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 2010-02-15

00:00:21pamaurygevaerts: On some targets, perhaps ram doesn't include bin
00:00:35gevaertspamaury: in these calculations, it always does
00:00:54*pamaury shuts up
00:01:44gevaertsthat's inaccurate for rombox, but it's true for all other cases
00:01:51mtBuschel: I think we should disregard joint-stereo for the time being. If someone finds out a sample that's playing incorrectly they might tell us.
00:01:56pixelmaJdGordon: the splash shows
00:03:01 Quit ender` (Quit: Anyone who thinks people lack originality should watch them folding roadmaps.)
00:03:31pixelmaand I can listen to the radio. If I then want to stop the radio I get a "Panic - Stack underflow... viewportmanager", not sure if it means something with that change
00:03:38amiconngevaerts: 'ram' isn't necessarily max
00:04:07amiconnIf some code change increases binsize but gets rid of a static buffer, 'ram' could even be negative while bin is positive
00:04:57*gevaerts spots confusion
00:05:34gevaertsI meant the total size, not the delta
00:05:44mthmm .. I guess the 2-bit shift for higher accuracy could be folded into the imlt step and done for free, on arm at least. (don't know about cf)
00:06:07pamaurygevaerts: we are interesting in the delta right there
00:06:13gevaertsyes, I know
00:06:16 Quit antil33t ()
00:06:26*gevaerts tries to find something to blame
00:07:44gevaertsI think we either should show the ram-delta, or both
00:08:03 Quit fml (Quit: CGI:IRC)
00:08:15Buschelmt: JS is at least 160 lines of untested code then
00:10:13mtI don't see a problem with that for now at least, until we have samples to test on.
00:10:57Buschelmt: you are right
00:12:24Unhelpfulamiconn: maybe it would make more sense to support either endianness in the DB tool, and only native on lowmem targets?
00:12:58pixelmaJdGordon: fun fun, the blank screen issue magically went away (after reverting the changes) but the recording info doesn't work and the peakmeter updates too slowly. If I understand correctly, then the prerecording time should show up instead if it is set and if I start the recording, the recording time should be displayed there - but the peakmeter stays there no matter what
00:12:59amiconnThat would require the tool to know the target's endianness
00:13:07gevaertshm, how about an endianness conversion plugin?
00:13:13amiconnHmm, is database creation part of rbutil these days?
00:13:27domonoky-r +e
00:13:42Unhelpfulamiconn: it wouldn't be that difficult to have a list of targets and their bytesex in the tool, would it.?
00:14:15domonokyif we had it in rbutil, target depended endianess wouldnt be a problem.
00:14:15pamauryGetting rid of endianess support in tagcache would probably simplify the code
00:14:20amiconnThat part is easy. SH1 and coldfire are big endian, arm and mips are little endian
00:14:40amiconnBut the tool needs to know what target it's creating the db for
00:14:56pamauryA plugin can do it
00:15:00pixelmaalso - if I understand correctly and if it works, it will just show the prerecording time setting which is different from what is there now and which I prefer much (counting up as time starts until the set time is reached and so you know how much of it is really available)
00:15:18Unhelpfulamiconn: right, why not have an option for that... and it can just map to −−endian [big/little] internally.
00:15:24amiconnThat kinda defeats the purpose of an external tool
00:15:43Unhelpfulpamaury: a plugin might run too late (if you load the DB before fixing it)
00:15:46amiconnUnhelpful: That's why I asked whether it's part of rbutil these days
00:15:54amiconnRbutil already does target detection
00:16:14*rasher doesn't see quite why it defeats the purpose of an external tool
00:16:30Unhelpfuljust add −−endian to the tool, then support running the tool from rbutil?
00:17:30Buschelmt: did you ever profile the different steps in the decoder parts which happen before the synthesis (e.g. decodeTonal, decodeSpectrum, addTonal)? iqmf and imdct should be well known and already were/are under optimization.
00:17:39pamauryWhat about tagcache calling a plugin to change endianness if required ?
00:19:10mtQuestion for CF experts : Can shift operations be done without consuming any extra cycle when folded with a load or a store for example ?
00:19:28mtBuschel: No, I didn't profile it.
00:19:38mtamiconn: thanks
00:20:22amiconnShifts are never free on coldfire (except a few special shift amounts in emac operations)
00:20:27 Quit domonoky (Read error: Connection reset by peer)
00:21:04Unhelpfularm can't shift on load/store, either. ;)
00:21:16pamauryUnhelpful: what do you think of ^^ ?
00:22:08Unhelpfulpamaury: it seems a bit silly to me, tbh... and potentially fail-tastic. :/
00:22:13mtUnhelpful: Bad memory ! :(
00:22:39Unhelpfuli'm not sure absolutely everywhere the DB accessed from is safe to run a plugin from
00:22:48pamauryUnhelpful: yeah but it can free memory. It's silly to have support for both endianness also ;)
00:22:55Unhelpfulespecially if you, say, only use the DB to be able to run pictureflow.
00:23:07Unhelpfulmt: you should replace it!
00:24:00pamauryUnhelpful: I don't understand. You can access the db directly, using open/read/write/close to change some endianness
00:24:59Unhelpfulpamaury: yes, but i'm saying that if a plugin is the first thing to access the DB (via the DB functions), what happens then? or do we run this plugin on each storage unplug or startup?
00:27:05pamauryUnhelpful: I didn't think about that problem. Perhaps on each unplug or startup yes. Not sure it's the best solution though.
00:27:23amiconnAlso rockbox has no file locking. So if both the plugin and the database thread access the file at the same time,
00:27:26 Join AndyI [0] (~pasha_int@
00:27:51amiconnthings will become seriously messed up
00:29:45pamauryThe plugin would call a special db function like tagcache_lock
00:31:30pamauryAnyway, either you support endianness on target or not. If you do, that's either in the core or either in a plugin. The hird solution is to have a piece of code in tagcache that's always modifies the db files in place to fit the endianness. Thus the tagcache code would use the target endianness. That could be a sensible solution
00:31:56pamauryNot sure it would free code, though
00:32:43 Quit dfkt (Quit: -= SysReset 2.53=- Ph'nglui mglw'nafh Cthulhu R'lyeh wgah'nagl fhtagn.)
00:34:53 Join stripwax [0] (
00:35:18 Join shaggy-h [0] (
00:36:36 Quit stripwax_ (Ping timeout: 256 seconds)
00:37:34saratogamt: joint stereo is rarely used for ATRAC3
00:37:46saratogafor some reason Sony decided that only the lowest bitrate should use it
00:37:55saratogaIIRC that is 64kbps ?
00:39:26saratogaBuschel, mt: I did profile ATRAC3 when it was first ported
00:39:45saratogaIIRC the tonal stuff took no time at all because its rarely used, the other parts added up to something like 20MHz on ARM
00:39:52mtsaratoga: I don't remember the bitrates
00:39:58saratogabut i didn't take notes so i could be remembering wrong
00:40:02stripwaxsaratoga - why can't liba52 use the imdct (why can it only use the fft part?)
00:40:14saratogastripwax: it can but its a bit harder
00:40:17 Quit shaggy-h (Ping timeout: 240 seconds)
00:40:23saratogaa52 short blocks are done together in pairs
00:40:43saratogaand the current codec already folds the windowing into the post rotation, so that needs rewriting too
00:41:07 Quit Sajber^ (Read error: Connection reset by peer)
00:41:11saratogaIMO its not worth the effort just yet, the current liba52 code is actually very efficient
00:41:19stripwaxyep, I imagine it is
00:43:04Buschelsaratoga: if the tonal stuff is not used much, most of the decoding time must be spent in decodeSpectrum
00:43:10stripwaxsaratoga - I tried rewriting the various TRANSFORM macros to use ldmia/stmia but it doesn't seem to make a whole lot of difference
00:43:35stripwaxand I don't *think* they're very inefficient (although gcc might have put lots of crap elsewhere to counteract the benefit, who knows)
00:43:56saratogabasically to do the short block efficiently you need to do both blocks at the same time since they're interleaved: block 1-sample1, block2-sample1, ...
00:44:31saratogaBuschel: i think thats what i concluded but my memory is foggy
00:44:49saratogai remember thinking it wasn't worth the time since then the filterbank took 3-4x as much cpu time
00:44:52 Quit efyx_ (Remote host closed the connection)
00:45:18saratogastripwax: did you ever try benchmarking with the transform macros commented out?
00:45:25saratogai wonder how much time they really use
00:45:34 Quit bmbl (Quit: Bye!)
00:45:59stripwaxsaratoga - I haven't tried that, no. But - quite a lot, right? since the transform macros implement the fft steps?
00:46:17stripwaxyou comment them out, it means you comment out the butterflies and the twiddle multiplications
00:46:34stripwax(well, except fft4/fft8)
00:46:37JdGordonpixelma: excellent. thanks
00:46:43JdGordonthe panic is expected
00:46:56JdGordonand the peak meters not updating fast enough is an easy fix
00:46:56saratogaiirc fft4/8 took a large portion of the entire run time already
00:47:44 Part lead ("Leaving")
00:48:08saratogastripwax: any thought about when you want to merge the mdct branch?
00:48:10stripwaxsaratoga - probably, since you're guaranteed to use at least those steps; it's only the large fft sizes that end up using the transform macros
00:48:26saratogawell the large transform sizes are the only ones most codecs use
00:48:34saratogablocks less then 512 points are quite rare
00:48:47stripwaxsaratoga - well, it now compiles on coldfire at least .. But I'm not in a position to write optimised coldfire. Perhaps merge when the coldfire is also faster than trunk?
00:48:56freddybShould there be #WARNING in the clickable background patch if there's no button map defined?
00:49:13saratogafrom the tests before CF was probably within 1MHz of trunk
00:49:20saratogaso its almost there
00:49:42 Join froggyman [0] (
00:49:46stripwaxI also want to get rid of the terrible revtab; but haven't come up with a good way yet.
00:49:58saratogahave you worked out the init functions you want?
00:50:08pamauryUnhelpful: do you want me to try something for tagcache endianness ? To see if we can win something
00:50:08saratogamdct_global is now useless i believe
00:50:18Buschelmt, saratoga: we can forget about addTonalComponents -> needs only 0.1 MHz
00:50:19stripwaxFeel free to merge now if you like :) (The extra memory requirement of revtab *might* mean some targets won't compile?)
00:50:29pixelmaJdGordon: I wouldn't call it excellent
00:50:39saratogaBuschel: sounds good
00:50:40stripwaxsaratoga - you need to init the revtab. That's the only init function you need.
00:50:48 Quit sudoman (Quit: sudoman)
00:51:00stripwaxmdct_init_global is indeed useless now.
00:51:03saratogais it worth conveting the revtab to a constant in a header?
00:51:09stripwax(hadn't realised I'd left it in)
00:51:26stripwaxsaratoga - maybe, but as we discussed it's 8KB.
00:51:32 Quit saratoga (Quit: CGI:IRC)
00:51:38 Join saratoga2 [0] (
00:51:49stripwaxputting it in a header is a good idea actually, yeh.
00:51:54saratoga2the 512 revtab could be much smaller
00:51:59saratoga2iirc its just 1KB
00:52:11saratoga2and that would speed up atrac3, wma and cook a good bit
00:52:19 Join anewuser [0] (anewuser@unaffiliated/anewuser)
00:52:43stripwaxsaratoga - hm, you only need one revtab right? or you mean make it codec specific .. ?
00:52:52saratoga2bah stupid cgiirc, freenode's client needs to come back onlinne
00:53:09saratoga2stripwax: just put the small ones in IRAM and leave the rest in DRAM
00:53:25stripwaxso you have twice as much data?
00:53:33JdGordonpixelma: well, its alot better than a blank display
00:53:48saratoga2stripwax: its only 2x as much for wma, for everything else the waste is much smaller
00:53:49stripwaxthe big table is a superset of the small table. not sure what you'd gain by splitting it up into a big table and a small table.
00:54:09***Saving seen data "./dancer.seen"
00:54:23saratoga2for tremor is something like a 16KB table and a 2KB waste i think
00:54:37perfectdrug_if anyone needs testing for coldfire I'm happy to help out btw
00:55:12 Nick dmb is now known as dmbq (~Dmb@unaffiliated/dmb)
00:55:19 Nick dmbq is now known as dmb (~Dmb@unaffiliated/dmb)
00:55:31stripwaxsaratoga2 - for tremor, which table are you proposing putting into iram?
00:56:01pixelmaJdGordon: I don't know why it didn't work before and suddenly does, the only thing I changed in the meantime was changing the statusbar to the inbuilt one but even selecting an sbs again now doesn't bring the blank screen back. Maybe you need another tester.(another change is that I have the bmp viewer plugin patch applied but I doubt it is related)
00:57:43saratoga2stripwax: just want the table used for the 512 point mdct in IRAM so its fast for the codecs that need it
00:57:56saratoga2i don't think it will help tremor but it will also barely use any extra memory
00:59:12saratoga2although if you can get rid of it entirely thats even better
00:59:12stripwaxsaratoga - go for it. I think iram usage is fairly tight for tremor but see what happens. I've made the chaneg to remove mdct_init_global by the way - but svn commit seems down.
01:00:29stripwaxI think getting rid of it entirely would require changing the butterflies somewhat, so that they work in a regular bitreverse indexing scheme, rather than the current complex mixed digit-reverse
01:00:39saratoga2stripwax: svn is up is just very slow due to the cia bot timeout
01:00:51saratoga2it'll eventually commit just let it sit for 5 minutes or so while it waits for the time out
01:01:12stripwaxindeed - seems like it's just worked.
01:01:15saratoga2how much time does the bit reverse actually use?
01:01:22saratoga2it seems ugly but if its fast maybe its not an issue
01:01:47perfectdrug_saratoga2: what file should i test, wma_192.wma again?
01:01:59saratoga2perfectdrug_: thats fine
01:02:10saratoga2ac3, wma and vorbis all use the new code
01:02:43saratoga2stripwax: if you're planning to remove the last init function, then we should wait to change the other codecs over to the new mdct
01:07:01stripwaxsaratoga2 - it's ugly but probably very fast
01:07:54stripwaxso maybe leave it (you can put it precomputed and in iram, but not change anything?)
01:08:06 Join dmbq [0] (
01:08:21stripwaxand I will continue experimenting with improving some of the macros etc
01:08:34 Nick dmbq is now known as Guest31556 (
01:09:03saratoga2i just figured eventually we were going to have to try and rewrite all of it in ASM
01:09:45 Quit pamaury (Quit: abort();)
01:10:09saratoga2also maybe try doing that 3mul,5add complex mul in asm and see if it looks like it might be fast enough to be worth the trouble
01:12:47amiconnYou may be really better off writing some performance critical routines in asm
01:13:36amiconnWhen doing the mmx stuff for ape, I tried using gcc intrinsics first. Using asm() blocks made it twice as fast...
01:13:58 Nick fxb__ is now known as fxb (
01:14:38stripwaxamiconn - was that to me? a bunch is already in asm.
01:14:51Buschelmt, saratoga2: atrac3 profiling on arm -> total=65.0MHz, iqmf=43.9MHz, overlap_add=2.7MHz, imdtc=11.6MHz, other=6.8MHz (of that inverse quantization of spectrum needs 1.5MHz)
01:15:04amiconnThat was referring to [01:09:04] <saratoga2> i just figured eventually we were going to have to try and rewrite all of it in ASM
01:15:07Unhelpfulpamaury: i don't even have an archos, i was just suggesting some possibilities...
01:15:24stripwaxamiconn - thanks. saratoga - it is already, right?
01:15:38 Quit Guest31556 (Quit: Gone)
01:15:47stripwaxexcept the actually looping, I guess
01:15:59mtBuschel: So it's still mostly in iqmf..
01:17:11perfectdrug_ok tested wma_192.wma, mdct: current svn:
01:17:49 Join dmbq [0] (~quassel@unaffiliated/dmb)
01:18:59mtperfectdrug_: Thanks for the test. :)
01:19:08Buschelmt: yes.
01:19:16stripwaxmt - (is that for mdctexp?)
01:19:57 Join sudoman [0] (
01:20:02mtstripwax: I assume so .. it's saratoga who sent the file. saratoga2 ^ ?
01:20:52stripwaxperfectdrug_ - so about 1Mhz faster than trunk - excellent.
01:21:41 Quit dmbq (Client Quit)
01:22:09sudomanany idea on how i could contact the maker (Demetrios Stylianopoulos) of a particular theme about a small fix i made to the .wps file of his theme?
01:22:21 Join dmbq [0] (~dmb@unaffiliated/dmb)
01:26:47 Quit dmbq (Client Quit)
01:29:37 Join dmbq [0] (~dmb@unaffiliated/dmb)
01:33:16saratoga2sudoman: either email, or pm on the forums are your best bets
01:33:31saratoga2stripwax: so it can go in whenever we think its ready
01:33:38mtBuschel: Wouldn't it save something if the index in iqmf is hardcoded instead of incrementing i after each mul ? (The loop is unrolled already, so what's the need for i anyway ?)
01:34:18stripwaxsaratoga2 - yep, go for it. like I said I haven't tried building on all targets, just ipod video and iriver h120.
01:34:56saratoga2stripwax: lets wait until the init stuff is cleaned up
01:35:18amiconnmt: On coldfire incrementing is better. Storing literal offsets needs extension words
01:35:39Buschelmt: that should not have any effect. maybe it is even easier for the compiler this way, because the loads are auto increments.
01:36:25mtamiconn, Buschel : thanks.
01:36:30 Quit dmbq (Quit: Gone)
01:37:08 Join dmbq [0] (~dmb@unaffiliated/dmb)
01:37:18stripwaxsaratoga2 - fine - are you doing that? (it's just that I'm not ....)
01:42:03 Quit FlynDice (Remote host closed the connection)
01:42:22saratoga2stripwax: sure I can, but not tonight, have to get back to work
01:47:29stripwaxsaratoga2 - cool (yes, doesn't have to be tonight!! :-)
01:54:03 Quit m3dlg (Ping timeout: 252 seconds)
01:55:55 Join JdGordon_ [0] (
01:57:33 Join m3dlg [0] (
01:58:17 Part froggyman
01:59:59 Quit stripwax (Quit:
02:01:43 Quit mt (Quit: ChatZilla 0.9.86 [Firefox 3.5.7/20091221164558])
02:02:43Buschelmt: can you compare wav output of two codec versions at your PC now?
02:04:08 Join AndChat- [0] (
02:04:09 Quit JdGordon_ (Read error: Connection reset by peer)
02:05:49Buschelmt: you could test this patch ->
02:06:23Buschelmt: it saves ~1.2 MHz on arm and does the <<=2 shift during the calculation of the scalefactors.
02:13:55 Join JdGordon_ [0] (
02:15:32 Quit AndChat- (Ping timeout: 258 seconds)
02:16:15 Quit petur (Remote host closed the connection)
02:23:12 Join kramer3d [0] (~kramer@unaffiliated/kramer3d)
02:23:13 Quit JdGordon_ (Read error: Connection reset by peer)
02:24:09 Quit sudoman (Quit: sudoman)
02:24:38 Quit dmbq (Quit: Gone)
02:24:54 Join dmbq [0] (~dmb@unaffiliated/dmb)
02:27:13 Quit Kitar|st ()
02:28:30Buschelmt: another issue to optimize is within gainCompensateAndOverlap() -> gain1 and gain2 are mostly ONE_16. Therefor fixmul16(x, gain) can be exchanged with simple x.
02:30:01Buschelmt: all other gains can be modified to shifts as all gains are of the type 2^n. so, we can delete the table lookup and use the index for the shifts.
02:30:48 Join GeekShad0w [0] (
02:33:05 Quit GeekShadow (Ping timeout: 240 seconds)
02:33:49Buschelgotta get some sleep now. bye
02:33:50 Quit Buschel ()
02:35:39 Join Kitar|st [0] (
02:38:41 Join JdGordon_ [0] (
02:38:54 Quit TheSphinX^ (Quit: XChat@Linux)
02:42:18 Join GeekSh4dow [0] (
02:44:45 Quit GeekShad0w (Ping timeout: 245 seconds)
02:48:57 Quit MethoS- (Remote host closed the connection)
02:49:32 Quit m3dlg (Ping timeout: 252 seconds)
02:54:13***Saving seen data "./dancer.seen"
02:54:56 Join Rob2223 [0] (
02:58:50 Quit Rob2222 (Ping timeout: 256 seconds)
03:06:13 Quit bzed (Read error: Operation timed out)
03:08:40 Join bzed [0] (
03:09:48 Quit anewuser (Quit: =ooo ϢINTER ϾHIP 5iVE is OOON!!)
03:12:35 Quit JdGordon_ (Ping timeout: 256 seconds)
03:13:00 Join JdGordon_ [0] (
03:13:41 Quit dmb_ (Quit: Leaving)
03:13:54 Quit GeekSh4dow (Read error: Connection reset by peer)
03:14:30 Quit dmbq (Read error: Connection reset by peer)
03:15:37 Quit Adnyxo (Ping timeout: 276 seconds)
03:15:49 Quit dmb (Quit: Leaving)
03:16:13 Join dmb [0] (~dmb@unaffiliated/dmb)
03:23:57 Join anewuser [0] (anewuser@unaffiliated/anewuser)
03:26:39 Part stooo
03:27:40 Join AndChat| [0] (
03:27:45 Quit JdGordon_ (Read error: Connection reset by peer)
03:27:47 Join dmb__ [0] (~Dmb@
03:29:40 Quit Schmogel (Read error: Connection reset by peer)
03:31:49 Join djoume [0] (
03:32:05djoumehi there
03:32:47 Join antil33t [0] (
03:32:48djoumeI just installed rockbox on an ipod video 30GB
03:33:00 Join JdGordon_ [0] (
03:33:06 Quit AndChat| (Read error: Connection reset by peer)
03:33:19djoumeand I get a PANIC: "Unsupported physical sector size: 4096"
03:33:20 Quit Strife89 (Quit: Bed.)
03:34:34djoumeI changed MAX_PHYS_SECTOR_SIZE to 4096 in firmeware/export/config/ipodvideo.h
03:34:46djoumerecompile, reinstall
03:34:59djoumebut I still get the same error :-(
03:35:15djoumeany suggestion?
03:35:54djoumeis it enough to copy the files on the mounted device?
03:36:05djoumeafter recompiling the firmware?
03:38:20 Nick fxb is now known as fxb__ (
03:38:35 Quit JdGordon_ (Read error: Connection reset by peer)
03:40:16 Join karashata [0] (
03:44:49karashatacould someone please take a look at whatever it is that checks the themes for proper syntax on the theme site..?
03:45:00karashatait's telling me a comment isn't a valid theme setting
03:47:06saratoga2djoume: did you upgrade the hard drive on your player?
03:48:43saratoga2then you shouldn't need to change anything
03:48:47djoumeit's the default 30GB
03:48:58saratoga2are you sure you installed the right build in the first place?
03:49:13djoumeI used the gui automatic install
03:49:56 Join flyback [0] (
03:50:04saratoga2you could try downloading a new build from the website and extracting it to your player
03:50:16saratoga2are you sure no one else swapped the disk on your player?
03:51:08djoumeno I'm not 100% sure, I just bought it "second hand" (but like new).
03:51:21djoumeI don't think it has been upgraded because the disk is still 30GB
03:51:37saratoga2you may need to come back and ask someone who knows the ipods better
03:51:43saratoga2daytime in europe is a good bet
03:52:00djoumeok, thanks for your help
03:54:50 Quit DerPapst (Quit: Leaving.)
03:55:02karashataI'm trying to upload an update to one of my themes and it's not allowing me to upload it because it thinks my comment "#Theme Name: <name>" is not an allowed theme setting
04:00:17saratoga2a lot of changes to themeing just went through
04:00:21saratoga2it probably hasn't been updated
04:00:30saratoga2ping JdGordon about it
04:01:48 Join froggyman [0] (
04:02:01karashataokay, and what's the best way to do that? leave a message here with his nick, or..?
04:05:43 Quit n17ikh (Ping timeout: 256 seconds)
04:06:09 Quit froggyman (Client Quit)
04:08:30karashataJdGordon: ping! Could you please check on the theme site's theme checking, it's telling me comments aren't allowed theme settings and I can't upload my themes because of it
04:09:39 Quit djoume (Quit: leaving)
04:11:38 Join n17ikh [0] (
04:13:17 Quit kramer3d (Read error: Connection reset by peer)
04:13:44 Join kramer3d [0] (~kramer@unaffiliated/kramer3d)
04:16:23 Quit TheSeven (Disconnected by services)
04:16:38 Join The_Seven [0] (~theseven@rockbox/developer/TheSeven)
04:16:49 Nick The_Seven is now known as TheSeven (~theseven@rockbox/developer/TheSeven)
04:17:05 Quit karashata (Quit: The fluffy dragon has left entirely!)
04:17:42 Join karashata [0] (
04:28:31 Quit Rob2223 (Quit: Rob2223)
04:29:43 Join Rob2222 [0] (
04:32:07 Join kramer3d_ [0] (~kramer@unaffiliated/kramer3d)
04:32:29 Quit kramer3d (Ping timeout: 240 seconds)
04:38:07 Join Barahir [0] (
04:38:10 Join Llorean [0] (
04:38:11 Quit Llorean (Changing host)
04:38:11 Join Llorean [0] (~DarkkOne@rockbox/user/Llorean)
04:40:30 Quit Res1 (Read error: Operation timed out)
04:41:21 Quit Barahir_ (Ping timeout: 248 seconds)
04:41:50 Quit dmb__ (Quit: Leaving)
04:43:41 Quit freddyb (Remote host closed the connection)
04:52:23 Join Res1 [0] (
04:54:17***Saving seen data "./dancer.seen"
05:01:29JdGordonkarashata: you'd need to talk to rasher about that. it seems a script somewhere is broken
05:03:17 Join S_a_i_n_t [0] (S_a_i_n_t@
05:03:20karashataah, okay. I noticed his simulator builds haven't been building recently either so I'll bug him to fix that too
05:06:48JdGordonyeah, I have no idea where those settings scripts live
05:09:18JdGordonah, found it. unfortunatly my php sucks
05:10:15JdGordonwhats the line that fails?
05:10:47karashata#Theme Name: <name of theme here>
05:10:59JdGordon# is the first char of the line?
05:11:00karashatavery first line
05:11:30JdGordonwierd, that is handled
05:11:33JdGordonor appears to be
05:11:48karashatathemes work fine in the simulator build I have
05:12:11karashatathough that's not very recent anymore since rasher's site doesn't have any up-to-date builds
05:13:06S_a_i_n_tthe builds on rasher's sit were updated bang on 3.5 I believe...a lots changed since then.
05:13:25JdGordonyeah, the theme site is very strict with the .cfg files
05:13:35JdGordonits thinking #comment is a setting which it obviously isnt
05:13:42 Quit panni_ (Read error: Connection reset by peer)
05:13:52JdGordonhow do you simply check if a line matches a regex in php?
05:14:08JdGordonif (preg_match("/\s*#/", $line) == 0) continue;
05:14:24*S_a_i_n_t 's php sucks...sorry
05:14:54JdGordonthats what the site is coded in :p
05:15:41JdGordonapparently yes
05:17:39 Join bob77 [0] (
05:18:53JdGordonbug created, hopefully it will be fixed soon. otherwise you can just remove that line
05:19:00JdGordonare you sure the # is the first char?
05:19:10JdGordonthere is no sily utf8 BOM?
05:19:37karashataI wouldn't know for sure but I'm as certain as I can be that it is the first character
05:19:50karashataif there's anything before it it's nothing I can edit out anyway
05:20:46karashatathat theme uploaded fine before too and I didn't make any changes to that part of the theme
05:20:55saratoga2what did you edit it with
05:21:05karashataWindows notepad
05:21:15karashatasame program I originally created it with too
05:21:55karashatasaved it with ANSI encoding
05:22:11*S_a_i_n_t likes can let you know when you've messed up your syntax
05:23:03bob77hi all, with current r24663 on an 80GB slim 5.5G I got *PANIC* unsupported physical sector 4096 .. what to do?
05:23:34S_a_i_n_t*well, it has syntax highlighting rather, its still up to you to notice if your syntax is incorrect :p
05:24:17JdGordonbob77: did it used to work?
05:24:28JdGordonI tihnk you need to compile a custom build with support for that
05:24:35bob77no, first time install
05:28:34 Join kimi [0] (
05:28:57JdGordonyeah, you need to do a custom build, I'm not sure what needs to be changed though
05:29:12JdGordonthere might be an answer in the forums
05:29:33 Quit karashata (Disconnected by services)
05:29:41 Nick kimi is now known as karashata (
05:31:23 Quit ps-auxw (Read error: Operation timed out)
05:31:40 Join ps-auxw [0] (~arneb@2001:470:c807:0:1532:4e5f:2ad3:4123)
05:31:40saratoga2JdGordon: someone came in here an hour ago and asked about that
05:31:45saratoga2with a 30GB
05:31:57JdGordonbob77: you going to be around for a while?
05:32:09saratoga2maybe something is broken in the current build
05:32:38JdGordonit might be a 2 line change in the config, but i need to eat so I'll hel you in a bit
05:32:46saratoga2bob77: does this happen with 3.5?
05:33:05bob77yes, tried both stable and current
05:33:35bob77the ipod is *slim* 5.5G 80GB just got it from ebay
05:33:43JdGordonI tihnk you just need to add #define HAVE_LBA48 and change the MAX_PHYS_SECTOR_SIZE 4096 in the video config
05:33:54saratoga2ok so it may have been repaired with a new disk drive
05:34:18bob77true, refurb
05:34:29S_a_i_n_tJdGordon: from what I'm looking at, that seems correct.
05:35:13 Quit Horscht (Quit: Verlassend)
05:38:48JdGordonbob77: 64mb or 32mb version?
05:39:03bob77I think I got the 64MB
05:39:26JdGordonyou've got the bootloader installed yeah?
05:40:21bob77I installed the bootloader first then .rockbox under
05:40:27bob77I installed the bootloader first then .rockbox under "installation" tab
05:41:40JdGordonunzip"> to the ipod
05:42:23bob77okay, lemme try ..
05:46:44S_a_i_n_tJdGordon: Now *that's service ;-D
05:48:07JdGordonif that works I'll fix up configure to allow it as an advanced build option
05:48:24JdGordonI don't know why we dont do that build already.. arnt those disks fairly common?
05:48:57S_a_i_n_tprovided of course that the user has a build environment set up...
05:49:22S_a_i_n_tthere probably should be a "big disk" build available.
05:50:15 Quit akur (Quit: Leaving.)
05:50:33saratoga2i think it assumed that if someone can upgrade an ipod they can compile their own rockbox
05:50:43saratoga2which was probably true until someone started selling these things on ebay
05:51:39bob77same error came up when booting
05:52:49bob77I actually bought this one to put rockbox in :)
05:53:24JdGordonsearch the forums :/
05:53:32S_a_i_n_tsaratoga2: that is probably correct, but like you say, there is the issue with'd be nice to be able to get a percentile of who has which disk know if having the build on the main page would be worthwhile.
05:56:45S_a_i_n_tbob77: may help...
05:58:38*bob77 is looking
05:59:18S_a_i_n_tis there a chance it's actually a 32MB model? the 64MB build wouldn;t work in that case...would it?
05:59:32S_a_i_n_twhereas 32MB would work on the 64.
05:59:55saratoga2it'll work but crash after playing for a while
06:01:04bob77hmm could be, lemme try the 30GB model
06:11:15bob77no go, I'm trying now
06:12:19 Quit dmb (Quit: Gone)
06:18:38 Join dmb [0] (~dmb@unaffiliated/dmb)
06:20:09JdGordonyou probbaly need a bigdisk bootloader also
06:21:42JdGordonyou need to manually install it
06:21:44 Quit anewuser (Quit: =ooo ϢINTER ϾHIP 5iVE is OOON!!)
06:21:55S_a_i_n_thmmm, that's a good point.
06:22:08bob77YAY!!! it's go go go
06:22:40S_a_i_n_tbob77: which build did you use?
06:22:49S_a_i_n_tand which bootloader?
06:23:00bob77this one:
06:23:12bob77the 64MB one
06:24:20*bob77 sincerely thank JdGordon, saratoga2, and the aptly named S_a_i_n_t for this
06:24:25S_a_i_n_tHmmmm, in shouldn't be too hard to see what's touched by those patches, and apply them to current build.
06:24:34S_a_i_n_tI have a feeling you'd prefer 3.5
06:24:41S_a_i_n_ta lot has changed since 3.4
06:24:45bob77I see
06:25:12JdGordoncan you try my bootloader and build?
06:25:23bob77of course hang on
06:25:43JdGordonyeah, looks like it should work
06:27:28bob77damn you guys, can;t I play with this for a while FIRST ? :)
06:27:47S_a_i_n_tlater... :P
06:27:49bob77nah, I'll go straight to JdGordon's stuff, hang on
06:27:58S_a_i_n_twe need to know we can get a current build working :D
06:35:52bob77Model name in the input file doesn't match.. tried renamed it to bootloader-ipodvideo.ipod too
06:36:55S_a_i_n_tare you installing this manually?
06:37:01S_a_i_n_tthe bootloader I mean.
06:37:25bob77yes ipodpatcher -a bootloader-lba64.bin
06:41:22bob77ah ipodpatcher -ab *.bin
06:44:12S_a_i_n_tyep... -a only works for .ipod files.
06:44:21S_a_i_n_t*and .ipodx
06:45:31bob77bootloader-lba64.bin *PANIC* 4096
06:45:54S_a_i_n_ttry the 3. one JdGordon linked earlier
06:45:57bob77only with white background ?
06:46:00JdGordonwhat do I call the configure option?
06:46:28S_a_i_n_tin "advanced" for the 5.5g?
06:46:34JdGordon(B)ig Disk, or (4)Kb sectors
06:46:52S_a_i_n_t(B) fits in more readily
06:47:00S_a_i_n_tas all else are
06:47:43S_a_i_n_tthat one...try that.
06:48:14S_a_i_n_tsorry for the mess around btw...
06:48:16S_a_i_n_tand thanks.
06:48:48 Join kramer3d [0] (~kramer@unaffiliated/kramer3d)
06:49:09*S_a_i_n_t wonders what's broken...
06:49:10 Quit kramer3d_ (Ping timeout: 246 seconds)
06:49:38S_a_i_n_tbob77: well, at least you know that there's *one* current(-ish) build you can use.
06:49:50JdGordonneither of mine work?
06:49:52bob77same with 32MB
06:50:58bob77both white screened with *PANIC
06:53:10bob77I read somewhere Steve Martin (lol) suggested commenting out MAX_PHYS_SECTOR_SIZE 4096 altogether ... and only have thet #define HAVE_LBA48
06:54:07JdGordonbob77:"> try this one again
06:54:17bob77I don't mean to be rude of course.. but could it be that
06:54:18***Saving seen data "./dancer.seen"
06:54:23*S_a_i_n_t also found Steve Martin amusing...(I doubt it's *the* Steve Martin however :P)
06:56:20JdGordonit seems odd that that #define isnt used
06:57:37JdGordonbob77: woops, download that link again
06:57:43 Part toffe82
07:02:57bob77any 64MB I should try?
07:03:23*S_a_i_n_t scratches his head...
07:04:32bob77did ya -#define MAX_LOG_SECTOR_SIZE 2048 +#define MAX_LOG_SECTOR_SIZE 16384
07:05:26bob77coz nugged did,
07:07:41JdGordonthat one shouldnt make a difference :/
07:08:31JdGordonarg, what state is the bootloader in?
07:08:52bob77rockbox flashed a little than white screened
07:09:06JdGordonwith the panic?
07:09:06bob77*PANIC* pysical 4096
07:09:45bob77should I try nugged's 32MB?
07:09:53 Quit Lss__ (Read error: Connection reset by peer)
07:10:04bob77what's the sign of wrong amount of RAM
07:10:33S_a_i_n_t32mb build will work fine on the 64mb
07:10:38S_a_i_n_tbut not in the reverse
07:11:00 Quit liar (Ping timeout: 245 seconds)
07:11:04JdGordonwhats the exact wording of the panic message?
07:11:29bob77*PANIC* unsupported physical sector 4096
07:12:23JdGordonnot *PANIC* unsupported physical sector size <something>?
07:13:56bob77Unsupported physical sector size: 4096
07:16:12JdGordontry it once more... if that doesnt work im out of ideas
07:16:24JdGordonthe download link that is
07:16:27bob77btw before doing anything I used itunes to format the thing
07:16:36JdGordonthats fine
07:20:37bob77the last one doesn't work ... btw the link always 404 Not found, so added the t myself
07:21:14bob77the last one 5dfbbc004f7a5a78ac8902d5d789b2eb still panic
07:22:16JdGordonthats the md5sum?
07:22:35JdGordonthats very wrong...
07:24:17JdGordon"> 2d3382341729ea0ba6f29881a5bb19c3
07:25:52bob77flashing hold on
07:25:55S_a_i_n_tif you go by nugged's .diff only +#define MAX_LOG_SECTOR_SIZE 16384, +#define MAX_PHYS_SECTOR_SIZE 4096, and +#define HAVE_LBA48 are touched...this shouldn't be too hard.
07:26:03*S_a_i_n_t wonders whats going wrong.
07:26:41JdGordonwget -O -"> | md5sum - gives me the correct md5sum
07:27:18bob772d3382341729ea0ba6f29881a5bb19c3 *bootloader-lba32.bin
07:27:18S_a_i_n_tJdGordon: FWIW...same here.
07:28:00bob77still no go :)
07:28:31JdGordonwhich rockbox build have you got installed?
07:28:35 Join FlynDice [0] (
07:28:42JdGordonnow im wondering if the bootloader is correct but the main build is panicing
07:29:40bob77could be, do you have any build I should try
07:30:06bob77this one:">
07:30:18S_a_i_n_tmy Video has no HD atm...otherwise I'd test myself.
07:30:51JdGordonyeah that one
07:31:07JdGordonand yes, this shouldnt be this hard :p
07:31:41bob77I got that build in place for the 32MB.bins
07:32:31S_a_i_n_tbob77: JdGordon: FWIW...someone else trying to apply the bigdisk patch is/was having the same trouble in Dec '09 " "
07:33:22bob77no not me :) although I'm making a fool of myself often.. diff bozzo
07:34:03JdGordonno answer though
07:34:27JdGordonbob77: you need to be in here when euorpe (GMT+1) is awake...
07:34:28S_a_i_n_tI thought it may be relevant to suggesting that the main build may be broken, as he mentioned that the patch used to work fine for him...then just dstopped.
07:34:33JdGordonamiconn: can probaly help
07:35:49S_a_i_n_tbob77: for npw, at least you know there is one build you can use...and this will *probably* get sorted at some point in the not-so-distant future.
07:36:45bob77okay.. np guys, I'd like to help too
07:37:16bob77am looking at nugged's config-ipodvideo.h
07:37:41S_a_i_n_tthere is only 3 changes there that I can see
07:37:45S_a_i_n_t*at a glance
07:38:14S_a_i_n_t#define MAX_LOG_SECTOR_SIZE 16384, #define MAX_PHYS_SECTOR_SIZE 4096, and define HAVE_LBA48
07:39:26S_a_i_n_tbut he seems to suggest later in the thread that #define MAX_PHYS_SECTOR_SIZE 4096 may be neglected altogether?
07:41:16bob77and #define HAVE_ATA_DMA
07:41:41bob77the original
07:42:03bob77there's no HAVE_ATA_DMA in the .orig
07:42:39 Quit TopyMobile_ (Ping timeout: 265 seconds)
07:43:52S_a_i_n_tI have to pass on this...code-wise, I have to assume JD knows his stuff to a degree far beyond that of myself.
07:44:16S_a_i_n_tand that is more a fact than an assumption :P
07:44:43 Join Zagor [0] (
07:44:44 Quit Zagor (Changing host)
07:44:44 Join Zagor [0] (~bjst@rockbox/developer/Zagor)
07:46:57bob77other than those three discussed, nugged also added /* Define this to add support for ATA DMA */ #define HAVE_ATA_DM
07:54:03 Join BHSPitMonkey [0] (~stephen@unaffiliated/bhspitmonkey)
07:54:09 Quit BHSPitMonkey (Client Quit)
07:55:15 Quit Tomis (Quit: Tomis)
08:00:16 Join Tomis [0] (~Tomis@
08:01:59 Quit Tomis (Client Quit)
08:04:21 Join TopyMobile_ [0] (
08:16:41 Quit TopyMobile_ (Ping timeout: 240 seconds)
08:18:03 Join bmbl [0] (~Miranda@unaffiliated/bmbl)
08:23:41 Join TopyMobile_ [0] (
08:27:29 Quit AndyI (Ping timeout: 248 seconds)
08:28:17 Join Bagder [0] (~daniel@rockbox/developer/bagder)
08:29:50 Join Tomis [0] (~Tomis@
08:32:51 Join AndyI [0] (~pasha_int@
08:35:47S_a_i_n_tare the sims/check wps all good with multifont?
08:36:04S_a_i_n_tI mean, the sims wont complain will they?
08:36:43*S_a_i_n_t is trying to work out whether to build a sim, or painstakingly start testing on-device.
08:38:04JdGordonthe sim bloody bbetter work
08:38:47S_a_i_n_tOK, I'll try it.
08:41:13S_a_i_n_tSimulator build selected
08:41:13S_a_i_n_tUsing source code root directory: /cygdrive/c/Cygwin/Rockbox_Source/rockbox
08:41:13S_a_i_n_tconfigure didn't find sdl-config, which indicates that you
08:41:13DBUGEnqueued KICK S_a_i_n_t
08:41:13S_a_i_n_tdon't have SDL (properly) installed. Please correct and
08:41:13S_a_i_n_tre-run configure!
08:41:46S_a_i_n_tit *used* to
08:43:51 Join flydutch [0] (
08:44:34 Quit phanboy4 (Read error: Connection reset by peer)
08:45:14 Join einhirn [0] (
08:49:22 Join kramer3d_ [0] (~kramer@unaffiliated/kramer3d)
08:51:29 Quit kramer3d (Ping timeout: 248 seconds)
08:51:37 Nick kramer3d_ is now known as kramer3d (~kramer@unaffiliated/kramer3d)
08:54:20***Saving seen data "./dancer.seen"
08:55:27 Join TopyMobile__ [0] (
08:56:31 Quit bob77 (Quit: CGI:IRC)
08:58:50 Quit TopyMobile_ (Ping timeout: 276 seconds)
09:01:12S_a_i_n_tJdGordon: Do you think it would be possible to get dict.rock to use multifint? On small screen players, and even on larger screens (depending on the length of the definition) the definition often draws way off screen.
09:01:36S_a_i_n_tbeing able to scroll it would be better, but multifont is perhaps easier?
09:04:21 Quit einhirn (Read error: Connection reset by peer)
09:07:35 Join ender` [0] (
09:15:39 Join evilnick [0] (~evilnick@rockbox/staff/evilnick)
09:21:52 Join r0b- [0] (
09:23:50 Join bob77 [0] (
09:23:59 Join TopyMobile_ [0] (
09:24:36 Join petur [0] (~petur@rockbox/developer/petur)
09:26:32 Join LinusN [0] (~linus@rockbox/developer/LinusN)
09:26:52 Quit r0b- (Read error: Connection reset by peer)
09:27:09 Quit TopyMobile__ (Ping timeout: 256 seconds)
09:35:15 Join shaggy-h [0] (
09:39:21 Join notByan [0] (
09:39:25CtcpIgnored 1 channel CTCP requests in 0 seconds at the last flood
09:39:25*notByan just had an idea
09:39:41notByanWPS's should be able to know whether or not the backlight is on
09:39:53notByanthat way, they could change colors for the proper situation..
09:40:00JdGordonS_a_i_n_t: yeah, giving the plugins a simple API to load a font is on the todo list
09:40:01notByanmy WPS is hard as hell to read without the backlight
09:40:56S_a_i_n_tnotByan: (generally speaking) wouldn;t it be more feasable to alter the theme to be more readable in that case?
09:41:48S_a_i_n_tI just can't really see a purpose on changing colours if the backlight is on/off
09:42:01notByangenerally? IDK. But, in my case my WPS is lightist blue on black
09:42:08S_a_i_n_tone easy to read scheme seems a lot better to me.
09:42:13notByanso, it's easy on the eyes when using it
09:42:24notByanbut, when the backlight is off.. it's hardly visible at all
09:42:34 Quit bob77 (Quit: zzzz)
09:42:36S_a_i_n_teasy on the eyes seems debatable if its "hard as hell to read"
09:42:51notByanit's only hard to read without the backlight..
09:42:57notByanotherwise it's very easy to read..
09:43:23notByanbut, the constrast is purposefully lower, so that it doesn't blind you when the light is on >_>
09:43:58notByanI do admit.. this seems to be special case.. I doubt many people use a WPS like mine you'd like to change the *contrast* if backlight is on/off?
09:44:15notByanI wonder how hard it'd be to hack into
09:44:24S_a_i_n_tand not the colour?
09:44:30 Quit kramer3d (Quit: Leaving)
09:45:29JdGordonthe WPS cant change the colours, you have to use conditional viewports which means a very big wps if you do that
09:45:48S_a_i_n_tnotByan: the easiest way to do this would be to have two .cfg's you can easily switch between.
09:45:48notByanS_a_i_n_t: err.. how often do you use devices without backlights on? if the background is black and the foreground is white.. full contrast, it's still not going to as readable as white background and black foreground
09:47:01S_a_i_n_tyou could also make conditional viewports that switch schemes between am/pm if you were keen
09:47:09*S_a_i_n_t debates trying that himself...
09:47:10notByanerrr... the idea that I am going to touch the buttons.. which turns the backlight back on.. to switch to a differen cfg.. so I can wait for it to turn off, that'd be silly
09:47:39notByanI thought about that idea too.. but I'd rather just adjust brightness based on daytime
09:48:03S_a_i_n_tstart coding ;D
09:48:35notByanthe problem with that idea is the fact that my time is never right.. cause my battery is so poor
09:49:10notByanbetter more practical idea: rockbox driver aquires time, so anytime it's connected it syncs the time?
09:49:42S_a_i_n_tdoes it not do that already?
09:49:47S_a_i_n_twhat target do you have?
09:49:58notByanipod 5.5G
09:50:08notByanusing bleeding edge
09:50:14S_a_i_n_tit should always be correct then
09:50:19S_a_i_n_tipod nano 1/2g
09:50:25S_a_i_n_tnever set the time once
09:50:29S_a_i_n_talways correct
09:50:35notByanthats cause your battery doesn't die..
09:50:47notByanas long as it has some left in it.. the RTC on the ipod will keep time
09:50:54notByanbut my battery dies.. alot..
09:50:54S_a_i_n_twell...there's no software fix for a shitty battery :P
09:51:18notByanbut there is for it losing time all the time >_>
09:51:33S_a_i_n_tthere is?
09:51:49notByanas I said.. whenever I plug it in.. it syncs the time >_>
09:52:11notByanevery time I charge it, I plug it into the computer.. so, after each time it dies. the time would be fixed
09:52:27S_a_i_n_tit should do that already, pluhg it in under the OF
09:52:33S_a_i_n_tit should sync the time then
09:52:47notByanerr.. but I don't use the OF?
09:52:57S_a_i_n_tapparently..RB doesn't, but I was fairly sure it did.
09:54:24S_a_i_n_tyou could plug it in under OF once in a while to set the time...if its that big an issue...FYI the OF actually draws 500mA from the USB, but (atm) Rockbox only draws 100mA for the ipods
09:54:36S_a_i_n_tso you charge will be a lot quicker under OF anyway
09:54:57notByan90% of the time, charging and playing are one in the same
09:55:04notByanso, thats not super useful for me
09:55:17notByanbut, as you said, it seems rockbox supports.. at least under linux
09:55:31S_a_i_n_tcharging/playing+possibly not charging at all...and probably why your batery is messed up
09:55:48S_a_i_n_tas it only draws 100mA, and your target needs a lot of juice
09:56:33notByanno, it was messed up before rockbox
09:57:04notByanwhy on earth does it only draw 100mA anyway?
09:57:36S_a_i_n_tnot sure *why* exactly, just that that's how it is...
09:57:49S_a_i_n_tyou're welcome to fix it :P
09:57:53notByanI wonder how hard that is to change
09:58:02notByanor if they didn't do that on purpose at all
09:58:20 Join einhirn [0] (
09:58:21S_a_i_n_tapparently quite hard, as several people have been trying to fix it for some time now IIRC
09:59:03S_a_i_n_tit is on purpose...its because rockbox has no way of knowing whether or not it *can* craw the full 500mA
09:59:03notByanI also wonder if thats why it seems like my USB charger things chargers so much faster
09:59:17S_a_i_n_tie. it may be in a hub, and not in a high-power port
09:59:49notByanerr.. I'm no expert, but doesn't it request as much as it likes, and whatever it gets, it gets?
10:01:12S_a_i_n_tI'm no expert either...but until it knows how much it can get, or where its coming from...its safest to assume the lowest amount I believe.
10:01:30S_a_i_n_tIts not like noones working on it.
10:02:51 Join Sajber^ [0] (
10:04:25notByanthe time stuff doesn't seem very hard
10:04:31notByanmaybe I can write something for windows
10:06:31notByanor probably not.. looks like it depends on libgpod damn
10:12:38 Quit CaptainKewl (Quit: ( :: NoNameScript 4.22 :: ))
10:19:42S_a_i_n_tis there any aprticular reason why the ipod bootloaders on are from Oct '08?
10:20:17Bagderinstead of?
10:20:21S_a_i_n_tsurely they have changed somewhat since then, no?
10:21:03 Join bob77 [0] (
10:33:31bob77guys, I just wanna report that nugged's r23166 and bobbozzo's r23339M can boot this 80GB but after that it's really messed up... even keypress and scrolling are hit and miss.
10:38:17S_a_i_n_tbob77: does the OF behave correctly?
10:38:49bob77what's OF
10:39:02S_a_i_n_toriginal firmware...the Apple side
10:39:33bob77yes, smooth as.. scrolling and select.
10:40:31S_a_i_n_tHmmmmmm :/
10:54:24***Saving seen data "./dancer.seen"
11:10:34AlexPWhat are nugged's r23166 and bobbozzo's r23339M ?
11:10:44AlexPUnsupported builds?
11:12:49LloreanI'm sure.
11:15:59gevaertsTorne: have you seen FS #11016?
11:26:29bob77I tried them because release and current builds kept giving me *panic* unsupported physical sector: 4096
11:29:32bob77this model:!_W0QQitemZ250571262741QQcmdZViewItemQQimsxZ20100129?IMSfp=TL100129193015r21915
11:37:28*S_a_i_n_t is beginning to wonder if the drive is indeed the size it says it is...or whether its been dodgily hacked to represent a different volome size...
11:38:07 Join robin0800 [0] (
11:38:43S_a_i_n_tbob77: out of curiousity...have you tried to 30GB video build on your player?
11:39:28S_a_i_n_tas it is actually a 30GB video is it not...but with a larger disk?
11:39:30bob77yes, if the 64MB failed I tried 32MB've actually tried the 30GB video build...not the 32mb, but the 30GB?
11:40:32S_a_i_n_t <-this one?
11:40:56bob77yes, but via rbutilqt.exe
11:41:03gevaertsS_a_i_n_t: there are no separate 32mb *and* 30GB builds. Those are the same thing
11:41:38 Join MethoS- [0] (~clemens@
11:42:47 Join wind [0] (
11:43:29bob77Now I'm gonna format it with /a:2048
11:43:42bob77then try the current build :)
11:46:20 Join jeffp [0] (
11:46:39 Part jeffp
11:47:02 Join jeffp [0] (
11:47:19 Quit jeffp (Remote host closed the connection)
11:47:24 Join jeffp [0] (
11:47:55 Quit jeffp (Remote host closed the connection)
11:51:21FrampisSID files often have more than one song, is it possible to play anything but the first one in rockbox?
11:53:17linuxstbFrampis: I _think_ you can, by seeking in the file.
11:53:38Frampisseeking eh? I'll be sure to try that out
11:54:43 Join dfkt [0] (dfkt@unaffiliated/dfkt)
11:55:36Frampisit works! thanks linuxstb
11:56:14Frampisit's just brilliant that rockbox even supports sids
12:13:34 Quit Sajber^ (Read error: Connection reset by peer)
12:18:54 Join m3dlg [0] (~m3dlg@
12:34:22Tornegevaerts: i have now.. hm.
12:34:36Tornegevaerts: i couldn't reproduce horscht's problem at all, and he had the issue before DMA went in
12:34:49Tornegevaerts: but it could conceivably be related to my previous ATA change
12:35:02TorneI am inclined to think not, though, since the change to the behaviour of reads by the unification should hav ebeen zero
12:36:04gevaertsmaybe the multisector one?
12:37:27Tornethe unification only changed the behaviour of reads (intentionally, at least)
12:37:30Torneof writes
12:37:34Tornereads were already multisector
12:39:04Tornei guess the test would be to have horsct go back to the revision before
12:39:11Tornesince he seems to be able to do it reproducably
12:39:25Tornewe experimented with different album art before and it looked like it was related to AA rescaling
12:39:30Tornebut the results didn't make sense.
12:39:44Tornehe was using bitmap AA scaled to the exact size the skin needed, which the bmp loader doesn't yield while loading
12:39:59Torneand it went away if he used a different size, so that hte scaler could yield
12:40:09Tornebut then it came back for some kinds of jpg, which should've been yielding as well
12:43:11gevaertsyes, we need to know the exact revision that changes things
12:46:20Tornei've asked them to try it
12:52:16 Join _zic [0] (
12:54:27***Saving seen data "./dancer.seen"
12:54:39S_a_i_n_tin regards to AA...does using bmp over jpg (or vice versa) yeild any benefits?
12:54:55AlexPbmp is easier to decode
12:55:02AlexPbut takes more diskspace
12:55:31Bagderalso bmp is lossless, jpg is not
12:55:36AlexPunhelpful may say something about scaling at this point :) jpg is less to buffer? or just faster?
12:55:56AlexPbmp is easier to decode
12:56:03S_a_i_n_tooops, yes, slower.
12:57:07 Quit m3dlg (Ping timeout: 240 seconds)
12:57:41Tornejpg can be faster for some cases: if your album art is very big
12:57:54Tornethe jpg pre-scale by 8/16/whateveritis saves a lot of cycles that the bmp loader has to spend
12:58:25 Quit ThomasAH (Ping timeout: 248 seconds)
12:58:52 Join S_a_i_n_t_ [0] (S_a_i_n_t@
12:59:28TheSevenwhat about PNG btw?
13:00:28 Quit MethoS- (Remote host closed the connection)
13:00:29TheSevenis it supported? and if not: what would be the approximate binsize/ramsize hit in order to do so?
13:00:47 Quit S_a_i_n_t (Ping timeout: 256 seconds)
13:01:10 Join MethoS- [0] (~clemens@
13:01:14 Join archivator [0] (
13:03:00archivatorNow that fft has been committed, can someone please take a look at FS #10816? It's making more difficult than they should be..
13:05:39 Quit karashata (Quit: The fluffy dragon has left entirely!)
13:05:51 Join ThomasAH [0] (
13:13:58TheSevenare there any targets with a 2048 or bigger sector size, besides nano2g?
13:14:54 Join perfectdrug__ [0] (
13:15:41S_a_i_n_t_TheSeven: Isn;t it just Nano2g specifically that has 2048? besides "monster builds" etc. ?
13:16:03TheSevenIIRC there were some other non-512 targets at least
13:16:28S_a_i_n_t_Hmmmm, probably, but I can;t recall offhand.
13:16:34TheSevenbut those use some weird trick (remember that certain 80gb ipod video drive)
13:18:45 Quit perfectdrug_ (Ping timeout: 258 seconds)
13:18:47S_a_i_n_t_TheSeven: are you talking about bob77? he has an 80GB video with 4096, and can't seem to get a build running on it.
13:18:57 Join teru [0] (
13:19:25TheSevenno, iirc there was a 1024-byte-sector drive that's commonly used in some ipods
13:20:45S_a_i_n_t_not sure sorry...not to my knowledge (which is however quite limited in this area).
13:28:49TheSevenhm, if he wants to use a filesystem that also has 4096-byte sectors, and not that 512-byte emulation, he should be able to use the same method I used on the nano2g
13:29:10TheSevenhe would probably run in a lot of STKOVs (like the dircache one) though
13:29:56 Quit archivator (Remote host closed the connection)
13:31:45bob77TheSeven: I managed to format with fat32format with 4096.. nugged's unsupported port can run ok now
13:32:17bob77Select still miss once or twice, with apple's firmware all fine though
13:33:25 Join GeekShadow [0] (~Antoine@reactos/tester/GeekShadow)
13:43:39GodEaterTheSeven: the ipod 80GB is a funny beast. It uses 1024 bytes sectors on device, but exposes the drive as 2048 byte over USB.
13:44:44bob77GodEater: what sector size you recommend using?
13:45:08GodEaterI don't understand the question
13:45:53 Join DerPapst [0] (
13:46:05bob77currently I formatted by 80GB with 4096 sector size, should I use 1024?
13:46:24 Join Schmogel [0] (
13:46:29GodEaterthis is with the original drive in it?
13:46:54bob77I'm not sure, probably not
13:47:45GodEaterfrankly, I'm surprised you even have an option to format it with a size the drive doesn't support
13:48:06Tornethe sector size in the FAT is kinda irrelevant, though
13:48:20Tornewell, as long as it's big enough
13:48:47Tornemaking it bigger than it is just makes sectors per cluster lower, or makes cluster size bigger.
13:49:36 Quit killan (Ping timeout: 264 seconds)
13:50:14 Join Buschel [0] (
13:50:21S_a_i_n_t_it should let you select *any* sector size, but format will fail if the drive can't support it.
13:50:46S_a_i_n_t_it *should* just blank the unsupported ones out, but doesn;t.
13:50:46Torneit will fail if it's too small, yes
13:50:49Tornebut not if it's too big
13:50:56TorneATA doesn't work that way
13:51:01TorneATA commands are always 512 byte sectors
13:51:33Torneif the OS thinks the sector size is 1024 it just happens to only ever do reads/writes that start at an even sector number and contain an even number of sectors
13:52:39 Join Riku [0] (
13:53:08 Nick Riku is now known as Guest94132 (
13:53:14S_a_i_n_t_from what I gathered from the description from ebay, bob77's iPod is actually a 30GB video, with a small form factor 80GB HD in it.
13:53:57GodEaterso not the same drive as in the original 80GB at all then
13:54:06S_a_i_n_t_whatever combination it actually is...RB doesn;t seem to like it.
13:54:07GodEaterand therefore less likely to be as fundamentally broken
13:54:46S_a_i_n_t_well, it was called the 80GB *slim* (never heard of it either)
13:54:56*bob77 would like to upgrade to 30GB
13:55:00S_a_i_n_t_and boasted being the same size as the 30GB
13:55:08S_a_i_n_t_so I figure that's what was done.
13:55:49S_a_i_n_t_bob77: *upgrade* to 30GB? isn;t that a DOWNgrade?
13:56:30TheSevenupgrade in functionality :-)
13:56:35linuxstbbob77: I think it depends on whether you have the "5g" or "5.5g" ipod. The difference is that the apple firmware uses a 2048-byte sector size on the 5.5g. ipodpatcher should report the logical sector size used by the Apple firmware.
13:57:22bob77I check on apple, my serial ends with V9R so 5.5
13:58:03linuxstbI wouldn't always trust that, especially with refurbished models. But then it sounds like you should use a 2048-byte logical sector size, which Rockbox should be happy with.
13:58:07bob77For some reason I reckon this is 80GB mobo, in a slim case, and slim hdd
13:58:24*GodEater is willing to bet it isn't
13:58:24S_a_i_n_t_*this* is why refurbs can suck ass sometimes....
13:58:25TheSeven*panic* unsupported physical sector: 4096
13:58:30TheSevenwhere does rockbox get that value from?
13:58:30bob77linuxstb: okay, lemme try to use 2048
13:58:36TorneTheSeven: Should be from ATA IDENTIFY
13:58:39Tornei.e. from the disk itself
13:58:50TheSevenand what is it that it doesn't like about it?
13:59:00Torneit's bigger than the value the ata driver is compiled for
13:59:00GodEaterTorne: which isn't to be trusted!
13:59:05S_a_i_n_t_wont boot
13:59:06TheSevenbecause it doesn't match the SECTOR_SIZE define?
13:59:19linuxstbBecause it doesn't match the MAX_SECTOR_SIZE define (IIUC)
13:59:40TorneMAX_SECTOR_SIZE needs to be 4096 or more, yes
13:59:45S_a_i_n_t_JdGordon tried to edit that for him, and sent him the builds...but it *still* didn't boot.
13:59:50Torneer, MAX_PHYS_SECTOR_SIZE
14:00:03Tornewith the same message?
14:00:11*TheSeven doesn't get what that whole emulation layer is for and why we can't just use 4096-byte sectors directly like we do with 2048 on nano2g
14:00:16Tornebecause that's totally impossible, right? :)
14:00:28S_a_i_n_t_not like I was there or anything
14:00:39Tornethe condition where it panics with that, it prints the same value that it compared to
14:00:48Torneso if MAX_PHYS_SECTOR_SIZE was 4096 that panic really can't happen
14:01:01Torneso I would ugess that JdGordon or the user made a mistake
14:01:06GodEaterTheSeven: that layer is there because our FAT code always assumed 512 bytes sectors
14:01:16GodEaterand no-one could be bothered to rewrite it to be more celver
14:01:23GodEaterit was easier to hack the ATA layer
14:01:46linuxstbTheSeven: One difference is that _every_ nano2g has 2048-byte sectors, and SECTOR_SIZE is a constant.
14:01:52S_a_i_n_t_weird thing is....*apparently* (again), the "big-disk" builds don;t work something is going wrong with someone's info somewhere.
14:02:04TorneThe big disk builds don't necessaraily have 4096 as the limit
14:02:11Tornethey migh tbe built for 1024 or 2048...
14:02:18GodEateror just for LBA48
14:02:22Tornebut yes. i'm not saying it will definately *work*
14:02:23GodEaterand not fiddle with the sector size at all
14:02:24 Join stoffel [0] (
14:02:28S_a_i_n_t_Um...he found a couple of unsupported build that do.
14:02:32Tornebut that specific panic message can *only* happen if MAX_PHYS_SECTOR_SIZE is too small
14:02:36bob77they do boot just flaky as
14:02:38Tornethere is no other way it can possibly occur
14:02:40S_a_i_n_t_both of them *actually* did specify 4096
14:02:59GodEaterS_a_i_n_t_: he downloaded the source for each to check?
14:03:16Tornebob77: wait, the large disk builds do boot?
14:03:19S_a_i_n_t_well..running off the .diff from the thread
14:04:07GodEaterdefine "flaky as" then please
14:04:29bob77I had to press select 3 4 times to register
14:05:05Tornethat sounds like the disk isn't big enough to prop up the logic board properly :)
14:05:12bob77and check again and again on apple firmware the hardwares fine
14:05:20*GodEater loves it when we're debugging the wrong plroblem
14:05:30Tornewell, then i would guess that the build you used is old/crappy :)
14:05:42Torneand that you should try the latest svn but with MAX_PHYS_SECTOR_SIZE changed to 4096
14:05:53Tornesince the unsupported builds posted there don't ge tupdated very much
14:05:56S_a_i_n_t_Torne: I had that idea too..but it failed when he said OF runs fine.
14:06:33S_a_i_n_t_*logic board unsupported/supported incorrectly that is.
14:07:39GodEaterI'd also go with a 30GB ipod build until you're certain the extra 32MB are there
14:07:51 Join panni_ [0] (
14:07:56bob77JdGordon tried patching latest build with MAX_PHYS_ and it still the same panic
14:08:20S_a_i_n_t_he did, yes...but for the 80GB
14:08:27S_a_i_n_t_I have a feeling its a 30
14:08:33GodEaterthat wouldn't cause that panic though
14:08:34Torneas I just said: if it panics on start with that message then the build does not hava MAX_PHYS_SECTOR_SIZE defined properly
14:08:38TorneEither you or he made a mistake
14:09:09Torneusing the 64mb build on a 32mb logic board doesn't cause problems until you start buffering or load a plugin/codec
14:09:54bob77atm I'm listening flacs with 64MB large-disk build
14:10:05S_a_i_n_t_I'd of built this guy a build to check out myself by now...but I'm reinstalling CygWin atm...and the mirror is *crawling* (slow as)
14:10:18Tornebob77: then it's a 64mb logic board.
14:10:37Torneotherwise that would definately not be working :)
14:10:46GodEaterwell, not for long anyway
14:10:48Torneso the only problem you have with that build is that the buttons don't respond reliably?
14:10:51bob77I reckon so
14:11:05bob77I reckon so
14:11:13Torne"yes" would do, but hey
14:11:17S_a_i_n_t_...and scrolling *apparently*
14:11:19Tornethen you need to try the latest svn
14:11:29Tornewith the appropriate sector size change :)
14:11:33S_a_i_n_t_you said scrolling sucked too right bob77?
14:11:38Torneeither JdGordon did it wrong, or you installed it wrong
14:11:45bob77sorry, a bit laggy here
14:11:54Tornei would build you one but my build machine isn't accessible atm
14:11:59linuxstbS_a_i_n_t_: Doesn't that come under "buttons don't respond reliably"
14:12:51S_a_i_n_t_probably, but with definitions like "flaky as", it pays to check ;D
14:13:11*GodEater is building a MAX_PHYS = 4096 build right
14:13:17GodEaterwill upload somewhere when it's done
14:13:34S_a_i_n_t_well..this should be definitive...
14:18:22GodEaterbloody hell, can't have done this for a while
14:18:26GodEaterccache is hella outta date">
14:20:16Torneand the bootloader?
14:20:24GodEatereasy tiger ;)">
14:22:25*bob77 on it
14:25:21 Join CIA-8 [0] (cia@
14:26:41S_a_i_n_t_yay! cia is alive again...
14:28:13bob77HEY HEY HEY..... it's booting ....!! looking good
14:36:48 Quit shai (Quit: Leaving)
14:44:39bob77buttons still won't register straightaway, but flacs play fine
14:45:13 Join m3dlg [0] (~m3dlg@
14:47:36 Join killan [0] (
14:50:06Tornewell, the buttons are not related to the disk :)
14:50:11Torneexcept physically, as mentioned above
14:50:21Tornebut if it work sin the OF then that's *probably* not it. but who knows.
14:51:07bob77I'm gonna replace the buttons at a shop tomorrow, I suspect the seller used dodgy parts as well
14:53:34*bob77 would like to thank Torne, GodEater, and S_a_i_n_t_ .. sorry to cause so much trouble guys :)
14:54:03*GodEater thinks the button thing could probably be improved by turning on dircache
14:54:28***Saving seen data "./dancer.seen"
14:54:30bob77okay.. I'll be back :)
14:54:46 Quit bob77 (Quit: thanks!)
15:02:41 Join johnf1911 [0] (
15:02:51johnf1911I've got a lovely spot of rockbox trouble
15:02:52S_a_i_n_t_can some nice soul please build/upload me a nano1g SIM please? I need to debug a WPS, and my build environment is still down.
15:03:03 Quit _zic (Remote host closed the connection)
15:03:28johnf1911I'm set to auto resume playback on startup, the problem is that the file I'm playing is corrupt and crashes the player
15:03:34johnf1911so every startup I crash
15:03:49johnf1911any advice? anyone know what file is used to store the resume track?
15:06:42Torneor, er, something similar. in .rockbox
15:08:24Frampisjohnf1911, have you tried renaming or deleting the file?
15:08:47Frampisthe corrupt one
15:08:57johnf1911I'm not sure if the file is corrupt
15:09:04johnf1911or if it's just stuck in some strange state
15:09:10johnf1911it appears to have played past the end of the file
15:09:28GodEaterTorne: can't you also startup with the "clear config" settings method to stop this?
15:09:39GodEater(i.e. switch on hold if on an ipod etc.)
15:09:40Torneyes, but then it'll clear your config, no?
15:09:48GodEaterit'll still fix the problem!
15:10:06johnf1911hmm, GodEater; gibson reference?
15:10:14GodEaterjohnf1911: good catch
15:10:17Frampisjohnf1911, worth a shot I think, renaming I mean
15:10:34johnf1911I've read those books a few times :)
15:10:54GodEateror you could edit the config file to change the startup screen from "resume playback" to something else
15:11:11Torneseriously, delete the playlist control file in .rockbox :)
15:11:12Tornethat will do it
15:11:22johnf1911I'm working on it, I'm working on it
15:11:25 Quit m3dlg (Ping timeout: 246 seconds)
15:12:00 Join sim2409 [0] (~kvirc@
15:12:33johnf1911it's healed!
15:12:57 Join fml [0] (
15:13:38fmlJdGordon: hello. Have you seen my question about the buffer signedness? I wonder why the compiler doesn't complain there.
15:13:54 Quit GeekShadow (Read error: Connection reset by peer)
15:14:26johnf1911Torne: thanks a lot for your help
15:14:32johnf1911man, I love rockbox
15:14:48johnf1911my little H120 lasts forever
15:15:03johnf1911and when the hard drive finally failed, I replaced it with a larger disk
15:24:33 Join Sajber^ [0] (
15:29:39 Part LinusN
15:30:02johnf1911thanks again
15:30:05 Part johnf1911
15:33:34S_a_i_n_t_JdGordon: does multifont still work as %Fl|2|etc. etc.?
15:34:02*S_a_i_n_t_ is trying to figure out what's breaking his WPS...but without a SIM, it's a lot harder than I thought
15:36:38mc2739S_a_i_n_t_: do you still need a sim build?
15:37:05 Join Adnyxo [0] (
15:38:26 Quit S_a_i_n_t_ (Ping timeout: 252 seconds)
15:41:36 Join S_a_i_n_t [0] (S_a_i_n_t@
15:44:14 Quit Bagder (Quit: It is time to say moo)
15:47:07mc273915:36:38 < mc2739> S_a_i_n_t_: do you still need a sim build?
15:51:27S_a_i_n_teach cygwin mirror has dropped on me around 80% so far....and are all running *very* slow. otherwise I'd build one myself
15:56:15 Join biengo [0] (
15:58:09biengoHi, I'm stuck in an infinite loop with a theme upload
15:58:31biengofirst it asks to remove a font included in the rockbox install
15:58:55biengothen it murks that a specified font is missing
15:59:12Torneyou probably specified the font using a name that isn't the name in the rockbox font pack
15:59:46Tornewhat does your theme say for the font?
16:00:14biengofont: /.rockbox/fonts/14-Nimbus.fnt
16:01:48Tornehm, that should be ok.
16:02:01biengoI mean it did recognise that a font by that name should be included anyway,
16:02:04S_a_i_n_twell, that's in the font multifont breaking checkWPS?
16:04:27Torneaw, disappointingly the theme site only checks by filename
16:04:32biengoIt's in the default fonts, not the one in extras
16:05:01Tornerasher: can we get the theme site to detect themes that include fonts in the font pack even when they have been renamed? since that seems to happen quite a lot (Nimbus14 and so on)
16:10:47 Join watto [0] (~watto@
16:11:42biengoTorne: I'm not shure if you agree with me that this is not a naming issue.
16:11:50biengocheckwps doesn't seem to ignore fonts that are included anyway by default
16:12:02Tornei said it was okay. it was just an aside, unrelated to your problem
16:12:06 Quit teru (Quit: Quit)
16:12:11Torneand no, it's not checkwps which issues this error but the theme site code
16:12:15biengook, thanks
16:19:22 Quit sim2409 (Quit: KVIrc 3.2.0 'Realia')
16:21:51 Join jgarvey [0] (
16:26:44 Quit wind (Quit: CGI:IRC (EOF))
16:28:46*amiconn wonders why this MAX_PHYS_SECTOR_SIZE problem is so difficult to understand
16:29:01*amiconn would try to fix it if he had access to an 80GB G5.5
16:36:10 Join shai [0] (
16:37:13 Quit shai (Max SendQ exceeded)
16:37:54 Join shai [0] (
16:38:31 Quit shai (Read error: Connection reset by peer)
16:38:40 Join shai [0] (
16:39:42 Quit shai (Max SendQ exceeded)
16:40:05 Join shai [0] (
16:40:39 Quit shai (Read error: Connection reset by peer)
16:42:00 Join shai [0] (
16:42:19 Join m3dlg [0] (~m3dlg@
16:43:39 Join freddyb [0] (
16:44:24 Join toffe82 [0] (~chatzilla@
16:45:31 Quit freddyb (Client Quit)
16:46:13 Join freddyb [0] (
16:48:01 Quit m3dlg (Ping timeout: 246 seconds)
16:48:08 Join m3dlg [0] (~m3dlg@
16:48:16freddybJdGordon: I think FS #11008 is good to go.
16:49:14UnhelpfulTorne/AlexP/S_a_i_n_t: supported jpeg scalings are 1/8, 1/4, and 1/2. the jpeg loader will decode the file at the smallest size which is greater than or equal to the requested size, and the scaler will do the rest of the work. *Only* the imdct, some scaler work (due to having a smaller input size) and some small portion of entropy decode and dequantization are actually reduced.
16:51:44UnhelpfulThe huffman-coded image data must be decoded almost completely to skip the unneeded coefficients. JPEG uses huffman codes to encode lengths of coefficients, followed by the actual bits reperesenting the coefficient, which must be sign-extended and dequantized. sign-extension and dequantization are skipped, huffman decoding cannot be.
16:52:03Unhelpful..ugh, i've been listening to too much of the mdct_exp work. not imdct, idct. ;)
16:53:26Unhelpfulthe idct is significantly faster as it goes down in size - the 1/8 scaling consists of a right-shift.
16:53:30 Join ShadWeasel [0] (
16:53:36ShadWeaselHi all
16:54:32***Saving seen data "./dancer.seen"
16:55:39ShadWeaselJ'utilise RockBox dans mon iPod-Nano2g.
16:56:23Buschelanyone with an coldfire target willing to do a performance test for the atrac3 codec?
16:57:06ShadWeaselAnd I have not too many problems ..
16:57:35UnhelpfulShadWeasel: ..but there are some? :)
16:57:53 Quit m3dlg (Ping timeout: 252 seconds)
16:58:26ShadWeaselBut I want the "Alarm"...
16:58:41 Join m3dlg [0] (~m3dlg@
16:59:12Buschelmt, saratoga2: I could speed up atracs gainCompensateAndOverlap() by nearly a factor of 2 without asm'ing. This saves another 1.2 MHz on ARM.
16:59:50ShadWeaselCan you help me ?
16:59:54*Buschel is interested in the saving on coldfire targets
17:00:20 Quit m3dlg (Remote host closed the connection)
17:00:20linuxstbShadWeasel: The alarm feature hasn't been implemented on the nano2g yet.
17:00:48 Join m3dlg [0] (~m3dlg@
17:01:21ShadWeaselAh ok thank you very much ...
17:01:53 Quit fml (Quit: CGI:IRC)
17:07:37 Quit m3dlg (Ping timeout: 258 seconds)
17:09:34 Quit ShadWeasel (Quit: CGI:IRC (EOF))
17:10:58 Join LambdaCalculus37 [0] (~rmenes@rockbox/staff/LambdaCalculus37)
17:11:51CIA-8New commit by Buschel (r24668): Further optimization of atrac3 codec. Refacturate gainCompensateAndOverlap(), avoid multiplication if not needed, unroll loops. Speeds up codec by 1.1 ...
17:12:24 Join kugel [0] (~kugel@rockbox/developer/kugel)
17:14:20Buschelmt, saratoga2: atrac3 profiling on arm -> total=62.7MHz, iqmf=43.9MHz, overlap_add=1.6MHz, imdtc=11.6MHz, other=5.6MHz (of that inverse quantization of spectrum needs 1.5MHz). In total 2.3 MHz less than last night :o)
17:20:07AlexPTorne: ...
17:20:12AlexPSome people...
17:20:50perfectdrug__Buschel: i can test with coldfire if you provide the files
17:21:34TorneARGH JESUS FUCK
17:21:38Torneoops sorry
17:21:45Tornethought this was -community :)
17:21:51Tornecan someone *else* try and explain it to that guy?
17:23:10AlexPi thought about it, but I'm not sure how
17:23:44pixelmacalling it a fix might be debateable </semi-serious>
17:24:53S_a_i_n_ti could *try*...
17:25:08kugelAlexP, Torne: :'(
17:25:16*S_a_i_n_t thinks how to word this...
17:26:12AlexPI tried to answer
17:26:18Unhelpfulso, the "problem" is that you see a low battery icon on shutdown?
17:27:07TorneHe doesn't understand that the low battery icon is a side effect of a fix for a *totally unrelated problem he knows nothing baout*
17:27:16Tornei have tried to explain this multiple times to no effect
17:27:27TorneHe thinks the low battery icon is the problem being discussed in the thread
17:27:37 Join ecio [0] (
17:27:43TorneMaybe. I don't know.
17:27:53 Join Farthen [0] (
17:28:01 Join pamaury [0] (
17:28:26S_a_i_n_the fails to see it as a sideeffect of shutdown, but sees it as the cause of the problem
17:28:29S_a_i_n_t*the icon
17:28:43linuxstbAlexP: Shouldn't "Some people suffer from a very annoying start up problem that means that it is difficult to turn on their ipod." be in the past tense (i.e. suffered) ?
17:28:48kugelTorne: I think he doesn't realize that the shutdown effect didn't affect all users and that the side effect affects all
17:29:16AlexPkugel: I've tried to spell that out
17:29:25AlexPlinuxstb: yeah, changed
17:29:33kugelI noticed :)
17:29:41*S_a_i_n_t doesn't get any icon on shutdown FWIW
17:29:56 Quit Zagor (Quit: Clint excited)
17:29:59 Join liar [0] (
17:30:04TorneS_a_i_n_t: you probably do
17:30:05linuxstbAlexP: Maybe fix the tense of the whole sentence? ;)
17:30:09Torneit's just very hard to notice on some of the LCDs
17:30:23Torneafaik it does show up on all of the affected models.
17:30:27AlexPlinuxstb: picky :)
17:30:42kugelcan't the lcd just be turned off or is (parts of) the apple system turning it on again?
17:31:35S_a_i_n_tI've tried and tried and tried to see it, even asked the GF if she can (as my eyes suck), I've seen it before (in its actual usage) so I know what it looks like. but on Nano 1 and 2g ( I have 4, 2 of both) I can't see any icon on/during sghutdown
17:32:04kugelbacklight isn't enough AFAIK as some ipods don't have it and the others have rather transflective displays
17:32:37S_a_i_n_teven tilting the screen, to see the contrast difference if backlight is off by that time, still no icon.
17:32:53S_a_i_n_tyet everyone else complains about it (some more than most :P)
17:33:15kugelTorne: ^?
17:34:22Tornekugel: I haven't experimented, actually
17:34:25*kugel needs to try out JdGordon's patch
17:34:27Tornekugel: it might be that that would work..
17:34:54Tornekugel: i didn't make any change to the code, i just committed what dreamlayers wrot e:)
17:35:19kugelIMO the icon should be avoided if it's avoidable
17:36:55 Quit Farthen (Ping timeout: 272 seconds)
17:37:02Buschelperfectdrug__: hi, here are some sample files -> please test with FUN_RM_132.RM and mc_sich_ra8_264.rm
17:37:18Buschelperfectdrug__: just use plain svn, the patch is committed
17:37:34Tornekugel: well, i can try
17:37:37Torneat some point.
17:37:37Buschelperfectdrug__: what kind of target do you use?
17:37:38kugelBuschel: can you make files from the "official" test track and hand it to one of the swedes?
17:37:40Torneor someone else can :)
17:38:22Buschelkugel: I'll do my very best :o)
17:38:45 Join Farthen [0] (
17:39:39perfectdrug__Buschel: i updated my svn yesterday so i can make a test from before this commit as well so that you can compare. iriver h300
17:40:17Buschelperfect :)
17:42:07perfectdrug__which detail of the test codec result does actually matter?
17:42:21Buschel% and MHz
17:44:50kugelBuschel: in applyVariableGain(): is the difference between start and end always a multiple of 8?
17:45:03kugelah, the comment before tells it
17:45:05kugelsorry :)
17:46:30kugelsadly this unrolling makes code ugly :( I wish there was a preprocessor mechanism to duplicate lines X times
17:47:10 Quit stoffel (Ping timeout: 246 seconds)
17:47:28Unhelpfulkugel: it would be nice for LUTs :)
17:48:04*Buschel is preparing the next (and final?) unrolling patch for atrac3
17:48:37*kugel wonders if anyone actually proposed such a thing. I believe the committee might actually accept it
17:51:20 Quit robin0800 (Remote host closed the connection)
17:51:26kugelJdGordon: are there known problems with your fms patch? I would like to give it a test
17:52:13JdGordonumm, not that i can tihnk of off the top of my head
17:52:32JdGordonnot on swcodec anyway
17:54:34BuschelI cannot find any free atrac3 encoder... i guess that is why there are no sample files for rockbox available
17:55:45linuxstbBuschel: What about realproducer?
17:56:34Buschellinuxstb: checking...
17:56:37perfectdrug__Buschel: r24662M-100215: 103,65% realtime, 119,81MhZ, current svn rev24668: 108,27% realtime, 114,69MhZ
17:57:43 Quit petur (Quit: work->home)
17:58:30kugelnot bad :)
17:58:40Buschelperfectdrug__: :o) sounds like realtime
17:59:33kugelI think 120%+ is considered realtime
17:59:50perfectdrug__yes but the testfile sounds like ugly noise:D
18:00:58kugelperfectdrug__: on both revision? on a pc too?
18:01:30Buschelperfectdrug__: ? with what rev ?
18:02:09 Join komputes [0] (~komputes@ubuntu/member/komputes)
18:02:37perfectdrug__oh i thought it was supposed to be that way:/
18:02:44perfectdrug__on both revisions
18:03:58perfectdrug__just tested on the pc and i see that it is not that way there, sounds like really ugly noise on the h300
18:04:21Buschelcan you check with which revision this was introduced?
18:04:26JdGordonkugel: oh, apparently %C break a fms
18:04:32JdGordonit shouldn't though
18:04:47 Quit Farthen (Ping timeout: 272 seconds)
18:04:53perfectdrug__Buschel: I can but i have to learn how to check out older revisions
18:05:26kugelJdGordon: shouldn't the playlist tags indicate the current and max presets?
18:06:44Buschelperfectdrug__: To go back to a specific revision, use -> svn update -r 12345
18:06:47kugelmaybe not, but it sounds natural to me
18:07:05perfectdrug__ok will do
18:07:15JdGordonwe don't want the tags to become too confusing with doubleups
18:07:18Buschelperfectdrug__: thanks. I am puzzled
18:07:47kugelalright, I'll just fix my sbs then :)
18:08:33kugelit won't work with 3.5 anymore then though
18:12:02Buschelperfectdrug__: if playback sounds weird with r24622 the next interesting change came in with r24648. does this error also happen with other audio files or with atrac3 only?
18:12:22JdGordonwhat wont work with 3.5 anymore?
18:12:54perfectdrug__Buschel: never heard noise like that before, i try 24641 atm
18:13:14 Join tomers [0] (
18:14:48 Nick perfectdrug__ is now known as perfectdrug (
18:15:00Buschelperfectdrug__: the changed c-version of the function works fine on my arm target...
18:15:30Buschelwill leave for a while (need to spend some time with my son now). will be back later to check what is going on
18:16:23 Quit einhirn (Quit: Miranda IM! Smaller, Faster, Easier.
18:19:39 Quit Schmogel (Read error: Connection reset by peer)
18:20:44 Join mt [0] (~mtee@rockbox/developer/mt)
18:21:37Buschelperfectdrug: if you any results just keep posting it here
18:21:51Buschel*have *them
18:22:01kugelJdGordon: can I specify a backdrop if yes: will it take additional space if it's the same as the wps one?
18:22:37kugelJdGordon: the theme won't work in the 3.5 anymore (more specifically the sbs)
18:22:50kugelunless I can do better and move parts to the fms
18:22:52 Quit kaniini (Read error: Connection reset by peer)
18:23:52 Join kaniini [0] (
18:24:20 Quit notByan (Read error: Operation timed out)
18:24:27 Join notByan [0] (
18:24:59 Quit thegeek (Ping timeout: 265 seconds)
18:25:22 Join thegeek [0] (
18:25:44mtBuschel: I noticed you have committed the patch, do you still need the wav comparison ?
18:26:59 Join GeekShadow [0] (~Antoine@reactos/tester/GeekShadow)
18:28:46kugel"The file /.rockbox/fonts/12-Adobe-Helvetica.fnt from the setting 'font' doesnt exist." :'(
18:28:48jobecit does seem that some clips have some problems with the radio. after explicitly cycling it on and off, 1 got 1.5h longer runtime than earlier (though still less than half of OF)
18:30:01 Join Strife89DS [0] (
18:31:26 Quit biengo (Remote host closed the connection)
18:31:53Buschelmt: no
18:32:01Buschelbut thanks for asking :)
18:32:13 Join bertrik [0] (
18:32:19kugelthe theme site is broken :(
18:32:27 Join JdGordon1 [0] (
18:33:11kugelmaybe think the .cfg checking was committed too early
18:33:20kugelthere's also no exception for the default icon set
18:33:56JdGordon1that just needs to b added
18:34:07JdGordon1its in the admin stuff
18:34:15AlexPkugel: I vaguely remember someone saying something like it refuses a zip with a font that comes in the font pack inside, but it refuses a zip if it doesn't contain the font specified in the theme?
18:34:26 Quit mt (Ping timeout: 240 seconds)
18:34:27AlexPNot sure if that is true or not though
18:34:30kugelit does both currently
18:34:36Tornesomeone was complaining about that earlier, yes :)
18:34:43AlexPkugel: That's what I meant
18:34:45kugelthe fontpack exception doesn't work
18:36:32 Join TheSphinX^ [0] (
18:38:03 Join CPICH_ [0] (
18:38:03 Quit ecio (Read error: Connection reset by peer)
18:38:33 Join m3dlg [0] (~m3dlg@
18:38:48 Join rawhed [0] (
18:39:20 Quit Strife89DS (Quit: ClIRC - IRC client for Nintendo DS)
18:39:34rawhedwhats up :)
18:39:44Adnyxonot much
18:39:55 Join mt [0] (~mtee@rockbox/developer/mt)
18:40:27kugelwould anyone mind reverting? I'd like to update my theme but I cannot fix it (or how to test a possible fix)=
18:40:51 Quit rawhed (Client Quit)
18:42:12JdGordon1you cant make the changes live anyway can you?
18:42:30kugelarg, I don't think I can
18:42:54gevaertsjobec: that might point to OF version being the difference I guess
18:44:24Buschelgevaerts: would you mind checking whether current svn plays atrac3 files correctly on your h300? perfectdrug reported that only noise is audible...
18:44:40JdGordon1kugel: anyway, I dont think it needs reverting, just a few minor fixes in
18:44:47kugelJdGordon1: saw my backdrop question?
18:44:49perfectdrugBuschel: false alarm, it works now maybe my .rockbox folder or svn was dodgy, test-codec was accurate though:)
18:44:57JdGordon1#comment: in .cfg apparently breaks it also
18:45:20jobecbertrik: i did the run and got slightly longer runtime
18:45:37perfectdruggevaerts: nevermind;)
18:45:40Buschelperfectdrug: good news :o)
18:45:54*gevaerts goes back to sleep :)
18:50:50kugelJdGordon: I find the casing of the new tokens confusing
18:51:41*JdGordon1 is open to suggestions for them
18:51:49JdGordon1and what backgorund question?
18:54:34***Saving seen data "./dancer.seen"
18:54:49kugelJdGordon: can I specify a backdrop if yes: will it take additional space if it's the same as the wps one?
18:55:22kugelfor the tags, is there a reason to have upper case ones at all? %T* is also used for sme touchscreen tags
18:55:25JdGordon1yes %X and yes
18:56:02JdGordon1there isnt much reason to any of the capitlisation outside of %I,F,D
18:57:03JdGordon1I dont want to have a backdrop mess between skins, but if the wps and sbs, or fms and sbs use the same backdrop then they should use the same buffer
18:57:08JdGordon1not wps and fms htouhg
18:57:45kugelit seems more likely to me that fms and wps share the same though
18:58:19JdGordon1yes, but thats not something I want to add
18:58:26kugelbecause of?
18:58:26JdGordon1especially when we dont currently save the filenames
18:58:56JdGordon1skins are seperate
18:59:10dionoeayou could compare on load and dump one of the buffers if they're identical
18:59:21dionoea(that would be quite ugly)
18:59:32JdGordon1and any->sbs should only work if both are using the setting
18:59:50JdGordon1i.e not loading one manually with %X
19:00:09JdGordon1depending how its done, I could have my mind changed because it does waste alot of RAM
19:01:46kugela hash of the filename would be sufficient I guess
19:02:33JdGordon1and a get_backdrop_(int* hash, char* buffer, int screen) for each skin user?
19:03:31JdGordon1we could add a backdrop manager like the font manager in skins, but thats just as bad
19:03:43kugelthat, or just store it within the skin engine
19:03:51 Quit mikroflops (Quit: "ERROR: Division by zero")
19:04:16kugelpreventing duplication of the backdrop isn't really something the skins should be concerned with, that can be handled by the engine imo
19:04:27JdGordon1I'd be more interested in resizable skin buffer on boot
19:04:38JdGordon1back in 10
19:05:06 Quit m3dlg (Ping timeout: 265 seconds)
19:06:19kugelJdGordon1: have_rds doesn't seem to be implemented in the skip_to_false_branch thing
19:07:49 Join domonoky [0] (~Domonoky@rockbox/developer/domonoky)
19:09:01 Join aaron__ [0] (
19:09:23 Quit JdGordon1 (Ping timeout: 272 seconds)
19:10:20 Join Horscht [0] (~Horscht2@xbmc/user/horscht)
19:11:23 Quit Adnyxo (Ping timeout: 265 seconds)
19:12:32Horschtis there an archive for older revisions?
19:12:42 Join JdGordon_ [0] (
19:13:01JdGordon_kugel: a posible gotcha with any backdriop manager... the sbs can change the backdrop on the fly, no other skins can
19:13:25JdGordon_which means if the sbs + wps both load the same filename with %X and then its changed manually, it would change both
19:14:11JdGordon_that might be something not worth worrying baout though
19:14:17krazykitHorscht, svn is the archive. you'd have to build older revisions yourself
19:14:38Horschthm... ok.
19:14:38pamauryNote to all and myself: changing all ecread->read and ecwrite->write, suppressing all endianness code in tagcache saves 1015 bytes on ondio sp (that's without even removing the ec* functions totally of the core because I think there are not use elsewhere)
19:16:05kugelmy fms fails to parse apparently
19:16:42*JdGordon_ tells pamaury not to bother
19:16:58 Quit shai (Quit: Leaving)
19:17:31kugelpamaury: there's little point in saving rombox
19:18:24kugelit *will* break in the sometime, unless we find something that makes a real difference. saving 1k (or even 5k) doesn't help at all
19:19:14pamaurykugel: true, but that's also to see what is the cost of supporting endianness correction in tagcache
19:19:16 Join JdGordon__ [0] (~Miranda@
19:19:27 Quit JdGordon_ (Quit: CGI:IRC)
19:20:03pamauryBut I also feel there is lots of redundancy in tagcache. The code could be cleaned probably but that's seems hard work
19:20:22kugelJdGordon: "fms: /path/to/skin.fms" is correct, isn'T it?
19:21:13JdGordon__. /.rockbox/wps/you.fms
19:21:22JdGordon__without the first .
19:22:03 Join shai [0] (
19:23:52saratoga2kugel: were you still thinking about doing GSOC this summer?
19:24:20kugelJdGordon__: how does %Tf work?
19:24:24kugelsaratoga2: sure
19:24:42JdGordon__whats Tf?
19:24:46saratoga2thought about a target?
19:24:53kugelJdGordon__: preset frequency
19:25:20kugelsaratoga2: not yet; but raap seems wanted and I was always interested in it
19:25:21JdGordon__%Tfx where x is the frequency offset number
19:25:30JdGordon__this was added befoer the playlist viewer remmever
19:25:40saratoga2did you see bagder's link above?
19:25:40kugelcan I use %Tn for the current preset?
19:25:51kugelsaratoga2: yea
19:26:02JdGordon__%Tf0 if the current one
19:26:26freddybJdGordon__: is ready if you want to look at it. Keymaps are in for everything that works.
19:26:30JdGordon__I need to redo those and just force ppeople to use the playlist viewer
19:26:51JdGordon__freddyb: ok, I'll commit tonight it i remember unless someone beat me to it
19:28:05kugelsaratoga2: so I would spend most of the gsoc money for devices the app runs on
19:28:12freddybI built all the targets and tried them out, so I feel pretty good about it. The only thing I can think of is maybe some not so obvious buttons on some players being swapped.
19:28:34 Quit aaron__ (Quit: Leaving)
19:28:45 Join Adnyxo [0] (
19:30:25freddybSome of the button labels don't show up very well on the UI background. I tried to build the manual for the ones I wasn't sure what to map them to but there were still a few I wasn't 100% sure about.
19:31:10 Join stoffel [0] (
19:31:40kugelthe radio in the sim apparently got cought in a endless loop after downloading a preset file
19:36:17kugelthe %cs conditional doesn't seem to function correctly
19:36:31domonokykugel: thats a expensive device to run Raap, if you spend nearly all of gsoc money for it :-)
19:36:53JdGordon__the %cs conditional really isnt good enough to know why skin stuff to be displaying
19:37:03kugeldomonoky: I meant multiple devices
19:37:37domonokykugel: ah, you think you can get it to multiple device in one Gsoc summer ?
19:37:40JdGordon__we need a simple tag (onlg the lines of %mp) with just "show nothing, show music stuff, show radio stuff, show rec stuff, etc"
19:37:41kugelJdGordon__: it should be fixed then. but it doesn't work well for lists when you want to show stuff of the last music screen
19:38:10kugelthat's what %cs should do
19:38:13S_a_i_n_tdo additional fonts have their own buffer, or do they steal from the skin/some other buffer?
19:38:23kugeldomonoky: no idea
19:38:38JdGordon__S_a_i_n_t: they steal from the skin buffer
19:38:47JdGordon__10Kb each
19:39:00S_a_i_n_tthat explains why I can only load 1 additional font
19:39:06S_a_i_n_tHUGE wps :P
19:39:17JdGordon__disable the backdrop
19:39:21kugeldepends on whether sdl is ported already (or on whether it's prefered to implement native calls over sdl calls instead of porting to another device)
19:39:23JdGordon__that will let you load 9 fonts
19:39:36S_a_i_n_tthat's a big part of the WPS though
19:39:41kugelJdGordon__: what's the difference between your tag and %cs?
19:39:47S_a_i_n_tthe bloat comes from animation
19:40:16JdGordon__%cs loses track of the last screen when you are in the menu doesnt it?
19:41:21domonokykugel: i think the biggest part of Raap is to convert the sim to the target directory, so you get SDL/etc targets.
19:41:57*JdGordon__ agrees
19:42:16JdGordon__remove the SIMUKLATOR dependancy for a sdl build, and then worry about building for a real target
19:42:23AlexPdomonoky: Have you seen the theme site brokeness?
19:42:39domonokyjup, give me some time to fix it
19:42:48AlexPcool, no worries :)
19:43:08kugelJdGordon__: yes but your tag would as well
19:43:35JdGordon__no it wouldnt
19:43:49kugelhow would "show nothing, show music stuff, show radio stuff, show rec stuff, etc" solve it?
19:43:54perfectdrugshouldn't it be raaap?
19:44:14domonokyRaaAp :-)
19:44:40JdGordon__kugel: ? it would have logic in it to make sure it didnt lose track... or be some mashup between current_screen() and audio_status()
19:44:51perfectdruggangsta raaap, yeah
19:44:52 Nick CPICH_ is now known as ecio (
19:46:18B4gderas I write, lots of raaa is also things like threading and windows resizing etc
19:46:36B4gderand to get "hooked" in correctly into frameworks
19:47:13JdGordon__do we need to worry about window resizing though?
19:47:23JdGordon__stick that in the too hard basket for the time being
19:47:27B4gderwell, android runs on many different screen sizes for example
19:47:36B4gderyeah, as a start it can be fixed
19:47:53*gevaerts refuses to focus on Android!
19:48:10B4gderright, all maemo devices have the same size ;-)
19:48:24JdGordon__all android devices did too at the start!
19:48:48gevaertsI don't think we need runtime resizing
19:48:56amiconnBuschel: For these two extra bits - wouldn't it be better to have a fixmul29(). Or would this extra precision be overkill?
19:49:05B4gderwell, an app on a desktop OS might want resize
19:49:38JdGordon__I think the biggest step for RaaA is to just get a !SIMULATOR sdl build going on lunix
19:49:41gevaertsyes, but I think that that's a different market
19:50:13B4gderyeah, and less interesting - to me
19:50:28JdGordon__well yeah, but baby steps...
19:50:50*JdGordon__ wants rockbox on android also
19:50:57*domonoky wonders about this themesite cfg bug. i am sure i tested it with a theme which has comments in the .cfg
19:51:19B4gderand of course, no matter what target gets first hit, the initial raaa work will benefit lots of other potential raaa targets
19:51:28domonokybut maybe it chokes if the comment sign isnt the first char of a line ?
19:51:38JdGordon__he said it was
19:51:53JdGordon__did your test one do "#comment: something" ?
19:52:17JdGordon__did you see the patch I did to maybe fix the regex
19:53:08domonokyyes i saw the patch.. just looking at the changes.
19:54:48linuxstbB4gder: Android isn't just java nowadays then?
19:55:02B4gderno, you can build native stuff
19:55:16B4gderbut there needs to be some kind of frontend in java
19:55:25linuxstbAnd phone companies let you install native stuff?
19:55:30B4gderyes sure
19:56:24 Join efyx_ [0] (
19:56:32B4gderthe only downside is that apps get arch-specific, but that's not a real problem for us at least
19:57:16domonokyJdGordon__: i tested with this config:
19:57:28JdGordon__linuxstb: there is a doom port for android, so we shuold be able to do it also
19:59:01JdGordon__all those comments fail the regex, #comment out one of the setting lines
19:59:31JdGordon__hmm.. Last Update should match actually
20:00:54 Join mikroflops [0] (
20:04:58domonokyJdGordon__: i just tried with a setting commented out, and it works fine.
20:05:27domonokyso unless i get a test config which shows the faulty behavior, i am not gonna fix it.
20:05:27JdGordon__ok, blame the user then :)
20:05:49 Quit stoffel (Read error: Connection reset by peer)
20:05:50JdGordon__maybe its better to go with the regex anyway because its fine to have whitespace before the #
20:06:45domonokyit fails if it has whitespaces before the comment sign.
20:07:39 Quit tmzt (Ping timeout: 272 seconds)
20:07:44 Join Barahir_ [0] (
20:07:56JdGordon__IIRC he said he used notepad which means \r\n endings which doesnt really make sense that it would cause problems
20:10:06domonokyyour regexp patch doesnt seem to be correct, it gives me a empty $name.
20:10:24JdGordon__I didnt test it :p
20:10:57 Quit Barahir (Ping timeout: 248 seconds)
20:13:28kugeldomonoky: can you also make an exception for the icon set that comes with cabbie (is that even wanted?)
20:13:36kugelI never included those icons in my themes
20:14:13domonokyrashers fix is also not correct.. i will let it just ignore all lines which contain #
20:14:25rasherMy fix for what?
20:14:32JdGordon__thats probalby not a good fix
20:14:46kugelare inline comments allowed in .cfg files?
20:15:07JdGordon__you'd have to check the settings.c file, but i dont tihnk so
20:15:16JdGordon__but its possible # could be a legal value
20:15:21JdGordon__or part of a legal value
20:16:30LambdaCalculus37Having trouble building the rockbox toolchain for some reason on my Xserve:
20:16:44LambdaCalculus37PPC64 CPU running OS X 10.5.8.
20:16:56domonokythen someone please give me a proper regexp. or i will just declare # as a invalud config value :-)
20:17:03kugelJdGordon__: I'll tell you how my fms works, it seems boring so far without any kind of fm-AA though :)
20:17:04JdGordon__LambdaCalculus37: broken url
20:17:48 Quit amiconn (Disconnected by services)
20:17:50 Join amiconn_ [0] (quassel@rockbox/developer/amiconn)
20:18:02domonokyrasher: your patch for the comments in cfg file on the themepage.
20:18:14 Nick amiconn_ is now known as amiconn (quassel@rockbox/developer/amiconn)
20:18:38rasherI have a patch for that?
20:18:39LambdaCalculus37JdGordon__ Oops, sorry.
20:19:29 Join pixelma_ [0] (quassel@rockbox/staff/pixelma)
20:19:29 Quit pixelma (Disconnected by services)
20:19:46 Nick pixelma_ is now known as pixelma (quassel@rockbox/staff/pixelma)
20:19:47domonokyrasher: sorry, i thought the patch was from you, but its only the bug which is assigned to you (and me). :-)
20:20:00 Join phanboy4 [0] (
20:20:18*JdGordon__ kicks LambdaCalculus37 into ##gcc
20:22:33 Join anewuser [0] (anewuser@unaffiliated/anewuser)
20:22:36saratoga2kugel: picking a device with a good emulator would make the RAAP work a lot easier
20:22:51saratoga2and then working on a hardware later
20:22:51 Join tmzt [0] (~tmzt@
20:23:09saratoga2plus other people are more likely to help you if they can do it without hardware
20:23:27kugelisn't there an andrioid emulator?
20:23:42B4gderthere is, yes
20:24:00gevaertsyou don't need an emulator. The architecture work can be done just as well under plain linux
20:24:00saratoga2IMO focusing on the android emulator as the offical target for GSOC makes a lot of sense
20:24:08saratoga2its linux based and freely available
20:24:20saratoga2but that assumes you want to go android
20:24:25 Join kugel_ [0] (
20:24:39 Quit kugel (Disconnected by services)
20:24:43 Nick kugel_ is now known as kugel (
20:24:48 Quit kugel (Changing host)
20:24:48 Join kugel [0] (~kugel@rockbox/developer/kugel)
20:24:48saratoga2there is also a very nice free emulator for Windows mobile, but when I tried the tool chain it was a lot of work to get it to do anything
20:24:55gevaertsyou only need a development environment and/or emulator for the final step of integrating it in the java environment
20:24:58saratoga2gcc targeting windows arm is not fun
20:25:10saratoga2i don't even think we need to worry about java
20:25:15kugelporting rockbox to msvc is easier I guess :)
20:25:15saratoga2theres SDL interfaces for these devices
20:25:25JdGordon__sdl+linux makes far more sense for the initial work
20:25:29domonokyJdGordon__: what should the \s in you regexp do ?
20:25:31saratoga2you should be able to just use SDL and let it worry about binding to the OS
20:25:33JdGordon__or even linux+qt or something
20:25:40JdGordon__domonoky: whitespace?
20:25:41 Join Schmogel [0] (
20:25:45kugelsaratoga2: as B4gder said, andriod makes most sense currently
20:25:45gevaertsI'm assuming that on android you need a java wrapper for everything
20:25:54saratoga2i don't think you do
20:26:02gevaertsmaybe not
20:26:09 Join CaptainKewl [0] (
20:26:27saratoga2i think sdl works by calling the underlying os system functions directly [because they're in c]
20:26:32saratoga2but i'm not certain
20:26:43gevaertsyes, but you need to start the app
20:27:02saratoga2the impression i get is that android works more like SWT then Swing on java
20:27:03gevaertsAnyway, that's a detail. If we go for sdl, 99% of the work doesn't need anything apart from a PC
20:27:14saratoga2lots of c libraries with java wrappers to make them look like they're java
20:27:22JdGordon__you use java to start the app.... I have no idea how you'd do things like threads and passing back to the os from JNI though
20:27:45kugelI need to go, but keep talking for the logs :)
20:27:49 Quit kugel (Remote host closed the connection)
20:29:27saratoga2looks like you can use the keyboard and 3d hardware entirely from in c
20:29:51CIA-8New commit by FlynDice (r24669): Sansa Clip+: Implement button driver. ...
20:30:54saratoga2"Both video and audio are supported, also Java wrapper code will download and unpack application data from net on the first run (its not bundled into .apk), so to port it you only have to ensure that your application supports 320x200, 320x240 or 320x430 screen resolution, 16 bits per pixel. And don't do audio/video init in static constructors - it should be done in main(). "
20:31:10saratoga2sounds like the wrappers we'd need already written
20:31:35JdGordon__what about allowing the app to go to the background?
20:34:22 Join grndslm [0] (
20:35:52saratoga2i'm not sure, but i don't see why not
20:36:04saratoga2regardless for an initial port i don't think that critical
20:36:14saratoga2once you get something workable we can explore how to do it better
20:36:34saratoga2probably removing SDL and using direct os libraries wherever possible for performance reasons will be a good idea in the long run
20:38:29*LambdaCalculus37 decides to play with the clipv2 build and see what it does
20:41:39saratoga2i think only the bootloader builds
20:41:45gevaertsmaybe. I'd expect SDL to be pretty fast on most platforms
20:42:08gevaertsAnd if it isn't, it might make more sense to optimise SDL
20:42:12gevaertsAnd if it isn't, it might make more sense to optimise SDL
20:43:02 Join manotroll [0] (
20:43:53LambdaCalculus37saratoga2: Yep, compiling the build failed for me.
20:44:15 Join petur [0] (~petur@rockbox/developer/petur)
20:45:19GodEaterlet me guess, you felt that was so important, it was worth mentioning twice?
20:45:36gevaertsmy keyboard did :)
20:45:48 Quit manotroll (Client Quit)
20:45:55 Join manotroll [0] (
20:47:13 Quit manotroll (Client Quit)
20:47:27GodEaterso, when you going to start running android on your n900 then gevaerts ? :)
20:47:49gevaertsprobably not today :)
20:47:53 Join Zagor [0] (
20:47:53 Quit Zagor (Changing host)
20:47:53 Join Zagor [0] (~bjst@rockbox/developer/Zagor)
20:48:42 Quit S_a_i_n_t (Quit: There are 10 types of people, those who understand binary, and those who don't.)
20:50:40CIA-8New commit by Domonoky (r24670): try to fix comments in cfg file validation.
20:50:58 Join MuscleNerd [0] (
20:52:31CIA-8New commit by tomers (r24671): Brickmania: Fix continue message for Cowon D2
20:54:33gevaertsisn't there a cabbiev2 wps for mr500?
20:54:37***Saving seen data "./dancer.seen"
20:57:13 Part watto
20:59:00saratoga2i guess we should try and get the word out more about rockbox GSOC
20:59:06saratoga2since last year we couldn't fill our slots
20:59:17saratoga2but i'm at a loss for were to recruit
21:00:18JdGordon__well we need to come up with a good list of ideas
21:00:20 Join webguest90 [0] (
21:00:29JdGordon__and a proper interview/test plan
21:01:22saratoga2we have a partial list of ideas on the GSOC2010 page, but more would be nice
21:01:43webguest90egiste um metodo de rodar o rockbor no sansa c200v2 ou sansa c240 v2 ? e possivel passar link com mas explicações?
21:01:43tomersMaybe we should have a presentation that combines Rockbox and GSoC. We already have the Rockbox presentation, and there is going to be a GSoC presentation (if there is not already one)
21:01:52webguest90egister a method of running rockbor in c200v2 sansa or sansa c240 v2? and possible link to go but explanations?
21:02:27domonokywebguest90: as the frontpage states, no rockbox
21:02:31domonokyon c200v2
21:03:16webguest90mas antes havia uma versão intavel disponivel onde foi parar a mesma
21:03:30domonokywebguest90: please speak english here.
21:03:46webguest90but there was a version available intavel where has the same
21:05:20saratoga2rockbox on the c200v2 doesn't really work
21:05:28domonokywebguest90: some work has started on c200v2, but its not ready. And nobody knows when it will be ready.
21:06:11tomerssaratoga2: A call for spreading the work about GSoC was sent in the GSoC discussion group:
21:06:30webguest90asim but it is not possible to release a beta?
21:06:43domonokymore ideas for our gsoc2010 page would be good. Also, maybe the students here could try to recrut other students at their uni.
21:07:05*JdGordon__ wants to put buflib on the list
21:07:07webguest90but even asin not possible and release a beta?
21:07:37domonokywebguest90: no, its not in a state where non-developers can play with it.
21:08:33domonokyJdGordon__: then do it, and describe the buflib ideas which were discussed here to it. :-)
21:08:40tomerscan we have a GSoC slot for some 'dirty chores' like working on the manual?
21:08:49webguest90ok will wait anxiously for a version for the sansa c200v2
21:09:24JdGordon__domonoky: yeah, I want to write up the points from the discussion... i sort of think its too much work for gsoc though (and dunno if i'd trust it to a new comer anyway....)
21:10:03saratoga2tomers: i don't think thats allowed in GSOC
21:10:07pamaurytomers: what job can you do on the manual during three months ?
21:10:21domonokyJdGordon__: you never know those newcomers :-) And its just a possible idea, students dont have to choose it.
21:10:24saratoga2students must code, not do documentation
21:10:27 Nick danopia_ is now known as danopia (~danopia@fullcirclemagazine/developer/danopia)
21:10:48saratoga2ok added another GSOC codec project
21:10:55domonokythere is a differnt google thingi for documentation i think.
21:10:57saratoga2i suppose an mpegplayer project would make sense too
21:11:03tomerspamaury: just the job of a technical writer... whatever this might involve - language, writing about plugins more, screenshot...
21:11:16domonokysaratoga2: more codecs for mpeg player ? sure
21:11:36saratoga2AC3 support would make sense, as well as doing .vob
21:11:42saratoga2some targets could probably play DVDs
21:11:44pamaurytomers: that's not enough work for a gsoc, is it ?
21:11:56saratoga2its not allowed for GSOC so no
21:11:58domonokytomers: gsoc projects should be clearly defined, so its possible to judge the students performance..
21:12:23tomersGSoC project: implementing USB controller driver! (for any target that we don't have it implemented already)
21:12:28JdGordon__redoing the radio code would be great. right now its a mess in recorder/radio.c with no nice api to the driver
21:12:32saratoga2anyone feel like putting a note on the front page about GSOC projects?
21:12:37tomersdomonoky: I agree
21:12:49saratoga2no sense waiting to get ideas and interest
21:13:03JdGordon__a project blog would be a good place for these ads/suggestions :p
21:13:07domonokytomers: USB-driver is a good gsoc project.
21:13:23saratoga2we need a blog thing for the front page
21:13:40gevaertstomers: is that enough work for gsoc?
21:13:51saratoga2but i have no idea how to make one since i'm not very good at the web two point oh
21:13:54pamaurythen usb driver+host ;)
21:13:57*tomers wonders how Electrical Engineering students might use a software controlled DAP for...
21:13:59gevaertsit depends I guess
21:14:08domonokybut usb-driver for new targets might be also risky, depending on available documentation.
21:14:19saratoga2we've tried that before and it didn't work so well
21:14:25*gevaerts decides that it would be enough work
21:14:25 Quit webguest90 (Quit: CGI:IRC (EOF))
21:14:27saratoga2imo hardware specific projects are bad idea
21:14:31gevaertsbut yes, risky
21:14:47JdGordon__usb host support :)
21:14:49saratoga2unless perhaps the student is exceptional
21:14:59tomersgevaerts: I think it is... It is very important to the users, and it can be easily measured. We might ask the student for implementing two different controllers, if that's not enough ;-)
21:15:30saratoga2who runs the forums anyway?
21:15:56saratoga2scorche i think?
21:16:09saratoga2maybe i can bug him about some kind of rudimentary blog software
21:16:11 Join Tomis2 [0] (~Tomis@
21:16:16 Quit Tomis (Ping timeout: 260 seconds)
21:16:16 Nick Tomis2 is now known as Tomis (~Tomis@
21:16:38tomersI was interested in getting xaos to be compiled as a Rockbox plugin (
21:16:51Buschelamiconn: I also thought about fixmul29(). But I prefer this dirty solution as this still is an obvious "todo". It also won't save to much cycles...
21:16:57*Buschel opens another Beck's
21:17:19domonokya note on the frontpage about searching ideas and students for a possible gsoc2010 might be a good idea. (also the frontpage needs the new Packard Bell Target mentioned)
21:17:37tomersmaybe that's a nice GSoC, having open source applications that use the GNU configure script be compiled for Rockbox - we need to have a Rockbox library file, and make the application compile as .rock file
21:19:05 Quit shaggy-h (Ping timeout: 240 seconds)
21:19:06domonokyhm, you normally can just compile a normal app to a plugin, it has to be adapted. Think about the memory constraints, the api mechanism, etc.
21:19:20JdGordon__can we also make each student work out of a branch in svn from the very start?
21:19:35domonokythat would be good.
21:19:48tomersdomonoky: the xaos application has compile flag that does avoids use of malloc
21:20:08saratoga2JdGordon__: sounds like a good idea
21:20:14domonokyalso the weekly reports to the mailing list, should be enforced directly from the start.
21:20:15tomersi think it doable for that specific application
21:20:16Tornegetting configure scripts to build against custom hosted environmens that aren't ANSI C is spectacularly difficult
21:20:34Tornefor specific apps you are more likely to succeed by arbitrarily writign your own config.h and just crossing your fingers
21:20:54CIA-8New commit by Buschel (r24672): Minor atrac3 codec optimization. Refacturate requantization of spectral lines, unroll loops.
21:21:09domonokytomers: true, it might work for specific apps, but it doesnt work generally. (also we already have a fractals plugin in rockbox)
21:21:12saratoga2how do you create a new news item on the front page?
21:21:21 Join audio-i [0] (
21:21:40tomersTorne: thanks for this info. I thought it is doable :-) So I guess for xaos I have to play a bit more with the code, and it is irrelevant for GSoC
21:21:41JdGordon__freddyb: I dont see the point in your latest patch. Why should the sim look for the backdrop in the root folder? it should look in the folder with rockboxui
21:22:10Tornetomers: yeah, just run configure on the closest platform you can find (i.e. same arch/compiler) and then edit config.h and the makefile when things don't work
21:22:22Torneit's more or less how you get python to crosscompile :)
21:22:33 Quit Tomis (Quit: Tomis)
21:22:37domonokysaratoga2: change index.t, the news items are hardcoded there.
21:22:47audio-iI'm trying to upload a theme to the site but I think the theme check is doing something wrong
21:23:29tomersif I'll try adding xaos as a new plugin, can I use svn:external to have the whole trunk of xaos synchronized? I guess it is not a good idea. It will make it highly coupled with the xaos project, and will not allow me to add Makefile files
21:23:30 Join shaggy-h [0] (
21:23:58Tornethat is a good question indeed
21:24:09audio-ithe theme uses 12-nimbus font, I don't include it in the zip, it throws "The file /.rockbox/fonts/12-Nimbus.fnt from the setting 'font' doesnt exist"
21:24:33domonokyaudio-i: when did you try ?
21:24:53tomersTorne: also, the xaos code repository contains some directories which will not be relevant for our target...
21:24:56audio-iI just did
21:25:07Torneyah. i had pondered how to sensibly build python as a rockbox plugin :)
21:25:15audio-iipod video, if it makes a difference
21:25:22*domonoky will take a look at the themepage code again..
21:25:35Tornefor pycorn I have makefile rules which get the tarball, unpack it, patch it, configure it with magic, patch it some more, and then build it by invoking *its* makefile
21:25:41Tornewhich is slightly nightmarish
21:25:44saratoga2"Rockbox will be applying for Google Summer of Code 2010. Please suggest possible projects, mentors or students."
21:25:46saratoga2sound ok?
21:26:08JdGordon__suggest where thoguh?
21:27:19saratoga2it links to the wiki
21:27:22domonokyperhaps mention irc ?
21:27:34JdGordon__FLAC embedded cue support and ogg subtrack support would be nice
21:28:01saratoga2are subtracks the same as chaining?
21:28:06 Quit Schmogel (Quit: Miranda IM! Smaller, Faster, Easier.
21:28:17JdGordon__I tihnk so, I could have the wrong terminology
21:29:19 Quit ecio (Read error: Connection reset by peer)
21:29:21 Quit TheSphinX^ (Remote host closed the connection)
21:29:33saratoga2chaining is basically not going to work with rockbox because its stupidly implemented
21:29:34JdGordon__forum integration with the themesite would be nice also, create a thread when themes are added for discussion
21:29:56saratoga2if subtracks have a header or something, that might work
21:30:11JdGordon__how stupidly implemented?
21:30:44GodEaterI don't think themesite integration with our forums is a valid rockbox SOC project
21:30:51GodEaterit smacks more of a SMF project
21:31:25JdGordon__well, not really... SMF wouldnt care about our custom sites
21:31:36gevaertsIs there a good reason for the sim to ask for a 24 bit window if rockbox uses 16 bit everywhere?
21:31:36saratoga2it just concatenates tracks together
21:31:41saratoga2no header or anything else
21:31:53saratoga2so you have to constantly watch to see if your file became a new file while decoding
21:32:00saratoga2thats problematic for devices with real malloc
21:32:09JdGordon__thats bloody stupid :p
21:33:31CIA-8New commit by Domonoky (r24673): try to fix the font exemption for fontpack fonts in cfg file checking.
21:34:10 Join Tomis [0] (~Tomis@
21:34:28saratoga2<tr><td nowrap>2010-02-15</td><td>Rockbox will be applying for Google Summer of Code 2010. Please <a href="">suggest</a> possible projects or <a href="http$="">ask</a> to be a mentor or students on our IRC channel.</td></tr>
21:34:42domonokyaudio-i: can you try again please ? and if it still doesnt work, please send my the theme zip for testing.
21:34:53 Quit shaggy-h ()
21:35:38 Quit BeFalou (Quit: leaving)
21:35:51JdGordon__domonoky: and you handle remote font also?
21:36:01saratoga2pulling trigger if no one minds
21:36:44domonokyJdGordon__: the exemption comes only from the file extention..
21:37:21CIA-8New commit by saratoga (r24674): Add mention of GSOC2010 to the front page.
21:37:25domonokyie any .fnt mentioned in the .cfg is allowed to be missing, if a fontpack font with the same name exists.
21:37:31saratoga2B4gder: can you update the front page?
21:37:41saratoga2Zagor: or you
21:37:55domonokysaratoga2: you could add the new target too :-)
21:38:11saratoga2whats it called?
21:38:33domonokyPackard Bell Vibe 500
21:39:08 Quit komputes (Ping timeout: 260 seconds)
21:39:14domonokyit should go atleast under unusable (but its probably very shortly before unstable, only automatic bins missing i think).
21:40:17Zagorsaratoga2: done
21:40:29saratoga2if its about to go under unstable, lets just wait
21:40:35saratoga2Zagor: thanks
21:41:13 Quit grndslm (Remote host closed the connection)
21:41:25saratoga2Zagor: if you get a chance could you change "students" to "student"
21:41:26freddybJdGordon: It only looks for the image in the same dir as rockboxui if you call it from that directory. ./rockboxui works. /something/rockboxui doesn't
21:41:32 Join froggyman [0] (
21:41:53Zagorsaratoga2: done
21:42:10CIA-8New commit by zagor (r24675): Spell fix
21:42:44saratoga2hopefully this year we wont' have to give back GSOC slots for lack of students
21:43:08JdGordon__for those that control the website.. has there been any thoughts on a project blog?
21:43:37gevaertsHas anyone thought about devcon recently?
21:44:31saratoga2JdGordon__: yes I asked Bagder about it and we agreed that we want one
21:44:51saratoga2the problem is that I don't know how to do it and we don't have anyone else with the time and motivation
21:45:03*JdGordon__ votes for .au around june :)
21:45:17 Quit audio-i (Quit: CGI:IRC (EOF))
21:45:19*tomers Wonders what to do with the guy from FS #10552 who wouldn't give his name :-) I would like to commit his patch
21:47:01 Join audio-i [0] (
21:47:28audio-iDomonoky: I didn't work, I'll send you the zip file
21:47:50freddybJdGordon__: Do you see what I mean?
21:48:16JdGordon__no, because I'm at work and cant play, but if that patch fixes it then cool
21:49:21saratoga2tomers: ask him for it and if he doesn't give it just close his task
21:50:02saratoga2skimming it that task looks trivial so you could just implement it yourself
21:50:04tomerssaratoga2: I already asked. I'm will wait several days for him to response. But can I commit the patch and close it?
21:50:16tomerssaratoga2: thought so.
21:50:44domonokytomers: you cant commit his patch without real name, but you are ofcourse free to reimplement it yourself.
21:51:14JdGordon__*cough* or say he gave you a real name in email and its Joe Hairy*cough*
21:51:57saratoga2its like 2 lines of code and immediately obvious from his description
21:52:06saratoga2you can safely reimplement that
21:52:32 Join robin0800 [0] (
21:52:35 Quit Zagor (Quit: Clint excited)
21:52:53 Join ecio [0] (
21:55:01 Join guest_10 [0] (
21:56:14 Quit audio-i (Quit: CGI:IRC)
22:00:25 Quit anewuser (Quit: =ooo ϢINTER ϾHIP 5iVE is OOON!!)
22:00:31 Quit tomers (Quit: ChatZilla 0.9.86 [Firefox 3.5.7/20100106054534])
22:01:57 Quit Xerion (Quit: )
22:03:03 Quit guest_10 (Read error: Connection reset by peer)
22:06:27 Join komputes [0] (~komputes@ubuntu/member/komputes)
22:07:11gevaertsI think that for RaaA, one of the first things to do is dump the SDL threads. They don't do priorities
22:09:31 Join stripwax [0] (
22:09:53bertrikjobec, nice, thanks for testing! that's nearly 1.5 hours saved :P but still a bit short compared to other benchmarks
22:11:22saratoga2so the radio isn't powered off correctly?
22:11:44 Quit phanboy4 (Read error: Connection reset by peer)
22:16:25 Join stripwax_ [0] (
22:17:42 Quit freddyb (Quit: Konversation terminated!)
22:17:49 Quit stripwax (Ping timeout: 256 seconds)
22:21:32 Join pyro_maniac_ [0] (
22:23:48 Join Bullet` [0] (
22:24:06 Join TheUser [0] (
22:24:08Bullet`Hi all,
22:24:17TheUserhi all
22:24:54Bullet`I have a problem, I build rockbox for h1x0 but I have an iriver h320
22:25:01 Join webguest89 [0] (
22:25:32Bullet`and now i have I04 : IllInstr at XXXXXXXXX
22:25:32 Quit webguest89 (Client Quit)
22:25:42Bullet`I don't know what to do
22:25:49JdGordon__Bullet`: so rebuild for h300 and reinstall it
22:25:56Bullet`i did it
22:26:03Bullet`but the problem is still there
22:26:10JdGordon__did you extract the whole zip? or just rockbox.iriver?
22:26:27Bullet`i tried purge the .rockbox dir but...
22:26:33linuxstbBullet`: Are you using the latest SVN? There was a recent bug causing that error on the irivers, which I think was fixed.
22:26:59Bullet`I tried to fix my problem with the current build on
22:27:20Bullet`so i don't understand why the bug is happening :/
22:27:43*Buschel needs a coldfire target tester again :)
22:27:57Buscheland again −− of course −− for atrac3
22:28:15*LambdaCalculus37 has a coldfire target but no atrac3 files
22:29:40Buschel-> (take FUN_RM_132.rm and/or the mc_sich_ra8_264.rm)
22:30:08Buschelbtw, what target do you have?
22:30:47linuxstbBullet`: You downloaded the latest h320 build from and that is giving you an IllInstr error?
22:30:50gevaertsBullet`: you didn't unpack the new .rockbox inside the old .rockbox or anything like that?
22:31:19JdGordon__Buschel: I can test this evening on h300 if noone else does (if I remember/am reminded)
22:31:22Bullet`I purged the old .rockbox
22:31:29Bullet`i'll tried one more
22:32:24CIA-8New commit by Domonoky (r24676): try to fix the fontpack fonts issue again.
22:33:57 Quit robin0800 (Ping timeout: 272 seconds)
22:34:27 Join freddyb [0] (
22:35:36 Quit dfkt (Quit: -= SysReset 2.53=- Ph'nglui mglw'nafh Cthulhu R'lyeh wgah'nagl fhtagn.)
22:35:56Unhelpfulwould it be hard to transcode the standard testfile to atrac3?
22:37:59linuxstbUnhelpful: No-one can find an atrac3 encoder...
22:38:46Unhelpfullinuxstb: didn't the first-gen sony DAPs *require* one? or can that only produce files with DRM, or only after verifying the player, or something?
22:39:25 Join shaggy-h [0] (
22:40:02linuxstbUnhelpful: I don't know, but doesn't Rockbox just support atrac3-in-rm, meaning we need a Real encoder?
22:40:26*linuxstb guesses the problem is that no-one wants to installl realplayer...
22:40:29Unhelpfullinuxstb: hrm. i wonder if ffmpeg could remux that?
22:40:41Unhelpfulhah. that is definitely *a* problem. ;)
22:45:22LambdaCalculus37Buschel: H340.
22:45:24stripwax_not necessarily free but a quick google turns up some sample files - has an atrac sample file (although oddly in a .wav container)
22:45:35BuschelLambdaCalculus37: that's perfect :o)
22:46:14mtstripwax_: atrac3 can be in wav too
22:46:18BuschelLambdaCalculus37: I just need a short test codec run with the result in % and MHz
22:46:27stripwax_mt - does that work on rockbox?
22:46:51mtNot yet ;)
22:47:05LambdaCalculus37Buschel: Let me build with test_codec enabled.
22:49:27 Join Xerion [0] (
22:49:30*Buschel is hoping to have some positive effect via dedicated iram usage
22:49:47LambdaCalculus37Oh crap, I forgot that I can't build m86k because of issues in building the m68k toolchain.
22:49:59LambdaCalculus37Buschel: Can you build me an H300 build with test_codec enabled?
22:50:05 Quit pamaury (Quit: abort();)
22:50:19LambdaCalculus37Thanks. :)
22:54:39***Saving seen data "./dancer.seen"
22:55:02 Join S_a_i_n_t [0] (S_a_i_n_t@
22:57:14 Quit ecio (Ping timeout: 258 seconds)
23:00:09BuschelLambdaCalculus37: does it work?
23:00:32LambdaCalculus37Rapidshare isn't allowing me to download right now. :/
23:01:25 Join ecio [0] (
23:01:27perfectdrugBuschel: I can try;)
23:01:48perfectdrugsame test file again?
23:01:53 Quit petur (Ping timeout: 240 seconds)
23:01:56Buschelperfectdrug: yes :)
23:02:41BuschelLambdaCalculus37: try this ->
23:03:04 Quit pyro_maniac_ (Remote host closed the connection)
23:04:09perfectdrug160,51% realtime, 77,36MhZ
23:04:16perfectdruggood work!:)
23:05:13*Buschel didn't expect _that_ much effect
23:05:29Bullet`gevaerts, linuxstb : i just fix my problem =) thx
23:05:31AlexPWhat was it before?
23:05:46Buschelperfectdrug: and no noise? ;-)
23:05:54gevaertsBullet`: what was wrong?
23:06:05gevaertsAlexP: just below 100%
23:06:16AlexPblimey, that is nice
23:06:17perfectdrugBuschel: no all fine:D
23:06:36Buschelso, this _finally_ is realtime :)
23:06:50*Unhelpful still wonders if the recip table hack for arm would be useful on CF for APE...
23:07:16LambdaCalculus37Buschel: Got it now.
23:07:39LambdaCalculus37I need to finish something first so I can free a USB port.
23:07:58Unhelpfuli know CF has hardware division but it still takes *many* cycles, and a couple of compares, a table lookup (from iram) and a few multiplies might still be faster
23:08:08Bullet`I had permissions problems between my iriver and my ubuntu, so my rockbox.iriver hasn't been erased and lots of others bug due to this (my ubuntu is a trash :/)
23:08:30gevaertsah, ok
23:16:58JdGordon__how would people feel about live lyrics in the wps?
23:18:06gevaertsI'd feel that it's clearly about time to start thinking seriously about visualisation plugins :)
23:18:19LambdaCalculus37Buschel: What tests would you like performed?
23:19:52JdGordon__gevaerts: well yes, the problem there is only allowing 1 of those plugins running at a time :(
23:20:02gevaertsJdGordon__: that's what needs thinking
23:20:39gevaertsI don't think that this is worth doing if you can only use one
23:20:39BuschelLambdaCalculus37: well, you were overtaken by perfectdrug. if you like you can of course re-test the speed of the atrac decoder :o)
23:20:39JdGordon__with the skin buffer the RAM usage is a non issue, its just having it keep track of the current line, which shouldnt actually be to difficult I think
23:21:23gevaertshm, how about wps visualisation in lua?
23:21:40JdGordon__thats where I tihnk we will go eventually
23:21:50gevaertsI'd imagine that lua doesn'
23:21:55LambdaCalculus37Buschel: Speed testing the folder now. :)
23:22:18gevaertst have fundamental issues with running two "plugins" in the same interpreter code
23:22:53gevaertsYou can't really do fft or oscilloscopes in lua though. That would be too inefficient
23:23:04JdGordon__is it possible to load functions from a .rock into a random location? so we could register something like a tag handler which comes from a .rock/dll but is not a full blown plugin?
23:24:07 Quit bluebroth3r (Ping timeout: 260 seconds)
23:24:22gevaertshm, maybe
23:24:32mtBuschel: Is that 160% for CF ??
23:24:41gevaertsOnly if it doesn't use globals though
23:24:43Buschelmt: yes :o)
23:24:48JdGordon__I'm not sure how that would work though, you'd still need globals of some sort
23:25:00gevaertsno, you can give it a pointer to some data
23:25:02mtBuschel: *Respect* !! :)
23:25:37gevaertsat least I think so
23:25:48 Join bluebrother [0] (
23:25:48 Quit bluebrother (Changing host)
23:25:48 Join bluebrother [0] (~dom@rockbox/developer/bluebrother)
23:25:53Buschelmt: that is more than a factor of 10 compared to the first implementation
23:25:54JdGordon__it would be nice we could somehow unload a chunk of the main binary for feastures that arnt used
23:26:02*gevaerts now expects one of the experts to hit him over the head with a ton of clue
23:26:21perfectdrugi would like lyrics in the wps, even if it would mean a lot of work to get the syncronized lyrics at the firt place
23:26:24domonokymaybe we should just design a different form of plugins for the wps which is relocatable
23:27:08 Quit martian67 (Ping timeout: 258 seconds)
23:27:13*Buschel fears that committing his atrac patch will result in lots of red (full iram)
23:27:26mtBuschel: Amazing ! :)
23:28:12domonokythis visualisation plugins also dont need to be limited to the wps, any themed screen could have some :-)
23:28:46*JdGordon__ is using wps as the generic skin name
23:28:56*JdGordon__ slaps self for this abuse
23:29:10 Join anewuser [0] (anewuser@unaffiliated/anewuser)
23:29:46gevaertsSo we'd be looking at visualisation *functions* (not full plugins) that can call the plugin API (or a restricted variant), can access one blob of data (passed as one argument), can draw to one viewport (another argument), and can *not* access globals or other functions, right?
23:30:19JdGordon__except it needs globals somehow for its local store?
23:30:30gevaertsthat's the blob of data
23:30:59JdGordon__no, it would probably need to be more full than that
23:31:09gevaertsin what way?
23:31:14JdGordon__it needs a parse/init function, update and cleanup
23:31:37domonokyusing a given struct for userdata is probably fine. but you probably need more functions.
23:31:42JdGordon__the init could allocate its needed memory from the skin buffer without problems
23:32:15gevaertsWhy? I'd promise to zero the data before the first call so it can know it should do some init stuff, and I don't see why you need cleanup
23:32:27gevaertsah, ok
23:32:35domonokyalso it should expose some info, like "i want to be update often"
23:32:52JdGordon__well init would need to figure out what %xx|d|53|gfdb.bat|234|7|4 means
23:33:29gevaertsSo you'd basically have a set of functions, for each of which those conditions apply
23:33:52*gevaerts first wants to hear from experts if this is possible
23:33:59gevaertsTorne? Anyone else?
23:34:18JdGordon__I tihnk the easiest way of doing this is having a single plugin which can handle as many of these as we need and just make it really easy to add to
23:34:55JdGordon__a subfolder in apps/plugins/wps-steroids or something.
23:35:05BuschelLambdaCalculus37: test still running?
23:35:23gevaertsif you do it that way, all those limitations go away, but you also are limited by the plugin buffer size
23:35:51gevaertsThe other way, you could actually even consider loading the functions into the skin buffer
23:36:24JdGordon__plugin buffer size isnt exactly small, and if you use the skin buffer for allocations you have even more room to play with
23:36:43JdGordon__it would mean you could never batt bench your skins thuogh
23:36:49gevaertsAnd that will allow you to run those visualisation plugins from an sbs while showing a plugin menu :)
23:36:53domonokytheoretically that should be possible if we link this functions in a relocatable way, and have a loader which handles this format.
23:37:16 Quit Bullet` (Quit: Bye les gens =))
23:37:19gevaertsdomonoky: the idea of my limitations is to try to avoid having to do that
23:37:37 Join phanboy4 [0] (
23:37:51gevaertsYou'd still want them to only use relative addresses for local things, but I guess the compiler will do that anyway
23:38:02CIA-8New commit by Buschel (r24677): Major optimization of atrac3 codec for Coldfire targets. Moving several number cruncher arrays to IRAM. Decoder is sped up by +50% on h300, decoder ...
23:38:09domonokyJdGordon__: it would also prevent any other tsr plugin in normal use (dont know if there is any)
23:38:15*Buschel crosses fingers
23:38:38gevaertsthere's one I think. This car adapter thing for archos
23:38:39JdGordon__well, we could always see if the wps plugin is loaded and if not then just not update that viewport
23:39:15*gevaerts proposes exploring both methods
23:39:38*domonoky thinks using only skinbuffer would be really nice. We need some linker expert.
23:40:02JdGordon__that buffer is getting awefully cramped though
23:40:10JdGordon__untill someone does autmatic sizing on boot...
23:40:12*domonoky thinks it should be possible, with some minor fixing of adresses with the loader.
23:40:37*gevaerts thinks that with the restrictions, it should be possible without any fixing at all
23:42:16JdGordon__at the very least any new tags (which these would be) needs a parser function, and a value getter (so it can work with conditionals) and an updater
23:42:43 Quit jgarvey (Quit: Leaving)
23:43:24gevaertsSince these would be some sort of light plugins, I propose to use the ".pebble" filename extension for them
23:43:37JdGordon__oy vey
23:43:52domonokyJdGordon__: that could still be done with the data struct and one single "process(struct)" function. Just put all info into the struct.
23:44:49 Join martian67 [0] (~martian67@about/linux/regular/martian67)
23:45:27domonokyie put a "action" field in tot the struct, and call process for what every you need, "update","getValue","init" etc
23:45:40Unhelpfuldomonoky: battery_bench is a tsr plugin
23:46:12domonokyUnhelpful: yes, i meant any other then that.
23:46:47mtBuschel: \o/ all green
23:47:32LambdaCalculus37Buschel: Average decoding time is 168% realtime.
23:47:38JdGordon__so FYI *ages* ago I did actually get a TSR wps plugin working, apparently its not on the tracker though
23:47:49 Quit phanboy4 (Read error: Connection reset by peer)
23:48:04JdGordon__I tihnk it was while we were still parsing the wps during the update!
23:48:06domonokyand if we manage to link this function (and its subfunctions) to address 0, we can just put the struct on the stack and jump to the pebbles address. :-)
23:48:10Buschelmt, LambdaCalculus37: \o/ :o)
23:48:34stripwax_saratoga - optimising TRANSFORM_xxx somewhat (although still relying on gcc a little bit) saves about another 1MHz in vorbis ;)
23:48:40stripwax_on arm, that is
23:48:53stripwax_96kbps is now 25.28MHz :-D
23:49:34stripwax_However I've now made the code *incredibly* ugly. But I'm going to commit it (to mdctexp) unless you can think of a good reason not to..
23:50:12Buschelstripwax_: I am most interested in seeing the new mdct in trunk :o)
23:50:36stripwax_Buschel - patience, patience :-)
23:50:57stripwax_192kbps vorbis now 28.58MHz . Excellent
23:51:33gevaertsWe can use all those saved cycles for fancy visualisations!
23:54:55mtstripwax_: Nice :) .. It seems there's an optimization fest today !
23:55:38 Quit flydutch (Ping timeout: 252 seconds)
23:57:00 Quit ender` (Quit: The Web is a procrastination apparatus: It can absorb as much time as is required to ensure that you won't get any real work done.)
23:57:50saratoga2stripwax: go for it

Previous day | Next day