw_printf thing is only a temporary solution; it won't help if we ever happen to have a native win32 port
02:52:15gevaertskugel: maybe, but that's a problem for whoever does that port
02:53:03 Join RPG-Master [0] (
02:53:48RPG-MasterHey, last time I used Rockbox on my Fuze, it tended to suck the battery quicker then the default firmware. Is it doing any better these days?
02:54:06*kugel this mingw mess distracted me from updating my pth work :/
02:54:32kugelRPG-Master: better than before but still worse than the OF I think
02:56:01RPG-Masterkugel: L(
02:58:51RPG-MasterMy poor friend has a Zune G1. Has anyone attempted porting Rockbox to it?
03:02:36kugelgevaerts: adding a mingw target to won't help for cygwin, will it?
03:04:05saratogai thought you could build the compiler on cygwin with it
03:04:46kugelyea, the cross compiler, but it would be strange to build a "native" compiler with it
03:05:01gevaertskugel: which particular issues does cygwin have?
03:05:16kugelthe same as mingw I think
03:05:21gevaertsIt seems to do %z correctly at least
03:07:53 Part RPG-Master ("Ex-Chat")
03:10:12 Join mobile [0] (~evilnick|
03:10:26gevaertskugel: mingw-runtime seems to be at 3.17 or 3.18 in cygwin
03:10:54gevaertsso for cygwin-native builds I don't see a problem at all, and for mingw builds there shouldn't be any either
03:16:17kugelok, that sounds good
03:16:44 Quit kugel (Remote host closed the connection)
03:16:44 Quit mobile (Remote host closed the connection)
03:17:52 Join mobile [0] (
03:27:15 Quit adnyxo (Ping timeout: 264 seconds)
03:38:25 Join ischeriad [0] (
03:40:19 Join Boldfilter [0] (
03:42:15 Quit ischeria1 (Ping timeout: 264 seconds)
03:51:01***Saving seen data "./dancer.seen"
04:10:22 Quit JdGordon (Quit: Leaving.)
04:12:59 Quit TheSeven (Ping timeout: 276 seconds)
04:16:05 Join TheSeven [0] (~TheSeven@rockbox/developer/TheSeven)
04:19:04 Join pixelma_ [0] (quassel@rockbox/staff/pixelma)
04:19:04 Quit pixelma (Disconnected by services)
04:19:08 Quit amiconn (Disconnected by services)
04:19:11 Join amiconn_ [0] (quassel@rockbox/developer/amiconn)
04:19:23 Nick pixelma_ is now known as pixelma (quassel@rockbox/staff/pixelma)
04:19:30 Nick amiconn_ is now known as amiconn (quassel@rockbox/developer/amiconn)
04:21:56 Join Forsaken_Boy [0] (~chatzilla@
04:26:06 Join JdGordon [0] (
04:26:06 Quit JdGordon (Changing host)
04:26:06 Join JdGordon [0] (~Miranda@rockbox/developer/JdGordon)
04:31:45 Quit krabador (Remote host closed the connection)
04:46:27JdGordonthe 2 cons for APE in mp3 is utter crap
04:49:46 Join dys` [0] (
04:51:16 Quit GeekShadow (Quit: The cake is a lie !)
04:51:23 Quit Forsaken_Boy (Quit: ChatZilla 0.9.86 [Firefox 3.6.3/20100401080539])
04:51:50 Quit dys (Ping timeout: 276 seconds)
04:51:56 Join Barahir_ [0] (
04:55:33 Quit Barahir (Ping timeout: 264 seconds)
04:57:10 Quit Strife89 (Ping timeout: 260 seconds)
04:59:13 Quit Bagder (Ping timeout: 258 seconds)
04:59:46 Join Strife89 [0] (
05:05:17 Join Forsaken_Boy [0] (~chatzilla@
05:08:16 Part Boldfilter
05:13:42 Join Bagder [0] (~daniel@rockbox/developer/bagder)
05:14:58 Quit elcan (Read error: Connection reset by peer)
05:18:22 Part TFGBD
05:28:30 Join elcan [0] (
05:40:23 Quit anewuser (Quit: for SELL 2 by the price of 1 now!)
05:51:04***Saving seen data "./dancer.seen"
06:16:01 Quit JdGordon (Ping timeout: 260 seconds)
06:23:33ranmaUgh. usb_core seems to sometimes pass UNCACHED_ADDR() ptrs to usb_drv_send or usb_drv_recv? That looks broken to me...
06:25:28CIA-5New commit by jethead71 (r26074): Make sure to include audiohw.h in settings.h or the definition of struct user_settings could get out of sync amongnst various #includes. Might be the ...
06:27:16jhMikeSranma: not if it needs physical, uncached addresses. careful with that! :)
06:27:23 Quit Strife89 (Quit: Bed. ZZzzzzzzzz....)
06:28:44jhMikeSif this is for AMS, then that memory setup is dodgey (mapping cached right on top of physical, and making uncached address a virtual one)
06:30:40jhMikeSother targets map cached=physical, cached=virtual. personally, if I started working on AMS, I'll probably fix that blunder
06:34:59 Join JdGordon [0] (~jd@
06:34:59 Quit JdGordon (Changing host)
06:34:59 Join JdGordon [0] (~jd@rockbox/developer/JdGordon)
06:37:48 Quit JdGordon (Client Quit)
06:37:57 Join JdGordon [0] (~jd@rockbox/developer/JdGordon)
06:37:57 Join Xerion_ [0] (
06:39:43 Quit Xerion (Read error: Operation timed out)
06:39:43 Nick Xerion_ is now known as Xerion (
06:40:02 Quit JdGordon (Client Quit)
06:40:11 Join JdGordon [0] (~jd@rockbox/developer/JdGordon)
06:42:28 Quit JdGordon (Client Quit)
06:42:47 Join JdGordon [0] (~jd@rockbox/developer/JdGordon)
06:55:51 Quit JdGordon (Quit: Bye)
06:56:09 Join JdGordon_ [0] (~jd@rockbox/developer/JdGordon)
07:00:36 Quit JdGordon_ (Ping timeout: 246 seconds)
07:00:45 Join FlynDice [0] (~FlynDice@
07:05:04S_a_i_n_tSomething I've noticed recently (got some new IEMs) is that my DAPs (Nano 1/2G) produce a weird white-noise/hiss when they are first started up (audible even at very low volumes) and that this hiss will persist untill either the first keyclick, .voice/.talk, or audio file is played. Any ideas?
07:05:15S_a_i_n_tNew IEMs have made this *really* noticable.
07:06:00S_a_i_n_tAfter even one single, solitary keyclick the hiss/noise stops dead...
07:06:24jhMikeSsomething not set up properly until audio is first played?
07:06:41*S_a_i_n_t shrugs
07:08:03jhMikeSgigabeat F/X once had something like that where the first audio out fixed a problem, only it wasn't noise but heavy current draw
07:08:14 Join bieber [0] (
07:08:31S_a_i_n_tI've always noticed a "click" or a "pop" when the DAP first starts up, but only just now noticed that it persists with a hissing sound until keyclick/voice/playback etc.
07:08:42S_a_i_n_tthen silence is actually silent.
07:17:19 Join Horscht [0] (~Horscht2@xbmc/user/horscht)
07:18:05 Quit Horschti (Ping timeout: 246 seconds)
07:22:16 Join Buschel [0] (
07:22:52 Quit Zarggg (Quit: Zarggg)
07:29:21 Join Zarggg [0] (
07:30:25 Join funman [0] (~fun@rockbox/developer/funman)
07:32:13Buschelkugel: your small patch to button_sdl.c helps. pcsim is building again with cygwin.
07:39:25 Join mikroflops_ [0] (
07:43:12 Quit mikroflops (Ping timeout: 240 seconds)
07:46:07LloreanS_a_i_n_t: Might it be flash access? See if doing something that involves several loads like switching themes also produces it?
07:46:58S_a_i_n_tLlorean: Already had that idea ;) And (unfortunately) the hiss/white-noise is only present on startup.
07:47:23S_a_i_n_tTheSeven had an explanation for the "pop" on startup, but it was a bit too tachie for myself.
07:47:30jhMikeSit persists after all activity from database and dircache stops?
07:47:32S_a_i_n_t*techie even
07:48:17 Join JdGordon [0] (~jonno@rockbox/developer/JdGordon)
07:48:19S_a_i_n_tjhMikeS: Yep, if I start the DAp and just leave it (assuming that idle shutdown is off) it will hiss all day if I let it.
07:49:15jhMikeSBuschel: is that ipodvideo problem fixed?
07:51:06***Saving seen data "./dancer.seen"
07:53:08S_a_i_n_tsomeone with very low quality IEMs could well miss this hiss, or if they just start up the DAP and start navigating the menu with keyclicks or .voice/.talk enabled. But (as with most things of this type) now that I *have* noticed it, I can't *not* hear it ;)
07:53:13funmanapparently the as3543 rtc has an auto-wakeup feature
08:11:05BuscheljhMikeS: just testing for pc sim (it also crashed) as my iPod is already hidden somewhere in the void of my luggage
08:14:23BuscheljhMikeS: hmm, sim still hangs.
08:14:59BuscheljhMikeS: but it looks different. at least the menu shows up
08:15:02 Join esperegu [0] (~quassel@
08:15:27saratogacheap fuzes on woot in the US today
08:18:34JdGordonis anyone not happy with the radio art system I've setup? I.e images are in .rockbox/fmpresets/<preset name>.bmp or .jpg ?
08:19:14jhMikeSif the sim hangs, then that's sort of good...I can at least play with that
08:20:27BuscheljhMikeS: yep. good luck! :o)
08:21:21 Quit Buschel ()
08:30:51 Quit S_a_i_n_t ()
08:32:14 Quit JdGordon (Ping timeout: 268 seconds)
08:35:51 Join S_a_i_n_t [0] (S_a_i_n_t@
08:38:22 Quit mobile (Read error: Connection reset by peer)
08:44:05saratogajhMikeS: charging on the fuzev1 seems to be broken in today, could you charging commits have caused that?
08:44:36 Quit esperegu (Ping timeout: 258 seconds)
08:45:56 Join esperegu [0] (~quassel@
08:46:21saratogaoddly the clipv2 seems fine
08:58:03jhMikeSI reverted the commits
09:00:06 Quit arbingordon (Quit: `)
09:00:44 Join JdGordon [0] (~jonno@rockbox/developer/JdGordon)
09:05:29funmanwoa: the (public) as3527 datasheet use an as3543 (or compatible) and all the bits are detailed
09:11:42funmanbtw, the as3527 mentions as3517, so the as3525(v1) doesn't use it
09:14:12 Quit [CGL] (Read error: Connection reset by peer)
09:16:17*jhMikeS can't even compile the ipv sim
09:17:05 Quit Beta2K (Ping timeout: 260 seconds)
09:18:00 Join Beta2K [0] (
09:20:21*jhMikeS gives it one more round, clean :\
09:22:52jhMikeSurgh, button-sdl.c 176, implicit declaration of function exit, EXIT_SUCCESS undeclared
09:34:31 Quit Galois (Ping timeout: 264 seconds)
09:39:58 Quit esperegu (Read error: Operation timed out)
09:42:06 Join hebz0rl [0] (
09:42:09 Join esperegu [0] (~quassel@
09:51:10***Saving seen data "./dancer.seen"
09:52:20 Join stoffel [0] (
09:54:52 Quit esperegu (Ping timeout: 260 seconds)
09:55:05 Join esperegu [0] (~quassel@
09:57:27 Quit wincent (Ping timeout: 276 seconds)
10:04:11 Join n1s [0] (~n1s@rockbox/developer/n1s)
10:23:24 Quit n1s (Read error: Connection reset by peer)
10:23:33 Join n1s [0] (~n1s@rockbox/developer/n1s)
10:23:53 Join mitk [0] (
10:24:09 Part mitk
10:24:20 Join mitk [0] (
10:35:17funmanthe 'as3514' in as3525 seems to be an as3515
10:39:45 Join DataGhost [0] (
10:39:45 Quit DataGhost (Changing host)
10:39:45 Join DataGhost [0] (~dataghost@unaffiliated/dataghost)
10:40:49 Join ender` [0] (
10:52:34 Join flydutch [0] (
10:53:09 Join bmbl [0] (~Miranda@unaffiliated/bmbl)
10:57:47 Join DerPapst [0] (
11:00:17funmanranma: ascodec_*_endofch_irq() is both an exported function & a static inline
11:03:40JdGordonfrom -community... is anyone looking at commiting 11270?>
11:04:20S_a_i_n_tJdGordon: I had wondered if you'd looked at that.
11:04:28 Join Lombax [0] (
11:05:01JdGordoni looked at the first version... i dont have any objections to it.... does anyone?
11:05:23 Join azazel [0] (
11:05:25S_a_i_n_tAm I right in thinking that oscilloscope, fft, other audio-relevant plugins could be done in the same way?
11:07:26 Join pamaury [0] (~c2c7a50a@rockbox/developer/pamaury)
11:07:33JdGordonsure... but thats not really the way to do it
11:08:03 Quit bluefoxx (Ping timeout: 240 seconds)
11:08:56 Quit Lombax (Ping timeout: 260 seconds)
11:09:47 Join merbanan [0] (
11:09:48S_a_i_n_tWell, I hardly have any sway regarding if it gets committed or not, but I think it's a damn fine idea.
11:10:42 Quit azazel (Ping timeout: 268 seconds)
11:14:02 Join bluefoxx [0] (
11:21:10 Quit Forsaken_Boy (Quit: ChatZilla 0.9.86 [Firefox 3.6.3/20100401080539])
11:22:06CIA-5New commit by funman (r26075): as3525v1 use an as3515 audio codec/PMU, not an as3517 as previously thought ...
11:29:09ranmafunman: But arm/ascodec-target.h does not get included on AMS, right? In as3525/ascodec-target.h it's not an inline.
11:29:46 Join Galois [0] (
11:30:09ranmaI think armv6 (clean|invalidate)_dcache range is broken
11:31:15funmanranma: ah right i was reading the wrong file
11:31:35funmanyou have an armv6?
11:32:33ranmaOh, ok, it's armv4
11:32:46ranmaThen armv4 (clean|invaliate)_dcache range is broken :)
11:33:06funmanwhat's wrong with it?
11:33:47ranmaI think it's just not working properly. If I use clean_dcache_range(buf, buflen) it doesn't work, clean_dcache() seems to work though.
11:34:41funmandoes it work if you align buf on 32 bytes?
11:34:52funmanalign start of buffer & buffer size
11:35:26ranmaI'm getting the buffer from usb_core, but I tried clean_dcache_range(buf-32, buflen+64)...
11:36:09funmanbuf is in 0x30xxxxxx range ?
11:36:19 Join JohannesSM64 [0] (
11:36:58funmani.e. not UNCACHED(something)
11:37:38ranmaYep. I have to explicitly catch uncached buffers and convert the address, because I can't DMA to/from an uncached address...
11:37:57funmanUNCACHED is only used in storage driver it seems
11:38:14ranmaI think it's broken that usb-storage passes in UNCACHED addresses.
11:38:46ranmaI'd rather it wouldn't do that :)
11:39:36funmanyes it should go in the target drivers
11:39:41funmanthe core doesn't know if we use dma or not
11:43:19ranmaHmm, I don't see why clean_dcache_range doesn't work.
11:43:54funmannothing accessing the data after you cleaned it ?
11:45:00ranmaNope, right now I'm copying it over into my own buffer and that's only accessed in that function.
11:45:23ranmaHmm, but I should check alignment...
11:47:02ranmaNo, even if it's not aligned, the buffer before that one is not used, so it shouldn't be a problem with some other data sharing the same cacheline...
11:47:34ranmaAnd thinking of cachelines, I wonder if I can even trust usb_core to pass in cacheline aligned buffer for receive and transmit...
11:48:03funmanwhat if you only access the uncached buffer?
11:48:33ranmaThen I definetly have to use a local buffer.
11:48:42ranmaThe samsung usb driver doesn't use one though.
11:49:01ranmaI'm using usb-s3c6400x.c as a reference here...
11:49:48ranmaWhich also uses clean_dcache() and invalidate_dcache() instead of the range() variants.
11:49:56 Join kugel [0] (~kugel@rockbox/developer/kugel)
11:51:13***Saving seen data "./dancer.seen"
11:51:19kugelranma: I think the usb buffers are cache aligned
11:51:47CIA-5New commit by kugel (r26076): Fix building on cygwin.
11:53:11ranmaI just found a potential issue with my dma descriptors sharing cachelines with other data...
11:56:11 Quit mitk (Quit: Leaving)
12:01:42 Quit shaggy-h (Ping timeout: 240 seconds)
12:02:43 Join JdGordon1 [0] (
12:02:48 Join MethoS- [0] (~clemens@
12:14:44 Quit bieber (Ping timeout: 260 seconds)
12:15:08 Join bieber [0] (
12:21:55jhMikeSkugel: what's the proper off_t min/max definition?
12:22:15kugeldon't know
12:22:57jhMikeSI thought you might with all the sim stuff...hmm
12:23:36kugelyea, but off_t is something I never understood ;)
12:24:33CIA-5New commit by funman (r26077): as3525: make sure we don't use a negative number of sectors ...
12:25:05funmani think it's important for handling files > 2 or 4GB on 32bits cpus
12:25:12jhMikeSjust a signed offset, related to the whole size_t and ssize_t group I guess
12:26:27jhMikeSknow the max would save me checks
12:26:30funmani don't see off_t in posix/limits.h
12:27:03funmanjhMikeS: size_t might be smaller than off_t though
12:27:14 Join Jaykay [0] (
12:27:17kugelI thought the opposite
12:27:46funmansize_t can't go past 4G on 32 bits cpu but you can still be at position 8GB of a very large file
12:27:47jhMikeSI'm handling files so I'm keeping it consistent.
12:28:22kugel <−− this is on a 64bit system
12:28:39kugelit only creates off_t as 64bit and off64_t when asked to
12:30:32kugeljhMikeS: how about if (sizeof(off_t) == sizeof(long)) OFF_T_MAX = LONG_MAX else {...} ?
12:31:01funmanbetter just cast off_t to long long
12:31:21funmanit's simpler no?
12:31:31kugelbut is it correct?
12:31:53AlexPJdGordon: I don't object to the principle :)
12:32:01funmani guess
12:32:26funmani can't find 'off_t' in n1256.pdf (ISO/IEC 9899:TC3 draft)
12:32:44kugelthat's because it's not iso c
12:32:58funmanis it posix?
12:33:13kugelposix/unix yes, defined in sys/types.h
12:35:22 Join pamaury_ [0] (~c2c7a50a@rockbox/developer/pamaury)
12:35:41 Join Llorean1 [0] (
12:36:40 Quit Llorean (Ping timeout: 260 seconds)
12:37:08 Quit bieber (Ping timeout: 260 seconds)
12:37:31 Quit pamaury (Ping timeout: 252 seconds)
12:37:34 Join bieber [0] (
12:40:59*kugel would like to commit his pth thread work
12:40:59 Join moos [0] (moos@rockbox/staff/moos)
12:41:43kugelgevaerts: ping :)
12:42:33jhMikeShmmm...of course there's always (~(type)0 - ((type)1 << (sizeof (type)*8 - 1)) :)
12:44:14jhMikeSor just the last term - 1
12:49:45funmanranma: usb serial works?
12:56:31ranmaDidn't try, but since usb-storage is nowhere near working yet...
12:59:12funmani'd have left storage for the end, simplest things first ;)
12:59:44JdGordon1anyone got a radio target able to do a test?
13:00:43JdGordon1does anyone know which targets have HAVE_NOISY_IDLE_MODE enabled?
13:00:57ranmaAh, damn. I see why control sends don't work without memcpy to my buffer. The core buffer is in iram and so the address is wrong, but doesn't give a bus error since it's an extram alias address
13:01:17pixelmaJdGordon1: what test?
13:01:21kugelgevaerts: uploaded the patch to FS #11234. jhMikeS, you're potentially also interested in it?
13:01:28JdGordon1a patch
13:01:43ranmaIs there 'virt2bus' or something like that?
13:01:55pixelmaJdGordon1: well, what patch?
13:02:27JdGordon111263.. but it needs some setup... meh never mind
13:02:53kugelranma: UNCACHED_ADDR for iram I think
13:03:09kugelhm, no
13:03:18*jhMikeS has to disagree about cooperative being advantageous, other than that I'll looksee :)
13:04:05kugelranma: but subtracting IRAM_ORIG from the start pointer should work
13:06:21 Quit esperegu (Read error: Operation timed out)
13:06:56kugelfunman: that's the start of the alias, he needs the phys. location for dma
13:07:00funmanhm no that's the reverse :o
13:07:09*JdGordon1 thinks there is a bug in the dependancy generation
13:07:25funmanthen -(what i said) ;)
13:07:33JdGordon1filetypes.c shuold be depending on english.lang but once every so often it will start compiling before the langs are done
13:08:03ranma if (addr & 0x10000000) { /* uncached address */
13:08:04ranma addr -= 0x10000000;
13:08:04ranma }
13:08:04DBUGSent KICK ranma to server
13:08:04ranma if (addr >= IRAM_ORIG) { /* iram address */
13:08:04ranma addr -= IRAM_ORIG;
13:08:04Kick(#rockbox ranma :No flooding!) by logbot!
13:08:53 Join ranma [0] (
13:09:17kugelsomeone should hug logbot a bit, it's quite sensitive these days
13:09:41AlexPseems to be 5 lines in no time
13:10:28kugeljhMikeS: not advantageous generally but in our situation
13:11:26funmanas3525v2 do NOT use as3517 (they have separate line1/line2 and mic1/mic2 registers), so assumption of as3543 still holds
13:12:27 Join esperegu [0] (~quassel@
13:13:29pamaury_funman: I just looked at the audio isr of the fuzev2 firmware and indeed it does some initialization but I don't understand how it decides ot init usb or not...
13:13:40 Nick pamaury_ is now known as pamaury (~c2c7a50a@rockbox/developer/pamaury)
13:13:44CIA-5New commit by jdgordon (r26078): FS #11263 - Radio Art support! %C and %Cl tags work in the radio screen and Base Skin when the radio is running. ...
13:14:13funmanpamaury: isn't it just unconditional?
13:14:46pamauryit reads a value in memory which is a pointer and then reads the content and compare to 0
13:14:51JdGordon1AlexP: hey, can you fix the manual for that ci? do I need to add a FS task for it?
13:15:05AlexPI can, and yes please as I can't right now :)
13:15:15 Join Schmogel [0] (
13:15:17AlexPand I don't want to forget
13:16:08*JdGordon1 spots the sbs being called the Custom Status Bar in the manual and decides to go against all his morals and fix it
13:16:19AlexPhehe :~)
13:16:21funmanpamaury: it writes 1 just after so I guess it's: static bool usb_detected = false; if(usb_detected) return; usb_detected=1;
13:17:02funmanbtw usb detection works the same way than on as3525, but i still don't get why we get thousands of audio interrupts
13:17:02 Join petur [0] (~petur@rockbox/developer/petur)
13:17:13funmanthey tend to just stop at random also
13:17:16 Quit esperegu (Ping timeout: 260 seconds)
13:17:24funmanthe 2 additional irq registers just read as 0
13:17:30JdGordon1AlexP: do you know which file chap 12.c would be in?
13:17:45pamauryBut i audio isr only used for usb detection ? Doesn't that defeat the purpose of 'audio' interrupt ?
13:17:48AlexPwhat is the chapter?
13:17:49JdGordon1ah found it
13:20:18funmanpamaury: 'audio' is named AFE in linux code
13:20:19JdGordon1yeah, no idea how to fix the manul for radio art...
13:20:37funmanit's used for interrupts of audio/pmu .. i guess the name is bad yes
13:20:39pixelmaAlexP: two manual questions: (1) I have a patch to update and make the start screen description more complete, would you have a look? (2) Do you think it is possible to make the \Actions use "\ActionBlah (remote: \ActionRCBlah)" except in the button table environment?
13:20:44CIA-5New commit by jdgordon (r26079): .sbs is base skin, not custom statusbar
13:21:11pixelmaI'm looking for a more general solution to this so I don't have to look for every text
13:21:26funmanranma: fixes -> is it ok?
13:22:03AlexPpixelma: 1) Sure 2) sounds like the best plan - you mean is there a way to automagically do it?
13:22:06pixelmaAlexP: I would think that's a cleaner solution and avoids things being forgotten later
13:22:52pixelmaAlexP: yes, I could *imagine* there is a way to set this up in preamble.tex etc.
13:22:55 Join dfkt [0] (~dfkt@unaffiliated/dfkt)
13:23:13AlexPI guess there are two options - a new macro, or to redefine the current one so it does one thing in button tables and one thing everywhere else
13:24:16AlexPI don't immediately know how, but it must be possible :)
13:24:18pixelmaI'd prefer the latter I think, the former means going through all texts and change to the new macro
13:24:33pamaurythat's a shame the fuzev2 write a undocument bit in PCGCCTL...
13:24:44AlexPpixelma: Yes, it would be good I think
13:24:58pixelmaand later have to rethink if you need \ActionAllBlah or not
13:25:37pixelmaAlexP: about (1) - pastebin good enough or the tracker?
13:26:54AlexPpastebin I'd have thought if we are looking now
13:28:08CIA-5New commit by jdgordon (r26080): yellow go bye bye
13:28:55pixelmaAlexP: there are some comments in it for myself
13:30:18JdGordon1rasher: ping
13:30:26rasherJdGordon1: yes?
13:30:36AlexPpixelma: "automatically start to record" −−> "automatically start recording"
13:31:12pamauryI'm a bit puzzled by the linux code for the as353x, it's incosnistent in the use of bitfields... something lsb to msb and something the reverse :( Furthermore, there are plenty of naming inconsistencies
13:31:16JdGordon1rasher: have you got a sec to look at a patch for the theme site to run the fms's through checkwps?
13:31:30JdGordon1trying to find it on the tracker...
13:31:43funmanpamaury: perhaps there will be similarities with the as352x code?
13:32:08AlexPOnly other thing - I can't remember what the convention is (if we have one) w.r.t. using capital letters after the item: e.g. [Item.] description - shoukd description be Description? The full stop suggests yes
13:32:11AlexP@ pixelma
13:33:19pixelmanot sure there is a convention, I just kept it the same style as before
13:33:24pamauryfunman: not really :(
13:33:30rasherJdGordon1: and checkwps parses .fms files?
13:33:41pixelmaAlexP: but maybe there should be one
13:33:54rasherJdGordon1: I don't see anyway that could possibly break anything
13:34:04JdGordon1it should, I dont tihnk it cares about file extensions
13:34:20AlexPpixelma: I think it'll look odd to have a full stop after the item name, then start with a small letter
13:34:36JdGordon1arg, checkwps is braindead checking for remotes
13:35:26rasherJdGordon1: it's okay, the theme site doesn't check them (I think) :)
13:35:46JdGordon1if that patch is ok ill commit it
13:36:13rasherIt should be, go ahead and I'll import it
13:37:23CIA-5New commit by jdgordon (r26081): make the themesite check .fms files through checkwps
13:37:45rasherJdGordon1: change is live now
13:37:57CIA-5New commit by jdgordon (r26082): make checkwps accept .rfms files as remote skins
13:38:44*JdGordon1 wishes there was a way to know which themes use what skins, and tag stats
13:39:11S_a_i_n_tJdGordon: I've often wished the same...
13:39:22JdGordon1also having it not download every theme for a target... grab them in lots of 10 or something
13:40:38pixelmarasher: the theme site also checks the cfgs, right? Could it reject greyscale themes for displays that exist as greyscale and colour - made for colour targets and setting global foreground colours? They are not rejected currently if the WPS does not define different colours then they are technically correct but since greyscale doesn't know foreground and background colours you can end up with an unreadable display
13:40:57pixelmahappens when the theme uses white text on a dark backdrop
13:41:28pixelmathe backdrop will be displayed but the text is still black on the greyscale display
13:41:50linuxstbkugel: Was there any recent discussion about the decision to move the sim SDL code into target tree? (I haven't been paying close attention the last few days)
13:42:14kugellinuxstb: yes, on the ML
13:42:44kugelI tried to ping because I remembered you wanted to take part but you weren't available
13:43:21AlexPpixelma: The problem I see here is that \ActionXxx is defined in the keymap file, not preamble.tex, and I don't know which is run through first. The second problem is that what is then wanted is after that definition to redefine \ActionXxx = \ActionXxx (remote: \ActionRcXxx) but only for specific places - I think we will need a different macro
13:43:45AlexPUnless latex has a sort of built in search and replace function
13:48:04pixelmaI thought of the spot where the definition of how to deal with \Action, is it not in preamble.tex?
13:48:16*pixelma should take a look herself
13:48:45CIA-5New commit by funman (r26083): as3525: don't use incomplete USB code when charging
13:49:24AlexPpixelma: Can't see anything - each \ActionXxx is a macro defined in the keymap file
13:51:16***Saving seen data "./dancer.seen"
13:52:56pixelmahmm... I thought it would be similar to making button tables 2 or 3 columns depending on HAVEREMOTEKEYMAP or not
13:53:37AlexPThat just changes the table though, you still need to type in both \Action \ActionRc
13:54:12linuxstbkugel: Sorry I missed that thread. I'm still not convinced it's a good idea - simulating Rockbox running on a real target is not the same as RaaA, so my feeling is that there won't end up being much in common (but I may of course be wrong - you've looked at this a lot more than me). So I think I would have preferred the sim SDL code was copied to create a clean SDL target - once the sim code is stripped out, I don't think there's much l
13:54:23pixelmaAlexP: yeah, I meant the principle "deal with things differently"
13:54:45AlexPI think each action needs a new macro - at least that way people don't have to litter the source with \opt{remote}, it is done automatically
13:55:11 Quit FlynDice (Remote host closed the connection)
13:55:53kugellinuxstb: That's unfortunate. I am convinced
13:57:03linuxstbkugel: Convinced of what?
13:57:04kugelfor now I only moved the "sdl device driver", the common sim-only related code is still in uisimulator/common
13:57:49CIA-5New commit by nls (r26084): Since we no longer use -fno-strict-aliasing in CFLAGS we don't need to strip it out for the codecs.
13:57:50kugelconvinced that it's a good idea
13:58:23 Quit funman (Quit: free(random());)
13:58:26 Quit TheSeven (Ping timeout: 276 seconds)
14:00:06kugeleven more so after doing the work
14:03:18 Join TheSeven [0] (~TheSeven@rockbox/developer/TheSeven)
14:03:58AlexPpixelma: works fine at the end of preamble.tex, but the downside is that it needs one for each action, and then people need to use different macros in the text and in the button tables. Also, if remotekeymap is defined but the specific key isn't, it'll fail to build
14:04:40pamauryWho is working on the as2525v2 and can answer my question ? There a function called during usb initialisation which set a bit in a register at 0xC8100020 but I can't find what it is, any idea ? It's just after the CCU ones...
14:05:49n1sis there any reason why not overwriting CFLAGS in root.make is a bad idea? we currently overwrite it but if we didn't, testing with different compiler flags is much easier.
14:08:12S_a_i_n_tdoes a .lang file need a newline at the end?
14:08:49 Join fml [0] (
14:08:55 Quit JdGordon1 (Ping timeout: 246 seconds)
14:09:00Tornen1s: overwrite what?
14:09:24Tornecan't you just do make CFLAGS="whatever" ?
14:10:23fmlpixelma: one more thought about displaying FM frequencies: what would you do if the excat value reported by the chip is 85,123,456? Would you display all the digits? I think no. S oI think we should assume that the chip gives us numbers accorsing to the step spec.
14:10:53fml...and hence the change is OK
14:11:28 Quit fml (Client Quit)
14:12:44n1sTorne: no, because tools/root.make overwrites it
14:13:16Tornen1s: how?
14:13:21n1smaking root.make just append stuff works of course but i wonder if it's likely that CFLAGS will contain junk
14:13:31Tornevariables on the make command line take precedence over variables in the makefile
14:13:34linuxstbkugel: Even after seeing your work, I'm not convinced - I still think the code will (or at least, should) diverge over time. But I'll agree to disagree, especially as I seem to be on my own here ;)
14:13:42Tornethe CFLAGS= line in root.make will be ignored if you specify it
14:14:16Tornen1s: i think you are confusing CFLAGS="foo" make, with make CFLAGS="foo"
14:14:21Tornethese behave very differently
14:14:46Tornevariable setting in a makefile overwrites the default values inherited from the environment, but variable setting in the make arguments overrides the makefile
14:15:21Torne(unless the variable is set with the "override" keyword in the makefile)
14:15:23n1sTorne: aha
14:15:51n1sbut i want to add stuff to cflags, not override it...
14:16:24Tornewell, that's trickier, yes
14:16:49Tornebut trusting the value inherited from the environment at all seems like a very bad idea
14:17:01n1smaybe i should just add a new var that gets included in CFLAGS
14:17:02Torneif you want a convenient way to append more flags, stuff $(EXTRA_CFLAGS) on the end or something
14:22:13n1sshould i then set EXTRA_CFLAGS inside root.make to an empty string to not accidentally get junk? (I am not very much at home in makefiles :/)
14:24:51pixelmaAlexP: I would have put the \opt{HAVEREMOTEKEYMAP} inside the first newcommand and around the remote: part - or is this not possible?
14:25:10AlexPprobably is, yes
14:25:17gevaertskugel: pong
14:25:27AlexPI'm just looking at checking whether a macro is defined or not
14:25:42kugelgevaerts: posted the pth patch to FS #11234
14:25:55AlexPAs not all targets have all actions (for both main and remote)
14:25:56gevaertskugel: yes, looking now
14:26:09AlexPAnd if you try to use one that doesn't exist it errors out
14:26:47kugelgevaerts: I pushed to the git repo as well now, but delete the branch before you pull because I messed it up :)
14:27:00pixelmaAlexP: then I'd think the keymap files need fixing. The same happens currently with all other \ActionBlah definitions IIRC
14:27:08*n1s thinks he's got a highscore :)
14:27:45AlexPpixelma: For the main keymaps, yes - but what about remotes? They don't always let you do everything through lack of buttons
14:27:53 Join anewuser [0] (anewuser@unaffiliated/anewuser)
14:28:33gevaertskugel: nothing a proper reset −−hard won't fix :)
14:28:53kugeln1s: according to yes, \o/
14:28:58pixelmaAlexP: hrrm
14:29:18AlexPI can check if a macro is defined fine, but it just makes each definition quite big
14:30:04AlexPone mo, I'll give you an example
14:30:40kugelthis reds are really impressive
14:30:53 Join einhirn [0] (
14:31:11n1sit's an easy fix but svn is weird
14:31:23gevaertsn1s: well done!
14:31:37n1si accidentally deleted the file and svn up doesn't get it back...
14:31:39pixelmaI don't like the fact that this solution needs one for each \Action anyways. I had hoped that it is possible to make latex just take the string after the \Action (e.g. StdNext) and put the pieces together
14:31:45n1sgevaerts: thank you, sir!
14:32:15 Join adnyxo [0] (
14:33:48gevaertskugel: looks ok to me
14:33:56pixelmaAlexP: not sure if that is possible and I see your point about the processing order now
14:34:45CIA-5New commit by nls (r26085): fix red
14:36:18pixelmanumber of affected targets might be interesting in the high score page too as the more ports exist, the sum can be easily higher
14:37:41pamaurymake an average
14:38:00*pamaury thinks of the best way to get a high score ... :)
14:38:08pixelmaor that
14:39:10kugeln1s: add your incredible high score to please :9
14:39:18n1salready donew!
14:39:22gevaertsThe current way is fine. It just means that you can't claim sole credit for a high score, you have to acknowledge the work of porters :)
14:39:26pixelmahooray for descriptive commit messages too :|
14:40:07*n1s thanks all the porters who's invaluable work he built upon
14:40:26kugelnot there yet, there's still a yellow :)
14:40:37n1skugel: that's not mine
14:41:08n1sor shouldn't be, it's about implicit declaration of memset so someone forgot to include the header or something
14:41:13AlexPpixelma: works
14:42:11kugeln1s: I can fix it
14:42:16AlexPpixelma: I think the problem is that different remote actions might be available on different targets
14:42:25AlexPpixelma: So you sort of have to test for them anyway
14:43:11CIA-5New commit by kugel (r26086): memset is in string.h, not memory.h.
14:45:24 Quit Jaykay (Quit: ChatZilla 0.9.86 [Firefox 3.6.3/20100401080539])
14:45:34kugeln1s: but it seems to be exposed by your commit, strange
14:45:43AlexPpixelma: It might be possible to build something with this to do what you want, I'm not sure :)
14:46:49 Nick Llorean1 is now known as Llorean (
14:46:50kugelspeaking of button actions, pixelma: did you get around testing the pla patch on other targets?
14:46:57 Quit Llorean (Changing host)
14:46:57 Join Llorean [0] (~DarkkOne@rockbox/user/Llorean)
14:47:26 Nick fxb__ is now known as fxb (
14:49:24pixelmaI tested on my M5 and if I remember correctly I had a problem in metronome with short Select being "tap" and long being "play", I'd also prefer that in this case "play" would be on the Play button (might apply to other keymaps too)
14:49:38pixelmathe other things were fine
14:55:11pixelmakugel: ^
14:55:43kugeldon't you need to be able to press tap quickly?
14:56:35pixelmaAlexP: I'll think it through but it doesn't look as straight-forward and elegant as I hoped but thanks for looking :)
14:57:08AlexPpixelma: The only advantage of doing it this way rather than manually is that it can't get forgotten
14:57:19pixelmakugel: tapping is fine but play is not and it would be easier if it was on a different button
14:57:33pixelma"play" that is
14:57:50kugeldifferent button is not something you'll do with pla :)
14:58:23pixelmayeah, which is why I think metronome is not a PLA plugin
14:58:33pixelmaor shouldn't be
14:58:53kugelI'm satisfied if it's better than before with my patch
15:02:06pixelmait could be that in the metronome case SVN is "better" on targets which have an extra Play button but I would need to check. But of course metronome is broken completely on other targets in SVN
15:02:38AlexPpixelma: I think the major barrier here is that when running through the text macros are just expanded - they can only be expanded to the thing they are defined as. You can define them differently depending on other options, but once defined they are static (unless you redefine them).
15:03:21S_a_i_n_tIts extremely unlikely anyone here cares about pong, but, is there any likelyhood of FS #5855 ever seeing SVN? I think pong is awesome, but I only have iPod targets and its far too hard (plus risks bending the clickwheel) to play 2 player on it, and without FS #5855 1 player is impossible (or, just *very* boring...)
15:03:54kugelpixelma: what was the plugin for I you a pastebin patch when asking you for testing the last time?
15:03:57S_a_i_n_tI've had it in my builds for ages, pong AI ftw!
15:04:56kugels/for I/for which I gave/
15:06:24pixelmakugel: if I understand you correctly then it was for metronome too and the tap/play thing (about the release action or so). But I think that it was in my M5 test build too, though I start to doubt now
15:06:25 Quit einhirn (Read error: Connection reset by peer)
15:07:53 Join JdGordon1 [0] (
15:07:55kugelpixelma: no, iirc the problem was that trying to play also generates a tap
15:09:01kugelthe action system is extensible on a per plugin basis so certain targets could behave differently
15:10:27 Join funman [0] (~fun@rockbox/developer/funman)
15:10:51 Join einhirn [0] (
15:11:09kugelnew patch up at fs#10387
15:13:30 Quit einhirn (Read error: Connection reset by peer)
15:13:32pixelmaok, I'll rerun tests with this later today
15:13:56funmangevaerts: any idea why we are using signed for number of sectors to transfer? (in fat & storage drivers)
15:14:11kugelpixelma: I changed nothing so that shouldn't be really needed, but I still welcome it anyway of course
15:14:50gevaertsfunman: no. I think that was there before I touched things
15:15:34 Join einhirn [0] (
15:15:55gevaertskugel: could you update the wiki a bit with what you plan to do next?
15:16:14kugelgevaerts: ok, I'll do it this evening
15:16:22kugelor earlier
15:17:24funmanfat driver use longs :o
15:17:44pamauryfunman: I saw in your work on fuzev2 that you commented a bit a function called at usb initialisation which read/writes a register at 0xC8100020, any idea of the purpose of this register ? (it seems ccu related)
15:18:38 Quit n1s (Ping timeout: 258 seconds)
15:18:55 Quit stoffel (Ping timeout: 246 seconds)
15:20:27 Quit einhirn (Ping timeout: 265 seconds)
15:25:48pamauryfunman: the usb init code seems to wait a bit between each register write, is that common to the rest of the code or not ?
15:26:24funmansometimes there are useless waits
15:28:33 Quit togetic (Read error: Operation timed out)
15:29:12 Join n1s [0] (~n1s@rockbox/developer/n1s)
15:31:09funmane200v1/c200v1 can detect the charger both by as3514 and by a GPIO ?
15:31:36funmanpower_input_status() uses the GPIO but powermgmt-ascodec.c check charger presence with as3514 bits
15:32:19funmansame for philips sa9200
15:32:46CIA-5New commit by jdgordon (r26087): fix a minor fms presetlist viewer bug with displaying prev when you are on the first playlist
15:34:59 Join robin0800 [0] (
15:35:05 Join togetic [0] (~togetic@unaffiliated/ibuffy)
15:35:50JdGordon1any suggestions on how and where to do a "what system is controlling audio now?" (playback, radio, recording, plugins)?
15:36:36jhMikeSpointer punning isn't an issue now, right? locally I get no complaints.
15:43:52funmanranma: r25299/powermgmt-ascodec.c: why replacing an ascodec_read() by disable_endofch_irq()?
15:44:04funmanit effectively remains disabled instead of just clearing the interrupt
15:48:32 Join stoffel [0] (
15:51:17***Saving seen data "./dancer.seen"
15:54:49kugeljhMikeS: I hope that was related to my pth patch? :)
15:55:27 Join mobile [0] (
15:55:47 Quit mobile (Client Quit)
15:56:04 Join mobile [0] (
15:56:42 Quit mt (Ping timeout: 240 seconds)
15:56:56 Part mobile
15:57:17 Join mobile [0] (
15:57:46 Quit mobile (Remote host closed the connection)
15:58:07 Join mobile [0] (
15:58:53 Quit mobile (Client Quit)
15:59:09 Join evilnick|ipad [0] (
16:01:42 Join yawny [0] (
16:03:31 Quit elcan (Ping timeout: 260 seconds)
16:04:21 Join mt_ [0] (~mtee@
16:04:37 Nick mt_ is now known as mt (~mtee@
16:05:37jhMikeSkugel: actually mine. it's preferrable to pun like: uint8_t x; fn(void **pp); fn((void **)&x);
16:06:16jhMikeSshould be: uint8_t *x;
16:06:40jhMikeSI needed to hide it before in mpegplayer
16:06:50n1sjhMikeS: there is a switch for more aliasing warnings that still warns about a few cases
16:07:15n1si have tried googling for the aliasing rules with regard to void pointers but have not found an answer...
16:08:08jhMikeSI"m not getting any complaints casting to void **. I think I did before.
16:10:24kugelI think you can do whatever you want with char pointers
16:10:37kugelincluding cast back and forth to any other pointers
16:11:33JdGordon1linuxstb: has the idea of making the sim buttons have the same combo limitations as real targets actually ever come up?
16:13:23funmanjhMikeS: /* Clear out interrupts (important!) */
16:13:31funmando you remember the story behind this comment ?
16:13:56linuxstbJdGordon1: Yes, many times.
16:14:32JdGordon1I wonder why its never been done? should be relativly simple and useful
16:14:34linuxstbNormally whenever someone commits code with button mappings that don't work on a real device...
16:14:59gevaertsJdGordon1: with the sim drivers not in the target tree it wasn't really straightforward ;)
16:15:11funmansince e200v1/c200v1 do not use interrupts it sounds weird
16:16:50funmanas i read it you can only check for end of charging once : if the bit is set, the next time you try to read it will be unset
16:19:09n1skugel: IIUC you may cast and pointer to a char and dereference it but you may not cast a char to any other type and dereference it
16:19:52kugeln1s: IIRC the aliasing rules don't apply to (unsigned) char*, which may or may not disagree with what you just said :)
16:20:35n1skugel: IIUC they do, you may cast stuff TO char* but not from
16:20:52jhMikeSfunman: probably, and it's probably important too :)
16:21:23funmani think it's harmful
16:21:25n1sthe problem with the whole strict aliasing is that there seems to be noone that understands it fully
16:21:28jhMikeSfunman: if you don't, it won't see end of charge
16:21:47funmanwhy that?
16:22:22jhMikeSreading it clears it and I think it's edge triggered, but if you need any of those other ints elsewhere, that read will clear them
16:22:40*JdGordon1 must be really naive re sim Vs RaaA... the only big difference I tihnk of is how threads are done
16:22:59JdGordon1or at least sim Vs SDL RaaA
16:23:04funmanjhMikeS: hm but if you clear it, then you won't see the end of charge ?
16:23:08jhMikeSfunman: ask the designers, but it won't get the signal properly and the charger won't stop
16:23:19 Quit stoffel (Ping timeout: 246 seconds)
16:23:55jhMikeSfunman: actually, I think you'll see a pending end of charge at the start, a false one (it's coming back to me)
16:24:54funmani moved enabling of interrupt after the HZ/10 delay
16:25:03funmandoes that remove the false one?
16:25:10jhMikeSI don't think it did
16:25:42jhMikeSI went through all sorts of different checks myself, I know that much
16:26:40funmanwas it consistent or random?
16:28:50funmanseems to work on as3525(v1) but the code use interrupts
16:28:59jhMikeSconsistent. I reckoned the low current (zero current) managed to trigger it
16:29:12funmanlow current?
16:29:26jhMikeSfrom it being off
16:30:10jhMikeSif you get end of charge that early, the batt would already be too high to trigger charging anyway
16:30:31jhMikeSthat is, once the regulator is enabled
16:30:39funmanoh ok
16:30:43ranmafunman: The thing is that the interrupt enables are write-only, you can't read the 'enabled or not' back.
16:31:02funmanranma: well i got it but reading the register will clear the interrupts
16:31:07ranmaOn read you get wether or not an interrupt is pending for that source and the interrupt request is also acked/cleared.
16:31:40ranmaYeah, but only the one will be acked. You don't have to write enable again, it's still set.
16:31:43funmansome files (including powermgmt-ascodec.c) read IRQ_ENRD0 directly so that doesn't go well with interrupts
16:31:44n1sJdGordon1: I *hope* the ui will be adapted to work nicely with the hos OS, i would want d&d etc if i used it on a desktop system
16:31:59funmanranma: what were you answering to?
16:32:23jhMikeSfunman: then it should be signaled through a dispatch instead, not read directly except by one reader
16:32:24gevaertslinuxstb: would it help if the target-tree move was described as "Provide a cleaner base to start splitting simulator and RaaA" instead of "Prepare for RaaA"?
16:32:25ranma16:23 < funman> jhMikeS: hm but if you clear it, then you won't see the end of charge ?
16:32:34 Join CGL [0] (~CGL@
16:32:46funmanjhMikeS: yes that's what i do
16:32:47JdGordon1n1s: well, at that point it starts sounding more like you want a new frontend to the playback engine more than RaaA
16:32:58gevaertsI just think that merging common code again later would be a nightmare
16:33:05JdGordon1I dont know what d&d is?
16:33:09jhMikeSfunman: I suppose this file needs to just have a callback set instead that sets a var
16:33:13funmanranma: if you ack it just after initializing the interrupt, then next times you read it it will be already acked?
16:33:42n1sJdGordon: yes, that is probably what i want d&d == drag and drop
16:33:45kugelmt: he posted to the -dev ml
16:33:50ranmaSure. I just misunderstood your sentence to mean you thought, you'd not get an end of charging irq then anymore
16:34:00funman <- this is what i have so far
16:34:13funmanwell that's what i mean
16:34:14n1sI don't see any reason why RaaA needs to look or behave anything like rockbox on a target
16:34:50mtkugel: I didn't receive the message, and saw Dave Hooper copying it to the -dev ml ?
16:34:59*kugel doesn't understand why linuxstb thinks that "sdl in the target tree" == "sim and app are the same"
16:35:00ranmaWell, if internally to the codec a new 'end of charging' event is triggerted, it will set the interrupt status again to 1 and because the 'enable' is still 1 it will trigger the interrupt.
16:35:32funman(time go this way ->) enable interrupt , sleep, interrupt is triggered, sleep ends, read blindly interrupt register and acknowledge the interrupt, read enrd0 -> bit is 0 ?
16:35:37kugelmt: his initial message is posted to the -dev ml here
16:35:49funmanok, i suppose it just happens once
16:35:52ranmaUnless you also write a 0 back to the enable bit, then of course the interrupt source is masked.
16:36:50ranmafunman: Ok, yeah. In that timeline you won't figure out the source for that one interrupt.
16:37:28funmanjhMikeS: i'll try with discharged fuze v1 & v2 to see if that happens
16:37:29linuxstbkugel: Because it's "sdl in the target tree", and not "sim in the target tree". I simply think that RaaA should be developed without being distracted (or influenced) by the sim. Improving the sim is an independent task.
16:37:32mtkugel: hmm, you're right. I wonder why I didn't receive it.
16:38:00kugelmt: looked at your spam folder?
16:38:28ranmafunman: Anyway, if somewhere else is blindly reading IRQ_ENRDx because it's doing a RMW cycle to set a bit, then you'd should probably add a wrapper like ascodec_(enable|disable)_endofch_irq()
16:38:36JdGordon1any objections to splitting apps/recorder/radio.c into lots of seperate files in apps/radio/ ?
16:38:54mtkugel: Yep, it's there .. and even cc'd to me
16:39:14gevaertslinuxstb: I think kugel's current work isn't really RaaA yet, more like "getting the code in a good shape to make it easy to work on OS-based rockbox-derived things"
16:39:24funmanranma: see the patch i just posted ;)
16:39:40gevaertsAnd that includes the sim, RaaA, and possibly even checkwps and the database
16:39:43kugelgevaerts: correct it's the groundwork thing we talked about
16:40:07funmanthat's what I did, i just saw this strange thing in powermgmt-ascodec.c
16:40:24funmanand fwiw it could help me understand what happens on as3525v2 when i try to use interrupts
16:40:26ranmaAh, I see
16:41:04CIA-5New commit by jethead71 (r26088): MPEGPlayer: Add a second layer of caching to help speed up byte-wise scanning and seeking a bit.
16:41:57 Nick CGL is now known as [CGL] (~CGL@
16:41:59ranmaBTW, you could just use ascodec_enrd0_shadow instead of adding two new bools
16:42:03funmando you remember why you skipped those bits from the interrupt handler and read them directl?
16:42:23JdGordon1gevaerts: kugel: there is a legitimate concern that eventually low level stuff (thread? sound?) will make assumptions for SDL, and that because the sim uses SDL the sim will get those assumptions which arnt nessecarily correct for the sim only
16:42:32jhMikeSfunman: I know it happened on as3514 on e200v1. if it's different, I wouldn't change it there.
16:42:56gevaertsJdGordon1: sorry, I don't understand what you're saying
16:43:06ranmaI think I wanted to go over the pwrmgmt code and see if theres other stuff that could use the interrupts, but I didn't get around to do it/forgot ^^;
16:43:08kugelJdGordon1: that's right, but I already ported a different thread and sound library so I know where the assumptions are and how to clear them up
16:43:09 Join detaos [0] (
16:43:19funmanjhMikeS: well this is supposedly the same, i'll make some tests and i'll ask you to test a patch before committing anything. if it's specific to e200v1 we can put an ifdef
16:43:40JdGordon1HAVE_SDL_SOUND is a good example.. that might one day be completly replaced later which makes perfect sense for RaaA, but would be a terrible mistake for sim
16:44:09 Quit detaos (Remote host closed the connection)
16:44:11JdGordon1kugel: and in 6 months time? a year? 2 years... just because *you* know them now doesnt mean you will be the one maiting it later
16:44:19 Join bluebrother [0] (~dom@rockbox/developer/bluebrother)
16:44:33funmannow i only tested if end of charge would trigger starting from 97 or 98% charged
16:44:37 Join detaos [0] (
16:45:24kugelJdGordon: in 6 month? That's out of question, I'm expected to be done with that by august
16:45:26funmanjhMikeS: is that really bad if end of charge doesn't happen? (because it doesn't in current svn)
16:45:53JdGordon1kugel: you are totally missing the point
16:45:54jhMikeSfunman: yes, then the charger runs and that's harmful to the battery
16:45:57 Join stoffel [0] (
16:46:10gevaertsJdGordon1: I think kugel answered that in his email. "There's no different way of doing it differently *within* SDL happening"
16:46:40jhMikeSwell, runs continuously anyway. it *must* stop
16:46:53funmanranma: i think r25299 disabled the irq for endofch
16:46:58gevaertsJdGordon1: Give us a real example of what you'd change in an SDL driver where your point would be valid :)
16:47:56JdGordon1I'm not saying there is a better way, or that the concern will actually happen.. I'm just trying to understand linuxstb's concern(?)
16:48:14 Quit bluebroth3r (Ping timeout: 276 seconds)
16:48:30JdGordon1but I guarentee that someone is going to replace #ifdef SIMULATOR with *_SDL_* something which will cause sim problems
16:48:37 Join arbingordon [0] (~w@unaffiliated/arbingordon)
16:48:41jhMikeScan you just turn off that bloody pun warning on colfire? wtf
16:49:21JdGordon1of course, there should be no #ifdef SIMULATOR's in apps/ anyway
16:49:35gevaertsexactly. As long as they're there, we have problems anyway
16:50:43funmanthere is still some target code in apps/
16:51:10ranmafunman: But AMS doesn't use an irq for that and PP should be unaffected by the r25299 IIRC
16:51:31*JdGordon1 thinks RaaA is silly and someone should work on stripping the playback engine out completly instead!
16:51:35kugelJdGordon1: I'll be investigating all the #ifdef SIMULATOR in a later step, hopefully resolving any (even possible) related problems
16:51:36jhMikeSkugel: what option does turn off that warning?
16:51:39funmanranma: powermgmt-ascodec.s is shared with the PP
16:52:11ranmaAh, wait, because of the 'clear out', hmm, missed that.
16:52:21*jhMikeS should just leave it there since it's a bloody stoopid warning
16:52:26funmani think you mistook a read() for a write()
16:52:36kugeljhMikeS: well, -fno-strict-aliasing is the only that comes into my mind, but that was removed just a few days ago
16:52:52n1sjhMikeS: -fno-strict-aliasing disables all the aliasing optimizations and warnings
16:53:02kugeljhMikeS: it could be that the old cf gcc isn't smart enough
16:53:20n1syeah, the old gcc gives some false positives apparently
16:53:22ranmaProbably because of the comment, I don't think of ACKing an irq when it says 'clear out' :)
16:53:41kugelI remember someone saying the abilities of gcc to detect aliasing-related quirks improved slowly over time
16:54:10ranmaThat would need a ascodec_clear_endofch_irq(); instead.
16:54:20funmanperhaps disabling and reenabling it after would do the trick?
16:54:38jhMikeSn1s: optimizations too? it can't just be shut up without consequences to code?
16:54:39ranmaWhich would do the read on PP and on AMS it's a noop
16:55:17jhMikeSmeh, back to ridiculousness then :\
16:55:21n1sjhMikeS: yes there's a -Wno-strict-aliasing
16:55:22ranmaDisabling and reenabling _might_ work, but I wouldn't be on that...
16:55:45funmanbah, i'll test more tomorrow
16:56:02funmanalso, it will be easier to test with usb serial
16:56:05*funman looks at ranma
16:56:08n1s...but it then *might* break on some code that it assumes follows aliasing rules AFAIU
16:56:36ranmafunman: I've been chasing heisenbugs today mostly.
16:56:44funmanwant some help?
16:56:52ranmaThe latest version of the code has regressed somewhat.
16:57:24jhMikeSn1s: you seems to imply it had other consequeces which were not desireable (optimizations)
16:57:55jhMikeSand W not f
16:58:16ranmaThis one works almost, it reliably gets until the point where it tries to read the partition table.
16:58:24ranmaGood chance that usbserial might work with that one.
16:58:40linuxstbgevaerts: I guess my concern is that I see see RaaA and the sim as fundamentally independent, so I don't see sharing code between them as any kind of priority - I see sharing code as something that will get in the way. By saying that there is only one way to do things in SDL isn't the point - the point is the details of the abstraction you provide by your SDL drivers may not be the same for the two purposes.
16:58:53ranmaAnd this is my current working version:
16:58:59n1sjhMikeS: i dont' think disabling the optimizations are tahat bad, i've never seen them give any significant speedup
16:59:27gevaertslinuxstb: but the driver API is defined elsewhere anyway
16:59:34 Quit robin0800 (Remote host closed the connection)
16:59:44kugellinuxstb: they are different, but not the sdl parts of it
17:00:34kugelthe sdl stuff should've been in the target tree from the beginning IMO, and that it's there makes RaaA easier for me
17:01:27kugelcopying and leaving the sim aside has no advantage except an upcoming nightmare at either maintaining or merging both later
17:02:03kugelthe nightmare is a disadvantage of course
17:04:24linuxstbBut that should be weighed up against the flexibility of being free from the sim when developing RaaA. But if, as you claim, there will be a loit of code in common between the sim and RaaA, I can't see why that would be a nightmare. It would only be a nightmare if the code diverged.
17:04:48kugeland you think I didn't weight it up?
17:05:25*jhMikeS goes back to brute force concealment (easiest)
17:06:00CIA-5New commit by alle (r26089): Rearrange statements for a small binsize reduction; delete unneeded braces
17:07:11 Quit detaos (Remote host closed the connection)
17:07:47JdGordon1that commit isnt going to work
17:08:09 Join Boldfilter [0] (
17:08:18JdGordon1what does % do with negative numbers?
17:08:34JdGordon1-1%5 ?
17:09:09jhMikeSsqrt(2) mod pi?
17:09:13kugelI think the mathematical definition and what programming languages do differs here
17:09:19 Join detaos [0] (
17:09:39*amiconn is clearly with linuxstb regarding sim code and target tree
17:09:48kugelI think the binsize reduction comes from making the while() an if() not from the rearangement, which in fact changed the logic too
17:11:23kugelthere isn't even a binsize reduction it seems
17:12:14*gevaerts grumbles about people being opposed to things after a commit which was announced a week in advance on the mailing list
17:12:21linuxstbkugel: To be honest, yes, I don't think you fully weighed it up ;) Your view right from the start was to use target-tree to share code between the sim and RaaA, and you never seemed to have seriously considered not doing that..
17:12:24CIA-5New commit by jethead71 (r26090): MPEGPlayer: reconceal the point puns
17:13:39 Quit JohannesSM64 (Excess Flood)
17:14:03LloreanWasn't our policy last year to have GSOC work happen in branches?
17:14:52 Join JohannesSM64 [0] (
17:15:18*linuxstb isn't complaining about how kugel went about the commit - it's his own fault for not paying enough attention during the last week.
17:15:28kugellinuxstb: my view was influenced by saratoga, but believe it or not I weighted it up. I just came to a different conclusion as you
17:15:30CIA-5New commit by alle (r26091): Improve spacing in comments
17:16:49gevaertsLlorean: for some projects that's doable, but with the RaaA project, at least for the current phase, if you don't commit it quickly to trunk, merging will be horrible
17:17:29CIA-5New commit by jethead71 (r26092): How punny. Just one more whack-a-mole to get.
17:17:54kugelthe german wikipedia article mentions 2 mod variants, the mathematical and the symmetrical one. C appears to use the symmetrical one (-a mod b == -(a mod b))
17:18:13kugelI also think fml's commit doesn't work
17:18:52kugeljust for the fact that preset_count is added *after* taking the mod, and not before, but well
17:19:43 Quit yosafbridge (Quit: Coyote finally caught me)
17:20:22JdGordon1if -1%5 == -1 then it should work
17:21:01 Join yosafbridge [0] (
17:21:06JdGordon1as long as preset is never less than -present_count
17:21:13 Join halmi [0] (
17:22:33kugelJdGordon: what is 1%5?
17:22:59 Quit evilnick|ipad (Remote host closed the connection)
17:23:11kugelaccording to that wiki article C does -1%5 == (1%5)*-1
17:23:30*gevaerts wonders if we disagree about the end result for RaaA/sim, or just the way to get there
17:23:33funmanranma: which revision is your current svn checkout ?
17:24:16Lloreangevaerts: I personally am not too concerned about the end result for RaaA, as long as it doesn't reduce the ability of the sims to be used for what they currently are supposed to do.
17:24:23pixelmagevaerts: from what I remember about its start then it wasn't clear
17:24:41gevaertspixelma: I mean specifically the target tree discussion here
17:25:47 Quit halmi (Read error: Connection reset by peer)
17:27:08 Join liar [0] (
17:27:31*kugel thought 1%5 = 5 or some reason
17:28:16Lloreankugel: You will never, ever get the number on the right side of a % out of modulus
17:28:25LloreanMaybe if 0s are involved
17:28:27LloreanI really don't know there
17:28:49JdGordon1%0 will give you a nice div0 exception
17:29:16funmani like the % pi of jhMikeS ;)
17:30:37kugelgoogle does "(-1) mod 5 = 4"
17:30:49funmanranma: yesterday it worked (linux reacted) on clipv1 but not today, i assume usb only works on odd days
17:32:45Lloreankugel: That's because, I assume, mathematically a remainder can only ever be positive.
17:33:12*kugel thinks r26089 doesn't help anything except creating confusion
17:37:08ranmafunman: 25990
17:37:09 Quit n1s (Ping timeout: 240 seconds)
17:40:04linuxstbgevaerts: Using terms like "RaaA/sim" is the main problem I have - I don't want to see the two things being lumped together. But as long as they are thought of independently, and there is the possibility to do things differently in one than the other, then I'm basically happy.
17:42:28 Quit funman (Quit: free(random());)
17:42:59 Quit JohannesSM64 (Quit: WeeChat 0.3.3-dev)
17:44:11 Join JohannesSM64 [0] (
17:45:26gevaertslinuxstb: ok, I think we can promise that
17:45:33*gevaerts looks at kugel
17:45:53kugelYep, I think so too
17:50:21 Join GeekShadow [0] (~Antoine@reactos/tester/GeekShadow)
17:51:21***Saving seen data "./dancer.seen"
17:54:42 Quit anewuser (Quit: for SELL 2 by the price of 1 now!)
18:02:03 Quit JdGordon (Quit: Leaving.)
18:05:08*kugel needs soon to think about #defines to use for RaaA
18:05:58kugel#define CONFIG_CPU [CPU_]SDL seems weird, but that would be inline with the current ports
18:06:51gevaertskugel: isn't CONFIG_CPU used to switch optimisations?
18:07:05kugelyes and no
18:07:30kugelmostly for optimizations as far as iram allocation is concerned
18:07:54kugelfor asm, the #defines CPU_ARM, CPU_COLDFIRE, etc are used
18:08:09 Join Jaykay [0] (
18:09:38kugelCONFIG_CPU is #undef'd in the sim so that work out as well
18:10:36kugelbut I could also imagine a higher CONFIG_PLATFORM (could be e.g. PLATFORM_NATIVE or PLATFORM_HOSTED)
18:10:51gevaertsyes, that seems better
18:11:16*gevaerts wonders about sims for RaaA targets
18:11:23kugela mix of both could also work
18:13:40 Quit S_a_i_n_t ()
18:16:47 Nick fxb is now known as fxb__ (
18:18:56 Join S_a_i_n_t [0] (S_a_i_n_t@
18:20:45Lloreangevaerts: That could be complicated, since you'd then want to simulate the native widgets of the device if they're used.
18:21:01gevaertsLlorean: don't shatter my dreams!
18:21:03LloreanDon't many RaaA targets have emulators or other development tools anyway, so a sim may not be necessary?
18:21:12Llorean*potential RaaA targets, rather
18:21:28gevaertsthey do, but they
18:21:33gevaerts're not always easy to set up
18:22:07LloreanWell, sims that had no additional dependencies beyond our current ones would be neat of course.
18:22:13 Quit S_a_i_n_t (Client Quit)
18:24:26 Quit bieber (Ping timeout: 260 seconds)
18:24:51 Join bieber [0] (
18:25:26*kugel wonders about the "native widgets" talk
18:25:30 Join S_a_i_n_t [0] (S_a_i_n_t@
18:26:01Lloreankugel: You'll note I said "if they're used"
18:26:33kugelyep, it hasn't been decided whether we want that or not
18:26:39kugelbut I would not want them
18:26:40LloreanAs an app, there are many situations where keeping the Rockbox UI metaphor is extremely restrictive.
18:27:02kugelthat depends on the device
18:27:05amiconnAnyone with a PP5020 target around that uses the colour lcd bridge?
18:27:40Lloreankugel: Yes, but it should generally be assumed that it could end up on such a device - while it doesn't need to use native widgets now, it should strive to not rule out the possibility to do so later.
18:27:49kugelfor instance on a smartphone I can imagine that our own gui could work extremly well (i.e. as well as it would if rockbox is run natively)
18:28:01Lloreanamiconn: "the color" meaning "in color" or "the same one as the iPod color" or?
18:28:23*pixelma somehow doubts many people would know if their player is a PP5020 target with the lcd colour bridge
18:28:28amiconnThe same bridge (in the PP) as the ipod color
18:28:33kugelbut on a PC it could be akward indeed
18:28:35Lloreanamiconn: Which targets is that, then?
18:29:03amiconnExamples: iPod Color, Nano G1, H10 (big and small), e200v1
18:29:08Lloreankugel: On some smartphones it might be very much preferred to use, for example, a native file browsing interface if available, etc. Rockbox's current one is pretty clunky, even with text, compared to what such users may be used to.
18:29:13Lloreanamiconn: I have a nano G1
18:29:27kugelbut do we agree that the native widget thing is out of scope for this gsoc project? I don't want to rule it out completely but I don't really want to work on it during the summer (especially since I have personal interest for it)
18:29:32Lloreankugel: But I'm just saying, we should try to assume we might want to in the future *if reasonably possible*
18:29:39amiconnI'd need an iospace dump from 0x70008000 to 0x7000ffff (i.e. 32KByte)
18:29:51amiconnI could do it myself if I had an appropraite target with me
18:29:58kugel*no* personal interest
18:30:09Lloreankugel: In my opinion, you shouldn't need to work on native widgets unless everything else works. But you should try to make sure that, if possible, replacing one of ours with a native one won't be harder than it needs to be (if it doesn't make your work significantly harder)
18:30:31 Join einhirn [0] (
18:30:32amiconnOh, hmm, nano g1 is not suitable
18:30:32kugelthat's fine with me
18:30:40Lloreanamiconn: Unless there's a way to do that without me compiling my own build, I don't think I can help. No build environment atm.
18:30:46LloreanAh, never mind then.
18:30:48amiconnIt doesn't use PP5020, but PP5022, and that has different io aliasing
18:31:12amiconnSo essentially just ipod color and H10 (big/ small)
18:31:35Lloreankugel: I think that's good then. I doubt we're going to decide if we want native widgets or not any time soon anyway, and some targets Rockbox widgets are simply going to be better anyway, so support for *our* widgets will always be necessary I think.
18:31:35amiconnIt looks like the elio might use an alias address - I'd do it myself, but I don't have a suitable target with me
18:34:47gevaertsamiconn: I can do it, but my H10 needs to charge a bit first apparently
18:37:57 Join mighty_ [0] (~mighty@
18:38:17*amiconn can do it later tonight
18:38:26gevaertsamiconn: I assume I just need to adjust dbg_save_roms() in apps/debug_menu.c a bit?
18:38:47mighty_anyone speak french ?
18:38:51 Nick mighty_ is now known as mighty (~mighty@
18:39:06 Nick mighty is now known as Angelichounet (~mighty@
18:39:50amiconnI did a rudimentary PP5020 emulation for skyeye. So far it only emulates core id, pp version and the usec timer in a proper way
18:40:02amiconngevaerts: Yes, that should work
18:40:37Angelichounetgevaerts, hello
18:40:47amiconnI found a bunch of accesses to 0x7000ad00, which our pp5020.h doesn't know. It might have to do with lcd
18:41:12amiconngevaerts: Hmm, better dump the whole 64KB starting at 0x70000000
18:41:19 Join Strife89 [0] (
18:41:42AngelichounetAnyone know if it is possible to install games on iPod Nano 3nd gen ?
18:42:45S_a_i_n_tyes, no.
18:43:05S_a_i_n_t;) No, sorry. It isn't.
18:43:06pixelmado you mean a 2nd or a 3rd generation Nano? Also installing any games is off-topic here, installing Rockbox is not
18:43:57Angelichounetje parle de Rockbox, mais j'arrive pas l'installer pour avoir des jeux sur mon ipod
18:44:03*gevaerts hopes that his H10 will be charged enough to actually boot rockbox soon
18:44:27 Quit GeekShadow (Read error: Connection reset by peer)
18:44:41AngelichounetI talk about Rockbox, but I can not install it for games on my ipod
18:44:54 Join GeekShadow [0] (~Antoine@reactos/tester/GeekShadow)
18:45:47pixelmais your Nano a 3rd generation one? The "short/fat" one?
18:46:41AngelichounetNano a 3rd generation
18:46:45Angelichounetand fat
18:47:04pixelmathen no, it's currently not possible to install Rockbox on it
18:47:27Angelichounetmouarf xD
18:47:43Angelichounetthank you anyway
18:49:54 Quit Angelichounet (Quit: Quitte)
18:52:52 Join anewuser [0] (anewuser@unaffiliated/anewuser)
18:54:21 Quit BeFalou (Ping timeout: 240 seconds)
18:56:00 Join n1s [0] (~n1s@rockbox/developer/n1s)
18:56:54 Join BeFalou [0] (~mamutoi@unaffiliated/befalou)
18:57:15 Quit kugel (Remote host closed the connection)
18:57:31 Join panni_ [0] (
19:04:29 Quit BeFalou (Remote host closed the connection)
19:07:37 Join BeFalou [0] (~mamutoi@unaffiliated/befalou)
19:12:12 Quit JohannesSM64 (Quit: WeeChat 0.3.3-dev)
19:12:32 Join JohannesSM64 [0] (
19:13:43 Quit stoffel (Remote host closed the connection)
19:13:46pixelmawhile the sim compiled for me under cygwin it isn't usable, background etc. appears and goes to the menu but it is unresponisve and I can't do anything. Does anyone else see that?
19:15:29pixelmarevision 26092, fm enabled M5 sim
19:16:58*pixelma could also try a crosscompiled sim
19:21:04gevaertsI'm trying a mingw sim here
19:23:49gevaertsamiconn: gevaerts/iospace.bin">
19:27:27amiconnHmm, that stuff at 0x7000ad00 isn't lcd... but there is something
19:28:09*amiconn spots a possible optimisation when dealing with the colour bridge though
19:28:48gevaertspixelma: a mingw-built sim in wine also seems to ignore all keypresses
19:29:00*amiconn wonders whether he could dump the iospace on the elio, even if lcd isn't working
19:29:25pixelmagevaerts: did you try a linux "native" sim also?
19:29:33gevaertspixelma: yes. That one works
19:30:36gevaertsThe good news is that one doesn't need cygwin to work on this :)
19:30:40pixelmaat least you can reproduce...
19:31:54amiconnHmm, ipod (and hence rockbox) already uses an alias for the colour bridge.... that one is avaliable at base 0x70008Y00 with Y = {0, 2, 4, 6, 8 , a, c, e}
19:32:42 Join saratoga_ [0] (~463f90ed@gateway/web/freenode/x-jjduostbjujjnlkh)
19:32:53*amiconn wonders why...
19:33:04saratoga_can an article about rockbox use our logo?
19:33:23gevaertspixelma: I suspect "and the implicit application thread will now act as our main thread (previously a separate one was created for this in thread initialization)" in r26065
19:37:06pixelmasaratoga_: I don't see why not but don't know a lot about licenses
19:37:24gevaertssaratoga_: sounds like basic fair use to me
19:37:42saratoga_wiki says they can, so sounds good
19:44:11 Join Xqtftqx [0] (
19:44:55Xqtftqxare there any present devices i can buy in a store that rockbox runs on
19:46:40gevaertsThat depends on what you mean by "runs"
19:46:59LloreanAnd possibly "buy in a store"
19:47:14gevaertspixelma: I've submitted FS #11280 for this. This is really something for kugel to look at
19:47:19 Quit anewuser (Quit: for SELL 2 by the price of 1 now!)
19:47:23amiconnHmm, PP502x might have two colour lcd bridges...
19:47:27XqtftqxIs there a device being manufactured that rockbox is usable to play audio on
19:47:37LloreanXqtftqx: Basically, there's a list of supported and unstable targets on the front page.
19:47:40LloreanLook for them in stores near you.
19:47:51pixelmagevaerts: thanks, just wanted to ask if you were looking into it
19:47:56gevaertsXqtftqx: I suspect some of the newer sansas
19:48:55gevaertspixelma: I looked a bit, but this is going to be related to threading changes, and I really don't know those well enough to look for this efficiently
19:51:22***Saving seen data "./dancer.seen"
19:52:14mtsaratoga_: Some math help. :) .. 1024 samples/channel * 2channels = 2048 samples .. for 2 bytes/sample, the total size should be 4KB .. I'm getting the total size from the wma pro decoder to be 8KB .. Is there something I'm missing ??
19:52:22 Join mikroflops [0] (
19:52:58saratoga_mt: does it returin 32 bit samples? many codecs do
19:53:47mtit returns floats yes .. but then the size is 16K .. I divided that by 2 because I'm converting them to 16-bit signed ints, so I'm getting 8K
19:54:37saratoga_i'll look at svn
19:54:43 Join esperegu [0] (~quassel@
19:55:50 Quit mikroflops_ (Ping timeout: 248 seconds)
19:59:22saratoga_WMA standard actually returns 2048 sample per channel per frame for 44.1khz audio
19:59:40saratoga_WMA Pro seems to use the same transforms, so maybe its 2048 per frame too?
20:00:15 Join efyx [0] (
20:00:16saratoga_actually looking at the code it seems like it can also return 4096 samples, but probably only at higher sampling rates
20:00:45mtI'll check .. although I have audio working already in the sim, but I wanted to know why I needed to divide the output data_size by 2..
20:01:16mtwell actually by 4 ..
20:01:22mt2 for 2 bytes/sample
20:01:38mtand then the other still-unknown 2
20:02:12saratoga_where do you divide it by 2/4?
20:04:29mtI'll submit a patch very soon ..
20:05:08mtbut for now .. I divide by 2 first when I convert the output format to int16_t in wmaprodec.c ..
20:05:09 Quit Jaykay (Ping timeout: 240 seconds)
20:05:17mtthen I divide by 4 in wmapro.c
20:05:21 Quit n1s (Ping timeout: 240 seconds)
20:05:51mtit's actually 2048 samples
20:06:44saratoga_if possible could you make wmapro.c return 32 bit ints?
20:07:50mtBut the sim uses s16bit pcm ?
20:08:04saratoga_the DSP code though runs in full 32 bit precision
20:08:26saratoga_so it'll convert everything back to 32 bit when it does EQ, replaygain, etc
20:09:37mthmm .. something isn't right
20:09:59mtthe decoder is actually only filling a total of 2k samples
20:10:24saratoga_so it is 1024 per channel?
20:11:30CIA-5New commit by funman (r26093): Clipv2/+ : ascodec register 0x25 is not related to backlight
20:11:35mtdump_context says it's 2048/frame
20:12:08mtso it should be 1k/channel
20:14:27 Join stoffel [0] (
20:15:09saratoga_mt: int total_samples = s->samples_per_frame * s->num_channels;
20:15:20saratoga_samples_per_frame is per channel i think
20:17:06mtStill it's only filling 2k locations .. if I pass pcmbuf_insert a count of 4k, I'm giving it just 2K of zeros.
20:17:55saratoga_can you check how many samples wmapro_window is generating?
20:22:50mtit seems to be 256/channel
20:24:00mtsaratoga_: ^
20:25:35saratoga_mt: in wma there can be many blocks per frame
20:25:55saratoga_its not like cook where they're all the same size, instead you just have to add up to samples_per_frame
20:26:10saratoga_so 2048 is valid, as is 256+512+256+1024
20:26:21saratoga_(sorry if you already knew that)
20:29:22mtyou mean 2K per channel ?
20:30:59saratoga_yeah i *think*
20:31:08saratoga_wma stadard is 2k per channel at 44.1k anyway
20:31:21saratoga_as is vorbis and aac
20:31:43mtEverything says it has to be 4K .. but what I'm hearing disagrees :)
20:32:01saratoga_can you print all the sizes used by wmapro_window in one frame?
20:32:07mtsaratoga_: I'll submit a patch so it would be easier to discuss ok ?
20:32:12saratoga_sounds good
20:33:01 Join wincent [0] (
20:39:52 Quit GeekShadow (Ping timeout: 268 seconds)
20:42:39 Quit stoffel (Remote host closed the connection)
20:43:37 Quit intrados (Quit: WeeChat 0.3.2)
20:46:12 Quit TheSeven (Read error: Connection reset by peer)
20:46:30 Join TheSeven [0] (~TheSeven@rockbox/developer/TheSeven)
20:54:10saratoga_mt: it looks fine to me, and very similar to wma
20:55:54saratoga_unfortunately i have no way to compile it for another hour or so
20:56:05 Quit esperegu (Ping timeout: 248 seconds)
20:56:12mtOk no problem ..
20:56:21 Quit ssorgatem (Quit: Konversation terminated!)
20:56:33 Join ssorgatem [0] (~ssorgatem@
20:58:36mtsaratoga_: Could that clicking noise be a precision problem ?
20:58:57saratoga_could be overflow, could be dropped samples somewhere
20:59:04saratoga_overflow usually sounds like a burst of static
21:00:16mt dropped samples would just give silence, no ?
21:01:33 Join GeekShadow [0] (~Antoine@reactos/tester/GeekShadow)
21:01:44saratoga_depends how many you drop
21:01:51saratoga_if its just a few its a click or a pop
21:01:54saratoga_outlen /= 4; //XXX: Why 4 ?
21:02:03 Join esperegu [0] (~quassel@
21:02:07saratoga_isn't this just converting from bytes from floats?
21:02:12 Quit esperegu (Remote host closed the connection)
21:02:12saratoga_since 4 bytes per float?
21:03:13 Join krabador [0] (
21:03:25 Quit wincent (Changing host)
21:03:25 Join wincent [0] (~wincent@rockbox/developer/wincent)
21:03:46saratoga_i'm probably missing something stupid, but it seems like the decoder tells you the number of bytes its returned, and it returns floating point samples, so you have to divide by 4 to get the number of samples
21:08:44mtsaratoga_: you're right .. but I'm converting the samples to int16 in wmaprodec.c (decode_packet() ).. so the output number of bytes should be /2 not /4
21:10:59 Quit guysoft22 (Read error: Connection reset by peer)
21:11:14 Join guysoft22 [0] (
21:11:54saratoga_sorry i was looking at the unpatched version . . .
21:13:25mtAh ok no problem.
21:13:38 Join dobson [0] (~rokas@
21:14:11dobsonhey guys, i have quite a big idea, that you may could bring to reality
21:14:40dobsonyou know when you go to recording and you can hear surroundings amplified
21:15:10dobsonso why not to make funcktion to invert the sound, that player would become noise canceling
21:15:34saratoga_its not really possible given how noise cancelation works
21:16:08dobsonwhel noise cancelation works by inverting sound wawe, so it couls ilaminte each other
21:16:28dobsoni think that shouldn
21:16:35dobsonbe hard to do
21:17:17saratoga_yes but it requires you to measure the noise field essentially at the point where you're canceling it, so you need special headphones that are designed to work with it
21:17:32saratoga_you can't just measure sound at the player and expect it to be in phase with sound in your ear
21:17:42dobsonyea you need to sinchronize it
21:18:01 Quit flydutch (Quit: /* empty */)
21:18:01dobsonthats where programing comes in
21:18:15saratoga_its not possible to do this in software
21:18:31dobsonall that noise canceling headphones do invert sound vawe
21:19:10dobsoni'm not talking about playing music at the same time
21:19:15saratoga_heh well if you've discovered a way to do this, then by all means submit a patch
21:19:54dobsoni'm don't know anything about rockbox code :)
21:20:11dobsonbut i will try to just invert sound with amplifier
21:20:16saratoga_i think thats the least of your problems :)
21:20:44dobsonand if that works it should be plausible to invert it in player
21:20:48dobsoni think :)
21:21:17dobsonthing like that would be incredible upgrade in rockbox
21:24:09gevaertsincredible indeed
21:33:45 Join domonoky [0] (~Domonoky@rockbox/developer/domonoky)
21:35:26AlexPI think we would all be interested to see you make it work
21:35:51AlexPI think it is fair to say that we are sceptical too
21:36:41 Quit domonoky (Client Quit)
21:37:02 Quit saratoga_ (Quit: Page closed)
21:37:06 Join stripwax [0] (
21:37:23 Part Boldfilter
21:39:39 Join domonoky [0] (~Domonoky@rockbox/developer/domonoky)
21:41:20pixelmaAlexP: I reworked the start screen description quite a bit according to your suggestion about item and explanation not starting lower case (also put things a bit differently to make the text shorter) -
21:42:29pixelmaoh, and linuxstb once said that he would drop the "will" in all those sentences so I tried that. I overlooked some it seems though
21:47:00AlexPpixelma: I changed a couple of minor things:
21:47:12AlexPpixelma: One thing is the verb form - Show vs Shows
21:47:38AlexPBoth are there - I changed them all to the Shows form, but I can't remember what the consensus was for the rest of the manual
21:47:45pixelmayep, found it too :\
21:48:49pixelmawhat do you think in general - is this version "better"?
21:48:56AlexPYep :)
21:50:35pixelmaok, after checking if there is some guideline regarding shows/show and potentially adapt to it, I'll commit that then. Improvements can still be done later but the current description is a bit coarse IMO
21:51:12AlexPpixelma: The manual was changed to be all one way (after some discussion), so it *should* be easy to just see what is used now :)
21:51:26***Saving seen data "./dancer.seen"
21:51:46 Join lpereira [0] (
21:52:51AlexPyeah, I'll see if I can find a date
21:53:17pixelmaI think I saw something in the guidelines once but I have my doubts that it is used consistently now
21:56:20 Quit bieber (Ping timeout: 268 seconds)
21:56:46 Join bieber [0] (
22:01:11 Join bholu [0] (
22:01:58pixelmaAlexP: hmm, nothing on LatexGuidelines... would be nice if you would find something :)
22:01:58 Quit bholu (Client Quit)
22:02:19AlexPI'll have to search the commits - give me a while :)
22:08:00 Quit dobson (Ping timeout: 276 seconds)
22:08:55AlexPpixelma: - I assume the IRC discussion was the same day. Just before that it was changed to the other form, then this task changed it to this. The commit is linked from the task, there was another commit shortly afterwards (by alle) to change a few missed ones
22:09:36 Join tenfoot [0] (
22:09:50 Join Blue_Dude [0] (~chatzilla@rockbox/developer/Blue-Dude)
22:10:17Blue_Dudekugel: any special reason you yanked out my logf debug lines from the sdl sound code? They were in there for a reason.
22:11:04pixelmaAlexP: thanks
22:11:11Blue_DudeMoving files is one thing. You gotta do what you gotta do. Moving then and then making undocumented changes is something else.
22:14:07Blue_DudeIn addition, putting the debugaudio code within conditional compile statements negates the value of having −−debugaudio on the command line.
22:14:21pixelmaAlexP: although it is for a different kind of description. The thing that got me a bit is the beginning of the first explanation - "Rockbox starts", then all the following sentences looked weird
22:17:02AlexPpixelma: Yes, I agree - I think here removing the "Rockbox" then having everyhting as "Start" would look better.
22:17:21AlexPAlso, "Go to the WPS and resume playback from where it was"
22:17:53pixelmafine, I'll trust your opinion as a native speaker :)
22:18:04AlexPActually, "Start Rockbox in the ..."
22:19:16 Quit Blue_Dude (Quit: ChatZilla 0.9.86 [Firefox 3.6.3/20100401080539])
22:19:16pixelmado I need the "in" in this sentence?
22:19:31pixelmaerr... I mean the second one
22:19:35AlexPMaybe "Set the screen that Rockbox will start in. The default is..."
22:20:02AlexPWhich sentence now? :)
22:20:08pixelmathere is a default?
22:20:41AlexPI don't know, I was just looking at the English
22:21:09AlexPline 14 in the pastebin
22:21:15pixelmaoh, I didn't mean the introduction but the explanation of the first item
22:21:37pixelmathe "previous screen" one
22:21:39AlexPyeah, I think you need the in
22:21:45AlexPthe second in
22:21:58AlexPIf the sentence stays like that anyway - I'll try rearranging
22:22:52AlexP"Start Rockbox in the same screen as when it was shut off" maybe
22:23:31 Join Strife1989 [0] (
22:23:37pixelmameh, I just realised that I point out the default twice in my patch (introduction and "main menu" item)
22:25:37 Join Strife89|Laptop [0] (
22:26:12 Quit Strife89 (Ping timeout: 276 seconds)
22:26:27 Nick Strife89|Laptop is now known as Strife89 (
22:28:39 Quit Strife1989 (Ping timeout: 276 seconds)
22:29:02 Quit Xqtftqx (Ping timeout: 252 seconds)
22:32:32 Quit stripwax (Read error: Connection reset by peer)
22:36:51 Join Boldfilter [0] (
22:37:20pixelmaAlexP: final version (I hope) -
22:40:15AlexPpixelma: A couple of tiny changes, but looks good!
22:44:43 Join n1s [0] (~n1s@rockbox/developer/n1s)
22:47:08 Quit bmbl (Quit: Bye!)
22:48:11pixelmamany thanks :)
22:48:39AlexPno worries :)
22:52:28 Quit dfkt (Ping timeout: 264 seconds)
22:54:37 Quit GeekShadow (Quit: The cake is a lie !)
22:56:00 Join max242 [0] (
22:56:32 Join Farthen_ [0] (
22:56:36 Quit Farthen_ (Client Quit)
22:57:26 Quit petur (Quit: Zzzz)
23:03:55 Quit Schmogel (Ping timeout: 265 seconds)
23:03:56 Join stripwax [0] (
23:04:49max242is there anybody in here who has seen my post on about PictureFlow locking up the Fuze V1?
23:05:01max242i posted this here:
23:05:44CIA-5New commit by pixelma (r26094): Update and complete the start screen option description in the manual. Thanks to Alex Parker for the native speaker help.
23:06:02 Quit JohannesSM64 (Quit: WeeChat 0.3.3-dev)
23:10:12saratogamax242: did you figure out which revision caused the problem?
23:11:13 Join Schmogel [0] (
23:11:25max242yeah, that was 25299
23:12:04max242quite a long way back, and current revisions still have the same issue
23:12:34saratogawhat do i need to do to reproduce it?
23:13:32max242that's a very good question, i don't know yet what causes PictureFlow to lock up the Fuze V1
23:14:21max242i thought it had to do with the size of the microSD card i've got plugged in
23:14:28max242which is 16G
23:14:57max242but apparently, my fuze also locks up every now and then having no microSD card inserted at all
23:16:10max242funman replied to the thread i started a couple of times
23:16:12pixelmadoes pictureflow lock up for you then without the card and just a bit of music on the internal memory?
23:16:24gevaertsmax242: maybe a stupid question, and probably not helpful, but is the filesystem OK?
23:17:18max242gevaerts: yes, the filesystem is ok, i've even reformatted the Fuze to be sure on this one
23:18:30gevaertsmax242: can you put your .rockbox/config.cfg on or similar?
23:19:16max242pixelma: pictureflow locks up randomly during the generation of the .pfraw files, at least that's what i think is wrong
23:20:11pixelmawell you mentioned a possible relation to your microSD card which I wanted to find out more about
23:20:24 Nick fxb__ is now known as fxb (
23:20:41max242gevaerts: i'm new in this thing, and never used before. do i need an account
23:20:57gevaertsno, you just paste some stuff there and you get a link
23:22:01max242does it matter which revision i currently am using? Because i'm on r26093 right now
23:22:05 Join anewuser [0] (anewuser@unaffiliated/anewuser)
23:22:07 Quit liar (Quit: Verlassend)
23:23:37 Quit tenfoot (Quit: CGI:IRC)
23:23:42saratogahow come picture flow doesn't update its album cache automatically
23:24:03saratogait recognizes new albums, but i have to tell it to rebuild the cache before they have album art
23:24:11 Quit anewuser (Client Quit)
23:25:28max242ok, my config.cfg is on pastebin now:
23:25:31saratogawell other then that it seems to work on my fuzev1
23:26:18amiconnlinuxstb: Do you think I should commit the colour lcd bridge un-aliasing?
23:26:21saratogado we really have database autoupdate off by default?
23:26:21gevaertsnothing too special there I'd say
23:26:49 Quit n1s (Ping timeout: 260 seconds)
23:28:11max242saratoga: what do you mean by pictureflow recognizing new albums? I always do rebuild the database, then trigger PictureFlow to generate a new cache
23:28:31*gevaerts doesn't have a fuze so he can't really try to reproduce this
23:28:47saratogai copied new songs to the player, then loaded pictureflow, they show up, but theres no thumbnails for them until i tell it to rebuild
23:29:19saratogait seems like pictureflow should rebuild its album art whenever it detects new songs
23:30:25max242so i assume you do let rockbox update the database automagically?
23:31:48amiconnlinuxstb: I tested it on my Color and my H10 and it works. Someone should dump regs on a colour PP5021/2/4 to verify (that *excludes* c200v1)
23:32:20amiconnOh, and it also excludes the Video, as well as the e200v1
23:33:01max242i also do remember that funman was thinking that there might be issues with the database, and that this maybe is the cause of these pictureflow issues?
23:33:23saratogayes I set it to update automatically
23:33:50max242i even thought at one point that it has to do with timing issues towards the storage
23:33:53saratogaamiconn: how close were you to having skyeye working?
23:34:30amiconnsaratoga: Depends. skyeye itself works, but depending on i/o emulation the boot rom may take wrong code paths
23:35:06saratogaso probably not very close to booting rockbox
23:35:19 Quit guysoft22 (Read error: Connection reset by peer)
23:35:22amiconnCertainly not, and that's not my goal at all
23:35:36 Join guysoft22 [0] (
23:35:36amiconnI hope to find the elio's lcd init
23:36:18 Quit merbanan (Ping timeout: 265 seconds)
23:36:19amiconnOr rather, the lcd driver.
23:36:45max242also something i should mention: i also do have some issues every now and then when generating the database
23:37:20 Join M3DLG [0] (
23:38:19max242i think it is skipping some steps when trying to commit the database
23:38:50max242but it's been a while when i had these issues, and i don't remember exactly
23:40:37max242also building the database was sometimes very slow
23:40:51stripwaxsaratoga - right, pictureflows 'database' of album art is different to the rockbox tag database; and in fact pictureflow seems to index album art by an ID that becomes invalid if you add more albums and rebuild the database
23:41:39saratogathat makes sense, but its odd that it will see new entries and not try to load the art for them
23:42:10stripwaxe.g. pictureflow has image A for album 1, image B for album 2, image C for album 3 .. and then you add a new album and rebuild the database, and now image B shows up for album 4 which happens to be alphabetically between album 1 and album 2 (or something like that)
23:42:13pixelmaare you sure, about it using a different album data base?
23:42:24stripwaxpixelma - see above
23:42:34stripwax*not* a different album database
23:43:06stripwaxdifferent 'database' of tag order. it seems to assign album art by an indexing scheme that is immediately invalid as soon as you rebuild the database
23:43:19stripwax^rebuild the rockbox tag database
23:43:23pixelmastripwax: I meant the album "list". I think it's taken from the database, the indexed pictures are not
23:43:32 Quit lpereira (Quit: Leaving.)
23:43:55stripwaxyes. whatever it is that tries to do the "left outer join" of those two lists, doesn't do a good job if the db updates
23:44:06saratogaah so the album art i saw was probbaly for the wrong album?
23:44:20amiconnstripwax: pictureflow uses the album index from the database, and that becomes wrong whenever albums are added or deleted
23:44:26stripwaxthat'd do it, yep
23:44:38max242my database has close to 6000 entries, could that be a potential issue?
23:44:43amiconn(unless this addition/ deletion happens at the very end)
23:44:44saratogapictureflow seems pretty close to being useful, maybe that guy who added the hotkey stuff for it will be interested in improving it
23:45:02max242wasn't that kugel?
23:45:08amiconnThe album list is sorted alphabetically, in order to allow chunked browsing
23:45:29amiconnThis problem exists since pictureflow has been introduced
23:45:44stripwax(i suspect I know the answer alread, but -) wouldn't it be better for pictureflow to record the album art id into the database itself?
23:46:08amiconnThe database is a core thing
23:47:01amiconnAll it needs to do is to use a better index instead of the database's internal album id
23:47:38max242when pictureflow is preparing it's album artwork, there's a green progress bar
23:47:48stripwaxthat's right
23:47:55max242wouldn't it be usefull to show where it is at already?
23:48:02stripwax"at" ?
23:48:21max242i mean, which directory or file it is processing
23:48:36stripwaxI don't know - would it? how would that help?
23:48:42max242i understood pictureflow is generating pfraw files
23:49:19saratogamax242: does your pictureflow still crash if you use the same files, but from the simulator?
23:49:58max242i've never really used the simulator before, but that definitely seems a good idea
23:50:44saratogatheme site needs the fuzev2 added to it
23:50:45max242however, i have a iPoD video around, and there i never got these pictureflow issues
23:50:59saratogadoes it have exactly the same folder structure as your fuze?
23:51:28***Saving seen data "./dancer.seen"
23:51:32max242no, it has different content, and i don't use it that much because of the battery draining too fast
23:52:13amiconnLlorean: ping
23:53:46max242ok, i'll try to use the simulator on the same batch of files later during the week, i'm off to bed now
23:57:08 Quit Xerion (Ping timeout: 240 seconds)

Previous day | Next day