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

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

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

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

#rockbox log for 2010-02-12

00:01:41stripwaxwhat's the best way to svn sync a branch with trunk?
00:01:41checkeranyone here on ubuntu?
00:02:10JdGordon_pixelma: do you think the new tags all make sense? I'm happy to add/remove/change as needed, and we shold get them mostly good before commit
00:02:17kugelpamaury: sounds good
00:02:21stripwaxI'm thinking of just copying all the latest files from trunk over the top of the (stale) files in my branch and doing a massive svn commit. But, that won't handle any stale files in my branch that have since been deleted from trunk right?
00:02:49kugelpamaury: pictureflow has a similar problem, it uses fill_tags() but not find_index()
00:03:12JdGordon_has anyone asked ohloh to fix up their enlistment? B4gder ?
00:03:12kugelwould be nice if you could tell how much faster find_index() compared to get_metadata() is
00:03:21pamaurykugel: so it doesn't work when db is not in ram ?
00:03:37kugelpf calls get_metadata() if fill_tags fails
00:04:37stripwaxchecker - is your WPS issue something to do with ubuntu? (I don't see the connection myself)
00:04:38kugelalthough find_index() also uses ramcache when it's enabled. if you need the index only I'd just use that in the first place
00:05:29checkeri added a file in to a .zip using the ubuntu archive manager, then unpacked the .zip onto my gigabeat, now i have problems :P
00:05:35stripwaxchecker - if your rockbox basically works fine, but shows 'BM6' on the screen, then maybe your wps contains just the text "BM6" perhaps? If on the other hand your rockbox is crashing, you probably need to tell us more info...
00:05:46 Quit JdGordon_ (*.net *.split)
00:05:46 Quit saratoga (*.net *.split)
00:05:56stripwaxchecker - unfortunately this is #rockbox not #ubuntu-archive-manager-bugs .. :(
00:05:59pixelmaJdGordon_: haven't looked into detail. I first thought it was a bit much maybe they are all needed (thought I remembered a discussion about reusing artist etc. tags but at the same time I remember being unable to decide if e.g. a station name equals artist or album)
00:06:19 Join JdGordon_ [0] (~836b0052@gateway/web/freenode/x-netztotbusamhqac)
00:06:19 Join saratoga [0] (~9803c6dd@gateway/web/freenode/x-uxeftoofekqjpbkk)
00:06:26pixelmais there a way to get the fmr name (without the full path)?
00:06:29stripwaxchecker - why didn't you just copy the file to your gigabeat? (why add it to the zip and then unpack the .zip onto your gigabeat?)
00:06:47stripwaxyou could just restore the original zip file, and then just copy your updated file over, no?
00:06:59checkerbecause luckily, i tested the .zip before uploading it to the theme site
00:07:16stripwaxah, theme zip. ok. (now we're getting somewhere :-)
00:07:22JdGordon_pixelma: the fmr is the presets file yeah? you want to display it on the skin or use it somewhere else?
00:08:04pixelmaJdGordon_: display it in the skin
00:08:22stripwaxchecker - so, um - you made a bad theme zip and now it doesn't work. how do we help?
00:08:36checker... you dont
00:08:38 Join tomers [0] (
00:08:41pixelmabut I need to get a basic one to be loaded first :\
00:08:46piotrekmi have a little eabi compilation problem (nano 2g target), and would be gratful for help -
00:09:00stripwaxchecker - I'd like to, just not sure how ..
00:09:01mtstripwax: sweet mdct ;)
00:09:07stripwaxmt - thanks! :)
00:09:16checkerwhat would a .zip to a file that would cause it to bug
00:09:25stripwaxmt - any ideas on how to sync the whole branch to trunk?
00:10:11CIA-88New commit by 03tomers (r24611): fft plugin: add touchscreen key-mapping and enable plugin for touchscreen targets
00:10:43stripwaxchecker - sorry, I can't understand the question... (was it missing a verb?)
00:10:55 Quit shaggy-h ()
00:11:03JdGordon_pixelma: I'll add it, it shuold be easy enough to get to
00:11:07mtstripwax: We're going to use this mdct from now onwards, right ? (i.e, the old one is going to be dropped now?)
00:11:24stripwaxchecker - well, you could always unzip it onto your ubuntu machine and see if the contents match what you thought they would be?
00:11:57stripwaxmt - there's at least one more thing to do - get rid of the awful fft initialisation (revtab)
00:12:27stripwaxchecker - e.g. using diff, etc.
00:12:46 Quit komputes (Quit: I haven't slept for ten days, because that would be too long.)
00:13:31FlynDicemt: ping
00:13:39stripwaxor just rebuild the zip file from scratch and start again (making sure that the files you zip up are correct to begin with). I don't think there's inherently something bad about a .zip file that could corrupt a file and/or make rockbox freak out with a weird wps, unless the files in the zip are just faulty. (But I can't speak for ubuntu archive manager)
00:14:01kugeltomers: ping
00:14:10JdGordon_reply from ohloh staff was damn quick, we shuold be getting updated again
00:14:11checkerfile wasnt, the zipper somehow messed up the encoding on it, fixed now
00:14:27checkerencoding on the .wps file
00:14:40pixelmaJdGordon_: ugh sorry, I misspelled the actual file name of the fms. It loads now...
00:15:08checkerso BM6 means the encoding is incorrect on the .wps :P
00:15:17pixelmaJdGordon_: on the M5
00:15:27JdGordon_but ondio is totally blank?
00:15:37 Quit checker (Quit: CGI:IRC)
00:15:53stripwaxchecker - "BM6" isn't an error message. It probably means that the first printable characters in the wps are B, M, and 6.
00:16:07stripwax(if it's messed up the encoding, that's possible, certainly)
00:16:29 Join Adnyxo [0] (
00:16:39*stripwax could be wrong, but usually rockbox error messages have words in them, and has never seen a wps error like that. JdGordon can probably confirm .. ?
00:17:10JdGordon_yea, BM6 is a error in the wps text
00:17:15JdGordon_or mistake
00:17:57mtFlynDice: pong
00:18:17FlynDicemt: did you see my forum reply?
00:18:23stripwaxJdGordon - as in, rockbox shows "BM6" if your wps is garbaged? are there any other "BM" error messages?
00:18:28*stripwax learns something new
00:18:40 Quit anewuser (Quit: Another edition of chiptune gig WinterChip5! :O =ooo)
00:18:54JdGordon_no, the wps text has an error which doesnt break the parser but displays that as text
00:19:03pixelmaJdGordon_: yes, Ondio shows a blank screen (there was no file name error on that one, I just checked again) and it also showed a blank screen with the default while no fms was set
00:19:06mtstripwax: What's the problem with syncing ? (I'm thinking of just copying what we did and apply the changes by hand then commit, if so, I can do that tomorrow)
00:19:13mtFlynDice: I'll check
00:19:50stripwaxmt - syncing FROM trunk TO branch. how do we make branch less-stale
00:20:00tomerskugel: pong
00:20:10JdGordon_pixelma: ok, thats wierd, I cant tihnk why nothing shows, except if the skin load is inside a !HWCODEC block, which would be bloody stupid :p
00:20:16 Quit kramer3d (Quit: Leaving)
00:20:20stripwaxmt - including deleting files from branch that are no longer in trunk
00:20:52stripwaxJdGordon - as in the .wps file probably has some garbage in it including a string that looks like BM6 - right?
00:21:08stripwaxJdGordon - phew, cool.
00:21:10pixelmaJdGordon_: uhm, no comment :P
00:21:13*petur updates a tracker entry of 2006 with a new patch
00:21:32FlynDicemt: have you tried holding the pwr button for 10 secs or so to reset I need to head out be back in 30 mins
00:21:35mtFlynDice: Yes I tried a long power-button press to reset it, but it didn't work.
00:21:53mtstripwax: Sorry, misread your words.
00:21:55*pixelma should be careful after making own stupid mistakes
00:22:21 Quit petur (Quit: Zzzzz)
00:22:30 Quit Adnyxo (Quit: Leaving)
00:22:34pixelmaJdGordon_: just throwing out an idea - could there be something wrong with viewport ("colour") handling?
00:22:44 Quit moos (*.net *.split)
00:22:44 Quit Sajber^ (*.net *.split)
00:22:44 Quit DerPapst (*.net *.split)
00:22:44 Quit ranmachan (*.net *.split)
00:22:45 Quit Topy44 (*.net *.split)
00:22:51 Part domonoky
00:22:57stripwaxmt - not ready for trunk yet, as I'm unhappy with the fft init. mdct should be totally stateless in my opinion and tables should be constants really
00:23:02 Join DerPapst [0] (
00:23:12 Join Sajber^ [0] (
00:23:18 Join moos [0] (moos@rockbox/staff/moos)
00:24:03stripwaxmaybe we just don't bother sync branch yet.. that's probably easier anyway.
00:24:10pixelmahmm... and what about RFMSs btw.?
00:24:15 Join ranmachan [0] (
00:24:19kugelstripwax: svn merge?
00:24:39JdGordon_pixelma: well, I thought of that a while ago, but isnt the ondio mono?
00:24:41stripwaxkugel - that works across branch/trunk? neat!
00:24:50kugelsure it's meant for that
00:25:01pixelmaJdGordon_: it is
00:25:03stripwaxkugel - ah, great!
00:25:13 Join Adnyxo [0] (
00:25:16JdGordon_OH.. um..
00:25:24JdGordon_does the inbuilt on use viewports?
00:25:27*JdGordon_ cant remember
00:25:58 Join Topy44 [0] (
00:26:15pixelmawhich "one"?
00:26:30 Quit Adnyxo (Remote host closed the connection)
00:26:37JdGordon_the hardcoded one
00:26:53JdGordon_if it does then its probably failing to load because of extra garbage on the %V line
00:26:59pixelmahardcoded FM screen, or something else?
00:27:16JdGordon_yes, that one
00:27:46CIA-88New commit by 03tomers (r24612): fft: fix red
00:27:53 Quit Schmogel (Read error: Connection reset by peer)
00:28:01JdGordon_no, it doesnt
00:28:29 Join Adnyxo [0] (
00:28:36pixelmaI still use there (and already went back to a patchless build now)
00:28:41 Quit pamaury (Quit: abort();)
00:29:05JdGordon_pixelma: can you build checkwps for the ondio and pass the inbuilt one to it? maybe its using a bad tag
00:30:21JdGordon_the first one is missing a >
00:30:25pixelmadon't know, haven't used it so far. How do I do that?
00:30:58JdGordon_ok, new plan
00:31:43JdGordon_ok.. build checkwps (its one of the configure options for the target)
00:31:57JdGordon_then just ./checkwps.ondio filename.fms
00:32:04JdGordon_the first past is correct, not the second
00:32:21pixelmacan you have %pm in a conditional?
00:32:32pixelmaI thought it had to be on its own line
00:32:58kugelwhy are the colons in the language strings?
00:34:10pixelmayou could still get the same effect with conditional viewports (the %pm/prerecord info)
00:34:19kugelthat's nasty!
00:34:29kugel(the colon thing I mean)
00:35:27pixelmayeah, unnecessary
00:35:48pixelmaand might hurt if you want to use it in a different way
00:36:02pixelma(the colons)
00:36:09JdGordon_kugel: to make RTL langs easier to deal with
00:36:22JdGordon_iirc anyway
00:36:59JdGordon_pixelma: oh, it doesnt like %pm? ok hmm
00:37:16JdGordon_and we cant use conditional viewports here because it should work inside the ui viewport
00:37:19JdGordon_wherever that is
00:37:25pixelmahaven't checked but that's what I always thought
00:37:32kugelfor RTL? really?
00:37:54kugelpixelma: exactly, I would like to do it in a different way
00:40:37 Quit JdGordon_ (Quit: Page closed)
00:41:51 Join checker [0] (
00:44:59 Join shaggy-h [0] (
00:46:51 Quit checker (Quit: CGI:IRC (EOF))
00:48:44stripwaxhrm, svn merge sounds easy but I imagine it would cause a few problems here. I'll just wait until branch mdct is "finished" and port it back into trunk
00:49:16kugelyou need to merge in trunk anyway
00:49:19stripwaxmt - one thing, then, to do is to move the remaining codecs to using the newmdct if they're currently using the old tremor mdct. Would you have time to look at that
00:49:32kugelotherwise you mess up trunk
00:49:39stripwaxkugel - true, or could make the (small handful of) changes manually no?
00:49:57kugelsure, but that means losing versioning
00:50:08kugelI can't imagine which problem svn merge would cause
00:50:10stripwaxah. true
00:51:07stripwaxkugel - well, it seems like svn doesn't track the merges, and that you have to manually tell it which revisions to merge diffs from. which sounds like a bit of a manual headache if a) the merge goes wrong or b) I specify the wrong revisions
00:51:16pixelmabuilding checkwps fails in firmware/export/debug.h for me. Since the patch doesn't seem to touch this file, my guess is "maybe cygwin?"
00:51:26stripwaxI'm not a hardcore svn user, so if it's easy I'll happily let someone else do it :)
00:51:28mtstripwax: Sure. will look at that tomorrow.
00:51:30kugeldepends on the version. newer svn tracks merges
00:51:37stripwaxoh, neat-o
00:51:41kugelnewer = >1.5 I think
00:51:42stripwaxmt - cool!
00:52:25kugelit's possible that you need to specifiy the revision for the first merge, but I doubt you need it after then
00:52:28stripwaxkugel - so I could just "svn merge trunk branches/mdctexp" with no other args and it would know what it is supposed to merge with what?
00:52:42***Saving seen data "./dancer.seen"
00:53:01kugelyou could just try. you do the merge locally so nothing really bad can happen anyway
00:53:12stripwaxkugel true
00:53:40stripwaxand on that note, goodnight ....
00:54:04 Quit stripwax (Quit:
00:55:28 Quit togetic (Read error: Operation timed out)
00:55:52 Quit Tomis (Read error: Operation timed out)
00:57:28 Join Tomis [0] (~Tomis@
01:00:28 Quit bmbl (Quit: Bye!)
01:05:00 Quit amiconn (Disconnected by services)
01:05:01 Join amiconn_ [0] (quassel@rockbox/developer/amiconn)
01:05:17 Join pixelma_ [0] (quassel@rockbox/staff/pixelma)
01:05:18 Quit pixelma (Disconnected by services)
01:05:23 Nick amiconn_ is now known as amiconn (quassel@rockbox/developer/amiconn)
01:05:29 Quit FlynDice (Remote host closed the connection)
01:05:35 Nick pixelma_ is now known as pixelma (quassel@rockbox/staff/pixelma)
01:07:02CIA-88New commit by 03tomers (r24613): fft: fix yellow
01:07:20perfectdrugkugel: your blacknblue_glass theme seems to be odd with current svn
01:07:39 Quit Tomis (Read error: Connection reset by peer)
01:08:10perfectdrugsbs elements seems to be missing
01:08:26 Join Tomis [0] (~Tomis@
01:10:06amiconnkugel: I would think that the colons are part of the lang strings for (non-latin) language where using a colon would be unusual or even wrong
01:10:35 Quit dfkt (Quit: -= SysReset 2.53=- Ph'nglui mglw'nafh Cthulhu R'lyeh wgah'nagl fhtagn.)
01:11:06*amiconn would not expect colons in e.g. chinese or korean
01:11:21 Quit tomers (Quit: ChatZilla 0.9.86 [Firefox 3.5.7/20100106054534])
01:12:03 Join komputes [0] (~komputes@ubuntu/member/komputes)
01:12:04perfectdrugnote playback is stopped and it doesn't stop the progressbar
01:12:12kugelperfectdrug: there's no sbs
01:12:40perfectdrugso it is from your other theme still?
01:12:42kugelthe blacknblue glass for e200 is a tad bit outdated. it doesn't have the relevant lines for deactivating the sbs
01:12:57kugelyea, leftover
01:12:59perfectdrugwhich i loaded before
01:13:19kugelI wanted to update it today which was the reason I messed with the download counter :)
01:13:47perfectdrugyes and your mention made me try your theme out :D
01:15:20 Part toffe82
01:17:28 Join FlynDice [0] (
01:19:56kugelJdGordon: save theme settings produces "sbs: /.rockbox/wps/" instead of "sbs: -" for no sbs, is that right?
01:23:17 Join Horscht [0] (~Horscht2@xbmc/user/horscht)
01:32:20 Quit dmb (Ping timeout: 252 seconds)
01:40:57UnhelpfulJdGordon: it's going to be a busy night, but what did you want? buflib-related stuff?
01:48:32kugelperfectdrug: should be updated now
01:49:08CIA-88New commit by 03mc2739 (r24614): Turkish translation update ...
01:51:03 Quit saratoga (*.net *.split)
01:56:19 Join saratoga [0] (~9803c6dd@gateway/web/freenode/x-uxeftoofekqjpbkk)
01:59:10perfectdrugkugel: helvR10.fnt is renamed in the font package
02:00:08JdGordonUnhelpful: no,
02:00:12JdGordonAA question
02:00:21UnhelpfulJdGordon: shoot :D
02:00:45JdGordonis there anything in playback which would cause things to break if you try getting an AA in the fm screen, so audio is stopped?
02:00:54JdGordonit looks like it shouldnt, but its apparently causing problems
02:02:00UnhelpfulJdGordon: i'm not completely sure how AA is retrieved for rendering.. i only know about loading :/
02:02:13JdGordonbah, ok
02:04:32Unhelpfulbut, no, i can't imagine why it ought to break, surely this is similar to wps while playing a track with no AA?
02:05:02Unhelpfuli guess the first question is, "how does it know what the buffer handle for the AA is?"
02:05:30JdGordonright, it should be getting -1 or something because there is no aa
02:07:34kugelperfectdrug: argh, thanks
02:10:14 Quit Sajber^ (Read error: Connection reset by peer)
02:10:48perfectdrugkugel: i can imagine a lot more themes are outdatet this way, but no one notice because they have the old fonts still around, i guess "works with 3.5" or checkwps doesn't check against this
02:11:21kugelno, because it could be a custom font as well
02:12:19perfectdrugyes but then it has to be included
02:12:30 Join moos_ [0] (
02:13:08 Quit moos (Ping timeout: 258 seconds)
02:13:10 Nick moos_ is now known as moos (
02:16:01kugelperfectdrug: should be good now :/
02:17:19kugelJdGordon: well, doesn't the albumart load function fail?
02:19:33perfectdruglooks fine now:)
02:19:58kugelstrangely the update function didn't work
02:28:29perfectdrugkugel: what is the cross icon for it is always on and looking at your wps it is for crossfade, but you just declare 4 modes and not 5
02:29:32Unhelpfulamiconn: if we made the audio buffer a buflib (with some extensions), and we wanted to use buflib for a more flexible glyph cache, i imagine that would have its own buffer and allocator?
02:31:29Unhelpfulmy idea regarding dynamic plugin buffer was that the plugin buffer would still have a fixed start or end address, and would be allocated by moving one end or the other of the main buffer
02:31:31kugelperfectdrug: is there a new mode?
02:32:00perfectdrugi don't know I'm looking at the customwps site
02:32:06kugelit's off here
02:32:27kugelit should be for crossfade yes
02:32:46amiconnUnhelpful: No own allocator
02:33:18Unhelpfulhrm? so each glyph would be a buffer object also?
02:33:19kugelit doesn't seem to go on actually
02:34:06perfectdrug"Automatic Track Skip Only" is new according to customwps history
02:34:23kugelyea, blue_dude added it some weeks ago
02:34:38Unhelpfulalthough perhaps buflib-as-buffer ought to be proven on its own before any worry is given to buflib glyph cache and plugin buffer
02:35:15 Quit DerPapst (Quit: Leaving.)
02:37:22Unhelpfuli guess i was mainly thinking that the glyph-cache buffer will fragment much more quickly than the main buffer, and need compaction far more frequently.
02:37:38kugelUnhelpful: as I understand: buflib would only provide the audio buffer, the buffering within that buffer would still be managed by buffering.c?
02:38:05Unhelpfulkugel: *something* like that, maybe? ;)
02:38:38Unhelpfuli don't think we're able, now, to move the buffer of the currently-playing track, are we?
02:39:25perfectdrugkugel: %?xf<|%xdv|%xdv|%xdv|%xdv> fixes it
02:43:09kugelstrange, it's never on here
02:45:52perfectdrugi have a clean build i formated 3 hours ago and always uninstalled your theme before testing a new one
02:46:20kugelnevermind, works now
02:46:38*kugel is too lazy to upload it again
02:47:03 Join Tomis2 [0] (~Tomis@
02:47:08kugelwill do tomorrow
02:47:16kugelthe fuze's version has the same issue then, I guess
02:47:25 Quit Tomis (Read error: Connection reset by peer)
02:47:26 Nick Tomis2 is now known as Tomis (~Tomis@
02:50:44 Quit piotrekm (Quit: Leaving.)
02:52:45***Saving seen data "./dancer.seen"
02:55:13 Quit komputes (Remote host closed the connection)
02:58:46 Join Rob2223 [0] (
03:02:00 Quit Rob2222 (Ping timeout: 240 seconds)
03:03:54 Join Strife89 [0] (
03:12:29 Quit FlynDice (Remote host closed the connection)
03:14:49 Join FlynDice [0] (
03:15:06 Quit Tomis (Read error: Operation timed out)
03:15:44 Quit FlynDice (Remote host closed the connection)
03:16:19 Join Tomis [0] (~Tomis@
03:18:57 Join FlynDice [0] (
03:19:04 Join togetic [0] (~togetic@unaffiliated/ibuffy)
03:20:47 Quit Casainho (Quit: ChatZilla 0.9.86 [Firefox 3.6/20100115133306])
03:21:29 Quit Tomis (Read error: Connection reset by peer)
03:22:24 Join Tomis [0] (~Tomis@
03:23:31 Quit ______dmb (Read error: Connection reset by peer)
03:24:31 Join dmb [0] (~Dmb@unaffiliated/dmb)
03:27:58 Quit MethoS- (Read error: Connection reset by peer)
03:29:12 Join TopyMobile__ [0] (
03:31:35 Quit TopyMobile_ (Ping timeout: 264 seconds)
03:41:24 Join BHSPitMonkey [0] (~stephen@unaffiliated/bhspitmonkey)
03:41:41 Quit Adnyxo (Ping timeout: 258 seconds)
03:49:36 Part froggyman
03:59:34 Join dmb_ [0] (~Dmb@unaffiliated/dmb)
04:00:18 Quit saratoga (Ping timeout: 248 seconds)
04:09:16 Quit moos (Quit: ChatZilla 0.9.86 [Firefox 3.6/20100115144158])
04:13:07 Join checker [0] (
04:13:34checkeranyway to load a progress bar vertically? or even better, radially?
04:14:26 Join CaptainKewl [0] (
04:18:16 Quit phanboy4 (Read error: Connection reset by peer)
04:18:47 Join kadoban_ [0] (
04:18:51 Quit kadoban (Write error: Broken pipe)
04:21:06 Quit TheSeven (Disconnected by services)
04:21:19 Join The_Seven [0] (~theseven@rockbox/developer/TheSeven)
04:21:25JdGordonchecker: not yet
04:21:30 Nick The_Seven is now known as TheSeven (~theseven@rockbox/developer/TheSeven)
04:21:40CIA-88New commit by 03mt (r24615): Add support for Sony OMA file format. Currently only supports ATRAC3 (without DRM), and seeks. Tested on sansa ...
04:22:10Unhelpfulradially is "better"?
04:22:39checkeryeah id say so
04:22:44checkerlooks cooler at least :P
04:23:46Unhelpfulby radially you mean what, an arc? a pie-chart sort of circle thing? either is wasting rather a lot of space :/
04:24:33checkercircle, can surround a button
04:24:54checkerbut vertical is more practical
04:25:44checker.gif support would be best :D
04:27:21Unhelpfulnot really, with a bar we can easily show as much resolution on the progress bar as the space allows. with a gif we only have X frames.
04:27:50checkera gif that loads the same as the .bmp currently does
04:28:50checkerit'd add a little life to it
04:29:16Unhelpfuli'm not even sure what you mean by that. :)
04:29:50Unhelpfulsurely "loads the same as a bmp does" would not give you any new options
04:29:50checkerfor the progress bar, basically imagine a .gif file loading the same way as they do now.
04:30:08checkerby load, i mean how it loads the image as the time progresses
04:30:15mtNow we beat VLC in OMA playback. B)
04:30:40Unhelpfulso, you'd get what, the ability to *animate* the completed portion of the progress bar?
04:30:57checkerno, it would be animated as it loads
04:31:49checkeryou could make a vista-like shine shoot across it for example
04:32:38checker check that out,
04:32:40Unhelpful... i'm still not really sure what you mean. so we'd load an animated gif, and draw a *portion* of it based on the amount played, and that portion would cycle through the frames of the animation?
04:33:05 Quit mt (Quit: ChatZilla 0.9.86 [Firefox 3.5.7/20091221164558])
04:33:19checkeryes cycle through all frames of animation, but only the correct portion (based on time), is visible...
04:34:18Unhelpfulwell, good luck convincing anybody that we need that :)
04:34:45checkerlol, im just saying its better than a radial progress bar :P
04:36:32Unhelpful...yes, but the progress bar we have now is also better than radial. it presents information. and you can make it a color or graphic. also updating the LCD repeatedly to run an animation will kill batteries.
04:37:24checkerfor some people, action on the screen is more desirable than excess battery life
04:37:45checkerthe .gif would also present the same information
04:37:59checkerand you could make it a color or graphic :P
04:38:52Unhelpfulthen go ahead and write a gif loader and progress bar animator :)
04:39:07checkerwould i had half a clue as to how
04:42:18 Join Barahir_ [0] (
04:45:35 Quit Barahir (Ping timeout: 252 seconds)
04:47:28 Quit antil33t (Read error: Connection reset by peer)
04:47:33 Join S_a_i_n_t [0] (S_a_i_n_t@
04:48:15checkerhey St., thanks again for checking that .wps, works great now :)
04:48:32S_a_i_n_tno problems man :D
04:48:36S_a_i_n_thappy to help
04:48:56 Quit piroko (Quit: Reconnecting)
04:49:10S_a_i_n_tand it does look good...I'll have to admit that, even *before* I fixed it up I could see it looked quite nice
04:49:14 Join piroko [0] (
04:49:20S_a_i_n_tnow it looks nice, and works :-P
04:49:26checkerlol yes
04:50:32 Quit checker (Quit: CGI:IRC)
04:50:41 Join checker [0] (
04:51:10 Quit checker (Client Quit)
04:51:19S_a_i_n_tAdded FS #11000 <−−- "slightly annoying kill you almost instantly" sky/terrain drawing issue in plugin/game chopper.rock
04:52:47***Saving seen data "./dancer.seen"
04:54:12 Join antil33t [0] (
04:54:51perfectdrugS_a_i_n_t: screenshot? :D
04:55:07 Quit panni_ (Quit: ( :: NoNameScript 3.81 :: ))
04:57:20S_a_i_n_tperfectdrug: I'll try and pull off a screenshot from the simulator (assuming the SIM has the same error (fingers crossed)), but my timing would need to be pretty bloody good lol, so no promises :-P
04:58:10perfectdrugthats what i thought, just a joke, but if you can manage fine:)
04:58:25 Join Recursive [0] (
04:58:39S_a_i_n_tAm I correct in the understanding that you can just format an ipod and then use ipodpatcher to install a new apple firmware image? Instead of going through the mess around with having to use iTunes (and not having to install iTuens again actually... :P)
04:59:09RecursiveIs there any project that supports alternative OS's on the iPod Nano (3rd generation, the fat one), or has it not been hacked at all yet?
05:00:33S_a_i_n_tperfectdrug: I figured it was a joke...there a *wee* hint of sarcasm there, but I actually think I *may* be able to do it if I get someone else to play chopper on the sim and hover over "PrtScr" myself waiting to strike :-P
05:02:31perfectdrugah ok and i figured there was "a *wee* hint of sarcasm there" also:) good night
05:02:58perfectdrugRecursive: no
05:03:10S_a_i_n_tRecursive: Well, there is no Rockbox port for it yet...if that's what you're asking.
05:03:25RecursiveYeah, but I'm asking if there's anything for it
05:03:33RecursiveGoogling around suggests it hasn't been cracked
05:03:53S_a_i_n_tthat would be offtopic here
05:04:14RecursiveYeah, but where else am I going to ask?
05:04:29Unhelpfulthat's hardly a reason to ask here?
05:04:37*S_a_i_n_t suggests google
05:06:03 Quit perfectdrug (Quit: perfectdrug)
05:06:09S_a_i_n_tyou could always "ask Jeeves" too :P
05:06:57 Quit AndyI (Ping timeout: 248 seconds)
05:07:18ranmachanIs there some app that will dump button events?
05:07:31ranmachanLike 'xev' for rockbox?
05:08:20Unhelpfulranmachan: i don't think so... and i'm not even sure how you'd do that, the available buttons are different on each target.
05:09:50JdGordonpixelma: oh crap, the rec sim shows the fms which means its a target problem :<
05:16:02JdGordonUnhelpful: so whats the current issues with buflib? did you read the logs before? is buflib in a state where we could do what we want?
05:16:59UnhelpfulJdGordon: well, i suspect there are one or two places where its bufer gets corrupted. it currently lives in pluginlib. and i have no idea how we'd let a codec know that its buffer has moved.
05:18:01Unhelpfulalso buflib doesn't currently support the "give me all available space, i'll give back what i don't use" usage, but could do so fairly easily
05:18:25JdGordonif it can shrink blocks then thats not really an issue
05:18:34JdGordoneven then we can work around it
05:18:51JdGordonreplacing every alloc system with a single one would be sweet though
05:21:39Unhelpfulfor buffer_alloc we'd want to add support for a group of contiguous, rarely-moved allocations at the start of the buffer, and specifically for collecting free space between those and other allocations in order to place a new allocation *there* instead of wherever it can fit
05:22:08*Unhelpful is pretty sure that no allocation resizing is implemented
05:23:25Unhelpfulresizable buffer_alloc would also need a callback mechanism - if a buffer_alloc user generally assumes the buffer won't move, then a callback can be used when it *does* move, to fix up any pointers etc.
05:26:15JdGordonnot automagic behind the scenes resizing, but being able to explicitly say resize this handle to size X
05:26:28Unhelpfuland there's the whole issue of moving the currently-playing file's buffer. on single-core this could be done just like buffer_alloc allocations, a callback in which the codec fixes all of its pointers. for multi-core the COP *must* be stopped and forced to update any of *its* pointers.
05:27:38UnhelpfulJdGordon: i wasn't suggesting anything automagic... but if you have a stack of "static" buffers and the one on the bottom requests a resize, you can either move all of the others, or re-allocate that buffer somewhere else and create fragmentation (which you wouldn't be able to fix without moving those buffers around)
05:27:55Unhelpfulright now buflib has no concept of a buffer you can't/shouldn't move
05:33:24 Join saratoga [0] (
05:33:44 Quit saratoga (Changing host)
05:33:44 Join saratoga [0] (saratoga@rockbox/developer/saratoga)
05:34:08 Quit yosafbridge (Quit: Coyote finally caught me)
05:34:15 Join yosafbridge [0] (
05:34:37*JdGordon has a "which screen?" menu working for font loader
05:34:53 Quit yosafbridge (Remote host closed the connection)
05:35:09 Join yosafbridge [0] (
05:35:26 Quit Horscht (Quit: Verlassend)
05:51:45 Quit evilnick (Ping timeout: 248 seconds)
05:56:08S_a_i_n_tOK, I've figured out (from the wiki) that I definitely *can* install an .ipsw file using ipodpatcher...but my question is this, "when an iPod displays the crappy 'Restore using iTunes' message" is it necessary to format the disk (iTunes appears to do so)? I'd rather not loose all my music/data if I don't *have* to.
06:09:24 Join kramer3d [0] (~kramer@unaffiliated/kramer3d)
06:21:29 Quit kugel (Ping timeout: 240 seconds)
06:21:36 Quit kramer3d (Quit: Leaving)
06:21:51 Join kramer3d [0] (~kramer@unaffiliated/kramer3d)
06:25:30 Join podkast [0] (
06:26:16podkastI saved my playlist in a .m3u file on my player but when i load it using rockbox, the order is randomized everytime i load the .m3u file... how do i keep it in the same order?
06:29:54JdGordonturn off shuffle mode
06:30:27podkastah forreal thanks :) but how do i keep my playlist playing over and over, repeat -> all or shuffle?
06:31:50podkastok i got, its repeat all
06:41:21 Join evilnick [0] (~evilnick@rockbox/staff/evilnick)
06:42:49 Quit podkast ()
06:46:22 Quit dmb (Read error: Connection reset by peer)
06:46:48*JdGordon is tempted to commit multifont already
06:52:50***Saving seen data "./dancer.seen"
06:56:27 Quit BHSPitMonkey (Remote host closed the connection)
06:57:50*S_a_i_n_t doesn;t see why not...
06:58:38S_a_i_n_tis it still using 2, 3 , 4 etc. ? Or did you end up defining it some other way?
06:59:08*S_a_i_n_t hopes it's still numbers so he doesn;t have to redo his new WPS's again :P
07:09:52 Quit S_a_i_n_t (Quit: [St.] has exited mIRC™)
07:10:27JdGordonyeah, it will stay numbers
07:10:33 Join S_a_i_n_t [0] (S_a_i_n_t@
07:31:16 Quit S_a_i_n_t (Read error: Connection reset by peer)
07:35:25 Join S_a_i_n_t [0] (S_a_i_n_t@
07:36:58 Part piroko
07:37:49pixelmathe Recorder has no radio... ;) I'm currently compiling an OndioFM sim just in case
07:38:01JdGordonthe fm recrorder does
07:38:21pixelmabut you didn't say so :P
07:38:30JdGordonactually, have you got time to play now
07:39:40pixelmano, got to get ready for work. I can let my computer compiling in the meantime but testing would need more attention
07:39:47JdGordoncan you change apps/recorder/radio.c line 984 to just "if (1)" and try again
07:40:45JdGordonthats the only line which could cause problems on target and not sim
07:40:57pixelmamultifont also has only seen testers on Sansas and Ipods, right? Nothing with an lcd remote even?
07:41:27pixelmaand no other architecture
07:41:42JdGordonI just tried on h300+remote
07:42:15JdGordonand i did the font for screen submenu thingy
07:47:26pixelmahmm... the radio screen works in the OndioFM sim but I first got crashes when going to the radio - with "bitmap too large for buffer: 45 bytes", unfortunately I haven't looked which theme was set before (built on top of a very old sim)
07:47:40 Quit liar (Ping timeout: 245 seconds)
07:48:04JdGordonthen its probaly not an issue
07:48:06pixelmabut the skin buffer usage info was very weird anyways as it contained three lines
07:50:18pixelmawhat do you mean with "it" there?
07:50:54JdGordonyour whole comment there. if you copied ontop of a very old sim who knows what could have caused that
07:52:14*Unhelpful still likes the idea of flags/types of some sort for a "core" buflib, which may introduce other bits into the buffer header that the buflib "knows" about, and perhaps does something with
07:52:58 Join einhirn [0] (
07:53:01Unhelpfulfor example there could be flags for things loaded from files, and for things with dimensions, and if either of those is set the buflib would check for an existing buffer object that matches :)
07:53:03pixelmaof the shipped themes on the Archoses, iCatcher is the biggest wrt skin RAM usage and even if I load that, the radio screen shows now and doesn't crash the sim
07:53:50JdGordonok cool, so its only on target :/
07:56:12Unhelpfulthis could lead to AA loading once per album and per size
08:01:05Unhelpfulbut... it would also require somewhere to *store* the flags field, although we could do something funny like limit length to 2^24 blocks and store flags in the low 8 bits of the size... otherwise every allocation would cost another 4B :/
08:01:49JdGordon2^24 should be enough blocks...
08:02:10JdGordonand do some sort of hash on the filenames and store that as extradata?
08:02:51Unhelpfulwell, we already store full filenames in the current buffer, don't we? i think they're part of the linked-list entry header :)
08:03:28JdGordonwe do?
08:03:42Unhelpfuli thought we did? :)
08:04:23 Join dmb [0] (~Dmb@unaffiliated/dmb)
08:04:40Unhelpfulyeah, struct memory_handle has a char path[MAX_PATH] field..
08:05:08JdGordonlets start simple though :p
08:07:25Unhelpfulyeah, refcounts and merging items are a bit much to start with... but i think that the initial version *should* have optional header fields that the audio buffer can use for the filename and such
08:07:51Unhelpfuland then other things that are bufalloc'd save MAX_PATH bytes :)
08:07:59 Quit Recursive (Ping timeout: 256 seconds)
08:08:29JdGordonhow about a 4 byte extradata field which if anything needs more than that, stores another handle id?
08:10:42Unhelpfuland the meaning of extradata is known only to the thing that allocated the item?
08:10:52 Join kramer3d_ [0] (~kramer@unaffiliated/kramer3d)
08:11:28pixelmaJdGordon: just realised that there is a big difference between this sim FMS and the target FMS (or how it should be). The sim doesn't show peakmeter or pre-recording. The info for drawing %pm is pulled from the MAS afaik , so not a big surprise it cam ne drawn correctly in the sim. There is a fake %pm for the WPS though but I don't know if there are differences between how radio/recording or playback draws the peakmeter and gets the info for doing
08:11:28pixelma so
08:12:02JdGordonyou got cut off, but yes, I know there is a difference
08:12:18JdGordonwhich is why I've been nagging to get real testing on target
08:13:34UnhelpfulJdGordon: so then the buffering code would be responsible for uniquifying files, if we want to do that later? and the extradata could hopefully be made optional somehow?
08:13:54*pixelma wonders how a %pm works in an sbs when in the radio or recording screen but has to postpone testing until later
08:14:12 Quit kramer3d (Ping timeout: 264 seconds)
08:14:29pixelmabtw. I wasn't able to build checkwps so maybe you'll try that yourself
08:14:56*TheSeven is confused by mt's red
08:15:04TheSevenis that a build system fluke?
08:15:28TheSevenit doesn't look like one at the first glance, but i can't explain how it could happen otherwise either
08:15:37JdGordonUnhelpful: we could do something like find_handle(callback) where the callback return true if the current handle is the one it wants?
08:16:13JdGordonsince when do we build the database as a lib?
08:16:20JdGordonor binary even
08:16:52UnhelpfulJdGordon: but then we would not only be walking the whole buffer to find a matching handle, but *also* calling a callback for each. :/
08:17:26JdGordonnot if handles are given a user id or something, then only those are sent to the callback
08:17:26Unhelpfulthe target appears to *be* the standalone database tool?
08:18:22UnhelpfulJdGordon: and that's even more data that might potentially be stored in *every* handle header :/
08:19:01JdGordonextradata is 4 bytes? 3 for a handle id, 1 for user id
08:19:04Unhelpfulstuff like this makes me think even more that a buflib-glyph-cache ought to have its own buffer... so that every buffer walk doesn't hit a few hundred glyphs in the buffer, etc.
08:20:05JdGordonwe could even keep a linked list of a "users" handles
08:20:11JdGordonor the user could
08:20:29JdGordonhow complicated do we want to make this? :D
08:22:19*Unhelpful hands JdGordon the Gloves. (
08:23:58Unhelpfulactually, buffering could initially allocate a list of handles-it-owns, and use that to do its own deduplication. and anything else that wants to could do the same.
08:25:06Unhelpfulif we make allocations resizable there is no good reason that this list needs to be a linked-list, it can be a simple array :)
08:25:41JdGordonI only care about shriking, growing could be difficult
08:26:10JdGordonwouldnt buffering store a linked list of handles anyway so it knows the correct order?
08:26:43 Join Zagor [0] (
08:26:43 Quit Zagor (Changing host)
08:26:43 Join Zagor [0] (~bjst@rockbox/developer/Zagor)
08:30:51 Nick kramer3d_ is now known as kramer3d (~kramer@unaffiliated/kramer3d)
08:39:42JdGordonunless anyone has any major objections I'm going to commit multifont (10984) over the weekend
08:39:53JdGordonif we decide its worth doing some optimisations that can happen after
08:39:58JdGordonbut I dont think its worth it
08:48:54UnhelpfulJdGordon: buffering right now stores the information itself in a linked list, doesn't it?
08:49:30Unhelpfuland growing wouldn't be terribly harder than shrinking - it should be very similar to compaction, as it's still a matter of moving existing allocations around.
08:50:27 Quit TheSeven (Quit: ChatZilla 0.9.86 [Firefox 3.5.7/20091221164558])
08:52:26*Unhelpful thinks the hardest problems will be mostly in doing this at all - making buffer_alloc allocations movable so that we can usefully clear or shrink them, making the currently playing track movable so that we can compact, and not just compact around, it.
08:52:52***Saving seen data "./dancer.seen"
08:52:54Unhelpfulstuff like deduping track/AA buffers, or using buffer objects for glyph cache, is all just window dressing :)
08:58:59 Join flydutch [0] (
09:01:04 Join ender` [0] (
09:11:19 Join bmbl [0] (~Miranda@unaffiliated/bmbl)
09:15:13 Join petur [0] (~petur@rockbox/developer/petur)
09:44:32 Quit bmbl (Ping timeout: 252 seconds)
09:56:52 Quit CaptainKewl (Ping timeout: 260 seconds)
09:57:37 Quit mc2739 (Ping timeout: 252 seconds)
09:59:26 Join mc2739 [0] (~mc2739@rockbox/developer/mc2739)
10:01:32 Join Sajber^ [0] (
10:29:32 Quit kramer3d (Ping timeout: 260 seconds)
10:31:38 Join LinusN [0] (~linus@rockbox/developer/LinusN)
10:37:15 Join _zic [0] (
10:44:50 Join watto [0] (~watto@
10:52:54***Saving seen data "./dancer.seen"
11:18:10 Join TheSeven [0] (~theseven@rockbox/developer/TheSeven)
11:19:06 Quit Sajber^ (Read error: Connection reset by peer)
11:31:58 Join Zarggg_ [0] (
11:35:00 Quit Zarggg (Ping timeout: 240 seconds)
11:46:51Unhelpfulgevaerts: you know that one of the proposed buflib uses is lots of small allocations/deallocations (glyph cache), right? ;)
11:48:48gevaertsUnhelpful: yes, and I don't think that's a good idea :)
11:49:22gevaertsalthough I might be wrong of course
11:49:39 Join GeekShadow [0] (~Antoine@reactos/tester/GeekShadow)
11:49:46Unhelpfulah. reasons? the main problems i see are that lots of objects in buffer means possibly a long walk to find free space
11:50:21gevaertsThe problem I see is that basically we want as much RAM as possible for the audio buffer, so in a trivial naive implementation, each allocation means that another one has to go/shrink
11:50:24Unhelpfuli don't think we'd see a horrible number of compactions happening if the glyph cache has its *own* buflib.
11:52:48Unhelpfulthe narrower the distribution in size of allocations, the more likely a recent free will fit a new allocation. otoh the audio buffer being fragmented by glyph cache abuse could mean quite a lot of compaction when a space can't be found for audio data and lots of tiny holes get squished together.
11:52:54 Join kramer3d [0] (~kramer@unaffiliated/kramer3d)
11:56:14gevaertsof course the glyph cache will never be very big in total, so compaction should be reaonably fast if those bits are the only issue
11:58:15 Join teru [0] (
11:58:41Unhelpfulyes, but in the best case compaction is one call to memmove per hole, worst case one per allocation above the lowest hole. i think the latter is how buflib currently does it, and to be honest i'm not sure the former is an improvement - it would require a second walk through each segment of contiguous allocations to fix up all of their handles, and that would probably be less cache-friendly than moving each buffer and then fixing its handle,
11:58:41Unhelpful since that way we're always moving up the buffer.
12:00:26Unhelpfuland i'm still not entirely sure how we will handle moving buffers that the COP accesses. although "COP never accesses buffer directly" might be feasible, if the COP threads can always be processing data that has come from some earlier calculation done on main core.
12:03:19gevaertsI'd actually handle glyph cache using a two-layer system, i.e. give it its own pool
12:04:36Unhelpfulthat's pretty much what i'm suggesting, buffer_alloc some space for glyph cache, and run an allocator there.
12:06:02gevaertsfor codec COP use, I'm pretty sure that we can probably get away with "don't move the buffer that the codec is currently working with"
12:09:34Unhelpfulif we make it a bit more flexible about how it finds space, i'd say definitely. things might get tricky in places - part of what i want to do is things like the ability to resize buffers that currently use buffer_alloc and can only be sized at startup, or to carve the plugin buffer out of the audio buffer on plugin startup and leave it for audio buffering when no plugin is running. these are both tricky and may *require* moving the
12:09:34Unhelpful currently-playing buffer, so i think we may need a method for that, but if it's called infrequently it won't be a big performance hit, and if it's usually the result of a user interaction, any skipping it might cause will likely go unnoticed. "oh, i made that happen" pretty much.
12:10:24 Join bmbl [0] (~Miranda@unaffiliated/bmbl)
12:11:13Unhelpfulwe already see this if WPS AA size changes during playback... but right now it forces a rebuffer to do that, and that's much worse than possibly having to stall the codec briefly.
12:12:57gevaertsresizing buffers doesn't have to mean moving a locked buffer, as long as you don't resize the locked one, and you assume that no buffer is larger than 1/3 of RAM
12:14:31gevaertsat least I think it's 1/3. Maybe my numbers are wrong and it's 1/4, but I don't think that matters right now
12:15:21gevaertsthat doesn't help with the plugin buffer of course. I don't really see a good solution for that one
12:16:37Unhelpfulgevaerts: i don't see how that would get you out of moving any specific buffer. the plugin buffer will be particularly tricky because unless we implement a relocating loader, it needs a fixed address.
12:17:58*gevaerts sees it, but he finds these things hard to explain :\
12:18:08 Quit _zic (Remote host closed the connection)
12:19:23*gevaerts thinks that maybe it's hard to explain because it's not entirely correct
12:19:44gevaertsCan I pile on more assumptions?
12:20:41Unhelpfulthe plugin buffer case i think makes it impossible to *always* avoid moving the current file buffer... since it could happen to overlap that fixed space. but a relocating loader has also been proposed, and would have other benefits (plugins could potentially short-call core functions, and at least would not have to load core function addresses from a table, among other things)
12:20:45Unhelpfulare they reasonable?
12:21:05gevaertsI don't know :)
12:21:23gevaertsI'm not going to solve the plugin buffer right now, but I'll have a go at the others
12:23:25 Join MethoS- [0] (~clemens@
12:23:27gevaertsIf (a) you never resize the currently-playing buffer, (b) no single buffer is larger than half of the RAM that's used for audio (*including* buffers used for audio), and (c) you're allowed to drop (or shrink) not-actively-used audio buffers, I think you can do it
12:27:28gevaerts(a) is only a requirement because obviously you can't resize the buffer that's at the end of the heap without moving it, and (b) and (c) together allow you to always find room for your resized buffer
12:28:30gevaertsI think (c) is reasonable because without that you have to reserve space for non-audio allocations which defeats the point of the entire exercise
12:28:44gevaertsI don't know how realistic (b) is
12:30:50Unhelpfulby "resize" you mean "grow"?
12:31:27gevaertsyes. Shrinking isn't very interesting :)
12:33:30Unhelpful(c) is a basic assumption of this whole plan, isn't it? the whole idea is that we can fill the buffer completely and then dump some audio when we load a bigger skin, or run a plugin that wants to grab some extra memory, or something like that.
12:34:20 Quit einhirn (Quit: Miranda IM! Smaller, Faster, Easier.
12:47:15 Quit xavieran (Ping timeout: 245 seconds)
12:49:28 Join Kitr88 [0] (
12:50:04 Quit Rob2223 (Quit: Rob2223)
12:50:48 Join DerPapst [0] (
12:51:53 Quit Kitar|st (Ping timeout: 240 seconds)
12:52:06archivatorIs it a known bug that scrolling text from the WPS doesn't disappear when USB is connected? I get the USB screen with the artist and album still scrolling...
12:52:39 Quit GeekShadow (Read error: Connection reset by peer)
12:52:57***Saving seen data "./dancer.seen"
12:53:44 Quit Kitr88 (Ping timeout: 252 seconds)
12:55:46 Join xavieran [0] (
12:57:53 Join moos [0] (moos@rockbox/staff/moos)
12:59:37 Join Kitar|st [0] (
13:07:36 Join dfkt [0] (dfkt@unaffiliated/dfkt)
13:11:04 Join kugel [0] (~kugel@rockbox/developer/kugel)
13:11:19 Quit kramer3d (Ping timeout: 246 seconds)
13:11:40CIA-88New commit by 03teru (r24616): chopper: fix FS #11000: Drawing issue with steep mode. don't change level while in game.
13:15:49 Join kramer3d [0] (~kramer@unaffiliated/kramer3d)
13:16:03archivatorAm I to assume I need to generate *all* manual screenshots for my plugin by hand?
13:16:35archivatorNo automated way to build all the simulators and take a screenshot?
13:17:41 Join teru0 [0] (
13:17:54 Quit teru0 (Client Quit)
13:18:37Zagorarchivator: unfortunately not
13:18:40 Quit teru (Ping timeout: 246 seconds)
13:18:44Zagorit's a nice idea though
13:19:28archivatorI guess you'll have to manage without screenshots for now.
13:19:48archivatorAlso, we need a way to convert #ifdef-based keymaps into LaTeX.
13:19:49 Join teru [0] (
13:23:42rasherI was working on it, but failed
13:24:14archivatorrasher: the first or second part? The ifdef parsing shouldn't be too hard..
13:24:16rasherFS #10575 and FS #10571
13:25:09archivatorAh, what was the problem?
13:26:18rasherI was trying to add a STARTPLUGIN event to the event queue, but it didn't quite pan out
13:26:33rasherI was getting it and starting the plugin from the wrong context or something like that
13:27:43rasherSo the plugin would crash, or not accept keypresses
13:27:58rasherBut starting the plugin by stuffing the appropriate keypresses in the button queue pretty much worked
13:28:09rasherNot terribly ideal though
13:29:31 Join Sajber^ [0] (
13:30:13archivatorI see. I might look into it, it'd be pretty cool if we got it to work consistently.
13:31:11rasherI've about exhausted my coding skills at this point
13:39:15 Join Rob2222 [0] (
13:40:12Unhelpfularchivator: log(x) isn't too hard. use __builtin_clz(x) to get log2(x), and then you can use some tricks to get a few more bits. if you know your possible input range then scaling to fit the screen won't be hard. i can give you a hand with this if you want, i've worked on fixed-point logs for aac a bit, and the algorithm i was working on could add a scale with fairly small cost if the scale can be determined at compile time.
13:40:23 Join wind [0] (
13:40:38 Join Casainho [0] (
13:41:31archivatorUnhelpful: thanks but I'm afraid it will have to wait. Can't put too much time into this plugin right now.. :(
13:41:47Casainhohello :-)
13:42:22Casainhocan someone please explain to me what are "manifols" on 3D? looks like that can give problems when processing the STL on skeinforge
13:43:00Unhelpfularchivator: well, i'll take a look anyway ;)
13:43:32archivatorOut of curiosity, is there a manual build for every target in tools/configure? I'm basing the manual entry for fft on the oscilloscope and I have quite a few more keymaps than the oscilloscope..
13:44:21Casainho(sorry worng channel)
13:44:38archivatorUnhelpful: I'm afraid you won't like the math in there. I never did clean it up properly and right now it just clips when the result overflows. If you figure it out, it would be most appreciated :)
13:46:10Unhelpfularchivator: hrm. do you know what the actual peak value would be?
13:47:06Unhelpfuloh, dur... this is a frequency/amplitude plot, isn't it? the peak value is determined by your fft size...
13:47:22Unhelpfulwell, the maximum coordinate, that is.
13:48:14archivatorThe maximum amplitude you mean?
13:48:34 Quit krazykit (Ping timeout: 258 seconds)
13:49:30archivatorIt shouldn't be determined by the fft size. However, it *is* bounded by the algorithm. 2^11 I think was the upper bound. I never did calculate it properly, it was just a good enough bound for practical reasons. I think it's tighter than that though.
13:49:52Unhelpfuli thought we were talking about log-scaling the frequency axis of the plot?
13:51:05archivatorAh, I thought you were looking at the calc_magnitudes bit. Then yes, the maximum frequency is determined by the transform.
13:51:21archivatorWell, not exactly.
13:52:02archivatorThe Nyquist frequency is the maximum frequency. The value in the last bin kinda represents Nyquist/FFT_SIZE frequencies
13:53:25 Quit kramer3d (Quit: Leaving)
13:53:27Unhelpfulwell, anyway, we're talking about log-scaling N input points to N output points, with both known. an LUT might cover this, really. ;)
13:54:05CIA-88New commit by 03teru (r24617): "remote_control: don't use goto.
13:55:06 Join krazykit [0] (
13:56:02 Quit kaniini (Quit: (@Syvere-Home) täytyy sanoo et noi applen tuotteet on niin ultimaalisen paskaa laatua et ihmetyttää kyl et miten ihmiset ostaa noita)
13:56:09 Join kaniini [0] (
13:58:07*Unhelpful wonders if a bresenham's-like formulation of log2(x) exists/is possible
13:59:41Unhelpfulthat would be ideal, you'd do some heavy math once to get initial values and then be able to update the bin to plot from for each screen pixel rather quickly
14:13:01CIA-88New commit by 03kugel (r24618): Convert RINGBUF_* macros to inline functions, saving binsize and improving type safety.
14:14:38B4gderhow does inline functions save binsize compared to macros?
14:15:39dionoeaby not being inlined anymore? :)
14:16:39 Quit Strife89 (Read error: Connection reset by peer)
14:17:06 Join Strife89 [0] (
14:17:43CIA-88New commit by 03mc2739 (r24619): Fix red caused by r24615
14:18:07 Quit Rob2222 (Read error: Connection reset by peer)
14:18:31pixelmathe H300 database red looks very weird but has been compiled on the same machine so far
14:18:40 Join Rob2222 [0] (
14:18:57pixelmaah, someone just committed a fix
14:20:32pixelmathere's only one "database" target (and I didn't even know such a thing existed)
14:25:21 Join Schmogel [0] (
14:29:08CIA-88New commit by 03mc2739 (r24620): Updated Chinese translation ...
14:31:34 Join dfkt_ [0] (dfkt@unaffiliated/dfkt)
14:32:24 Quit dfkt (Disconnected by services)
14:32:31 Nick dfkt_ is now known as dfkt (dfkt@unaffiliated/dfkt)
14:33:48kugeldionoea: it's still inlined
14:33:55kugelI checked bloat-o-meter
14:35:00 Quit kugel (Disconnected by services)
14:35:03 Nick fxb__ is now known as fxb (
14:35:06 Join kugel [0] (
14:35:18 Quit kugel (Changing host)
14:35:18 Join kugel [0] (~kugel@rockbox/developer/kugel)
14:36:56kugelB4gder: I think it's mainly by letting gcc optimize argument-expressions more cleverly. if you pass an expression (a+b) it's expanded everytime the parameter is used in the macro. with inline, it will be only computed once, then "passed" to the inline
14:37:58B4gderthat sounds like speculation to me ;-)
14:38:31kugelbut I'm fairly sure actually :)
14:38:45B4gderwhy would it do worse just because it does it many times?
14:38:49gevaertsit is spec-ulation in the sense that the spec guarantees that :)
14:39:27gevaertshm does it? Maybe I'm misunderstanding the exact issue
14:39:36kugelB4gder: because it's better to do an addition only once instead of thrice
14:40:14B4gderproves the compiler is still pretty stupid I guess
14:40:41rasherI think we can all agree gcc isn't exactly superhuman
14:41:03kugel(((p)+(v))<buffer_len ? (p)+(v) : (p)+(v)-buffer_len): v is used thrice. if v is an expression it's possible evaluated thrice, in an inline it's only evaluated once
14:41:24B4gderright, but it wasn't _because_ you made it inline
14:41:29B4gderit was because you changed the logic
14:41:31B4gderto be fair
14:41:50kugelI checked binsize with plain converting
14:41:52B4gderyou could've made that logic in the macros too
14:42:04kugeland after making it more readable. it didn't make a difference
14:42:06B4gdernot that I would favour that, I prefer inlines
14:42:46Torneyah, B4gder is right, but it's way uglier if you do that in a macro :)
14:42:52B4gderoh yes
14:42:57B4gderit would be terrible
14:42:58kugelas I said, bloat-o-meter was my friend. it showed the "logic change" didn't make any difference, and that it's still inlined
14:43:07 Join Adnyxo [0] (
14:43:22B4gderyour conversion to inline changed the logic - to the better
14:43:37kugelI hope so :)
14:44:08kugelbut w.r.t to binsize, plain converting gave the same saving
14:44:57B4gderyou mean converting and maintaining all the operations on multiple places?
14:46:09 Join froggymana [0] (
14:46:21kugelI mean
14:46:46rasherClearly gcc isn't right in its head
14:46:54B4gderso what did it save in that case?
14:47:37kugelthe same
14:47:59B4gderso then your explanation about the "less additions" is wrong?
14:48:28B4gderthis isn't very important, it just puzzles my curiosity
14:48:41kugelno, it shows that my explaination is right :)
14:49:05B4gderif that pastebin saved binsize, and it uses the same amount of operations, how can that then be?
14:49:36B4gderit would rather show that gcc is better to "collect" expressions within the scope of a function than within the scope where the macros were used
14:49:54kugelthe additions are evaluated before "calling" the inline functions. that doesn't happen in macros. since there's no other code change it must be the less additiopns
14:50:34B4gderin your pastebin they are not evaluated before calling them
14:51:00kugelsure, due to the fact that it's an inline function (what gevaerts noted)
14:51:14B4gderbut so is a macro
14:51:26B4gdermore or less
14:51:58kugela macro is text insertion. the parameters are inserted as they're written
14:52:31B4gderbasically like an inlined function!
14:52:58***Saving seen data "./dancer.seen"
14:53:26gevaertsan inline function should still be a function, which has guarantees about multiple evaluation of argunments
14:53:32kugelnah, inline functions work more like compiling the function as is, and then inserting that binary code where it would've been called. so in the callers all computations for arguments would happen
14:54:33B4gderthat's not exactly how inline functions work though
14:54:57B4gderbut I'll drop this now anyway and go to pretend I'm working
15:00:02 Join domonoky [0] (~Domonoky@rockbox/developer/domonoky)
15:04:15kugelgrml, something in buffering is horribly broken (not related to my commit)
15:13:13 Quit DerPapst (Quit: Leaving.)
15:17:59kugelI don't get it. I protect against overwriting the next handle but it's getting garbled
15:22:07kugelthe question is why the handle is added at all if it doesn't fit
15:22:36kugelit looks like there's always 1 handle at the end for next track metadata no matter of whether the previous one finished properly
15:37:44kugelwhich shouldn't happen according to add_handle()!
15:40:16 Quit Sajber^ (Quit: Leaving.)
15:40:41kugelit seems I can get around the segfaults/dataaborts but the next track info is still garbage
15:41:38kugelmaybe the wps doesn't handle it properly
15:41:54CIA-88New commit by 03teru (r24621): jpeg,png: some minor changes.
15:43:03 Join Sajber^ [0] (
15:43:15 Quit Sajber^ (Client Quit)
15:43:22 Join evilnick_B [0] (~0c140464@rockbox/staff/evilnick)
15:49:48 Quit froggymana (Quit: CGI:IRC)
15:53:17 Join GeekShadow [0] (~Antoine@reactos/tester/GeekShadow)
16:01:13kugelinteresting. the wps is using the static mp3entry struct because the next track id3 isn't buffered. when skipping to 5s before the end of the current track, and back again it's magically using a buffered mp3entry struct (even though it didn't fit when playing the track for the first time)
16:01:46kugelI would expect it to use the static struct again after switchting back
16:04:18kugelso it looks like the next track metadata is getting buffered when skipping to the end, but not invalidated again when skipping back
16:08:58 Quit moos (Quit: ChatZilla 0.9.86 [Firefox 3.6/20100115144158])
16:11:04 Quit teru (Quit: Quit)
16:22:06 Quit GeekShadow (Read error: Connection reset by peer)
16:33:04 Quit zumbi (Ping timeout: 260 seconds)
16:39:44 Quit wind (Quit: CGI:IRC (EOF))
16:45:54kugelI think I have a sort-of fix
16:47:13kugelthis is all strange, I'm not sure my fix is corret
16:47:17kugelbut it seems to work
16:47:30kugelJdGordon: ping
16:48:23kugelJdGordon: (for the logs) the wps isn't notified when the track is rebuffered (e.g. by skipping back with a low-mem audio buffer), it needs to refresh it's metadata
16:53:00***Saving seen data "./dancer.seen"
16:55:21 Join GeekShadow [0] (~Antoine@reactos/tester/GeekShadow)
16:56:23 Quit Tomis (Quit: Tomis)
16:59:21 Join wojtek1991 [0] (
16:59:47 Nick fxb is now known as fxb__ (
16:59:55wojtek1991could you tell me how can I make my own dictionary for rb?
17:02:02wojtek1991I have read on forum but theres not many
17:02:04wojtek1991I have read on forum but theres not many
17:02:06DBUGEnqueued KICK wojtek1991
17:02:06wojtek1991I have read on forum but theres not many
17:02:09wojtek1991I have read on forum but theres not many
17:02:10***Alert Mode level 1
17:02:10wojtek1991I have read on forum but theres not many
17:02:15***Alert Mode level 2
17:02:15wojtek1991I have read on forum but theres not many
17:02:23dionoeawojtek1991: that kind of behavior will get you banned
17:02:24wojtek1991I have read on forum but theres not many
17:02:25wojtek1991I have read on forum but theres not many
17:02:25***Alert Mode level 3
17:02:25wojtek1991I have read on forum but theres not many
17:02:25***Alert Mode level 4
17:02:25***Alert Mode level 5
17:02:25wojtek1991I have read on forum but theres not many
17:02:26***Alert Mode level 6
17:02:26***Alert Mode level 7
17:02:26wojtek1991I have read on forum but theres not many
17:02:27***Alert Mode level 8
17:02:27wojtek1991I have read on forum but theres not many?
17:02:28***Alert Mode level 9
17:02:28wojtek1991I have read on forum but theres not many
17:02:28***Alert Mode level 10
17:02:28wojtek1991I have read on forum but theres not many
17:02:28***Alert Mode level 11
17:02:28***Alert Mode level 12
17:02:28wojtek1991I have read on forum but theres not many
17:02:34wojtek1991I have
17:02:35wojtek1991I have
17:02:38pixelmastop it
17:02:50 Quit wojtek1991 (Client Quit)
17:09:08 Join HontoniLinux [0] (~chatzilla@
17:09:48kugelFS #11004
17:12:29***Alert Mode OFF
17:13:23HontoniLinuxHeya, earlier today I got a sansa e280 so I figured I'd install rockbox on it now. In linux it showed up as the device "sg0" which it wouldn't let me mount with the error "not a block device".
17:13:46Tornesg0 is indeed not a block device
17:13:50 Quit Zagor (Quit: Clint excited)
17:14:42HontoniLinuxSo I now booted into Windows 7 and it finds the device nicely and I can browse the system, however instead of assigning it a drive letter it lists it as the "portable device" Sansa e280 under which I can access the filesystem.
17:15:00Torneyou haven't followed the instructions. you need to set it to MSC mode instead of MTP
17:15:31HontoniLinuxOh woops. Sorry about that.
17:15:43HontoniLinuxLet me go back and read/have a go at it.
17:16:36 Join NickK [0] (
17:16:58 Join wojtek1991 [0] (
17:17:11HontoniLinuxTorne: Cheers that did it, thanks :).
17:17:52NickKHey guys I got a question. Every time Im playing a song that i want to skip, i can't do so
17:17:53 Join stavrob [0] (
17:18:13NickKi press the next button but it just fast forwards
17:18:39NickKoh and I got an ipod 5ht gen by the way
17:18:48stavrobhi everyone, i am using rockbox 3.5 with the latest bootloader on an ipod 5.5g 80GB. I am trying to get the HID working but whenever I plug in a usb cable it just immediately boots into the ipod firmware
17:19:04Tornestavrob: the release builds don't use our usb mode
17:19:12Torneinstall a current development build
17:19:16GodEaterstavrob: which means there is no HID
17:19:24Tornenote that usb file transfers may be slower, and charging may nt work properly
17:19:26wojtek1991Does anybody know how to make dictionary file for rb?
17:19:33stavrobTorne: someone might want to change the 3.5 release notes in that case i think then
17:20:10stavrobin the whats new section it says: Added switchable USB HID keypad modes. (FS #10468)
17:20:25Torneand we did
17:20:28 Join funman [0] (~fun@rockbox/developer/funman)
17:20:39Torneit just happens that your particular player doens't have it enabled
17:20:45Torneit is on other devices.
17:20:58stavrobah. i see. maybe a note should be added about that then i suppose
17:20:59NickKdoes anyone have the same problem not being able to skip to the next song?
17:21:16 Quit petur (Quit: beer time!)
17:21:16stavrobNickK: maybe you have "prevent track skipping" enabled?
17:21:34Tornestavrob: lots of features listed only work on players which can support them
17:22:08NickKwow I feel pretty stupid now. Thanks a million Stavrob
17:22:09stavrobTorne: well its listed in the ipod 5.5g manual as well so thats why i was confused when it didnt work
17:22:12evilnick_Bwojtek1991: You didn't ask a question
17:22:23wojtek1991scroll up?
17:22:27wojtek1991I did
17:22:28Tornestavrob: it shouldn't be
17:22:36Tornestavrob: which manual?
17:22:42GodEaterevilnick_B: he did, it's just the same question he asked earlier, which you already answered.
17:23:00stavrobits in section 8.5.8
17:23:03Tornestavrob: yes, that's the manual for the current development version
17:23:04wojtek1991I asked about dictionary for for plugin
17:23:08Tornewhich does indeed have USB HID
17:23:14evilnick_BGodEater: Ta
17:23:29stavrobTorne: aha i see
17:23:46Tornealas it is also in the release manual.
17:23:58 Quit parafin (Read error: Operation timed out)
17:24:04Torneit shouldn't be, but hey :)
17:24:09 Quit saratoga (Quit: Java user signed off)
17:24:19NickKwhy is it every time I select a song to play in the database, It says "searching" followed by a countdown
17:24:25stavrobTorne: well i doubt many other people will have the problem, its quite a niche use
17:24:38wojtek1991Its too complicated. I must prepare file but how to get it
17:25:00 Join parafin [0] (
17:25:06Tornegevaerts: apparently undefining USE_ROCKBOX_USB doesn't disable the hid feature
17:25:17Tornegevaerts: so the 3.5 release manual still thinks it has HID
17:25:47wojtek1991If I have binary file whats I should do then
17:25:56evilnick_Bwojtek1991: It should be explained on the wiki page, but if you say exactly which bit is causing you problems then someone may be able to help
17:26:10 Join pamaury [0] (
17:26:34 Join Farthen [0] (
17:26:34Tornegevaerts: should usb_hid maybe be inside the #if defined(HAVE_USBSTACK) && defined(USE_ROCKBOX_USB) in features.txt?
17:26:44GodEaterwojtek1991: if you already have a binary file then you don't have to generate anything, so I don't understand your question
17:26:58Tornestavrob: sure, but it should've been removed automatically, so the features file might be wrong :)
17:27:54stavroboh no, now rockbox has control of my mouse
17:27:57stavrobhow do i get it back lol
17:28:03GodEaterevilnick_B: although, that wiki page says nothing about where your place the dictionary file on the player.
17:28:06wojtek1991Well on this page is written that I must have file .dec and .index but I havent got prolog version
17:28:08GodEaterand I certainly have no idea.
17:28:20wojtek1991my version has only .binary files
17:28:36GodEateroh no, my mistake, it does.
17:29:00GodEaterwojtek1991: where did you get your file?
17:29:21 Join utchybann [0] (
17:30:39wojtek1991its Polish file from university wordnet project
17:31:17funmanranmachan: (answering to the forum) i think we should use i2c registers to detect usb
17:31:25 Quit NickK (Quit: CGI:IRC (EOF))
17:32:16evilnick_Bwojtek1991: Does this part help? "The input format for rdf2binary is very simple at this moment. It's one line per word, starting with the word, then a tab and then the description. The only thing you should be aware of when creating this files is that they must be in alphabetical order, and all words should be in lowercase. "
17:33:43wojtek1991ok I understand
17:34:03wojtek1991and where put it? in file anything.desc ?
17:34:13ranmachanfunman: That's the AS3514 register then, right?
17:36:05funmanperhaps we should also use this method in mkamsboot' dualboot.S but an i2c driver would be larger
17:36:16 Part wojtek1991
17:36:52 Join wojtek1991 [0] (
17:37:05wojtek1991sorry my connection f***d down
17:37:16ranmachanfunman: BTW I think the OF handles LCD brightness using software PWM on the c240v2
17:37:29wojtek1991could you repeat you reply?
17:37:43funmanwojtek1991: see
17:37:57 Join zumbi [0] (~zumbi@
17:38:04ranmachanDCD15 has no effect and if I wave the sansa I can see the flicker (except for the highes brightness setting, where it's on continously apparently)
17:39:16wojtek1991sorry for spamming before I had a silly client
17:39:40wojtek1991well where I have to put this data? I meant name of file
17:39:54 Quit TheSeven (Quit: ChatZilla 0.9.86 [Firefox 3.5.7/20091221164558])
17:40:14wojtek1991evilnick_B are you there?
17:40:55evilnick_Bwojtek1991: I am, but I haven't tried this myself yet, so am seeing if there's any other advice I can give
17:41:38stavrobwhat is wrong with current rockbox usb support that prevents it from being included in releases?
17:41:57AlexPOn ipods
17:41:58 Join komputes [0] (~komputes@ubuntu/member/komputes)
17:42:04AlexPIt is enabled on other players
17:42:10AlexPIt is mainly due to charging
17:42:12stavroboh ok i didnt know that
17:42:32AlexPIpods don't charge properly under Rockbox, so can actually slowly discharge when being used over USB
17:43:17stavrobah. what about if you plug it in and hold down menu at the same time so it goes into charge mode. does that actually not charge it then?
17:43:25AlexPIt does, but again slowly
17:43:38funmanranmachan: do you want to patch usb_detect() ? I can test on clipv1 and fuzev1, and some developers here have an e200v2
17:44:01ranmachanfunman: Already on it :)
17:44:10funmannice :)
17:44:51funmani just got a clip+ but i want to understand what went wrong with JdGordon and mt's Clip+ before trying any code on it
17:45:12kugelor with mt's one
17:45:33CtcpIgnored 1 channel CTCP requests in 0 seconds at the last flood
17:45:33*funman slaps kugel for not reading the full sentence
17:45:38evilnick_Bwojtek1991: From what I can work out, those instructions are to firstly convert two files and into an .rdf file
17:45:40 Join toffe82 [0] (
17:45:52AlexPstavrob: It is fine if you want to continue using it, but if you want a quick charge the Apple firmware is quicker
17:46:15stavrobAlexP: what about if you plug it into a charger
17:46:19 Join anewuser [0] (~anewuser@unaffiliated/anewuser)
17:46:20AlexPstavrob: But this is being worked on, so at some point hopefully it'll be possible
17:46:23AlexPstavrob: Not sure
17:46:24evilnick_Bwojtek1991: And then to run rdf2binary on that .rdf file to get the .index and .desc files - which are the ones to be placed in ./rockbox/rocks/apps on the player
17:46:25funmankugel: i will check my assumption that we can address 1MB of RAM, but I seriously doubt it would be different between 2 Clip+ since it worked for FlynDice
17:46:28*kugel puts his head in shame
17:47:07wojtek1991is rss channel with subversion posts?
17:47:14evilnick_Bwojtek1991: But I have no idea how to go from your .binary file the required .index and .desc
17:48:26 Join addikt1ve [0] (~addikt1ve@
17:48:54addikt1vemy Nano 2G tells me "Use iTunes to restore"
17:48:55 Quit Farthen (Read error: Connection reset by peer)
17:48:57addikt1velooks like it crashed :)
17:49:00addikt1vewhat shall I do?
17:49:22*funman would "Use iTunes to restore"
17:49:34wojtek1991one the other side. How to make .desc and .index file myself
17:49:59wojtek1991if I would write all phrases myself
17:50:44evilnick_B wojtek1991: Maybe you could try d/ling the .desc and .index files linked to from the Wiki page and see if you can figure out how to add to/alter them?
17:50:58AlexPwojtek1991: evilnick_B already told you
17:51:07ranmachanI'm having some trouble with the system locking up, let's see if it works properly if I use ascodec_lock/unlock...
17:51:07wojtek1991ok I try
17:51:19evilnick_BAlexP: To be fair, that wiki page isn't terribly clear
17:51:26AlexP<evilnick_B> wojtek1991: Does this part help? "The input format for rdf2binary is very simple at this moment. It's one line per word, starting with the word, then a tab and then the description. The only thing you should be aware of when creating this files is that they must be in alphabetical order, and all words should be in lowercase. "
17:51:34AlexP<evilnick_B> wojtek1991: And then to run rdf2binary on that .rdf file to get the .index and .desc files - which are the ones to be placed in ./rockbox/rocks/apps on the player
17:51:49 Join Farthen [0] (
17:52:06AlexPevilnick_B: I haven't read it, but it seemed fairly reasonable from that
17:52:17funmanranmachan: also don't forget rockbox will reboot to OF when usb is plugged unless you press center button (i believe it is center)
17:52:27wojtek1991ok I got
17:53:10 Quit Farthen (Read error: Connection reset by peer)
17:53:20wojtek1991What aobut writing plugins? Ansi C? Where can I get compiler for wndows?
17:53:25archivatorIs this taking it too far?
17:53:32ranmachanfunman: That part works so far.
17:53:38archivatorItutomatically generated from the source file.
17:53:44archivatorIt's automatically *
17:55:01ranmachanBut if go into my pimped 'show port status' which also accesses ascodec it locks up... D'Oh. adc_read of course automatically takes the lock, need to move the lock down...
17:55:58funmanlocking multiple times shouldn't be a problem afaiu
17:56:39funmani think locking the same mutex several times in the same thread is allowed
17:56:45ranmachanfunman: It is if you try to take the lock twice because you called ascodec_lock and then call adc_read :)
17:56:51funmanprovided you unlock it the same number of times afterward
17:57:32ranmachanHmm, ok, anyway I moved the lock down, let's see if it helps...
17:58:18funmanfirmware/kernel.c:mutex_lock() just increases refcount and returns if the mutex is alreayd locked by the calling thread
17:59:28*kugel pm'd matsch about fs#11004
17:59:54 Part LinusN
17:59:57wojtek1991thanks man
18:00:04wojtek1991I try
18:01:07funmanarchivator: i don't understand what this is
18:01:13ranmachanHmm, if I comment out usb_detect the port dump works. If I don't go into the port dump screen the usb detect works.
18:01:30kugelarchivator: nice
18:02:12funmanranmachan: hm perhaps usb detection is called from the kernel tick, where mutexes can't work
18:02:16archivatorfunman: a button table for the manual, generated from the source file.
18:02:21funmanbecause the isr isn't a thread
18:02:33ranmachanpower-as3525.c also doen't take the ascodec_lock for some reason
18:02:38funmanarchivator: then it is nice, writing manual button tables is boring ..
18:03:03archivatorTell me about it, I couldn't even do one :(
18:03:38funmanranmachan: ascodec_*() do take the lock
18:03:58ranmachanfunman: It does work with the lock in usb_detect as long as I don't use the port debug menu where I also take the lock.
18:04:24ranmachanfunman: Ah ok, I thought I'd have to take it because it's used in adc_read, but it makes sense if you can take it twice.
18:04:30ranmachanOk, so I don't need to take it.
18:05:11 Quit Casainho (Quit: ChatZilla 0.9.86 [Firefox 3.6/20100115133306])
18:06:01ranmachanusb_detect() seems to be called from apps/main.c...
18:06:20 Join DerPapst [0] (
18:06:31 Join Horscht [0] (~Horscht2@xbmc/user/horscht)
18:06:38funmanFlynDice: unlike the clipv1/v2, the clip+ OF enables a fixed set of peripherals (which doesn't include GPIO) in CGU_PERI immediately after reset, perhaps that could play and we should disable CGU_PERI before branching to OF and wait a bit ?
18:06:58ranmachanAh and also in firmware/usb.c
18:07:31funmanranmachan: usb.c is the usb tick, so perhaps we have a problem there
18:08:14ranmachanYep, it's called from usb_tick
18:08:54wojtek1991Icanr get setup.ini.sig from
18:10:27funmanif the tick isr happens in the middle of an ascodec operation we can't use ascodec_*() in the isr
18:10:51funmanand we can't use mutex_lock() since the operation can't finish before the isr has returned
18:11:28wojtek1991does anybod have setup.ini.sig for cygwin?
18:13:11evilnick_Bwojtek1991: Using VMWare should be a lot faster
18:13:26evilnick_Bwojtek1991: More of a download initially, but it'll build quicker
18:14:15kugelmutexes can't be used in isrs afaik
18:15:27funmanwe could use a status bit and do not run ascodec_*() if an operation is in progress
18:17:01ranmachanOr just check if the i2c_lock is taken and use a nonlocking i2c_read?
18:17:34funmanthere's no "nonlocking" i2c read
18:17:45funmanyou can only have one operation active at a time
18:17:50ranmachanOr maybe add a thread for the usb detection and
18:18:25funmani would add a bool ascodec_active(void); and use the previous state if it returns true
18:19:50ranmachanThat won't work since you still would have to use the lock-taking ascodec_read from within interrupt context if it's not active if I'm not mistaken?
18:19:51kugelcouldn't you just post to a thread from the usb tick; the thread would then read and change a variable which is read by the next usb tick?
18:20:45funmanhm i'm not sure what happens if you use mutex_* in isr
18:20:52 Quit HontoniLinux (Quit: ChatZilla 0.9.86 [Firefox 3.6/20100115144158])
18:21:04ranmachanI'd have suggested adding a thread which polls less often than usb tick, set the global from there and just return the variable state.
18:21:42ranmachanWhat's the tick frequency?
18:21:54kugelfunman: it will simply freeze
18:22:29kugelif a thread owns the mutex, then the isr will wait for the mutex being released. that can't happen in interrupt context so it will wait forever
18:22:41kugelranmachan: 100Hz
18:23:31funmankugel: but in this scenario we could only take the lock if it's not taken yet
18:23:40ranmachanI'd assume 10Hz would be sufficiently fast for usb presence polling :)
18:24:20addikt1veiPod nano 2g tells me I have to restore it using iTunes. What shall I do?
18:24:31funmanaddikt1ve: restore it using iTunes
18:25:03addikt1vefunman: wait
18:25:13*kugel thinks the thread is more promising
18:25:24addikt1vehow can I use iTunes under linux?
18:25:27funmankugel: and less of a hack
18:25:30funmanaddikt1ve: you can't afaik
18:25:53addikt1veand that's EXACTLY why I'm asking my question once again: what shall I do? :)
18:26:08topikfind someone with itunes
18:26:09funmanfind a computer running windows and install itunes
18:26:28addikt1vedo I really need to do such a thing?
18:26:56topikif you want it working again, kinda yes
18:26:57wojtek1991is there any rss channel with Subversion posts?
18:27:05addikt1vecan gtkpod restore an iPod?
18:27:21kugelwojtek1991: you mean wihch announces commits to the rockbox source?
18:27:22funmanwojtek1991: i don't think so but there is
18:27:39funmanaddikt1ve: ask gtkpod people but i don't think so
18:27:46kugelwojtek1991: there's a mailing list which is effectively the same
18:28:00addikt1vegonna take a look
18:28:13wojtek1991ok but I would like to add it to my webpage
18:28:19ranmachanOk, I've whipped something up, let's see if it works...
18:28:35addikt1veactually my iPod isnt showing up
18:28:38wojtek1991I know how to do it with rss but what about mail list?
18:29:12kugelwojtek1991: you can look at svn:// in our page does it
18:29:39topikaddikt1ve: google has a couple of hits on 'restore ipod without itunes'
18:29:49ranmachanfunman: BTW how about using an interrupt instead? After all the hardware can generated one on usb status change...
18:30:24funmani don't know how to use interrupts with as3514
18:30:29addikt1vetopik: that's what I was looking for
18:30:33funmanthe i2c module isn't documented
18:30:39addikt1velooking at*
18:30:43addikt1veplease forgive my english
18:30:53funmanthe registers names are in the datasheet but not their content (bits)
18:30:57kugelranmachan: you mean using the gpio change to initiate a i2c read?
18:31:06topikTheSeven would probably say it can be done by writing some of his special voodoo to the ipod directly
18:31:21topikit would still have to show up as an usb device though
18:31:47wojtek1991I give up
18:32:57 Join TheSeven [0] (~theseven@rockbox/developer/TheSeven)
18:33:28ranmachankugel: Well the problem is I couldn't find a gpio for usb status on my device.
18:34:12ranmachankugel: But as3525 datasheet says the as3514 part can generate an interrupt if the usb_status in irq_enrd_0 changes.
18:34:54funmanif you find how to setup interrupts with i2c, fine
18:35:34funmani think it's doable with a bit of reverse engineering
18:35:36kugelranmachan: have you tried changing the polarity of the pin before reading?
18:35:40 Join bertrik [0] (
18:36:14JdGordonkugel: more details on the issues you mentioned
18:36:18ranmachankugel: No, I tried debug view I/O ports as is.
18:36:33kugelthe old button driver of the fuze/e200v2 needed to write a 0 to all relevant pins before reading, otherwise it just wouldnt work
18:37:05 Part topik
18:37:16 Join topik [0] (awesome@
18:37:30kugelit's possible that a pin is 1 after reset or just changes to 1 after some time even if 1 would indicate usb, so I would try writing a 0 just before reading
18:37:44kugel(that can even be different from unit to unit)
18:39:22kugelJdGordon: does fs#11004 contain insufficient info?
18:39:23JdGordonUnhelpful: gevaerts: to avoid heaps of unallocs/rallocs for the glyphcache I thought the fonts would stay the same as now and just do a big allocation for each, the size of that block would vary with the font size, but average aroun 25K or something
18:40:16JdGordonoh, ok thats cool
18:40:49 Quit funman (Quit: free(random());)
18:41:21kugelwhat is cool?
18:41:36ranmachankugel: dualboot.S doesn't do that though.
18:41:57JdGordonI didnt see the task yet
18:41:59ranmachanIt only writes 0 to the direction register and then reads the port status.
18:42:10kugelthat could very well be because the c200v2s we had so far didn't need it
18:42:26ranmachanAnd it works fine for the right button that way
18:42:52kugelwe have like 3 c200v2s, so it's possible we simply missed it
18:44:16ranmachankugel: So you mean switch to output, write 0, switch to input and read the port status?
18:44:22kugeland, as I said already, the c200v2 is under unusable so we don't garuantee any correctness in the current code :)
18:44:41archivatorCould someone please take a look at and commit it if it's good enough? It's a little blurb in the manual about fft..
18:44:41JdGordonwe dont ever guarentee code correctness
18:45:06kugelranmachan: yes, you might need additional micro delays inbetween (like "volatile int i = 10; while (i−−);")
18:45:18archivatorThe touchscreen keymaps are not added but I am not sure how those work so I opted not to include them atm
18:45:21wojtek1991bye all
18:45:50 Part wojtek1991
18:47:36TheSevenarchivator: what's the difference between "IRIVER_H10_PAD" and "IRIVER H10 PAD"?
18:47:49TheSeventhis seems to have been used somewhat arbitrarily
18:48:00archivatorHmm, let me double check that :)
18:48:47archivatorThere isn't one with spaces on my end..
18:49:21bertrikin the c200v2 OF I do see a function that does something with both GPIO A5 and A7
18:49:32archivatorAhh! It's pastie's horrendous coloring that's doing it - there are actually underscores in there.
18:50:29JdGordoncan you call one lua script from another?
18:51:03 Join freddyb [0] (
18:51:10FlynDiceJdGordon: What size was your deceased clip+?
18:51:22JdGordon2G i tihnki
18:51:30 Join piotrekm [0] (~pm@unaffiliated/piotrekm)
18:52:54 Quit addikt1ve (Quit: WeeChat
18:53:04***Saving seen data "./dancer.seen"
19:00:59bertrikI'm seeing another function in the c200v2 OF where it chooses between GPIO A1 and A5 depending on some configuration value
19:01:21bertrikIt looks like there is more than 1 type of c200v2 around ...
19:01:28kugelJdGordon: I'm wondering if the change to playback.c is sane
19:01:49 Join m3dlg [0] (~m3dlg@
19:05:37freddybI wrote some code to enable mouse clicks on the background of the UI simulator. Would anyone else be interested in this? I mean should I post it on Flyspray?
19:06:44ranmachanbertrik: Might be that usb detect is supposed to be on A5 for me then.
19:07:36 Quit m3dlg (Ping timeout: 264 seconds)
19:08:10 Join m3dlg [0] (~m3dlg@
19:08:31 Join efyx_ [0] (
19:08:50ranmachankugel: I tried this: but still don't see a change on GPIOA when I plug in usb...
19:11:04 Join BHSPitMonkey [0] (~stephen@unaffiliated/bhspitmonkey)
19:11:32ranmachanOnly when I press menu or insert an microsd I see changes in GPIOA (even without the prior 0 write)
19:11:37kugel"GPIOA_DATA = data & dir;" ??
19:11:40archivatorIf anyone's interested, this is the current incarnation of the python ifdef-to-latex conversion script. It should work in 90% of the cases. The input should contain only the #if ... #endif block with the keymaps. The script will ignore touchscreen keymaps (or rather anything after the #endif)
19:12:23ranmachankugel: Well, GPIOA_DATA = data would probably suffice. Just to put the backlight enable back :)
19:12:51kugelwell, how should this work if you write the old data back before reading?
19:13:01 Quit GeekShadow (Read error: Connection reset by peer)
19:13:19ranmachanI've switched direction immediately before that, the bits in input mode should be unaffected by the write.
19:13:36ranmachanAnd they are for the menu switch and SD bit
19:14:29kugelso it shows GPIOA = 0x00 actually?
19:15:39ranmachanNo, it shows 0x86 or 0x8e (menu pressed) or 0x82 (microsd present) or 0x8a (microsd present and menu pressed)
19:16:38kugelso A7 is set, bertrik mentioned the OF does something with A7?
19:17:52ranmachankugel: A7 switches on the backlight
19:18:06bertrikthe OF seems to be very variable on what bits to set in GPIO A, depending on at least 5 variables that somehow define the player type
19:18:07ranmachan(DIR is 0x80)
19:18:59kugelthe dbop doesn't change too when usb is inserted, right?
19:20:13 Join panni_ [0] (
19:20:36bertrikIf anyone else wants to have a look, the type detection code is at 0x8000
19:20:49ranmachankugel: DBOP is unaffected by USB plugin, yes
19:21:12ranmachanStays at 0xF07F
19:25:02 Join mt [0] (~mtee@rockbox/developer/mt)
19:26:27gevaertsTorne: hm, from the manual point of view, I guess you're right. Technically, HID does not depend on USE_ROCKBOX_USB, since USE_ROCKBOX_USB should probably be USE_ROCKBOX_MSC. HID is really only noticeable on the USB screen, but it's always there
19:27:25 Quit m3dlg (Ping timeout: 256 seconds)
19:28:43 Quit kugel (Remote host closed the connection)
19:30:15 Quit pamaury (Ping timeout: 256 seconds)
19:32:19gevaertsJdGordon: perfect
19:34:06 Quit freddyb (Remote host closed the connection)
19:38:20archivatorAnyone willing to commit Or at least tell me what's wrong with it? Pretty pretty please? :)
19:38:46CIA-88New commit by 03mt (r24622): Remove svn:executable property
19:39:02 Join checker [0] (
19:39:17checkerare there currently any shuffle options?
19:39:26mtForgot to mention that this is in mdctexp branch ^
19:39:49 Join freddyb [0] (
19:42:06 Part watto
19:42:15 Join lantius [0] (~lantius@
19:42:20 Quit evilnick (Read error: Connection reset by peer)
19:42:44 Join evilnick [0] (
19:43:22lantiusi am trying to make a plugin for rhytmbox to support rockbox. I have found information about the database files, but I haven't found information about how the playlists are stored
19:43:29 Quit dmb_ (Ping timeout: 240 seconds)
19:43:39lantiusdo you know where is the specification?
19:43:52AlexPplaylists are just m3u files
19:43:55AlexPor m3u8
19:43:57archivatorlantius: the playlists are standard m3u
19:44:10lantiusok, thank you
19:44:32 Join JdGordon_ [0] (
19:45:44 Quit checker (Quit: CGI:IRC)
19:45:49JdGordon_kugel: which playback.c change?
19:49:13 Quit lantius (Quit: CGI:IRC 0.5.9 (2006/06/06))
19:49:19 Join lantius [0] (~lantius@
19:58:09freddybJdGordon_: I suspect he means FS #11004. I tried it on a couple songs and it seems to work...
19:58:11pixelmaarchivator: put it on the tracker - also I think this plugin is swcodec only (and I see no *_PAD of hwcodec targets) so the input is wrong and it looks to me that the *_PAD don't always match what we have in the manual (it differs sometimes from the c code), e.g. the for M5 and X5 the macro is only called IAUDIO_X5_PAD for the manual, currently. And sometimes also \Buttons are also called different, I believe.
19:58:57pixelmaI'd also wish that the button table code follows the new scheme which makes it easier to add a column for targets with remotes (which I think is even needed already for the H100/H300 and Gigabeat (F and X, not sure about the S)
19:59:38AlexPyeah, S too
19:59:46archivatorpixelma: will do. What new scheme are you referring to?
19:59:49pixelmaI really don't want to discourage you, already apreciate the work you put in there, if you stick it on the tracker someone else could finish it (maybe)
19:59:52AlexPIf there is no remote key then it needs to be a blank column
20:01:08pixelmaarchivator: at the end
20:01:18CIA-88New commit by 03bluebrother (r24623): Rockbox Utility polish translation update. ...
20:01:28archivatorpixelma: You're not discouraging me, I realize that the manual is an enormous project in and of itself. I will post on the tracker, along with the script. I might actually create a proper application to do the parsing...
20:01:59 Join phanboy4 [0] (
20:02:42pixelmabluebrother: do you mean a Polish translation update, or to polish some translation update ;)
20:04:01 Quit gevaerts (Ping timeout: 248 seconds)
20:04:05 Quit anewuser (Ping timeout: 252 seconds)
20:04:56 Join gevaerts [0] (~fg@rockbox/developer/gevaerts)
20:05:09pixelmaarchivator: great. Thanks. Maybe c code and manual PADs could be synced sometime
20:10:21 Join anewuser [0] (anewuser@unaffiliated/anewuser)
20:10:29 Quit gevaerts (Disconnected by services)
20:10:33freddybWhoops, it crashed: Data Abort @3068348C
20:10:38 Join gevaerts [0] (~fg@rockbox/developer/gevaerts)
20:12:38 Join Hacker258 [0] (
20:12:57ranmachanbertrik: Is there a size constraint on dualboot.S? I'm inclined to try using i2c to check usb status in dualboot.S tomorrow
20:14:08bertrikranmachan, I don't know but I don't think so. I do think we should keep the dualboot part very simple and I don't want to add i2c reading there
20:14:38 Quit komputes (Read error: Operation timed out)
20:15:03bertrikit's an extremely critical part that means the difference between bricking or not, so it should also be extremely simple IMO
20:15:23Hacker258Hi can anyone rewrite this code into a rockbox plugin please as i'm trying to lern . Its only 4 Lines , thankyou :)
20:15:48 Quit amiconn (Disconnected by services)
20:15:49 Join amiconn_ [0] (quassel@rockbox/developer/amiconn)
20:15:51 Quit pixelma (Disconnected by services)
20:15:51 Join pixelma_ [0] (quassel@rockbox/staff/pixelma)
20:16:09 Nick pixelma_ is now known as pixelma (quassel@rockbox/staff/pixelma)
20:16:11 Nick amiconn_ is now known as amiconn (quassel@rockbox/developer/amiconn)
20:16:58bertrikranmachan, I noticed something weird in your battery benchmark, it started at nearly 5V and ended much lower than usual for a liion battery, so the battery readout is different too.
20:17:15linuxstbHacker258: Have you read ?
20:17:51linuxstbHacker258: And then look at other plugins.
20:17:53Hacker258yes i know how to do the printf
20:17:59Hacker258its just the keyboard input
20:18:21Tornekbd_input is all you get
20:18:29Tornewhich is not very similar to ANSI C's input facilities
20:18:30 Join Tomis [0] (~Tomis@
20:18:39linuxstbHacker258: Your mp3 player has a keyboard?
20:18:58Torneany parsing you would have to do yourself; it' far preferable to avoid that kind of input entirely
20:19:20Hacker258no, can't I enter numbers using the onscreen keyboard??
20:20:40linuxstbYes, but the keyboard is designed for text entry. To be consistent with the rest of the rockbox UI, you would use a "settings list" - i.e. a long list of numbers the user chooses from.
20:21:10Hacker258how would I make a list from 1 to 31
20:21:39*AlexP knows people older than 31
20:22:17Hacker258I know but I want the persons date of birth, the code was just an example
20:22:27*linuxstb doesn't do UI...
20:22:42AlexPHacker258: Not including year presumably
20:23:03linuxstbBut you probably want to try and find the code for the date/time settings screen, and extract it from there.
20:25:39Torneyah, for that kind of thing it's probably better to have it displayed onscreen and to scroll through each set of digits
20:25:51Tornethat's a *lot* more code than scanf() though :)
20:26:45 Quit krazykit (Ping timeout: 265 seconds)
20:31:59 Join komputes [0] (~komputes@ubuntu/member/komputes)
20:32:30Hacker258can anyone post an example on how to do this? As time_menu.c is too complicated :S
20:33:59Tornethe point we were making is that this *is* a complicated thing to do.
20:34:08TorneAdapting user input systems to the interface of an mp3 player is quite hard
20:34:24TorneThere is no simple example we can give you because it's not possible to do it in a nice way and still have ti be simple
20:37:04 Join krazykit [0] (
20:37:32JdGordon_hacker258: you can use set_int from the plugin api to show a long list of numbers
20:37:36*bluebrother spots a problem with SystemInfo::platforms()
20:38:32Tornewell ok you can do that as well. but the UI from time_menu is nicer ;)
20:38:46gevaertsWill the sim use asm, or will it always go for the pure C version?
20:39:22Hacker258whats the best quick and dirty method available
20:39:39JdGordon_torne: you mean the split screen? or the ui to set the clock?
20:39:45Tornethe ui to set the clock
20:39:53JdGordon_thats terrible!
20:40:04Torne..maybe i am thinking of something else
20:40:06JdGordon_oh its painful to use except on a few select targets
20:40:47Tornei mean being able to move between components of the date and scroll through values
20:40:57Tornei haven't set the clock for ages so i may be misremembering ;)
20:41:05 Join advcomp2019 [0] (~advcomp20@unaffiliated/advcomp2019)
20:41:44 Quit advcomp2019_ (Ping timeout: 265 seconds)
20:46:15Hacker258what about the input in chessclock, how would i do that?
20:46:48Torneread the source code for chessclock
20:47:12linuxstbgevaerts: I think CONFIG_CPU is undefined on the sim, so no asm will be used.
20:47:42linuxstbgevaerts: You want rockbox-as-an-app, not a sim, so get working...
20:48:16gevaertslinuxstb: My first task is to find out why it's slow. I may have found it :)
20:49:02linuxstbgevaerts: I'm guessing this is the sim on ARM? (your n900) ?
20:49:04 Quit JdGordon_ (Quit: CGI:IRC (EOF))
20:49:44linuxstbI guess you could just define CONFIG_CPU and see what breaks...
20:50:31gevaertsor maybe just CPU_ARM and ARM_ARCH <something> (6?)
20:51:18linuxstbYes, that's probably better. I guess the two things you want are the arm-optimised generic functions (memset etc) in firmware/ and then the codec optimisations.
20:51:24 Quit krazykit (Ping timeout: 264 seconds)
20:51:26 Join krazykit [0] (
20:51:35*gevaerts tries
20:52:54 Quit Hacker258 (Quit: CGI:IRC)
20:53:06***Saving seen data "./dancer.seen"
20:53:11gevaertsyay, good start. apps/plugin.c:339: error: '__div0' undeclared here (not in a function)
20:54:47gevaertsmake -k to the rescue
20:55:51 Join JdGordon_ [0] (
21:02:45 Join dmb_ [0] (~Dmb@unaffiliated/dmb)
21:10:33 Quit komputes (Ping timeout: 256 seconds)
21:13:23 Join m3dlg [0] (~m3dlg@
21:14:36 Quit Tomis (Quit: Tomis)
21:17:01 Join Tomis [0] (~Tomis@
21:20:48 Quit BHSPitMonkey (Remote host closed the connection)
21:22:23 Quit utchybann (Ping timeout: 260 seconds)
21:22:28 Quit flydutch (Quit: /* empty */)
21:23:59gevaertshm, CPU_ARM doesn't really help much
21:24:16 Quit anewuser (Quit: =ooo Ï¢INTER ϾHIP 5iVE is OOON!!)
21:25:07CIA-88New commit by 03mt (r24624): branches/mdctexp : Modify cook codec to use the new mdct library, ~11% speedup on Sansa E200.
21:26:09JdGordon_gevaerts: I tried saying before that if SIMULATOR is defined then alot of the arm code would probably be ignored anyway
21:27:37gevaertsJdGordon_: in general, maybe, but not in the codecs I think. That means that this isn't a codec slowness as such
21:27:54 Quit ender` (Quit: It is difficult to produce a television documentary that is both incisive and probing when every twelve minutes one is interrupted by dancing rabbits singing about toilet paper.)
21:28:40gevaertsIt eats 17% CPU and causes pulseaudio to eat 17% CPU by just starting
21:28:58 Quit m3dlg (Ping timeout: 240 seconds)
21:30:03freddybI'm going to post the UI simulator patch if no one objects.
21:32:36JdGordon_maybe it just doesnt like the amount of threads or something?
21:33:28JdGordon_how much ram does the n900 have? and how much have you told the sim to use?
21:34:02JdGordon_I assume thats what you're building for?
21:34:07gevaertsit has 256MB, and I built a mostly-unmodified D2 sim
21:34:42gevaertsbut since it already eats lots of CPU by just sitting there, I think the big problems are either the sim audio driver or the sim lcd driver, or both
21:34:50JdGordon_maybe is swapping too much? (does it have swap space?)
21:35:15 Join Tomis2 [0] (~Tomis@
21:36:10gevaertsIt does have swap, yes. I doubt that that would be it though
21:36:12Strife89I edited the defines for the iPod Color before, but I forget where the file is.
21:36:27gevaertsSince pulseaudio is also using a lot, I tend to think it's the sound driver
21:36:30Strife89I need to change it again.
21:37:29 Quit Tomis (Ping timeout: 240 seconds)
21:37:29 Nick Tomis2 is now known as Tomis (~Tomis@
21:37:48 Join AlexP_ [0] (~ap@rockbox/staff/AlexP)
21:37:49gevaertshm, maybe it's just a priority issue
21:38:19CIA-88New commit by 03bluebrother (r24625): Fix problems with platform retrieval. ...
21:38:43 Quit AlexP (Ping timeout: 260 seconds)
21:40:07Strife89Never mind, found it.
21:44:14JdGordon_freddyb: nice patch! (havnt looked at the diff yet thuogh, only the description)
21:44:51JdGordon_do you rekon you can fix the problem where if you enable wps debugging in the sim and click in the lcd it should show the lcd pixel co-ordinates but really shows the window co-ords
21:50:15freddybJdGordon_: I don't know much about it. Is it just an offset for each player?
21:50:47JdGordon_yes, there is a #define for each player where the lcd should be
21:51:13JdGordon_also, your patch doesnt work for touchscreen targets?
21:51:54 Join Sajber^ [0] (
21:52:12freddybJdGordon_: I didn't get that far yet. I was looking for some interest before I did a bunch of work. I don't code as fast as you all.
21:53:03gevaertsfreddyb: I suspect that this is one of those patches that everyone has always wanted to do, only it's not extremely high on the priority list. Nice to see that someone finally does it :)
21:53:46gevaertsAs far as I am concerned, it's not wanted, it's needed!
21:53:46JdGordon_I bet im not the only one who clicks ont he sim buttons and wonders why nothing happens :p
21:54:04freddybOk, I try take some time and include all the players.
21:54:31*gevaerts has added a few comments
21:55:03JdGordon_also, remote buttons would be very useful
21:56:06CIA-88New commit by 03bluebrother (r24626): Update Rockbox Utility version to 1.2.5
21:57:59 Nick AlexP_ is now known as AlexP (~ap@rockbox/staff/AlexP)
21:58:13CIA-88New commit by 03bluebrother (r24627): Tag rbutil 1.2.5 release.
21:59:56 Join _zic [0] (
22:00:01 Quit _zic (Client Quit)
22:01:44 Join froggyman [0] (
22:03:35 Quit archivator (Remote host closed the connection)
22:05:09 Join moos [0] (moos@rockbox/staff/moos)
22:05:25freddybWhat target has buttons on remote?
22:06:10 Join archivator [0] (
22:06:36*domonoky would think that all remotes (in rockbox) have buttons..
22:11:14pixelmawhat remote would it be without buttons? ;)
22:12:15gevaertschances of that being useful are indeed remote
22:12:28freddybIt would be a great remote for my niece...
22:12:38AlexPgevaerts: Even for you.... :)
22:13:38JdGordon_this of course will eventually lead to swappable remote backdrops for the h300 (3 remote types)
22:13:40pixelmaIrivers H100 and H300, the Iaudios, the M:robe500 (100 uses the same remote I think but I don't know if it is implemented). There are some non-lcd remotes which are not simulated thoug afaik
22:14:02AlexPIs the gigabeat remote implemented in the sim?
22:14:08AlexPI don't think so
22:15:21pixelmathat's what I meant with "some non-lcd remotes", there are non-lcd remotes for the Archos Recorders too
22:16:19pixelmaand H10 (that's another one which I'm not sure if it is supported by Rockbox though)
22:21:32 Join astigous [0] (
22:23:37 Join castigous [0] (
22:23:57castigousis there anybody out there?
22:24:44B4gderlots of people are here
22:24:56TheSevenif you have a specific question, please just ask it.
22:26:31 Quit astigous (Quit: CGI:IRC (Ping timeout))
22:26:49castigousI just used the first time the new (?) *compressor* function of the 3.5 version and ...
22:27:20*JdGordon_ cant hold in his suspense
22:27:23castigousall the battery was dead in a matter of minutes (and I don't have my charger with me SNIFF) ...
22:27:41CIA-88New commit by 03mt (r24628): branches/mdctexp: Modify ATRAC3 codec to use the new MDCT library, ~1.2% speedup on Sansa E200.
22:28:09castigousis there any settings to the compressor (threshold, gain, knee, etc) that could help reduce the excessive cpu utilisation?
22:28:48archivatorcastigous: really? in a matter of minutes? or are you exaggerating a bit? :)
22:29:09castigous...because I really have a good use for this feature (L listened to classical and metal music, etc random mixed in a playlist)
22:30:21castigouswell it was at 47% when I turned it on, and +/-45 minutes later KABOOM, it depleted the whole battery and the player is brand new ...
22:31:11castigousI would REALLY love to use this but not if Iḿ going to be deprived of the use of my player...
22:31:36gevaertsWhich player is this?
22:31:36archivatorcastigous: what player is that and which version of rockbox are you running? (i.e., daily build, standard 3.5, etc)
22:32:43castigousso, any suggestions about it - the player is a brand new Sansa Fuze v1 with v3.5 and it runs pretty extensively (although I haven't measured it yet) without the compressor and with replaygain ...
22:33:17castigouswhich by the way kind of works but not as good as the compressor (it was working great for my situation but...)
22:33:26 Quit linuxguy3 (Ping timeout: 252 seconds)
22:34:30 Join linuxguy3 [0] (
22:38:16 Quit grndslm (Remote host closed the connection)
22:39:00castigousAdnyxo: so, no suggestions?
22:39:07castigousot comments?
22:39:43archivatorwell, you're the first to report anything like this afaik
22:40:26archivatorPersonally, i've been using the compressor since day 1 and can easily squeeze 10+ hours out of my ipod 5.5
22:42:25bluebrothercastigous: replaygain working worse than the compressor? You do have replaygain tags in your files, do you?
22:42:35castigousyeah, it was a *big* surprise to me when the voice told me I had 15% percent of battery and then a couple of minutes later (I dont remeber x%) and then shut itself off
22:43:03Unhelpfulcastigous: replaygain depends entirely on your files having tags. and it serves a different purpose than the compressor.
22:43:31castigousyes, I have replagain (track) applied by foobar and it works but not so much in a noisy environment and with the kind playlist genre mix that I usually do
22:44:53Unhelpfulwell, yes, in a noisy environment you might want the quiet *parts* of songs made louder, which is something replaygain is not intended to do
22:45:16castigousbut the compressor worjed pretty great (although with the max threshold), I could hear the pianissimos (super soft passages) of classical music, and later don become deaf with a modern mixed and mastered song
22:47:51*TheSeven will quickly check compressor power consumption on nano2g
22:48:29Llorean47% might not actually be 47%
22:48:44LloreanI know that several of my players don't report an accurate battery reading immediately upon boot.
22:49:05freddybJdGordon: I think I got your coordinates figured out on the simulator for −−debug-wps
22:49:09LloreanI'd suggest running a proper battery bench on the player with and without compressor just to see what the real difference in consumption is.
22:50:04castigousI like mixing genres because have lots of (different genres of) music to learn from and when in street I like to be surprised by the next song...
22:50:18TheSevenno noticable difference on nano2g
22:50:27JdGordon_freddyb: sweet :)
22:50:29TheSevenmaybe 1mA, but definitely not more
22:51:52TheSevenwith an without compressor: 18mA
22:53:07***Saving seen data "./dancer.seen"
22:53:31 Join komputes [0] (~komputes@ubuntu/member/komputes)
22:55:02 Join pamaury [0] (
22:57:44 Join perfectdrug [0] (
22:57:45castigousso, should I activate the debug mode? where I put the result, here or in the forums?
22:57:57AlexPWhat debug mode?
23:00:37togetichow do you get to work with rockbox?
23:01:43AlexPTurn on logging
23:01:54AlexPThen scrobble the resulting log file
23:03:17castigousHi, thanks I will do it (at home when I can charge it :)
23:04:31castigousThank you fellas, I need to go for now but I will try to consult the debug and post the results later
23:04:38castigousThanks you all, Bye!
23:04:51 Quit castigous (Quit: CGI:IRC)
23:04:52TheSevenAlexP: he probably means battery_bench
23:05:15AlexPyeah, maybe
23:06:29togeticAlexP: where do i turn on logging?
23:06:46perfectdrugand I think he thougt the log was his answer, and he tries to scobbles the batteryissue
23:06:55AlexPtogetic: It is in the menus somewhere, check the manual
23:07:10AlexPperfectdrug: Oh really :/
23:10:02togeticthanks AlexP
23:10:29AlexPno worries
23:10:36 Quit moos (Ping timeout: 265 seconds)
23:10:43 Quit komputes (Ping timeout: 272 seconds)
23:11:39CIA-88New commit by 03mt (r24629): branches/mdctexp : Remove svn:executable property.
23:14:28freddybGrrr. Why don't #ifdef's magically appear wherever they belong???
23:15:29archivatorfreddyb: there's a shortcut for that - C-x M-c M-ifdef
23:17:55freddybarchivator: Yeah, but you have to :wq emacs this_friggin_file.c first
23:18:43mtfreddyb: Nice patch. :)
23:19:18 Join petur [0] (
23:19:19 Quit petur (Changing host)
23:19:19 Join petur [0] (~peter@rockbox/developer/petur)
23:20:07JdGordon_just an FYI, I'm giong to commit multifont tomorow night or sunday unless someone objects
23:21:51 Quit krazykit (Read error: Connection reset by peer)
23:22:37AlexPJdGordon_: Did you still need whatever it was last night testing?
23:22:47 Quit bluebrother (Ping timeout: 260 seconds)
23:23:19JdGordon_I dont really remember what it was, so assume no
23:23:24 Join krazykit [0] (
23:23:43JdGordon_I tihnk it was to see if the latest fm patch broke it, but its muh more likely that its hwcodec only
23:23:49 Join komputes [0] (~komputes@ubuntu/member/komputes)
23:24:28AlexPOK :)
23:24:29 Join bluebrother [0] (
23:24:30 Quit bluebrother (Changing host)
23:24:30 Join bluebrother [0] (~dom@rockbox/developer/bluebrother)
23:33:01 Quit evilnick_B (Quit: Page closed)
23:34:53 Join moos [0] (moos@rockbox/staff/moos)
23:35:10 Quit komputes (Ping timeout: 256 seconds)
23:36:23 Join komputes [0] (~komputes@ubuntu/member/komputes)
23:38:30 Quit mt (Quit: ChatZilla 0.9.86 [Firefox 3.5.7/20091221164558])
23:39:20 Quit froggyman (Quit: lost my left side but I'm all right now...)
23:42:59*bertrik is struggling with the radio on the c200v2 again
23:44:53archivatorcan a plugin load a codec and decode a compressed file?
23:45:48 Join p3tur [0] (~petur@rockbox/developer/petur)
23:48:25JdGordon_in theory it should be able to, probably need to add a bunch of functions t the plugin api to do it though
23:49:00gevaertshave a look at test_codec
23:50:30archivatorI just had a crazy idea - if it's possible, it might be possible to turn fft into a viewer (as well as its current mode) - that way you could take all the time in the world to do a proper analysis.
23:58:39TheSevenshould be possible. and if not, just make it possible ;-)

Previous day | Next day