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 2011-02-27

00:00:04 Join ender| [0] (
00:02:15 Quit TheLemonMan (Quit: free(me))
00:05:31 Quit CapsAdmun (Ping timeout: 240 seconds)
00:09:26 Join CapsAdmin [0] (
00:09:31***Saving seen data "./dancer.seen"
00:12:50AlexPleavittx: Thanks :)
00:15:00 Quit leavittx (Ping timeout: 246 seconds)
00:15:43CIA-70New commit by 03alex (r29415): FS #11969 by Lev Panov - Russian translation
00:16:10 Join [Saint] [0] (S_a_i_n_t@
00:16:26CIA-70New commit by 03alex (r29416): FS #11969 by Lev Panov - Russian translation
00:17:19 Quit simonrvn (Quit: see ya - n'multes)
00:17:20pixelmaI thought fml could have an eye on the Russian translation as I understood he had Russian roots... don't know if he has time before tomorrow evening though
00:17:32 Join simonrvn [0] (simon@2001:470:8c85:11fe::c0a8:195)
00:17:41 Quit Keripo (Quit: Leaving.)
00:19:00AlexPpixelma: I've no idea where he is from tbh :)
00:19:20CIA-70r29415 build result: All green
00:20:31 Quit pamaury (Remote host closed the connection)
00:20:38 Quit simonrvn (Client Quit)
00:20:48pixelmaI know he lives in Germany now and speaks German, his name suggest something Russian though and I saw commits to both languages by him
00:20:55bluebroth3ranyone familiar with ImageMagick? I'm trying to build a new icon set from the tango svgs, but converting the transparency is giving me headaches
00:22:15 Join simonrvn [0] (simon@2001:470:8c85:11fe::c0a8:195)
00:22:38 Join leavittx [0] (~lev@
00:24:50 Quit Buschel (Ping timeout: 276 seconds)
00:25:53 Join olejorgenb [0] (
00:33:36 Quit kevku (Quit: KVIrc 4.0.2 Insomnia
00:36:56 Quit L-Strife89 (Ping timeout: 264 seconds)
00:39:24[Saint]bluebroth3r: Heh, that's on my ToDo...go for it ;)
00:39:30 Quit FOAD (Remote host closed the connection)
00:40:32[Saint]I've been using InkScape personally...things are a bit hectic here as you might imagine, but, I'll get there.
00:43:18pixelmabluebroth3r: I don't know but in this case it might be quicker and give a nicer result if you use a GUI editor
00:43:35*[Saint] nods.
00:56:47bluebroth3rpixelma, [Saint]: I have it _almost_ working. The problem is that ImageMagick wants to treat transparency as 8 bit, while I want it to be 1 bit :/
00:57:37bluebroth3rapart from that I can now create icon sets at almost any sizes from the svg icons. And a 24x24 looks much nicer on the Beast sim with a larger font than the stock 12x12
00:58:01[Saint]bluebroth3r: It's almost a certainty that if you want it to look any good you'll need to do some manual editing after a batch process anyway.
00:58:13[Saint]Which is why I opt for a "full manual" approach.
00:58:45[Saint]bluebroth3r: Is my iconset crap?
00:58:58[Saint]there's one on the tracker y'know...
00:59:21bluebroth3rthere is?
00:59:35 Quit Xerion (Read error: No route to host)
00:59:42bluebroth3r12x12 is definitely crap if you have a 35 font :P
01:00:02bluebroth3rhmm, missed that. Though my intent was to create a small script that will generate it.
01:00:44[Saint]I've thought of doing the same...though my cli-foo is lacking.
01:01:08[Saint]I figured it'd need to be fixed up manually to a point that made it almost irrelevant.
01:01:10bluebroth3rin my case it's the ImageMagick foo
01:02:28[Saint]I guess the iconset on the tracker is easily missed, as you'd never find it without stumbling across it or specifically searching RaaA
01:02:56[Saint](and my use of iconset as opposed to icon)
01:04:35 Quit {phoenix} (Remote host closed the connection)
01:06:40 Quit bmbl (Quit: Verlassend)
01:10:13[Saint]bluebroth3r: Thoughts on that iconset are welcomed on it's's a "trial run" of what the cabbie iconset for all targets is (hopefully) to become.
01:10:37[Saint](keeping in mind of course thatnot all icons are set in stone, it may not look exactly then as it does now)
01:13:40 Join S_a_i_n_t [0] (S_a_i_n_t@
01:13:46 Quit [Saint] (Disconnected by services)
01:14:11S_a_i_n_tgah...damn earthquake-mo-net :/
01:29:21 Quit ender` (Quit: cd /pub && get beer)
01:29:44 Join Xerion [0] (
01:43:47 Quit domonoky (Read error: Connection reset by peer)
01:46:06 Join Leif [0] (
01:49:07 Quit liar (Ping timeout: 240 seconds)
01:51:23 Join MethoS- [0] (~clemens@
01:54:11S_a_i_n_tIs it correct that an iPod should shutdown from "Long Play" even if it's in a hard lock, if it's able to?
01:54:26S_a_i_n_t"Long Play" like "Menu+Select" is handled by HW, no?
01:55:31 Quit leavittx (Ping timeout: 264 seconds)
01:55:55JdGordonI'd have thought hold would stop that
01:56:04JdGordonlong play is software iirc
01:56:42S_a_i_n_tyes, hold does cancel it IIUC, but I'm of the opinion that long play may be HW too.
01:57:16S_a_i_n_tI'm pretty sure that if the player is hard locked, there's no harm in trying to shut down with long play first.
01:57:33S_a_i_n_t(I'm writing another "Menu+Select is EVIL" post.)
01:58:03S_a_i_n_tie. "please avoid doing so unless absolutely necessary, if you like your disk/data"
02:02:40JdGordonin that case how will long play help?
02:03:02JdGordonand being of an opinion when the fact is verfiable is just stupid
02:07:30TheSeven[01:54] <S_a_i_n_t> "Long Play" like "Menu+Select" is handled by HW, no? << wrong
02:08:08S_a_i_n_tYeah, I wasn't sure it was. Or that it wasn't.
02:09:03*TheSeven just squashed a nasty bug in his fat drvier patch
02:09:32***Saving seen data "./dancer.seen"
02:09:42*TheSeven still wonders if he should create a development branch for that one
02:09:49S_a_i_n_tI just cringe at posts like "I rebooted using menu+select" when the case suggests there was no reason to, and I wonder what failing it is, and where it exists that makes people do this instead of trying to use long play when they can.
02:10:29JdGordonwhat was the issue?
02:11:07JdGordonI've had playback lockups where the wps would keep scrolling lines and buttons would work but long+play to stop would make it unusable and you'd have to reset
02:11:11TheSevena sizeof() that was now looking at a char* instead of char[SECTOR_SIZE], e.g. returning 4 instead of SECTOR_SIZE
02:11:23S_a_i_n_tThe way I read it, "something happened that I didn't I reset it"
02:11:28TheSevenah, you meant S_a_i_n_t ;)
02:11:52JdGordonTheSeven: I always enjoy reading other peoples snafu's though :)
02:11:55S_a_i_n_t(accidently started recording while playing audio)
02:12:21 Quit Jerom (Quit: Leaving.)
02:12:58 Quit bertrik (Ping timeout: 240 seconds)
02:14:31S_a_i_n_tIt also makes me think of the issue I see from time to time of people thinking that it's ok to use menu+reset when rockbox asks for a reboot, ie. to enable a feature.
02:15:00S_a_i_n_tIt just makes me wonder if there actually does need to be a huge note in bold somewhere that says "Don't do it man!"
02:15:57TheSevenS_a_i_n_t: show a big bomb icon after holding it for 2 seconds :)
02:16:24S_a_i_n_tA picture of piecies of your disk flying off into space ;)
02:16:34S_a_i_n_t*pieces even
02:16:53 Join sinthetek [0] (
02:16:53 Quit sinthetek (Changing host)
02:16:53 Join sinthetek [0] (~sinthetek@unaffiliated/sinthetek)
02:19:19 Quit Horscht (Read error: Connection reset by peer)
02:19:34 Join Horscht [0] (~Horscht@xbmc/user/horscht)
02:21:11JdGordonstarting recording with audio going breaks things?
02:22:00 Join L-Strife89 [0] (~Strife89@
02:22:53 Quit DerPapst (Ping timeout: 240 seconds)
02:23:24 Join DerPapst [0] (
02:28:44 Quit DerPapst (Read error: Connection reset by peer)
02:29:29 Quit sinthetek (Quit: leaving)
02:30:15 Join DerPapst [0] (
02:30:22 Join sinthetek [0] (~sinthetek@unaffiliated/sinthetek)
02:32:12 Part sinthetek
02:35:12 Quit matsl (Ping timeout: 252 seconds)
02:35:56 Quit L-Strife89 (Quit: Heading back, finally.)
02:42:28 Quit mshathlonxp (Quit: fall into sleep)
02:43:35 Quit sasquatch (Ping timeout: 250 seconds)
02:45:00 Join komputes [0] (~komputes@ubuntu/member/komputes)
02:59:50 Join sasquatch [0] (
03:14:05 Join DerPapst1 [0] (
03:14:36 Join L-Strife89 [0] (~Strife89@
03:15:51 Quit DerPapst (Ping timeout: 240 seconds)
03:21:04 Quit Topy44 (Read error: Connection reset by peer)
03:21:30 Join Topy44 [0] (~Topy44@
03:23:33 Join Keripo [0] (
03:30:42 Join Keripo1 [0] (
03:32:33 Quit Keripo (Ping timeout: 260 seconds)
03:41:35*TheSeven would like people to review/test
03:42:16TheSeven(FAT driver rework to get rid of on-stack sector buffers, and optionally force cacheline alignment of all storage accesses)
03:42:23*TheSeven pokes gevaerts
03:42:30S_a_i_n_tTheSeven: You might want to link to the task, and not the file.
03:42:52S_a_i_n_ta not so sane browser will just open that as a download.
03:43:25TheSevenfirefox just opens it as a text document and there is not much more than that file in the task
03:43:37TheSevenand you can just remove the ?getfile part
03:44:57TheSevenit seems to be running smoothly for me now, so unless we spot some more bugs soon, I'd like to get it into SVN (possibly as a branch if we don't want to have it in trunk yet)
03:46:57 Quit timccc (Ping timeout: 240 seconds)
03:47:18JdGordonstraight after 3.8 branch is a good time for something like this to go i
03:48:46 Quit Antibuddha (Remote host closed the connection)
03:49:15TheSevenJdGordon: the question is if we want to risk potential file system corruption for current builds of all targets
03:49:44TheSeveni think it deserves a good deal more testing
03:50:19 Quit tchan (Quit: WeeChat 0.3.4)
03:52:17 Join tchan [0] (~tchan@lunar-linux/developer/tchan)
03:56:26 Quit GeekShadow (Quit: The cake is a lie !)
03:57:59JdGordonTheSeven: can you realistically get the testing you need without it going into the current builds?
04:09:30 Join timccc [0] (~timccc@
04:09:36***Saving seen data "./dancer.seen"
04:11:45 Join kugel_ [0] (~kugel@rockbox/developer/kugel)
04:13:13*kugel_ thinks us git users should either get write access to the git mirror or collect each other on one of those git hosting sites
04:14:16 Quit kugel (Ping timeout: 250 seconds)
04:14:44kugel_TheSeven: git :-) (re: development branch)
04:15:53JdGordonkugel_: i see the stuff about using the native statusbar is going in... how does it affect the core's idea of the lcd dimensions?
04:16:45kugel_JdGordon: I'm unfortunately out of business in this area, but my plan was to create a "sbs with X pixels of empty space which the host statusbar takes"
04:17:34JdGordonso the core still tihnks it is a full 800 (or whatever) high?
04:17:45kugel_how it affects the core/lcd dimensions seems to be target dependant. I can do without fixup (but with a minor glitch currently) on android, bt maemo seems to need to apply an offset in lcd_update()
04:18:36kugel_the problem is that the statusbar also change X/Y, not only width/height
04:19:00JdGordonit doesnt have to
04:19:11kugel_in the maemo case it seems to
04:19:46kugel_well, it's still very early and we haven't considered everything. my patch was just a prove of concept
04:20:51kugel_I think faking an sbs is the easiest way, especially if we want to keep support for %we/%wd, but I might be wrong
04:21:08JdGordonthat would be fun :p
04:21:29JdGordonalso I agree with pause removing the notification icon
04:21:42JdGordonhaving to stop is annoying while the widget doesnt resume from stop
04:21:49JdGordonunless thats been fixed and i need a new build?
04:22:05kugel_that's been fixed and you need a new build
04:22:25kugel_just today/tonight :P
04:22:50JdGordoncan i get one?
04:23:01kugel_the widget resumes from stop, even if rockbox wasn't even started before, since a few days
04:23:03JdGordonalso, can you do a 1024*600 one? :)
04:23:20JdGordon500x500px AA :D
04:23:21kugel_and has album art with my local work!
04:24:09kugel_sorry, I can't make you builds right now
04:25:52JdGordonwe really should have a proper discussion re the dynamic screen size patch
04:25:57kugel_I tried the few plugins that are enabled for RaaA
04:26:00JdGordonhaving to do seperate builds is just stipid
04:26:32kugel_they're all list based. does anyone disagree with defauling to absolute point mode for RaaA for plugins?
04:27:02JdGordonthat is the wrong fix... those plugins should enable stylus mode on entry
04:27:03kugel_it's just 3, IIRC. credits, properties and rfa
04:27:21kugel_JdGordon: that doesn't make sense
04:27:44JdGordonim sorry, my english too advanced for you? of course it does
04:27:49kugel_sorry. but if the core defaults to absolute, why should plugins need to do so explicitely
04:28:03 Quit amiconn (Disconnected by services)
04:28:04 Join amiconn_ [0] (quassel@rockbox/developer/amiconn)
04:28:05JdGordonbecause you *will* break something later
04:28:21 Nick amiconn_ is now known as amiconn (quassel@rockbox/developer/amiconn)
04:28:29kugel_we don't code for "later"
04:28:37JdGordonand RaaA is a stupid condition for this change anyway
04:28:43JdGordonRaaA != touchscreen
04:29:02 Quit pixelma (Disconnected by services)
04:29:04 Join pixelma_ [0] (quassel@rockbox/staff/pixelma)
04:29:06 Nick pixelma_ is now known as pixelma (quassel@rockbox/staff/pixelma)
04:29:23kugel_well, if you weren't then all touchscreens had absolute as default by now
04:30:05kugel_but ok, s/RaaA/all targets that default to absolute mode/
04:30:18JdGordonyou're missing a verb
04:31:20kugel_raaa is a good condition because it has a separate SOURCES for plugins
04:31:56JdGordonit sholdnt have a seperate SOURCES but does because of the majority of plugins never gonig to be fixed for it because people are lazy
04:31:59kugel_any plugin which is added to it will need proper absolute point mode conversion first
04:32:26JdGordonso what you're saying is you dont care about any touchscreen target except RaaA
04:32:42JdGordonthere is no reason why properties should work in abs mode for raa but not the d2 or mr500
04:33:04kugel_I'm sorry you still didn't realize that nothing changes for touchscreen if we just keep going
04:33:54JdGordonyou want to "fix" those plugins why changeing the default touch mode for a subset of the touchscreen targets
04:33:57kugel_if plugin X is adapted to absolute for X, then it's also adapted for Y and Z, of course
04:34:42kugel_properties (and credits and rfa) are automagically fixed because they use core widets
04:35:37JdGordonso we should default to abs on theon RaaA only because you are too lazy to add 5 lines of code to 3 files?
04:35:54JdGordonand therefore make them forced to grid on the non RaaA but touch targets?
04:38:59kugel_huh? grid mode is forced currently
04:39:46kugel_my point is to enable abs as default for RaaA because it only builds a subset of plugins (all of which are known to work with abs)
04:40:05 Quit timccc (Quit: Leaving.)
04:40:32JdGordonyou are saying "I dont care about the touch targets which arnt RaaA, these plugins dont deserve to ever work in abs mode on them"
04:41:09kugel_no, I'm saying the separate SOURCES for RaaA is a chance to adapt plugins step by step for abs
04:41:19kugel_starting with the ones known to work
04:41:39 Join DerPapst [0] (
04:42:13kugel_you have to admit that you haven't done *anything* for plugins on touchscreenn even though you own both a d2 and mr500 IIRC
04:43:23JdGordonriddle me this.... how is adding a wrong conditional to the code better than fixing the plugins so they will instantly just work on all plugins
04:43:57kugel_and I don't want to blame you, I just want to assert that fixing all plugins is too much work for a single (or even a few) person at once
04:44:14 Join timccc [0] (~timccc@
04:44:25JdGordonyou identified 3 plugins which requires <5 lines each to fix
04:44:31JdGordonthat is nothing but you being lazy and selfish
04:44:57kugel_alright, discussion stopped by now
04:45:01 Quit DerPapst1 (Ping timeout: 272 seconds)
04:45:54 Part timccc
04:46:07JdGordonfyi, I *will* revert any commit which adds anything along the lines of changing defaults based on RaaA
04:46:35 Join Barahir_ [0] (
04:47:33 Quit Barahir (Read error: Operation timed out)
04:48:32kugel_JdGordon: why?
04:48:54JdGordonbecause it is just wrong
04:49:20JdGordonsetting the touchscreen mode is not based on sdl or meamo, its based on it being a touchscreen
04:49:20*kugel_ thinks this will be a exciting log for others :)
04:49:42 Quit MethoS- (Remote host closed the connection)
04:50:01JdGordonthat is as stupid as suggesting that the default shuffle mode should be "on" for ipods and off for everything else
04:50:08kugel_JdGordon: the difference is that RaaA has separate sources, allowing us do to things step by step and not in a huge batch
04:50:21JdGordonthat is totally irrelevant
04:50:35JdGordonyou yourself said those THREE plugins use only core widgets
04:50:38kugel_if it's irrelevent why aren't plugins fixed already?
04:50:45JdGordontherefore its a fix once thing for all targets EVER
04:50:51kugel_*nobody* has done work on them
04:50:53 Join sinthetek [0] (~sinthetek@unaffiliated/sinthetek)
04:51:00kugel_because it's just too much to ask for
04:51:17kugel_step by step is a good plan in my book
04:51:18JdGordonjust admit that you dont care about the d2/mr500 and be done with it
04:51:52JdGordonor give me a good reason why credits shold work in abs mode on raa and grid on other touch targets
04:53:10kugel_there's no good reason; other than forcing all plugins to work immediately isn't feasable
04:53:31JdGordonread what I type, what what you think I'm typing
04:53:37JdGordonhave I ever said ALL plugins?
04:53:46JdGordonim talking about the THREE plugins you specifically mentioned
04:54:47 Quit TheSeven (Ping timeout: 264 seconds)
04:54:51kugel_oh there's another reason. having to hack into plugins to override what the core decides (even though the core works otherwise)
04:55:18kugel_the d2/mr500 default to grid, and now some plugins should work with abs?
04:55:28kugel_without even a possiblity to change that?
04:55:46kugel_the plugins should follow the core
04:55:51JdGordonfine, then they shold only change to abs if it was in abs before entering the plugin
04:56:17JdGordonyou're inventing a rule that the only touch targets which default to abs are raaa
04:56:24JdGordonwhich is wrong and stupid
04:57:09kugel_I wish you stoppped to insult
04:57:35JdGordonand if you even looked at settings_list.c you'd see how wrong you are
04:59:34 Join TheSeven [0] (~TheSeven@rockbox/developer/TheSeven)
04:59:52kugel_"04:30: <kugel_> but ok, s/RaaA/all targets that default to absolute mode/"
05:03:00JdGordonif (rb->global_settings->touch_mode == ABSOLUTE) "change to absolute mode" at the start of those plugins is too much effort?
05:03:34JdGordonthose THREE (3) plugins...
05:18:04 Quit DerPapst (Quit: Leaving.)
05:19:41 Join Rob2222 [0] (
05:24:01 Quit Rob2223 (Ping timeout: 276 seconds)
05:41:23 Quit JesusFreak316 (Ping timeout: 240 seconds)
05:42:09 Join JesusFreak316 [0] (
05:49:44 Quit Leif (Quit: Leaving)
06:09:37***Saving seen data "./dancer.seen"
06:11:25 Quit JesusFreak316 (Ping timeout: 240 seconds)
06:17:16 Join timccc [0] (~timccc@
06:31:55 Join Horschti [0] (~Horscht@xbmc/user/horscht)
06:35:47 Quit Horscht (Ping timeout: 260 seconds)
06:59:28 Quit timccc (Quit: Leaving.)
07:06:06 Join timccc [0] (~timccc@
07:19:21 Quit timccc (Read error: Connection reset by peer)
07:21:43 Join timccc [0] (~timccc@
07:28:46 Quit L-Strife89 (Quit: Bed.\)
07:31:01 Quit panni_ (Read error: Connection reset by peer)
07:53:23 Quit factor (Read error: Connection reset by peer)
07:54:28 Join factor [0] (~factor@
08:00:38 Join adiamas [0] (
08:01:52 Part adiamas
08:09:39***Saving seen data "./dancer.seen"
08:15:54 Join Llorean [0] (~DarkkOne@rockbox/user/Llorean)
08:16:04 Quit bluebroth3r (Ping timeout: 250 seconds)
08:18:01 Join bluebrother [0] (
08:18:01 Quit bluebrother (Changing host)
08:18:01 Join bluebrother [0] (~dom@rockbox/developer/bluebrother)
08:28:31 Quit factor (Read error: Connection reset by peer)
08:29:44 Join factor [0] (~factor@
08:42:36 Join Guest6621 [0] (
08:42:36 Quit Guest6621 (Changing host)
08:42:36 Join Guest6621 [0] (~bjst@rockbox/developer/Zagor)
08:42:49 Nick Guest6621 is now known as Zagor (~bjst@rockbox/developer/Zagor)
08:48:52 Quit t0rc (Quit: Give someone code, help them with one project. Teach someone to code, help them rule the world.)
09:01:28 Join robin0800 [0] (
09:07:52 Join matsl [0] (
09:13:10 Join stoffel [0] (
09:17:17 Quit Llorean (Read error: Connection reset by peer)
09:17:46 Join chocolaate-maan [0] (~cwo_F4@
09:17:47chocolaate-maanTHIS IS THE BEST U CAN GET
09:17:47 Part chocolaate-maan ("")
09:18:01 Join Llorean [0] (
09:19:24 Quit Llorean (Changing host)
09:19:24 Join Llorean [0] (~DarkkOne@rockbox/user/Llorean)
09:20:55S_a_i_n_tGah...that guy/bot/thing seems to be stalking Freenode...advertising a dodgy link for a ~3 year old IRC bouncer.
09:43:03 Join Buschel [0] (
09:48:18 Join T44 [0] (~Topy44@
09:51:25 Quit Topy44 (Ping timeout: 240 seconds)
09:53:51JdGordonpity we cant scrub the log :/
09:55:22S_a_i_n_tIt *can* be done, can't it? Though I understand it needs a Sweed.
09:55:41JdGordonwell sure
09:55:57S_a_i_n_tNot to be confused with a Thneed, the thing that everyone needs.
09:57:32S_a_i_n_tArgh...gotta love those guys.
09:57:45S_a_i_n_t"Hey, I saw you that one time, at that thing....friend me?"
09:57:54JdGordonwrong channel :)
09:58:00S_a_i_n_tWhoops...wrong channel ;)
09:59:03 Quit sasquatch (Quit: WeeChat 0.3.2)
09:59:29 Join sasquatch [0] (
10:08:15*JdGordon *highly* doubts credits will work in absolute mode
10:08:25JdGordonabsolute touchscreen mode that is
10:09:42***Saving seen data "./dancer.seen"
10:10:49JdGordoncan I change the sdl install dir on the configure command line?
10:17:29JdGordonhas anyone actually tried compiling the sdl (or any RaaA even) with the plugins?
10:17:33JdGordoncredits.raw missing
10:20:10JdGordongevaerts: pre-wakeup ping...
10:25:46amiconnS_a_i_n_t: Maybe it would be a good idea to replace the reboot hints with a yesno screen offering to reboot right away?
10:26:28amiconnGuess we'd also need to add the required firmware-level reboot function
10:27:34S_a_i_n_tamiconn: I had actually been thinking about that, those reboot prompts seem to be one of the times in particular where users think it's ok to hard reset.
10:27:36JdGordonhaha "warning: low battery" on sdl shutdown :D
10:29:38S_a_i_n_tThat would at least remove/hinder one of the major cases where I hear tell of menu+select reboot being missused.
10:29:42 Join TheLemonMan [0] (
10:30:39CIA-70New commit by 03jdgordon (r29417): Blindly forcing stylus mode on plugin entry for an arbitrary subset of touch targets is a *REALLY BAD IDEA* so do it properly so we dont have to ...
10:30:55 Join logiclost [0] (
10:31:35JdGordoncan anyone try building a full build on sdl/android/meamo please? credits.rock isnt compilng for me and I wonder if it compiles for anyone
10:32:12 Quit lostlogic (Ping timeout: 240 seconds)
10:34:15CIA-70r29417 build result: All green
10:38:36JdGordonah crap.. my appologies... wats the status of 3.8?
10:38:40JdGordondid we branch yet?
10:38:49S_a_i_n_tSure did.
10:38:53*JdGordon will happily call that a bug fix though :)
10:38:58JdGordonok, phew
10:39:19S_a_i_n_tbranch was...not yeaterday, but the day before? Something like this.
10:40:31Buschelwe branched last Sunday
10:41:04JdGordonBuschel: really unscientific but I havnt had any issues at all with the dma disabled patch
10:41:43Buschelhmm, how long did you run this patched build?
10:44:11JdGordon25hr straight last weekend, and then on and off since then
10:46:25 Quit CapsAdmin (Ping timeout: 250 seconds)
10:46:53 Join CapsAdmin [0] (
10:47:15Buscheland how often did you experience the lockups before?
10:47:50Buschel(just out of interest. this is still unscientific :)
10:48:26JdGordonvery randomly... when it owuld happen it would happen constantly while I tried plaing a certain album
10:48:37JdGordonwhich is why I thought it was a faulty sector or filesystem issue
10:49:38S_a_i_n_tis this still 11863 craziness?
10:58:06 Join Stummi [0] (~Stummi@rockbox/developer/Stummi)
11:01:00JdGordonBuschel: is there anything I can do to figure out what the problem is or make it repeatable?
11:04:06 Join mshathlonxp [0] (
11:04:35mshathlonxpAlexP I hope you haven't started to build 3.8 yet?
11:05:27mshathlonxpwhat why?
11:05:40JdGordonif it is already branched then it doesnt matter when the builds are built.. the code isnt changing
11:06:05mshathlonxpok, whatever
11:07:15 Quit Keripo1 (Quit: Leaving.)
11:09:14*mshathlonxp last night wrote down about three dozen of needed changes for he's translation
11:09:44mshathlonxpduring that process I also encountered 2 bugs and I wonder whatever these are known or not
11:09:54JdGordonah, welll that can be added
11:09:56JdGordonwhat bugs?
11:10:47mshathlonxp1) sometimes sound my "become lost" after recording mode has been started and quit - afterwards when playback is resumed, it plays, but no sound, unless player is restarted
11:10:56 Quit powell14ski (Ping timeout: 246 seconds)
11:11:17mshathlonxp2) my nano managed to crash when I started pictureflow
11:12:43mshathlonxpah, and actually I'm wondering why after language change to english it splashes "new language"
11:13:14mshathlonxpit would be better if in place of this would be "language changed" as has done russian translator, and I'll do the same :
11:16:09*mshathlonxp starts to implement needed changes in latvian translation, whatever these are being included in 3.8 or not
11:20:12mshathlonxpby the way, why equalizer again shows dotted squares in place of latvian signs - different default font for it or unability to use font that could support these signs there?
11:21:24TheSevenIIRC the equalizer screen always uses the sysfont for some reason
11:22:08 Quit matsl (Remote host closed the connection)
11:23:06S_a_i_n_tYeah, it'd be really nice to get rid of sysfont in all cases if there's a userfont set.
11:23:18JdGordonthose LANG strings should be LANG_SYSFONT_<something> which means you are stuck with using ascii
11:24:22S_a_i_n_tThough I'm not sure what's requred in doing so, or why the few screens that still use sysfont continue to do so. If it's just no one's done it, or if it's immensly difficult for some reason.
11:25:05S_a_i_n_tI'd say JdGordon probably knows at least half of that.
11:25:15JdGordonbit of both
11:25:28JdGordonnoone has been bothered to make sure the screen will scale with user font to any screen size
11:26:03mshathlonxpand why there's a couple of texts that aren't in lang files but is in firmware?
11:26:40JdGordonthat part of code cant access the translations
11:28:16TheSevenJdGordon: so you basically said that I should just commit FS #11958 tonight?
11:29:20JdGordonif you realistically wont get the testing you need otherwise sure... another code review probably wouldn't go astray though
11:29:40JdGordonand I'm not accepting blame for corrupt files (but we don't guarentee anything like that anyway so meh)
11:33:29*TheSeven wonders why the people usually dealing with that part of the code are hiding away
11:33:57JdGordonself preservation :)
11:54:21 Join liar [0] (
11:54:59*mshathlonxp sacrifices that text, that is shown during usb connection for that text, which is shown in menu
11:56:21mshathlonxpusual problem - text, that is identical in english, is created as just one entry :>
11:58:51 Join bertrik [0] (
11:58:51 Quit bertrik (Changing host)
11:58:51 Join bertrik [0] (~bertrik@rockbox/developer/bertrik)
11:59:10 Quit robin0800 (Quit: Leaving)
12:01:11 Quit user890104 ()
12:02:15 Join neoteny [0] (
12:03:24neotenyis 3.8 going to drop today? i was going to upgrade to the daily release but if 3.8 is going to drop i'll wait.
12:04:02JdGordonif not today, in the next few
12:04:26Zagorneoteny: yes, later tonight
12:05:19neotenythanks. what an awesome project. i can't donate enough.
12:05:36 Join ender` [0] (
12:06:29 Join ender1 [0] (
12:06:29 Quit ender` (Read error: Connection reset by peer)
12:08:23 Part neoteny ("Leaving")
12:08:24AlexPmshathlonxp: No, I haven't - if you have some last miinute transaltion changes pease let me have the asap
12:08:40mshathlonxphow ASAP?
12:08:54mshathlonxpI've done, but I would like to check them
12:09:03*mshathlonxp is going to compile lng now
12:09:05AlexPHow long do you need??
12:09:12mshathlonxp10-20 mins
12:09:16ZagorTheSeven: I'll see if I can dig up my old fat test code
12:09:39AlexPmshathlonxp: No problem, take your time
12:09:46***Saving seen data "./dancer.seen"
12:10:00mshathlonxpwhen I'll finish, again, new task in tracker?
12:10:11AlexPOr just on pastebin
12:10:25mshathlonxpI'll use tracker
12:10:31linuxstbTheSeven: IIUC, you've moved the sector buffer out of IRAM (the stack is generally in IRAM) ?
12:10:52BuschelJdGordon: if you should find any way to reproduce those lockups this would be perfect! I tried as well, and failed...
12:11:40TheSevenlinuxstb: most notably this allows me to reduce the ipod nano 2g stack size by 8K
12:13:33linuxstbTheSeven: Yes, but with a detrimental effect on other targets... I can't say how important it is for that buffer to be in IRAM, but my guess is that on some targets it's going to make a difference.
12:13:36TheSevenwe can probably reduce a couple of other stack sizes as well now, but currently that patch increases ramsize by a bit
12:13:59TheSevenlinuxstb: because of what? that buffer isn't really needed often
12:14:02linuxstbBut is there a reason to reduce stack size?
12:14:11TheSevenand if it is, we could move it to IRAM again with a simple define+ifdef
12:14:21TheSevenlinuxstb: to reduce ramsize?
12:14:33linuxstbTheSeven: Ah, so it's just a sector buffer that's used occasionally?
12:14:46TheSevenyes, only for various kinds of write accesses
12:15:12 Nick ender1 is now known as ender` (
12:17:35mshathlonxpwhat does do "Car Adapter Mode"?
12:18:24TheSevenlinuxstb: exhaustive list: mounting partitions, updating fsinfo, writing directory entries, expanding directories, renaming directories and usb storage presence detection
12:19:13mshathlonxpah, found
12:19:33S_a_i_n_tmshathlonxp: It powers the player on/off (and starts/stops playback?) based on the detection of power at the dock.
12:19:42mshathlonxpsounds bad
12:19:54mshathlonxpI mean, useless
12:19:55S_a_i_n_tIt does?
12:20:00mshathlonxpfrom my viewpoint
12:20:11*mshathlonxp owns 1987 Audi 90 quattro
12:20:17TheSevenmshathlonxp: it basically makes it behave like your car radio
12:20:30S_a_i_n_tIf you say's quite useful for a vehicle.
12:20:42mshathlonxpand as an audi, it has permanently powered cigarette lighter sockets
12:21:08TheSevenwell, there's nothing we can do about that
12:21:36mshathlonxpbut it's good - you can charge your mobile whenewer you need, not only with ignition on :)
12:21:53S_a_i_n_tmshathlonxp: Well, if you've wired the accessory power in a sane way, it's negated, that's quite obviously not the standard "accessory" setup for a vahicle.
12:22:16S_a_i_n_t*vehicle too
12:22:53 Quit stoffel (Ping timeout: 272 seconds)
12:23:11mshathlonxpwell, all audis so far I've seen has permanently powered cigarette lighters and I like it, but anyway, it's offtopic now ;>
12:24:07TheSevenmshathlonxp: there are not only cigarette lighters, but also some cars with ipod docks or something similar, which usually behave differently
12:26:43*mshathlonxp thinks how to alter he's firs translation of "car adapter mode", then changes are done
12:26:51JdGordonBuschel: how many targets seem to be affected? is it only CF-ed mini's?
12:27:51S_a_i_n_tJdGordon: It may or may not have been the reason why I had so much difficulty with my CF'd Color...but, we'll never know now :/
12:27:56linuxstbTheSeven: OK, then I agree performance isn't an issue.
12:28:23S_a_i_n_tI had lockups and FS corruption *real* bad on that thing.
12:28:41JdGordonI wonder if it can be runtime configured?
12:28:55JdGordoncompile both in and disable dma if its a CF or the user disbales it
12:29:50BuschelJdGordon: I had this very few times on my stock iPod Video. also there were issues reported with other PP targets as well
12:31:29 Join dfkt [0] (dfkt@unaffiliated/dfkt)
12:31:53mshathlonxpAlexP I've finished
12:34:45bluebrotherS_a_i_n_t: I've beaten ImageMagick!
12:34:56 Join Jerom [0] (~jerome@
12:35:29bluebrotherS_a_i_n_t: the auto-created image strip can need a bit of manual work afterwards, but now it's quite easy to create a new version just by running a script
12:35:47S_a_i_n_tbluebrother: Oh? You have a "make cabbie iconset work now" cli line?
12:35:51bluebrotherand it looks _way_ better with bigger icons with a big font :)
12:36:39S_a_i_n_tbluebrother: you just cut a few days solid off my ToDo
12:36:58S_a_i_n_tIf you do some default sizes, I'll be happy to clean them manually.
12:37:16bluebrotherbasically extract the Tango archive to the script folder, set the wanted result size and run it.
12:37:33S_a_i_n_t12x12, 16x16, 24x24, 36x36 and 64x64 were discussed I believe.
12:37:45CIA-70New commit by 03dave (r29418): FS #11966 - use separate modelnames and config files for RaaA. This adds an APPLICATION define that can be used in Makefiles, and target-name defines ...
12:38:10S_a_i_n_tI don't have imagemagick setup presently, or I'd have a play myself.
12:38:24S_a_i_n_tI'm a few machines down presently.
12:38:37JdGordonlinuxstb: I dont suppose you did a full sdl compile recently?
12:38:43JdGordon(or meamo, or android)
12:38:53bluebrothermy plan was to eventually commit that script to svn as helper for generating / updating iconstrips. I don't mind someone else to improve the iconstrip manually afterwards since the goal was to have an easy way for the initial (not necessarily beautiful) version
12:39:04AlexPThe 24x24 on the tracker is still much too small for 480x800 android, o I'd like some enormous ones for that please :)
12:39:14TheSevenS_a_i_n_t, JdGordon, Buschel: RE: CF DMA: are you sure that unaligned transfer buffers are being handled correctly? Occasional DMA corruption seems to point towards that direction
12:39:25TheSevenalso, do we know what exactly the corruption looks like?
12:39:58linuxstbJdGordon: Yes, I think so. What's the problem with the credits? I did get an error recently about credits.raw not existing, but deleting my build directory and starting again fixed it.
12:40:29S_a_i_n_tAlexP: Yes, this is definitely a good thing...I can't produce the strips here, but I'm happy to spend some time cleaning them up if someone else does.
12:40:31JdGordonTheSeven: I have no idea what the symtoms was... when it happened it looks like buffering dies, all the buffers are full according to the debug screen bnut the buffering or playback thread dies
12:40:44AlexPS_a_i_n_t: Cool, good stuff
12:40:45JdGordonlinuxstb: ah ok, ill nuke the dir and see if its fixed
12:41:12TheSeveninteresting. if the buffers are full, why should the problem be coming from storage?
12:41:28JdGordonI think I need to updat :/
12:42:10S_a_i_n_tAlexP: bluebrother: This is also good for giving new default icon sizes to targets that deserve it like the Video-ish sized targets etc.
12:42:22S_a_i_n_twhich is why I thought 16x16 would be handy
12:43:00S_a_i_n_tPresently, the iconsize is hardcoded at 24x24...but that can be changed.
12:43:07CIA-70r29418 build result: All green
12:51:52bluebrotherit's obviously visible that the smaller need more manual fixing. Maybe the easiest way would be to simply take the big version and scale it with Gimp or similar instead
12:52:19bluebrotherchanging the script to operate on big exports, then scale down afterwards might also be a good idea.
12:52:25JdGordonlinuxstb: ah, fixed, thanks
12:52:33 Quit Jerom (Quit: Leaving.)
12:53:06bluebrotheroh, and I changed a few icons simply because I didn't find the appropriate svg in the Tango iconset. Feel free to tell me what else to use :)
12:53:45S_a_i_n_tbluebrother: thanks, I'll pull them and have a look at them over the night/tomorrow.
12:54:17JdGordonwell here's a massive surprise.... credits in absolute mode is impossible to *do anything* in...
12:54:24*JdGordon loves being right
12:54:35bluebrotherbut as for ImageMagick, that's pretty standard on any linux installation. So I would guess you already have that installed
12:54:46JdGordon603 lines in credits! go team! :p
12:55:39S_a_i_n_tbluebrother: I had two dedicated linux boxes, they both took a tumble during the quake.
12:55:48S_a_i_n_tI've yet to revive them ;)
12:57:04 Join FOAD [0] (~dok@
12:57:44gevaertsTheSeven: pong?
12:57:50gevaertsJdGordon: gnip
12:57:57ender`who's in charge of build clients?
12:58:34gevaertsender`: anyone, technically, but mainly Zagor and me I guess
12:58:37JdGordongevaerts: dont worry, all good
12:58:45 Quit liar (Ping timeout: 240 seconds)
12:58:47bluebrotheris it possible to put .rockbox to the SD card on Android by default?
12:58:54ender`ok, because i'm getting "Server message: Hello deepthought-ender. Your build client has been temporarily blocked by the administrators due to: Some weird error. Build logs are very short. Please go to #rockbox to enable your client again."
12:59:34gevaertsender`: ah, yes. That one...
12:59:44*gevaerts looks if he can find an example
13:00:02 Join liar [0] (
13:00:03JdGordonbluebrother: .rockbox on android is /sdcard/rockbox
13:01:02bluebrotherJdGordon: hmm. Why am I failing to find .rockbox/icons on my device then?
13:01:15gevaertsender`: hm, it seems I can't get at build logs that old (even if it's only four days or so...) from the server...
13:01:29JdGordonbluebrother: is .rockbox inside /sdcard/rockbox perhaps?
13:01:59ender`yeah, looks like i was blocked on 21st
13:02:02bluebrotherJdGordon: no. Browsing with Rockbox and Show Files set to all, so I assume Android isn't hiding it from me
13:02:09gevaertsender`: what happened was that the build logs stopped at IIRC "Generating dependencies", but the build log scanning script found lots of errors and warnings. This happened three times on your client with various targets
13:02:58JdGordonbluebrother: ok, so your icon file is in /sdcard/rockbox/icons ? and in the config it is : /.rockbox/icons/foo.bmp?
13:03:20*linuxstb does a count of Android phone LCD screen sizes from - 23 are 480x800, 22 are 320x480, 13 are 240x320, 6 are 480x854, 3 are 240x400 and then 5 other LCD sizes have one device each
13:03:26bluebrotherJdGordon: no, that's the problem. I would expect the icons to be in /sdcard/rockbox/icons, but that folder isn't there at all
13:03:43bluebrotherhmm, maybe I should try creating it then :)
13:03:57JdGordonare you trying to find the stock icon file?
13:04:22ender`is there a way to force a build like the server triggers it?
13:04:24JdGordonif /sdcard/rockbox/icons exists it wont look in the /data/data/or.grockbox/icons folder
13:04:54JdGordonlinuxstb: dont forget 1024*600 :)
13:04:57bluebrotherthis android file structure is weird. I can't browse /data/data :/
13:05:05 Quit Buschel (Quit: ChatZilla 0.9.86 [Firefox 3.6.13/20101203075014])
13:05:30JdGordonthe filebrowser user doesnt have read access to that dir
13:05:38JdGordonandroid "security" ftw
13:05:41mshathlonxpis anyone going to accept FS #11970 ?
13:06:13gevaertsender`: I'm not sure what the best way is
13:06:24gevaertsMaybe Zagor can give some hints?
13:06:35 Join stoffel [0] (
13:08:55Zagorender`: not without writing a fake buildmaster and connecting to it
13:09:05CIA-70New commit by 03alex (r29420): FS #11970 by Mārtiņš Šimis - update Latvian
13:09:18Zagorbut it's not doing anything special
13:09:25ender`ok. i'll try doing a build by running the commands manually then
13:10:04gevaertsMaybe some special characters end up in the log somehow?
13:10:37TheSevengevaerts: ?gnop
13:10:37CIA-70New commit by 03alex (r29419): FS #11970 by Mārtiņš Šimis - update Latvian
13:10:59gevaertsTheSeven: you pinged me?
13:12:08bluebrotherhmm. I can load my custom cfg, but loading the icon strip fails.
13:12:17mshathlonxpmy name and surname looks good
13:12:19*mshathlonxp laughs
13:12:33amiconnJdGordon: Fyi, sysfont is not limited to us ascii, it supports latin1
13:13:22AlexPRight, I'm about to start buiilding - anyone got any last minute stuff?
13:13:28CIA-70r29420 build result: All green
13:14:22Zagormshathlonxp: yeah the web pages are still latin1
13:15:17TheSevengevaerts: FS #11958
13:15:41gevaertsTheSeven: yes. Some FAT experts would be useful :)
13:15:50pixelmaAlexP: does anyone remember/fixed something about this filesystem corruption problem, not sure about the detail. I believe it had to do with Rockbox USB and sideral was looking into it
13:17:24mshathlonxpby the way, how long is this thing that when you plug in ipod that's powered off, it starts rockbox and goes usb mode, not starts default firmware?
13:17:33AlexPpixelma: I don't remember that
13:17:39mshathlonxplast time I messed with ipod it did like that
13:18:15mshathlonxponly that was several months ago
13:18:57ender`can't really see anything weird here - can you reenable my client, and then poke me if it happens again?
13:19:53 Join MethoS- [0] (~clemens@
13:20:17JdGordonbluebrother: what size is the icons?
13:20:36CIA-70New commit by 03gevaerts (r29421): Re-enable deepthought-ender to help investigate the issue
13:21:42ender`says i'm still blocked
13:21:42 Quit mudd1 (Read error: Connection reset by peer)
13:22:10 Join mudd1 [0] (
13:22:20gevaertsI'm not sure how often the server checks
13:22:35 Join kevku [0] (~kevku@2001:7d0:0:f9af:feed:feed:feed:feed)
13:22:47 Join Jerom [0] (~jerome@
13:24:40ender`ok, it stopped complaining
13:29:54 Join DerPapst [0] (
13:30:11TheSevengevaerts: you know any of them?
13:30:27TheSevenTorne? pamaury? or who else is dealing with that?
13:31:17 Quit mudd1 (Ping timeout: 272 seconds)
13:33:07 Join mudd1 [0] (
13:41:23Zagorthe fat test code is still there (firmware/test/fat) but it doesn't compile since a couple of years. I'll look into that tonight.
13:41:29 Nick Horschti is now known as Horscht (~Horscht@xbmc/user/horscht)
13:44:54amiconnHmm, last time I tried it did compile... looks like that has been quite a while ago
13:51:59 Join teenbeat [0] (
13:52:35kugel_linuxstb: aren't the new defines redundant? there's also config_platform
13:53:59CIA-70New commit by 03alex (r29422): Tweak dansk.lang from rasher
13:54:10CIA-70New commit by 03alex (r29423): Tweak dansk.lang, from rasher
13:58:07kugel_also, I may have missed it, but when did we rename the port from maemo to n900/n8xx?
13:58:54CIA-70r29422 build result: All green
14:01:43gevaertsAlexP, Zagor: is the fontpack that rockbox utility downloads up to date or automatically synced?
14:02:02gevaertsIt would be a pity to announce a new font in the release notes that people don't actually get
14:02:07linuxstbkugel_: That was my previous commit - the one which added the separate targets to the configure menu.
14:02:30AlexPgevaerts: The release scripts build a fonts zip
14:02:41AlexPgevaerts: So I assume that needs putting somewhere :)
14:02:50AlexPBut I don't know where rbutil looks
14:05:43linuxstbkugel_: And I agree there is some duplication amongst all the application-related #defines - that's what I intend to look at next. But at the very least, the target-specific defines (declared as $target in configure) are needed to choose the correct config file.
14:06:15 Join n1s [0] (
14:06:15 Quit n1s (Changing host)
14:06:15 Join n1s [0] (~n1s@rockbox/developer/n1s)
14:09:50***Saving seen data "./dancer.seen"
14:12:18 Quit teenbeat (Quit: Leaving)
14:12:54linuxstbkugel_: There was also an unrelated change that I committed by mistake in that last commit - I removed your (I think) lines from apps.make to include the "hosted" directory, and used the existing "appextra" mechanism in configure instead.
14:13:17 Join domonoky [0] (~Domonoky@rockbox/developer/domonoky)
14:14:19AlexPZagor: Later with the builds, is there somewhere I can upload them to directly?
14:14:47AlexPOr would you prefer that I put them somewhere for you to grab them?
14:16:24 Join lixxus [0] (
14:28:55amiconnHmm, voice building only uses a single core. Potential speedup...
14:34:10n1samiconn: even with a make -j ?
14:35:49amiconnYes, because it's basically all handled by
14:36:23*amiconn could just start two voice builds in parallel though (e.g. two targets, or German and English)
14:37:56n1sbuilding a voice file here nether nor festival goes over 1% cpu usage
14:38:45amiconnIn cygwin using sapi it takes quite a while, and I see 50% cpu load quite often (one core fully loaded)
14:38:48gevaertsn1s: possibly the processes are short-lived enough for top to not make sense of them?
14:39:17amiconnParallelizing targets is probably not a good idea, because of the pool
14:40:19amiconnBut sharing the pool across languages doesn't make sense, so that split would work well
14:40:53n1sgevaerts: ah yes, looks like it creates a process per clip
14:41:18 Join benedikt93 [0] (~benedikt9@unaffiliated/benedikt93)
14:45:29kugel_isn't that quite expensive especially on cygwin?
14:47:48n1syes i think so
14:48:57bluebrotherAlexP: Rockbox Utility doesn't know about release font pack (or release bootloaders), it only knows about the main Rockbox release vs. current
14:49:12bluebrotherthis needs fixing, and it's on my todo list since quite a while
14:49:25AlexPbluebrother: So it'll just install the current font pack?
14:49:40AlexPNot a problem as long as the current one is up-to-date
14:49:41bluebrotherbut IMO the restucturing of the UI is a precondition for that, unless we want the UI to become a lot more complex
14:49:58gevaertsAlexP: indeed, but that probably needs to be checked
14:49:59bluebrotherAlexP: yes. Since fonts aren't depending on a special Rockbox build this shouldn't be an issue
14:50:00AlexPI wonder when the font pack is built?
14:50:33bluebrotherwould be good to have it built at least daily, so updated fonts get distributed soon :)
14:50:48AlexPand it might well be for all I know :)
14:53:57S_a_i_n_tit's ages old.
14:54:08S_a_i_n_tI've mentioned it quite some times ;)
14:56:20pixelmaAlexP: what's a "high CPU format"? ;)
14:57:11pixelmahmm, should have been in the other channel, it's forum related though ;)
14:57:23AlexPpixelma: a format that uses lots of cpu
15:04:33amiconnHmm, the Maemo port doesn't seem to exist in our wiki (and not on the frontpage either)
15:09:45pixelmathere seems to be a very weird bug in metadata handling after skips on my Ondio (testing the release branch). Going to test an swcodec target too but it'll take a but
15:09:49pixelmaa bit too
15:12:25pixelmaafter a skip e.g. the track number tag info contains part of the title tag string, year tag info just vanished, other tags I show on my WPS were ok though - and no problem on "normal" track changes
15:17:05CIA-70New commit by 03gevaerts (r29424): Simplified Chinese update by Li Jie
15:18:08AlexPNow if anyone else submits a language update once I've already done the builds I'll be very cross :)
15:18:15AlexPgevaerts: Thanks for committing :)
15:18:20CIA-70New commit by 03gevaerts (r29425): Simplified Chinese update by Li Jie
15:21:44 Join leavittx [0] (~lev@
15:22:48CIA-70r29425 build result: All green
15:23:21linuxstbamiconn: Yes, I think the only documentation is on the maemo wiki.
15:25:16AlexPaaargh, the bloody manuals are broken
15:25:55pixelmameh, the signal strength line addition makes the inbuilt hwcodec radio screen slightly unusable
15:26:03AlexPwe really really need the build system to report broken manuals
15:26:30linuxstbAlexP: Which ones? Or all of them?
15:26:45pixelmaI "surfed" some manuals today on the site and they looked ok (Clip+ and OndioFM I believe)
15:26:53AlexPlinuxstb: Not all of them:
15:27:07AlexPYou can see the date they were last built
15:27:33bertrikCan I help with that somehow?
15:29:21AlexPWork out what broke them
15:29:22pixelmabertrik: me? Hmm, it seems ok with the default font (I use a 1 pixel taller one but it looks really crammed
15:29:39AlexPThe odd thing is that it was in the last few days, and I don't see any manual committs there
15:29:43bertrikpixelma, I was talking to AlexP, sorry
15:30:01pixelmaI thought me because you added the signal strength to the radio
15:30:07linuxstbAlexP: I seem to be missing something - can you give me an example of one which is broken?
15:30:57AlexPlinuxstb: I can't no, as I'm being a muppet :)
15:31:04linuxstbAlexP: Thought so... ;)
15:31:32AlexPSorry, they actually look OK - there must be another issue at my end, and then I misread the dates
15:32:11AlexPbertrik: Sorry, ignore me
15:32:21bertrikok :)
15:32:28pixelmaI still dislike non-stop mode as default for manual building :\
15:33:45AlexPhah, mshathlonxp's name is breaking the manual building here
15:33:51bertrikpixelma, let's change that then, if it bothers you and is easy to fix! :)
15:34:05pixelmaAlexP: the online manuals seem to be missing the Credits and Licenses section though
15:34:05AlexPPackage inputenc Error: Unicode char \u8:ā not set up for use with LaTeX.
15:34:25AlexPpixelma: yep, as they stop building in credits due to mshathlonxp's name
15:35:09pixelmayeah, we had that before with some icelandic name and bluebrother fixed that by switching font encoding setting
15:35:20kugel_so a change to CREDITs breaks things?
15:35:21*AlexP summons bluebrother :)
15:35:47mshathlonxpwtf? :D
15:35:50AlexPkugel_: A new name in CREDITS has a character in it which breaks things
15:36:34*linuxstb wonders about the line "You are not allowed to distribute this software commercially" in part of the code in apps/plugins/chip8.c
15:36:44 Join straw [0] (
15:36:56kugel_so .lang/credits updates aren't as safe as we thought :)
15:37:09mshathlonxpwhat's the problem?
15:37:30AlexPmshathlonxp: Your name
15:37:40mshathlonxpyeah, I've noticed that loooong time ago
15:37:48*mshathlonxp hates to be called Martins Simis
15:37:54bertrikThat's what you get when demanding people to use their real names! :P
15:38:02mshathlonxphaha :D
15:38:09AlexPmshathlonxp: the "ā" breaks the manual :)
15:38:30strawplease help me, I want to increase font size in AF_SQUEEZE theme for Cowon D2, to listen music when driving, but I don't know what I have to do
15:38:38AlexPmshathlonxp: yes, the manual
15:38:44pixelmabertrik: got a different place to put it? The hwcodec radio screen has one more line because you actually record from the radio screen, so there are peakmeters (which only works on hwcodec currently anyway) and or prerecording info if set
15:38:50mshathlonxpah, there's manual building script that crashes because of my name? :>
15:39:02AlexPthe entire manual is written and built in latex
15:39:17AlexPAll the different manuals are built from the same source
15:39:30AlexPAnd the "ā" character in your name breaks them :)
15:39:45kugel_linuxstb: that's not quite compatible with gpl I'd say
15:39:52linuxstbstraw: It's not straightforward - the entire theme will be based around a particular font size. You will need edit the theme (probably a lot) - see
15:40:06linuxstbkugel_: Not at all compatible... I wonder how that got through originally.
15:40:30mshathlonxpI have no idea which one it is, because probably we are using different encodings here
15:40:38bertrikpixelma, I haven't really looked at how it looks for hwcodec, maybe we can abbreviate it, or just plain not use the RSSI tag for hwcodec radio screen
15:40:46mshathlonxpmy mirc just shows me a with umlaut and a square
15:40:59AlexPthe a with a bar over it
15:41:06AlexPI'm using UTF-8 btw
15:41:09mshathlonxpsecond letter in my name?
15:41:37n1sthere's some encoding issue with the release notes credit list too
15:41:43mshathlonxpand what are you going to do - change that script or change me to Martins Simis?
15:42:07n1sstuff like Tadeusz Py\x{015b}
15:42:11AlexPIf bluebrother is around in the next few hours I'll ask him if he can fix it
15:42:26AlexPIf not, I might have to just put an "a" :)
15:42:56AlexPn1s: That's his real name! :P
15:43:16n1soh, carry on then :)
15:43:26mshathlonxpAlexP are u sure it won't stop at "nj" then? :D
15:43:35*mshathlonxp is laughing
15:43:36AlexPnope :)
15:43:38gevaertsAlexP: does \bar{a} work?
15:44:07AlexPgevaerts: The problem is that this is generated from docs/CREDITS
15:44:15gevaertsLooking now
15:45:03mshathlonxpwell, if that really is from "aa" and nothing else, u guys could now regret that my mother didn't named me "Juris"
15:45:05*mshathlonxp laughs
15:45:49*gevaerts reminds mshathlonxp of the "real words" rule
15:46:25mshathlonxpand it is?
15:46:27kugel_linuxstb: it has a gpl notice too, and we distribute it under gpl, the author didn't complain yet
15:46:30*mshathlonxp has forgotten about that
15:49:54linuxstbkugel_: Well, it seems LinusN committed that code -;revision=7041
15:51:24pixelmakugel_: sorry I can't read the code, but could the last hunk in your embedded album art commit have to do with my weird tag issue on hwcodec (maybe other non-albumart targets)? I mean here_
15:52:14pixelmasee log at 15:09
15:53:05kugel_pixelma: I can't see how
15:56:13strawthank you linuxstb. Where I find fonts available foro Cowon D2 Rockbox to download?
16:00:34 Join hebz0rl [0] (
16:02:25 Quit antil33t (Read error: Connection reset by peer)
16:02:34 Join antil33t [0] (
16:02:51*bluebrother looks around
16:04:17n1sbluebrother: manual builds broken by unicode char in the credits
16:04:44bluebrotherjust spotted it in the backlog. Will check.
16:04:55 Nick kugel_ is now known as kugel (~kugel@rockbox/developer/kugel)
16:05:01AlexPbluebrother: thanks :)
16:06:02 Join PsyBlade [0] (
16:06:37bluebrotherhmm, we should really keep the LaTeX log files for manual builds
16:06:47AlexPyes :)
16:07:01linuxstbstraw: You should have a lot of fonts already installed - in the .rockbox/fonts directory. You can browse the fonts from the device itself - somewhere within the settings menu.
16:09:31PsyBladeyou are using latex inputenc(utf8)? try:" The \DeclareUnicodeCharacter takes UTF-8 code point as its first argument (in form of a a hexadecimal number) and the definition that this maps to as its second argument. For example, the code point 00E4 which is “LATIN SMALL LETTER A WITH DIAERESIS” would be set up via: \DeclareUnicodeCharacter{00E4}{\"a}"
16:09:52***Saving seen data "./dancer.seen"
16:12:02bluebrotherseems like it works with utf8x, but that is (afaik) not maintained anymore. IIRC we switched from utf8x to utf8 for exactly that reason
16:12:44AlexPbluebrother: We could just make mshathlonxp change his name :)
16:20:32bluebrotherAlexP: you never know what name the next contributor will have and how many people we will need to change their names in the future ... ;-)
16:20:33mshathlonxpwell, I won't be angry if you will edit my name for now, providing it remains intact in credits shown by player itself :)
16:20:55bluebrotherthat n with the dot below it is also causing problems.
16:21:07AlexPthis is true :)
16:21:37linuxstbMaybe simply asciify it (sorry mshathlonxp!) in the release branch, and look to fix it later?
16:22:18AlexPlinuxstb: Yes, if there isn't a quick fix I'll fudge it for release
16:22:43PsyBladewhy not try the manual declaration aproach I suggested
16:22:58AlexPPsyBlade: As the tex ius built from an external file
16:23:08AlexPthat is used for other stuff
16:23:25AlexPSo it needs to stay as the real character
16:24:01PsyBladeyes I know but then you map that char in the tex file
16:26:47bluebrotherAlexP: we can do that in the preamble
16:27:32AlexPOh, replace it that way round
16:27:38AlexPOK, I was thinking in reverse
16:31:07 Join pamaury [0] (
16:31:08 Quit pamaury (Changing host)
16:31:08 Join pamaury [0] (~quassel@rockbox/developer/pamaury)
16:37:59 Quit straw (Quit: CGI:IRC (EOF))
16:38:01 Quit Stummi (Quit: Bye!)
17:04:11bluebrotherhmm, so it seems inputenx is part of a standard LaTeX distribution. At least it's in my TeXLive 2007
17:04:48bluebrothercan anyone running debian check for inputenx.sty? If it's there I think I could go with that solution
17:05:10n1sbluebrother: sure, where should i check for it?
17:07:08bluebrothern1s: on my system it's /usr/share/texmf/tex/latex/oberdiek/inputenx.sty. Or just try locate :)
17:11:04amiconn/usr/share/texmf-texlive/tex/latex/oberdiek/inputenx.sty <= Debian squeeze x86_64
17:11:35 Join thomasjfox [0] (~thomasjfo@rockbox/developer/thomasjfox)
17:12:12n1syes, i have that too
17:12:33AlexPExists on a
17:13:01AlexPArch too at /usr/share/texmf-dist/tex/latex/oberdiek/
17:13:23 Quit kevku (Quit: KVIrc 4.0.2 Insomnia
17:13:38PsyBladeubuntu: texlive-latex-base: /usr/share/texmf-texlive/tex/latex/oberdiek/inputenx.sty
17:14:18bluebrotherok, then I'll switch to inputenx and also use ix-utf8enc.dfu
17:16:03CIA-70New commit by 03thomasjfox (r29426): RaaA: Don't specify the LCD size in the maemo package, it's done in the config header now
17:16:07CIA-70New commit by 03thomasjfox (r29427): Fix comment about LCD size
17:16:27amiconnthomasjfox: I have some questions regarding raaa-maemo
17:16:37thomasjfoxamiconn: Shoot!
17:16:58amiconnFirst of all, didn't you mention that 'make deb' should work now?
17:17:23thomasjfoxamiconn: You mean on a normal debian box or inside scratchbox?
17:17:32amiconnIt doesn't seem to (latest svn)
17:17:42amiconnIn scratchbox
17:18:01thomasjfoxamiconn: Use this command line: dpkg-buildpackage -rfakeroot -sa
17:18:59thomasjfoxamiconn: That's how you build packages inside scratchbox IMHO
17:19:25amiconn'make deb' could just do that, then
17:19:32amiconnWould be easier to remember
17:20:00amiconnAnother thing: It seems that codecs don't use the arm asm optimisations?
17:20:00thomasjfoxyes. or maybe "make maemo" so "make deb would be" free if/when we would every do a native debian package
17:20:10CIA-70r29426 build result: All green
17:20:41amiconnThe Makefile is generated. For a native debian build, it would do whatever would be necessary there
17:21:40thomasjfoxamiconn: Have a look at firmware/export/config.h, it defines CPU_ARM and ARM_ARCH 7
17:21:59amiconnHmm, at least ape.codec doesn't seem to use it
17:22:22amiconn-c5000 isn't realtime - it should be about 2.6* realtime on my n900 (oc'd to 750 MHz)
17:23:01thomasjfoxamiconn: let me add a #warning
17:24:11CIA-70r29427 build result: All green
17:25:16thomasjfoxamiconn: apps/codecs/demac/libdemac/filter.c:45:2: warning: #warning include ARM asm
17:25:30thomasjfoxamiconn: Looks ok to me
17:26:26thomasjfoxamiconn: Doesn't the N900 clock the CPU down if it's idle?
17:26:44amiconnCould it be that the .deb from the maemo repository doesn't have it yet?
17:27:25amiconnYes it does - but it isn't idle...
17:27:30thomasjfoxamiconn: ASM optimizations are in there from the first release IIRC
17:27:45thomasjfoxhow about the mp3 codec?
17:28:49amiconnI dunno - that's always fast enough
17:29:06*amiconn will disassemble the included ape codec
17:29:24pixelmathere is also this ligature problem still (with some PDF readers or so)
17:29:44 Join user890104 [0] (
17:30:09pixelmaabout the manual encoding problem
17:30:20*pixelma also wonders about cygwin again
17:30:31amiconnthomasjfox: Another problem I found is that pictureflow doesn't work
17:30:43thomasjfoxamiconn: Pictureflow is currently broken for all app builds
17:30:50amiconnIt wants to create its cache folder in a place where the user isn't allowed to create it
17:30:59thomasjfoxamiconn: We need to fix the data store directory location
17:31:09thomasjfoxamiconn: Yes, exactly
17:31:46pixelmakugel: because I don't see this last hunk in a HAVE_ALBUMART but I may be missing some thing and don't know what it actually does
17:32:06CIA-70New commit by 03bluebrother (r29428): Manual: switch to inputenx and load additional characters. ...
17:32:06kugelpixelma: the #ifdef came in a later commit
17:32:12bluebrotherAlexP: ok, manuals should be fine again now.
17:32:28bluebrotherat least that fix works for me. It's a bit ugly IMO but well ...
17:32:41pixelmakugel: ah, hmm
17:33:06bluebrotherbetter than starting to maintain our own list of additional glyphs IMO
17:33:12AlexPbluebrother: thanks
17:33:19AlexPI'll try in a bit :)
17:33:27pixelmakugel: sorry I didn't see that
17:33:36kugelno problem
17:33:43thomasjfoxamiconn: I'll try to find the build log and check if it defined MAEMO_ARM_BUILD
17:34:04bluebrotherAlexP: should I "backport" that to 3.8 branch or are you working at that anyway?
17:34:25AlexPIf you could that'd be great
17:34:33AlexPOr I can do it in a bit when I get to it :)
17:35:43bluebrotherwill do
17:36:18CIA-70r29428 build result: All green
17:36:25 Quit stoffel (Ping timeout: 240 seconds)
17:37:50amiconnthomasjfox: The maemo shipped build uses compiled C, not the asm optimisations
17:38:02thomasjfoxamiconn: Mist
17:38:10thomasjfoxamiconn: I'm currently checking the build logs
17:38:26thomasjfoxfor the record:
17:40:07thomasjfoxamiconn: What size is your /opt/rockbox/lib/rockbox/codecs/ape.codec?
17:40:16amiconnMy own build does use asm, so there's something fishy @
17:40:41thomasjfoxamiconn: Guess it's the MAEMO_ARM_BUILD detection as we have to do non-asm builds for x86, too
17:40:51thomasjfoxamiconn: always builds two versions
17:41:19amiconn55620 bytes
17:41:41thomasjfoxamiconn: Ok, that's the C version. It should be 30k
17:42:17amiconnIs that stripped? My own is ~110K
17:42:24 Quit liar (Quit: Leaving)
17:42:47 Join {phoenix} [0] (
17:42:53thomasjfoxamiconn: Yes, it should be stripped by the .deb creation process as it creates a -debuginfo package
17:43:22amiconnHmm, "dpkg-buildpackage -rfakeroot -sa" doesn't work here
17:46:39amiconnerror: cannot open debian/changelog to find format: No such file or directory
17:46:48*pixelma wonders if the hwcodec tag weirdness and the playlist viewer tag weirdness seen recently are related somehow
17:48:06CIA-70New commit by 03bluebrother (r29429): Manual: switch to inputenx and load additional characters. ...
17:49:06thomasjfoxamiconn: Did you run it in the top level of the source tree?
17:49:18amiconnNo, I ran it in my build dir
17:49:48amiconnHow else would it find the build's result?
17:50:16thomasjfoxamiconn: That's one of the issue with the debian build system: You have to run it in the top level dir as you can't specify an alternative "debian" directory
17:50:24thomasjfoxamiconn: It will create it's own build dir
17:50:43amiconnHmm, how am I supposed to build a maemo build then?
17:50:58amiconnBtw, it doesn't help that the maemo port isn't mentioned at all in the wiki
17:51:16thomasjfoxGo to the top level source directory and run "dpkg-buildpackage -rfakeroot -sa"
17:52:28thomasjfoxI'll have to leave for about half an hour and will investigate the ARM build issue on then
17:52:44thomasjfoxThanks for pointing it out amiconn!
17:52:47 Quit pamaury (Ping timeout: 276 seconds)
17:53:39amiconnPut an ape -c5000 file on your n900, and try to play it. Then you'll notice easily when a build doesn't use asm - it won't play realtime
17:53:59thomasjfoxamiconn: and yes, it's still not in the wiki. A minimal page should be created soonish
17:55:47 Join robin0800 [0] (
18:00:21 Quit factor (Ping timeout: 264 seconds)
18:00:46 Join factor [0] (~factor@
18:01:05mshathlonxpis finally problem with my name fixed? :D
18:02:55mshathlonxpby fixing script or my name?
18:03:10AlexPby chqnging the input pqckqge used
18:03:25mshathlonxpwhen 3.8 will be ready?
18:03:42AlexPWhen it is reqdy
18:03:58pixelmashitty azerty ;)
18:04:02mshathlonxp...and that could be by...?
18:04:05AlexPpixelma: yep :)
18:04:18AlexPmshathlonxp: There are never any ETAs
18:04:28AlexPSoon hopefully
18:04:34pixelmawhat about ETQs?
18:04:38mshathlonxpwtf is ETA?
18:04:47AlexPpixelma: They are even rarer
18:04:50AlexPmshathlonxp: google
18:04:54kugelthomasjfox: how about moving debian into a maemo dir which can also act as a build dir?
18:05:36pixelmaI find the weird metadata handling bug on hwcodec quite important for a release. Currently I'm trying to find the revision that broke it
18:06:00AlexPall the builds are done, I'm just doing manuals, then we need to upload
18:06:45mshathlonxpAlexP do you mean - Basque Homeland and Freedom: a terrorist organization organized in 1959 by student activists who were dissatisfied with the moderate nationalism of the traditional Basque party ?
18:06:58AlexPmshathlonxp: This is well off topic for this channel
18:07:07mshathlonxpI guess so
18:09:23 Join vnl [0] (
18:09:27 Part vnl
18:09:53***Saving seen data "./dancer.seen"
18:10:22amiconnthomasjfox: Hmm, I can't seem to find the created .deb?
18:10:54thomasjfoxamiconn: Believe it or not, it one directory above the top level source dir. Dunno if that can be fixed
18:10:54 Quit TheLemonMan (Ping timeout: 246 seconds)
18:11:25thomasjfoxkugel: Unfortunately it's not supported. That's what I wanted to do in the first place
18:11:30amiconnAlso, where does it take the weird version from?
18:11:37thomasjfoxamiconn: debian/changelog
18:11:56kugelthomasjfox: why?
18:12:32thomasjfoxkugel: Debian build tools don't support alternative "debian" directory locations AFAIK
18:12:45thomasjfoxkugel: "debian" has to be in the top level source directory
18:13:13kugelbut the build tools don't know what the top level is do they?
18:13:47thomasjfoxkugel: You mean like "symlink everything in a sub dir" or some other trick?
18:14:34kugelsymlink tricks will likely also work, but I mean how do the build tools know that maemo/ isn't the top level source dir?
18:15:03kugel(I imagine you could temporarily ln -s debian ../debian at make deb time)
18:15:36thomasjfoxkugel: The build tools create a .tgz of everything in the "top level dir" before starting the build
18:16:18linuxstbthomasjfox: What does it do with that?
18:16:19thomasjfoxkugel: Dunno if those symlinks would be compatible with the .tgz creation process
18:17:03amiconnthomasjfox: My own build uses asm fine. .deb building is really weird...
18:17:27thomasjfoxlinuxstb: It's the source tarball it later on invokes the debian/rules files on
18:18:03thomasjfoxamiconn: It's likely a problem with tools/configure line 577
18:18:06linuxstbthomasjfox: So it unarchives it somewhere else to do the actual build?
18:18:32*thomasjfox starts a .deb build
18:18:55 Join liar [0] (
18:19:04linuxstbthomasjfox: Which reminds me, the lcd size can no longer be set for the N900/N8xx from configure - is that a problem? (it's easy enough to put back if you want it).
18:19:23linuxstb(so that can be removed from the CONFIGURE_OPTIONS line in debian/rules)
18:20:02thomasjfoxlinuxstb: It's already removed ;)
18:20:46 Quit robin0800 (Quit: Leaving)
18:20:55linuxstbthomasjfox: So it is ;)
18:21:50thomasjfoxlinuxstb: I think it creates the .tgz "source tarball" and then just invokes debian/rules for the actual build process
18:22:01*thomasjfox wishes maemo would use rpm like meego does
18:23:12 Join t0rc [0] (~t0rc@unaffiliated/t0rc/x-5233201)
18:23:42linuxstbthomasjfox: What is your intention regarding using the maemo status bar?
18:23:56thomasjfoxlinuxstb: Two steps needed here
18:24:18thomasjfoxlinuxstb: 1. Implement smooth fullscreen/windowed mode switching in the SDL code
18:24:41thomasjfoxlinuxstb: 2. Adapt/polish new OS status bar code from/with kugel
18:24:50linuxstbthomasjfox: OK, that answers my first question - it's going to be user-configurable?
18:24:57thomasjfoxlinuxstb: Yes
18:25:27thomasjfoxlinuxstb: kugels proof of concept code: kugel-rb.git/commit/bdaeda6a38fb81db03bfbe9fd35f1836989b5ef3">
18:27:18 Quit PsyBlade (Remote host closed the connection)
18:33:24 Join robin0800 [0] (
18:45:56 Quit bertrik (Read error: Connection timed out)
18:46:28 Join bertrik [0] (~bertrik@rockbox/developer/bertrik)
18:47:20 Join {-phoenix-} [0] (
18:49:27 Join esperegu [0] (~quassel@
18:50:44 Quit {phoenix} (Ping timeout: 240 seconds)
18:55:46 Join Keripo [0] (
18:56:03 Join kevku [0] (~kevku@2001:7d0:0:f9af:feed:feed:feed:feed)
19:05:22 Quit hebz0rl (Read error: Operation timed out)
19:11:08 Join Stummi [0] (~Stummi@rockbox/developer/Stummi)
19:12:14 Join saratogahome [0] (600afc5f@gateway/web/freenode/ip.
19:19:07 Join stoffel [0] (
19:19:49 Quit jordan` (Quit: Coyote finally caught me)
19:20:03 Join JesusFreak316 [0] (
19:22:08pixelmathe metadata weirdness occured with r 28672 (sadly some days old already but) ... still a commit by you kugel (;revision=28672)
19:22:18pixelmaon hwcodec
19:28:46kugelpixelma: I see nothing wrong right now. you could try, nothing else changed really for mp3
19:34:56pixelmaI'll try
19:35:57 Quit MaynardWaters (Remote host closed the connection)
19:36:31linuxstbkugel: I think that change will break other parsers - they (mis) use the path field as temporary storage. But that does look to be the bug - the mp3 parser probably relies on knowing the filename, which is why it was passed as a parameter.
19:37:09kugellinuxstb: no it doesn't, I checked that. but I now see what else changed for mp3
19:37:32linuxstbkugel: Yes, I was just about to say it doesn't...
19:37:40pixelmaok, I'll wait then
19:38:29kugelpixelma: try this please:
19:40:39*amiconn wonders why thi might help
19:41:12amiconnOr rather, if it does, why swcodec doesn't need the memset
19:42:35linuxstbamiconn: Presumably the swcodec playback engine clears that struct, but the hwcodec doesn't.
19:43:03linuxstbAlthough it may be sensible to always do it...
19:43:31kugelhwcodec calls mp3info(), swcodec doesn't
19:43:52kugelI wonder why hwcodec just also uses get_metadata()
19:44:05linuxstbAh yes, get_metadata does it...
19:44:21linuxstbkugel: I guess no-one has got round to unifying it...
19:44:32 Join jordan` [0] (
19:45:03pixelmaI also seem to remember one of the "exotic" tags which is supported by swcodec but isn't on hwcodec - grouping I think it was
19:45:21 Join pamaury [0] (
19:45:21 Quit pamaury (Changing host)
19:45:21 Join pamaury [0] (~quassel@rockbox/developer/pamaury)
19:46:12linuxstbThe API is different though - mp3info takes the filename, get_metadata() takes a file descriptor. Maybe mp3info should be a wrapper around get_metadata()
19:47:45pixelmathe patch doesn't seem to help though
19:49:05kugelpixelma: maybe this ?
19:49:46bluebrotheram I the only one with problems with playback on Android? In some cases playback stops shortly after starting it :/
19:50:24kugelbluebrother: some people reported problems. for me playback works just fine
19:51:06pixelmahrrmm... I think I took the wrong one to try (the very first)... ETOOMANYOPENTABS
19:51:48kugelpixelma: the last one is the "best" one even if the others also work
19:52:00 Quit pamaury (Ping timeout: 260 seconds)
19:52:04bluebrothertoo bad. Any easy way to check why playback has stopped if it stops?
19:52:06pixelmabluebrother: for me sometimes audio stops working and I need to kill the app and start again
19:52:08*bluebrother guesses not
19:52:28kugelbluebrother: adb logcat maybe
19:52:40bluebrotherpixelma: that "cures" it for me but it also makes Rockbox unusable
19:52:40kugel(you can also view logcat on the device if you're rooted)
19:53:00pixelmakugel: I'll try the last one then
19:53:40kugelwhen playback was unstable for me months ago stopping (i.e. long press play icon in the wps) once or twice helped
19:53:58bluebrotherthat unfortunately doesn't work for me
19:54:37kugeldid you try stopping twice? I seem to remember in some cases a second time was needed
19:55:00bluebrotherstop by long pressing the play icon?
19:55:29bluebrotheror is there another way on Android?
19:56:23kugelusing the widget
19:57:36pixelmakugel: looks like that one fixes it
19:58:00kugelgreat, can you commit it? I don't have the 3.8 branch
19:58:01bluebrotherdoesn't work. Rockbox "starts" playback but doesn't advance in the file, and no audio output
19:58:30 Join pamaury [0] (
19:58:30 Quit pamaury (Changing host)
19:58:30 Join pamaury [0] (~quassel@rockbox/developer/pamaury)
19:58:33CIA-70New commit by 03dave (r29430): Remove some unnecessary includes
19:59:32pixelmakugel: not easily now as I'm about to leave for a few hours... maybe I have some time in 20-25 minutes but it'll be very limited
19:59:37pixelmaand you should have ;)
19:59:58bluebrotherhmm. Trying it a couple of times resulted in it now playing back like half a second of music, then no audio again
20:00:00kugelAlexP: ? :)
20:02:06AlexPIf you are going to ask me to redo the builds yet again I'm going to kill you
20:02:26AlexPAnyway, how can I help? :)
20:02:38linuxstbOK, I'll ask - can you redo the builds?
20:02:47CIA-70r29430 build result: All green
20:02:54kugelafter someone committed the hwcodec fix
20:02:56AlexPThe promise of swift death applies to all :)
20:03:09linuxstbAlexP: OK, then I withdraw.
20:03:15AlexPSo I only need to rebuild hwcodec?
20:03:22AlexPThat is a small mercy :)
20:03:40kugelI'm not sure
20:04:10linuxstbTechnically I think you should, as it changes the code for all targets. Whether it fixes any bugs on swcodec is a different question.
20:04:14*bluebrother thinks Rockbox should show the Android status bar
20:04:16AlexPIt takes me an hour and a half to do all the builds, then I have to spend ages uploading
20:04:25AlexPIt is already 20:00
20:04:30AlexPAnd I have work tomorrow
20:04:30kugelvbrfix.rock also calls mp3info() on swcodec
20:04:37*AlexP is not pleased
20:04:45kugelI can imagine :(
20:04:47linuxstbkugel: But does it clear the mp3entry struct?
20:05:09bluebrotherAlexP: postpone the release until tomorrow then :)
20:05:17AlexPI will have to
20:05:19kugelit's on stack and passed to mp3info() directly
20:05:32AlexPBut it is a great shame that this was only found now
20:05:44AlexPWe've had two weeks, including release candidates
20:05:59bluebrotherthat's Murphy ;-)
20:06:00Zagorwell, it's good that it *was* found
20:06:10AlexPbut I'm still pissed off
20:06:30Zagoryes, it's a bit of a bummer
20:07:19*kugel could do the release builds on his build client
20:07:43AlexPDon't worry, I'll do them
20:07:46AlexPWhat is the fix
20:09:20kugelmy build client is fast and has fast upload speeds; we can still release today
20:09:31AlexPAnmd what is the problem it is fixing?
20:09:53AlexPDon't worry, it'll be OK
20:09:57***Saving seen data "./dancer.seen"
20:10:12kugel"15:12:25pixelma after a skip e.g. the track number tag info contains part of the title tag string, year tag info just vanished, other tags I show on my WPS were ok though - and no problem on "normal" track changes"
20:10:12saratogahomewhat does that fix?
20:10:15bluebrotherZagor: any idea why I get "Empty output. Bad input?" on ?
20:10:25saratogahomeah thanks!
20:11:03kugelso corrupted metadata on manual track change
20:11:20AlexPon hwcodec?
20:12:19CIA-70New commit by 03alex (r29431): Fix corrupted metadata on manual track change on hwcodec. Thanks to kugel for the patch.
20:13:32Zagorbluebrother: maybe because latviesu didn't exist in r29385
20:14:38AlexPOK, building again :)
20:14:38bluebrotherZagor: ah! Thought I checked when it was added, but seems like I misread the revision. D'oh!
20:14:52AlexPI really will hunt down anyone next time :)
20:14:54CIA-70New commit by 03kugel (r29432): Fix corrupted metadata on manual track change on hwcodec. ...
20:14:57bluebrotherusing the correct revision indeed does make it work :)
20:19:29CIA-70r29432 build result: All green
20:20:47 Join GeekShadow [0] (~Antoine@reactos/tester/GeekShadow)
20:25:58 Quit saratogahome (Quit: Page closed)
20:28:34 Quit {-phoenix-} (Remote host closed the connection)
20:31:10 Join panni_ [0] (
20:31:59mshathlonxpwhat's going on?
20:32:08*mshathlonxp seems to have caused much trouble :D
20:32:22 Join {phoenix} [0] (
20:32:27Zagormshathlonxp: no, it's just standard confusion :)
20:34:08 Join stripwax [0] (
20:34:21 Quit factor (Read error: Connection reset by peer)
20:34:51 Quit robin0800 (Quit: Leaving)
20:45:57 Join Sochiro [0] (~Sochiro@
20:49:41 Quit stripwax (Quit:
20:52:06 Join factor [0] (~factor@
20:52:18linuxstbkugel: There seem to be lots of files you've added to svn without any svn properties set...
20:56:45thomasjfoxlinuxstb: That me be true for me, too ;) Anything special needed?
20:56:51 Join stripwax [0] (
20:57:03 Quit stripwax (Client Quit)
20:57:04linuxstbthomasjfox: See the UsingSVN wiki page.
20:57:26thomasjfoxamiconn: Public build now features ASM optimizations. Thanks again for pointing it out.
20:57:26 Join stripwax [0] (
20:57:36*linuxstb isn't sure what the eol-style policy is though - most files seem to have it set to native, but the UsingSVN page doesn't state it.
20:58:13pixelmathe only thing I don't like about that fix especially with regard to the release is that it didn't get thorough testing, just a quick one...
20:58:15*linuxstb seems to recall n1s or bluebrother setting eol-style on a lot of files, but may be wrong...
20:58:43CIA-70New commit by 03thomasjfox (r29433): New way to detect build for arm target in the maemo port. ...
20:58:49CIA-70New commit by 03thomasjfox (r29434): Update maemo changelog
20:59:00pixelmaI don't have much time for this just now though :/
21:02:19pixelmalinuxstb: I seem to recall an ml discussion about svn properties some time ago but I think it was left without consensus
21:02:24 Join esperegu_ [0] (~quassel@
21:02:47CIA-70r29433 build result: All green
21:03:14 Quit esperegu (Ping timeout: 276 seconds)
21:05:12bluebrotheryep, there was a ML discussion.
21:06:45CIA-70r29434 build result: All green
21:07:02 Quit MethoS- (Remote host closed the connection)
21:07:06 Quit Jerom (Quit: Leaving.)
21:10:29 Join MethoS- [0] (~clemens@
21:12:02linuxstbbluebrother: Can you remember it? I've just read the last two posts, and it was about eol-style - is that all the discussion was?
21:13:16*linuxstb asks gevaerts, as he started it...
21:13:18kugellinuxstb: I'm using git, that's why
21:13:25bluebrotherlinuxstb: iirc the discussion was more general, starting with a discussion why svn properties are needed to get set at all
21:14:07linuxstbkugel: Huh?
21:14:13bluebrotherand git-svn doesn't support svn properties there are cases when people forget to set them. Not much of an issue with svn:keywords, much more of an issue with svn:mime-type
21:14:48 Quit sasquatch (Ping timeout: 276 seconds)
21:14:48kugelIIRC we agreed to drop svn:keywords anyway?
21:15:44linuxstbkugel: Then why has nobody removed it from existing files and edited the wiki page?
21:16:06kugelI don't know. I can be wrong also
21:17:04bluebrotherno, it was not agreed. Some people were in favour of doing so but (at least I didn't notice) no consensus was reached
21:17:17kugelanyway, even if I was using svn I would forget about the properties
21:17:22bluebrotherpersonally I wouldn't mind dropping svn:keywords.
21:17:24 Quit sinthetek (Ping timeout: 272 seconds)
21:17:28 Quit lixxus (Quit: Leaving.)
21:17:44*kugel thinks there's very little point in svn:keywords
21:18:02bluebrotherwell, it's all about getting the system you're using know. And making the process a habit.
21:18:45bluebrotherand in any case you can set svn auto-props so svn can do the obvious cases automatically
21:19:05 Join lixxus [0] (
21:19:24*bluebrother is using git as well
21:20:35 Quit lixxus (Client Quit)
21:20:37kugelthomasjfox: what was the problem with asm?
21:20:37 Quit stoffel (Read error: Connection reset by peer)
21:20:44 Quit Sochiro (Ping timeout: 250 seconds)
21:20:52*thomasjfox curses at the maemo build server
21:21:13kugelpixelma: the change was pretty safe
21:21:22thomasjfoxkugel: It had a target name without ARMEL in the name. So I had to find another way to detect if the current build target is for arm
21:21:35thomasjfoxkugel: I hope I've now found a future-proof way
21:22:51 Join sideral [0] (~sideral@rockbox/developer/sideral)
21:23:05kugelthomasjfox: doesn't gcc define __arm__ or so? can't that be used for enabling optimizations?
21:23:21sideralHi there, is it too late for a critical last-minute fix for 3.8?
21:23:56kugelonly if you're ok with getting killed by AlexP :P
21:24:13sideralsounds like a yes :)
21:24:32thomasjfoxkugel: Didn't check that. I now have a simple solution that should be future proof (calling "sb-conf" and query the current arch)
21:24:37kugelit probably depends on how bad the bug is
21:24:53sideralBTW, are there instructions somewhere for commiting to an SVN branch from git-svn?
21:25:17kugelAlexP is an expert for that
21:25:20CIA-70New commit by 03sideral (r29435): Fix database import and export ...
21:25:40AlexPsideral: That hgad better bloody be a joke
21:26:00CIA-70New commit by 03bluebrother (r29436): Rockbox Utility voicefile generation: add Latvian to supported languages.
21:26:29sideralAlexP: I just found this while trying to implement the extend-db-as-we-go feature you wanted :)
21:26:59*sideral hands AlexP a cookie
21:27:38AlexPWhat is the bug?
21:28:01sideralDB import/export has been broken since Jan 2 or so
21:28:55CIA-70r29435 build result: All green
21:29:14mshathlonxpby the way, is that glitch, where 1g nano loses sound after turning on record until next reboot, known?
21:29:19sideralAlexP: How bad is it?
21:29:37 Join sasquatch [0] (
21:30:57pixelmawith all those last minute fixes I wouldn't mind a one day delay or so. Just saying...
21:31:23AlexPI am seriosuly pissed off with this
21:31:55AlexPEverybody knew that today was release day, we have had a week of freeze and a week of branch, we have had a weeks worth release candiudates
21:32:07bluebrothermaybe it makes sense to schedule the next release for like a wednesday so people can fix the weekend before instead of spending time releasing
21:32:09AlexPI've spent all bloody day building and rebuilding, and have wasted the entire day
21:32:23linuxstbbluebrother: I was about to say the same.
21:32:27AlexPI don't have enough time on a work day evening
21:32:34sideralAlexP: I'm sincerely sorry. Feel free to ignore the fix.
21:32:34CIA-70r29436 build result: All green
21:32:36thomasjfoxIs database import/export a critical feature?
21:32:37AlexPPeople knew it was today
21:32:46AlexPThey should have spent last weekend fixing
21:32:52linuxstbDelay a week?
21:33:17pixelmaI am also sorry to only have checked a hwcodec build today...
21:33:27kugelsomeone else can do the builds
21:33:29sideralAlexP: I was offline all week and noticed the bug only today. I thought fixing this pre-release would make you happy :/
21:33:50AlexPsideral: in a way
21:34:00AlexPI'm just pissed off to have ended up wasting my entire day
21:34:10AlexPI gave so much warning this time round
21:34:16AlexPEmails announcing every stage
21:34:22kugelI don't think db import/export is very critical. I'd be ok with putting it into known issues and move forward
21:35:06AlexPI don't want to release with it broken really
21:35:06pixelmamaybe plan for an early point release?
21:35:38 Join hebz0rl [0] (
21:35:43AlexPsideral: I'm glad you found it before release, I just don't know what the point in panning and giving notice is if it makes fuck all difference
21:37:18bluebrotherAlexP: I think one problem is that quite a few people don't have much time to work on Rockbox during the week. At least that is the case for me. So as a result those fixes come in late, and hence the idea to release during the week
21:37:39sideralAlexP: There will be another last-minute fix if you delay you're release because of this one. So why not draw the line here?
21:38:36AlexPbluebrother: That is what the last two weekends are for though
21:38:40 Join robin0800 [0] (
21:39:17AlexPbluebrother: I published a timetable 4 months in advance
21:40:44bluebrotherAlexP: I know, just tried to think about a way how to improve it. I can really understand your frustration ...
21:41:19AlexPsideral: commit it, I'll build again
21:41:33 Join Buschel [0] (
21:41:51sideralAlexP: can you point me to instructions for committing via git-svn?
21:42:05AlexPto a remote branch you mean?
21:42:15pixelmame too, I'm honestly sorry
21:42:20AlexPIt depends if you checked them all out properly in the first place
21:42:33AlexPsideral: does git branch -r show you the remote branches?
21:42:36*bluebrother uses plain old svn for this :)
21:42:38sideralI cloned the official git repo
21:42:43AlexPpixelma: It's OK, I'm just frustrated
21:42:57AlexPsideral: then you probably don't have them
21:43:14AlexPIf you do, then you just do git checkout -b local_branch remote_branch and it just works
21:43:41sideralIt doesn't
21:44:03AlexPThat's because the git clone doesn't have them
21:44:11 Quit hebz0rl (Ping timeout: 250 seconds)
21:44:14AlexPI did git svn clone svn://blah
21:44:19AlexPas that gives you everything
21:44:29AlexPjust let me have the patch, I'll commit it to the branch
21:44:41sideral1 sec
21:44:57AlexPsideral: For reference,
21:47:43sideral(I'm on a slow and expensive GPRS link, is slow...)
21:47:52CIA-70New commit by 03kugel (r29437): Android: Show cover art in the widget (including option to hide it).
21:48:45 Quit t0rc (Quit: Give someone code, help them with one project. Teach someone to code, help them rule the world.)
21:49:15CIA-70New commit by 03Buschel (r29438): Remove some useless code and variables in the area of metadata parsing. Bump codec API.
21:51:47CIA-70r29437 build result: All green
21:56:48CIA-70r29438 build result: All green
22:01:36CIA-70New commit by 03alex (r29439): Backport r29435 to 3.8 branch.
22:03:45sideralThanks AlexP!
22:03:52AlexPno worries
22:09:42CIA-70New commit by 03kugel (r29440): Fix warning in android build.
22:10:01***Saving seen data "./dancer.seen"
22:11:56 Join Kitr88 [0] (~Kitarist@
22:14:43CIA-70r29440 build result: All green
22:15:44 Quit Kitar|st (Ping timeout: 264 seconds)
22:16:12 Quit Kitr88 (Ping timeout: 250 seconds)
22:20:58 Quit sideral (Ping timeout: 240 seconds)
22:21:01 Quit Keripo (Quit: Leaving.)
22:21:16CIA-70New commit by 03kugel (r29441): Make struct static.
22:21:32 Join Kitar|st [0] (
22:22:20linuxstbthomasjfox: I'm confused with the audio on the N900 - it seems to be a mixture of SDL, gstreamer and libplayback ?
22:23:22thomasjfoxlinuxstb: It's gstreamer + libplayback. libplayback is Nokia's library for audio policy management (audio output in silent mode, stop playback on incoming calls)
22:23:48linuxstbthomasjfox: So it's not using SDL? pcm-gstreamer.c includes <SDL.h>
22:24:39thomasjfoxlinuxstb: Seems like an oversight
22:24:44 Join einhirn [0] (
22:24:46linuxstb(and drivers/audio/sdl.c is also being built on the n900)
22:24:58 Quit bertrik (Quit: :tiuQ)
22:25:29CIA-70r29441 build result: All green
22:25:38thomasjfoxlinuxstb: drivers/audio/sdl.c consists mostly of stubs. There's one MAEMO5 ifdef in there
22:27:00 Quit Buschel (Quit: ChatZilla 0.9.86 [Firefox 3.6.13/20101203075014])
22:27:54linuxstbthomasjfox: Maybe you should create a firmware/drivers/audio/maemo.c - see the similar android.c
22:28:32linuxstbOr in fact, perhaps android.c could be application.c (kugel?)
22:28:51thomasjfoxlinuxstb: OTOH, maemo 4 uses SDL
22:29:01thomasjfoxlinuxstb: At least for now
22:29:02linuxstbOK, then maemo5.c
22:29:19kugelthomasjfox: did you work on pulseaudio already?
22:29:25linuxstbAnd then you can undefine HAVE_SDL_AUDIO, as you don't...
22:29:33thomasjfoxkugel: No :)
22:30:07thomasjfoxkugel: Nokia N8xx is not on my priority list...
22:30:42kugelI thought it would work better on the n900 as well? and on your openpandora?
22:31:36 Quit esperegu_ (Remote host closed the connection)
22:31:43thomasjfoxkugel: Openpandora uses alsa only :-( From a performance point of view that should be a :) though
22:32:08thomasjfoxkugel: gstreamer currently works good enough. Status bar support is more important
22:32:32thomasjfoxkugel: I want to push a release to extras-testing and then "extras" once we have proper status bar support. Right now rockbox is in "extras-devel" only
22:32:51kugelthat doesn't mean anything to me :)
22:33:07thomasjfoxJust different levels of user exposure
22:33:30thomasjfoxYou can upload anything you want to extras-devel. If the developer is happy with it, you can promote it to "extras-testing"
22:33:32kugeldo those imply stability?
22:34:05thomasjfoxIf ten people vote for it in extras-testing, it gets into normal "extras" and is available to every normal user without any special setup / download location
22:34:21thomasjfoxkugel: Never experienced any hickups
22:34:33Torneextras-testing is pretty much "i reasonably believe that installing this won't break your phone"
22:34:34 Quit leavittx (Ping timeout: 240 seconds)
22:35:13thomasjfoxAs long as rockbox can only be closed by pressing CTRL+BACKSPACE to invoke the tasks switcher, I don't want it in normal "extras" yet
22:35:38thomasjfoxThat's where the status bar support comes into play: Nothing to explain, just works out of the box
22:35:40kugelthomasjfox: I think releasing it to extras happen in coordination with the project
22:36:11thomasjfoxkugel: Yes, extras releases should be upstream releases or releases with "critical" fixes
22:36:26*gevaerts proposes to extend the freeze by one week per late fix
22:36:30gevaerts(next time)
22:37:21TheSevenyeah, maybe add one step more
22:38:05linuxstbthomasjfox: I think the question is when we release the first RaaA version. I don't think anyone knows what the criteria should be for calling a RaaA port "stable".
22:38:06TheSeventhe last week being just for trivial things like translations and similar cleanup, and if regressions are found during that period, re-start it
22:38:48gevaertsThe extension has to be *before* the branch, otherwise nobody will care
22:39:22TheSevenpushing things like "Fix database import and export" on the day of release certainly sounds dangerous, if it isn't something absolutely trivial (haven't checked the diff)
22:40:03TheSevengevaerts: i'm also wondering if more unit testing could help
22:40:09thomasjfoxlinuxstb: For maemo, I would consider status bar support + fixed plugin data dir a good release candidate
22:40:09AlexPTheSeven: It is a one liner anyway
22:40:24thomasjfoxkugel: Anything seriously left in android?
22:40:28TheSevenAlexP: one-liners can be rather evil as well :)
22:40:37 Quit pamaury (Ping timeout: 240 seconds)
22:40:43AlexPTheSeven: I know, hence the anyway :)
22:40:44gevaertsTheSeven: of course, but I don't think our source is really structured well enough for that
22:41:07linuxstbthomasjfox: Personally I would like to see more integration with the native UI. i.e. it should feel more like a native app.
22:41:43TheSevengevaerts: not neccessarily everything automated and simulated
22:42:17 Join esperegu [0] (~quassel@
22:42:25TheSevenwe have the apps/firmware split anyway, so why not write a second apps layer that tries to do as exhaustive as possible testing of the firmware layer on-target?
22:42:31thomasjfoxlinuxstb: That's a good development target, though I normally stick to "release early, release often"
22:42:46TheSevento stress test things like storage, PCM, or the kernel
22:42:51thomasjfoxlinuxstb: Of course only after testing and the usual QA
22:43:10TheSevenand detect misimplementations of some APIs on some targets :)
22:43:32gevaertsTheSeven: that would be possible I guess, but on the other hand it seems as if all those last-minute bugs tend to be in apps/
22:43:40AlexPthomasjfox: "usual QA" :P
22:44:01 Quit esperegu (Remote host closed the connection)
22:44:04TheSevengevaerts: yes, I'm seeing this as a separate goal :)
22:44:28gevaertsYes, for your purpose that would work
22:44:39gevaertsExcept for the usual Mr. Someone being lazy...
22:44:42TheSevencould be helpful while porting to new targets, and to weed out weird bugs like the one i'm currently experiencing on the ipod classic (occasional seemingly-random hard lockups)
22:44:46linuxstbthomasjfox: Perhaps. Although the Rockbox way of doing that is to make current builds available before the port is promoted to "stable" and is included in a release.
22:44:48thomasjfoxAlexP: Running valgrind on rockbox was worthwhile, see the fixes in the metadata parser ;)
22:45:37linuxstbthomasjfox: So I guess "extras-devel" is our equivalent of "unstable", and then "extras" will be when a release (3.9?) is made.
22:46:03thomasjfoxlinuxstb: Yes, that sounds reasonable
22:46:51thomasjfoxlinuxstb: In case some critical bug fix turns up after a release (...), I might still cherry-pick it and push it to extras, too.
22:51:43TheSevenAlexP: about how much time is left before the release will be tagged?
22:52:33TheSevenbecause I don't want to mess with trunk until we're all set
22:52:47AlexPI'm all in the branch anyway
22:53:02AlexPanway, not long
22:53:23TheSevennevertheless I don't want to red the trunk now, as usually fixes that are applied to the branch will be applied to trunk as well
22:53:27linuxstbthomasjfox: Well, we have point releases for critical bugs already. (3.7.1 exists)
22:53:46linuxstb(which can also only be for some targets I think)
22:54:14gevaertsWe tend to do them for all targets, even if they don't affect everything
22:54:23gevaertsJust to keep things manageable
22:55:29 Quit benedikt93 (Quit: Bye ;))
22:55:41 Quit einhirn (Ping timeout: 250 seconds)
22:56:02thomasjfoxWould r29395 qualify as critical? That's something I cherry-picked recently
22:56:19 Join sideral [0] (~sideral@rockbox/developer/sideral)
22:56:24linuxstbthomasjfox: Is it possible to automate updates to extras-devel/testing ?
22:56:26thomasjfoxThough I would stick to official release policy anyway
22:56:54gevaertsTheSeven: what's the impact?
22:57:06gevaertsthomasjfox: what's the impact?
22:57:16*gevaerts apologises to TheSeven for waking him up
22:57:24linuxstbHmm, that wasn't applied to the branch as well?
22:57:36*TheSeven points gevaerts to his fat patch
22:58:23 Join Rob2223 [0] (
23:01:19*gevaerts looks at TheSeven's fat patch and vows to never wake him up again
23:01:19 Join AlexP_ [0] (~alex@rockbox/staff/AlexP)
23:01:23 Join thomasjfox_ [0] (~thomasjfo@rockbox/developer/thomasjfox)
23:02:05 Join Mephistopheles [0] (
23:03:25 Quit AlexP (Ping timeout: 240 seconds)
23:03:45 Join knitt1 [0] (~knittl@unaffiliated/knittl)
23:04:17thomasjfox_linuxstb: extras-devel is accessible via scp + ssh pubkey if needed
23:04:22thomasjfox_linuxstb: Though pushing updates daily to extras-devel will annoy users I guess
23:04:29 Quit thomasjfox (Ping timeout: 240 seconds)
23:04:42linuxstbthomasjfox_: Ah yes, users will be prompted to upgrade?
23:04:49thomasjfox_gevaerts: IIIRC it was related to the playback/buffer freezes
23:04:52 Nick thomasjfox_ is now known as thomasjfox (~thomasjfo@rockbox/developer/thomasjfox)
23:05:01linuxstbthomasjfox: But on the other hand, they should...
23:05:02thomasjfoxlinuxstb: Yes, every time
23:05:26thomasjfoxlinuxstb: If an app wants to upgrade every day, I would uninstall it pretty fast :o)
23:05:29gevaertsthomasjfox: "related to" doesn't mean much :)
23:05:33 Quit Rob2222 (Ping timeout: 240 seconds)
23:05:34 Quit bluefoxx_ (Ping timeout: 240 seconds)
23:05:34 Quit MagusG (Read error: Connection reset by peer)
23:05:34 Quit knittl (Read error: Operation timed out)
23:05:53gevaertsthomasjfox: if that app is called "rockbox-svn-build", you know what you ask for :)
23:06:04 Join overnite [0] (
23:06:10linuxstbAnd it's more than once a day... ;)
23:06:28gevaertsYes, but the checking is only once a day
23:06:33 Join MagusG [0] (
23:07:37linuxstbthomasjfox: What do you upload? The source or a binary?
23:07:58thomasjfoxlinuxstb: Source tarball + some debian related control files
23:08:22thomasjfoxgevaerts: The "source" for r29395:
23:08:38gevaertsI suspect uploading ten times a day may annoy the maemo build server people
23:09:22gevaertsthomasjfox: I can't really judge. This fits in with all other stuff that jhMikeS committed after the branch, right?
23:10:29 Quit sideral (Ping timeout: 276 seconds)
23:10:37thomasjfoxgevaerts: Can't judge either. As I run SVN code, it was a good pick
23:11:22 Quit overnite (Quit: leaving)
23:14:24gevaertsTheSeven: I think I'd prefer the cachealign stuff split out for commit. Apart from that, if you didn't forget an error path somewhere (I didn't count them all!), and if our earlier analysis about there never being two sector buffers on the stack any more, that should be OK
23:14:44linuxstbthomasjfox: Is it possible to do a web-based install on the n900? i.e. just make current builds downloadable on the main Rockbox site?
23:15:00TheSevengevaerts: if there should be a path that needs the sector buffer twice, that will at least trigger a panic
23:15:25gevaertsTheSeven: exactly. And the same will happen if you missed a return path somewhere
23:15:33TheSevennot neccessarily
23:15:40 Quit Stummi (Quit: Bye!)
23:15:46gevaertsWell, not immediately
23:15:49TheSevenonly if there are more storage accesses coming from the same thread
23:16:08thomasjfoxlinuxstb: Don't know if it's enough to just click on a .deb file. You can definately scp it to the n900 and run dpkg -i. Sounds unpractical to me
23:16:09TheSevenif something comes from a different thread, that might end up as a mutex deadlock
23:16:27gevaertsAh, yes. We'll be told about those too :)
23:17:12gevaertsamiconn: does any flash stuff need to happen for 3.8, or is flash dead anyway due to binsize?
23:17:13thomasjfoxlinuxstb: Easiest thing would be to run an own apt channel and people can add rockbox to their "sources" list
23:17:43linuxstbthomasjfox: There are various issues - making stable releases available (easy - just put them in extras), and then making the daily and/or current builds available in a way that doesn't constantly annoy the user.
23:18:11linuxstbthomasjfox: But running our own apt channel would still have the constant prompts for upgrades?
23:18:34gevaertslinuxstb: yes, but it's easy to disable that one catalog
23:19:10TheSevengevaerts: so you think I should just commit that thing?
23:19:42linuxstbgevaerts: OK, then that seems sensible - someone should set it up now...
23:19:57gevaertsTheSeven: I think so, yes
23:20:07TheSeven(in 4 steps: 1. get rid of on-stack sector buffers and introduce shared sector buffer, 2. align the static buffers, 3. add that STORAGE_NEEDS_ALIGN thing, 4. S5L8702 ATA driver)
23:20:14AlexP_amiconn: Do we need any archos flash stuff?
23:20:31gevaertsTheSeven: what does STORAGE_NEEDS_ALIGN mean exactly?
23:20:57TheSeventhat the file code will ensure that all memory addresses passed to the storage layer are aligned
23:21:19TheSevene.g. copy through the file's sector buffer even for multiple-sector accesses if the buffer isn't aligned
23:21:47gevaertsAh, right. So nothing that will break anything in apps/
23:22:05gevaertsYes, that plan looks sane then
23:22:21TheSevenit just adds overhead if it's defined, but means that the storage layer doesn't have to care about that
23:22:29TheSeven(which would need another sector buffer)
23:23:41 Quit Unhelpful (Quit: No Ping reply in 180 seconds.)
23:24:06gevaertsYes. Not nice to have, but if the hardware needs it...
23:24:07 Join Unhelpful [0] (~quassel@rockbox/developer/Unhelpful)
23:25:54stripwaxany way to log how often that gets hit (if/when it gets hit)?
23:26:17stripwaxto help indicate which plugins might benefit from changes to align their own buffers
23:26:31amiconnAlexP_: Nothing new there, so no
23:26:37AlexP_OK, thanks
23:26:44gevaertsstripwax: there's a single if() that checks for it and patches things up, so that should be easy to add
23:26:45amiconnI wonder whether we should publish a new flash package from time to time
23:27:00amiconnSaves the user the follow-up flash of the current .ucl
23:27:35amiconnIt's only relevant for first-time flashers though, and I guess those are rare these days
23:27:39CIA-70New commit by 03alex (r29442): Tag 3.8 release.
23:28:43linuxstbAlexP_: \o/
23:29:08AlexP_linuxstb: Indeed :)
23:30:02 Join einhirn [0] (
23:36:48linuxstbkugel: I think you missed my question earlier, would you object to firmware/drivers/audio/android.c being renamed to something like "application.c" for use as a generic RaaA audio driver? I can't see anything specific to android there.
23:42:45CIA-70New commit by 03zagor (r29443): Point out the 3.8 release.
23:44:32CIA-70New commit by 03theseven (r29444): iPod Classic CE-ATA Support (Part 1 of 4: Cacheline align some statically allocated sector buffers)
23:44:56kugellinuxstb: I'm not sure
23:45:00CIA-70New commit by 03theseven (r29445): iPod Classic CE-ATA Support (Part 2 of 4: Remove on-stack sector buffers, and replace them with a single statically allocated sector buffer that's ...
23:45:19*TheSeven waits for r29444 to complete building before committing r29446
23:45:21kugelit's possible that other targets can offer more audio functions via APIs (or less, not even volume)
23:46:37thomasjfoxlinuxstb: Can you take a look at Hope I got the new RaaA build system stuff right
23:46:38CIA-70New commit by 03zagor (r29446): Updated front page
23:47:04n1sTheSeven: are stuff using the sector buffer ever performance sensitive?
23:47:19CIA-70r29443 build result: All green
23:47:26TheSevenn1s: depends on the use case, but not for read accesses at least
23:47:36CIA-70New commit by 03theseven (r29447): iPod Classic CE-ATA Support (Part 3 of 4: Introduce STORAGE_NEEDS_ALIGN, which ensures that no unaligned storage accesses are performed through ...
23:48:00CIA-70New commit by 03theseven (r29448): iPod Classic CE-ATA Support (Part 4 of 4: S5L8702 ATA driver)
23:48:35n1sTheSeven: ah good, since the main threads stack is in iram for most targets and that is significantly faster than dram for some
23:48:50gevaertsn1s: they're all doing disk I/O by definition
23:50:13n1sgevaerts: yes, just thought i'd ask in case some behind the scenes magic was going on :)
23:52:39CIA-70r29445 build result: All green
23:52:53gevaertsOh, and you're assuming they're all in the main stack, which I don't think is true
23:53:09TheSeventhey're duplicated amongst several stacks
23:53:49gevaertsnow, yes
23:54:08gevaertsI mean, the assumption that they used to be in IRAM isn't necessarily true
23:55:56linuxstbthomasjfox: I've only take a quick look, but it looks fine to me.
23:56:08 Quit {phoenix} (Remote host closed the connection)
23:56:34thomasjfoxlinuxstb: Compiles and runs fine too. I diffed the export/config/pandora.h against sdlapp.h and nokian8xx.h
23:57:10CIA-70r29448 build result: 0 errors, 198 warnings (theseven committed)
23:57:38kugelthomasjfox: what's is unique about pandora? isn
23:57:47kugelisn't it just generic linux actually?
23:57:49 Quit domonoky (Read error: Connection reset by peer)
23:57:55JdGordonAlexP_: why didnt you just say "ou've missed the deadline" and leave those extra fixes for 3.8.1 when you (or someone) has time?
23:57:59AlexP_Zagor: The font archive that RBUtil grabs, I think it isn't the 3.8 font zip
23:58:21gevaertsJdGordon: broken database restore is rather critical for a release I'd say
23:58:22thomasjfoxkugel: We know the LCD screen size, it will get a special button map and later on an ALSA audio backend
23:58:24amiconnTheSeven: Whoa, half a kilobyte additional code?
23:58:27 Quit einhirn (Ping timeout: 250 seconds)
23:58:30AlexP_JdGordon: Because it seems silly to release with a known problem
23:58:33gevaertsAs that's when people will actually use it
23:58:40kugelthomasjfox: alsa is generic linux, isn't it?
23:58:42amiconnAlso, all-yellow
23:58:46thomasjfoxkugel: Yes
23:58:47AlexP_Zagor: I'm not sure about that, just maybe worth checking
23:58:51TheSevenamiconn: no, not all-yellow
23:58:55TheSevenall except ipod classic :P

Previous day | Next day