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

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

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

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

#rockbox log for 2010-04-02

00:00:03Tornetagcache in ram just makes browsing tagcache faster
00:00:04gevaertsok, so first dircache, then database. I think that achieving this by some sort of priority thing still makes most sense though
00:00:13Tornedircache being scanned makes almost *everything* faster
00:00:20Torneer, scan beign done, i mean
00:00:40gevaertsMaybe update dircache if the disk is idle for 100ms, and do database things if it's idle for 200ms?
00:00:52gevaertsOr similarly different numbers
00:02:34Tornei guess.
00:02:46Tornei dunno what effect *updating* the db has, btw
00:02:51Tornei don't have autoupdate turned on
00:02:56Torneand i've never updated while actualyl using the plpayer
00:03:07Tornei just know that you can use it while it's not done loading to ram yet :)
00:03:56gevaertsI only have the database for occasional pictureflow testing/showing, so I don't have autoupdate either
00:05:00Tornei use it to pick random things :)
00:05:05 Join huelk_ [0] (~huelk@dtmd-4db78ea6.pool.mediaWays.net)
00:05:22Torneand occasionally to find songs on mix cds.
00:05:49 Quit fyrestorm (Quit: Ur skills' fireproof like a wooden panel -- U got feds talking leet on your IRC channel!)
00:06:51CIA-5New commit by 03bluebrother (r25430): Fix multiple warnings when network is unreachable. ...
00:06:54CIA-5New commit by 03bluebrother (r25431): Fix leaking file descriptors on error.
00:08:58gevaertsZagor: can you have a look at flyspray? It seems to not do attachments anymore
00:10:14 Quit dfkt (Quit: -= SysReset 2.53=- Ph'nglui mglw'nafh Cthulhu R'lyeh wgah'nagl fhtagn.)
00:13:07 Quit kio (Quit: Leaving)
00:15:28 Quit dockimble1 (Quit: WeeChat 0.2.6.3)
00:16:59 Quit Schmogel (Read error: No route to host)
00:17:10 Join dockimble [0] (~dockimble@77.227.1.24)
00:17:21 Join RoronoaZoro [0] (~vijay@202.3.77.11)
00:22:28 Quit funman (Quit: free(random());)
00:23:01 Quit komputes (Read error: Operation timed out)
00:23:06 Quit shaggy-h (Ping timeout: 240 seconds)
00:23:54 Join shaggy-h [0] (~kiwi@78-86-164-31.zone2.bethere.co.uk)
00:26:08 Quit jgarvey (Quit: Leaving)
00:28:09 Quit bertrik (Quit: De groeten)
00:28:12 Quit anewuser ()
00:30:17 Quit einhirn_ (Read error: Connection reset by peer)
00:31:56 Quit user124345 (Quit: Page closed)
00:32:14 Join froggyman [0] (~me@unaffiliated/froggyman)
00:41:34 Quit adnyxo (Ping timeout: 265 seconds)
00:42:35 Join komputes [0] (~komputes@ubuntu/member/komputes)
00:43:26CIA-5New commit by 03gevaerts (r25432): Add GSoC acceptance to the news
00:43:39gevaertsZagor, B4gder: maybe another homepage push?
00:43:54Zagordone
00:44:30ZagorI reverted the suexec change, so now the builds are back again, too. I'll give it another go when I have a few more hours spare.
00:44:53gevaertsAh, flyspray accepts attachments again as well
00:44:58 Quit perfectdrug (Quit: Leaving.)
00:45:02Zagorgood
00:49:36 Join xiainx [0] (~iain@wpa106051.Wireless.McGill.CA)
00:49:46 Quit Zagor (Quit: Clint excited)
00:52:17 Join Luca_S [0] (~5712526b@giant.haxx.se)
00:59:21 Quit ender` (Quit: With sufficient thrust, pigs fly just fine. However, this is not necessarily a good idea. It is hard to be sure where they are going to land, and it could be dangerous sitting under them as they fly overhead.-- RFC1925)
01:00
01:00:59***Saving seen data "./dancer.seen"
01:01:45 Quit dockimble (Quit: WeeChat 0.2.6.3)
01:03:32 Join bmbl [0] (~Miranda@unaffiliated/bmbl)
01:04:59 Join Guest62701F [0] (Status@gentoo.lonis.org)
01:05:06 Quit Status (Read error: Connection reset by peer)
01:08:46Luca_Sfuzev2 scrollwheel works great ;) time on my
01:08:55Luca_Sdevice seems offset by 65284500 seconds
01:10:01Luca_Sthat is, rockbox is 65284500 seconds in the future
01:10:01 Quit dmb (Quit: Leaving)
01:11:00 Join BHSPitMonkey [0] (~stephen@unaffiliated/bhspitmonkey)
01:11:34Luca_Sthe build table shows red for the as3525v2 boot loaders, related to the scrollwheel
01:20:03 Nick Guest62701F is now known as Status (Status@gentoo.lonis.org)
01:20:50Luca_Srecording seems to work, but I had a panic
01:21:03Luca_Si2sin error: 0x4e = push
01:21:17 Quit komputes (Remote host closed the connection)
01:24:13 Join robin0800 [0] (~robin0800@cpc2-brig8-0-0-cust964.brig.cable.ntl.com)
01:28:48 Join tarnzwerg [0] (~abc@188-195-103-184-dynip.superkabel.de)
01:30:07 Join komputes [0] (~komputes@ubuntu/member/komputes)
01:31:48 Join orby_alt [0] (~Miranda@ip28-116.kullen.RWTH-Aachen.DE)
01:32:37 Quit orby_alt (Client Quit)
01:34:46 Join totonono [0] (~thomas@196.207.219.211)
01:40:25 Join orby_alternative [0] (~oe@ip28-116.kullen.RWTH-Aachen.DE)
01:40:25 Quit Luca_S (Quit: CGI:IRC (EOF))
01:40:56orby_alternative/join #testlalalalal
01:41:33 Quit tarnzwerg (Quit: Verlassend)
01:42:44 Quit domonoky (Read error: Connection reset by peer)
01:45:45 Quit bmbl (Quit: Bye!)
01:48:41 Quit sevard (Ping timeout: 260 seconds)
01:49:52 Join Tomis2 [0] (~Tomis@70.134.81.146)
01:49:58 Quit orby_alternative ()
01:50:47 Quit Tomis (Ping timeout: 245 seconds)
01:50:47 Nick Tomis2 is now known as Tomis (~Tomis@70.134.81.146)
01:54:13 Quit arbingordon (Quit: `)
01:54:38 Quit xiainx (Quit: Good Bye!)
01:55:08 Join arbingordon [0] (~w@unaffiliated/arbingordon)
01:57:21 Join merbanan [0] (~banan@c-62-220-165-110.cust.bredband2.com)
01:57:44 Join Tomis2 [0] (~Tomis@70.134.100.51)
01:58:05 Quit Tomis (Ping timeout: 240 seconds)
01:58:05 Nick Tomis2 is now known as Tomis (~Tomis@70.134.100.51)
02:00
02:04:05 Quit huelk_ (Ping timeout: 240 seconds)
02:06:27 Join sevard [0] (sev@216.164.6.24)
02:06:28 Quit merbanan (Ping timeout: 264 seconds)
02:17:48 Join dmb [0] (~Dmb@unaffiliated/dmb)
02:24:19 Quit efgpinto (Quit: Lost terminal)
02:26:48saratogawe should release a bootloader for the fuzev2
02:34:15 Join CaptainKewl [0] (~jason@207-237-106-60.c3-0.nyr-ubr1.nyr.ny.cable.rcn.com)
02:38:37 Quit JohannesSM64 (Ping timeout: 264 seconds)
02:41:21saratogaRoronoaZoro: I would start out by outlining the steps you'll need to do and estimating how long you think they'll take
02:41:44saratogai would include a brief explanation of the benefits of your project as you see them
02:41:50RoronoaZorook
02:42:03saratogaand I would explain your reasoning behind chosing the project, and why you think its feasible
02:42:35saratogayour proposal is interesting, it'll mostly be about convincing people you are serious about it and able to do it
02:42:45saratogaand explaining your ideas in detail will help a lot with that
02:44:40RoronoaZorook
02:45:40saratogaunfortunately I can't really give you any specific advice for the project because I've never worked on mpegplayer and am not too familar with video decoding
02:46:17saratogai will suggest that for audio codecs, adding AC3 support might be a good goal though, since its a fast and efficient codec already in rockbox (for audio files only) that is commonly used with mpeg video
02:47:30RoronoaZoroas to why is it feasible meaning what should i tell....like currently i can only tell that as per my knowledge of c and asm are concerned i can tell that i can port the code but how do i tell that to the mentors
02:47:52RoronoaZoro*i mean related to ffmpeg
02:48:15saratogai would address the computational feasibility too, these are slow systems, so you should try to assure people that they're fast enough for the formats you want to add
02:50:25RoronoaZoroso i can list in maybe some example of fps or comparision with other codecs
02:50:44saratogayeah
02:51:09saratogatheres lots of papers showing mpeg performance on various arm cpus too, and some companies publish cpu requirements for embedded decoders too
02:51:24RoronoaZorok
02:51:28saratogai think it should be possible to estimate the minimum cpu speeds needed for 15 fps or 20 fps playback
02:51:37 Join JohannesSM64 [0] (~johannes@cm-84.215.75.42.getinternet.no)
02:51:40RoronoaZoroyes
02:53:37RoronoaZoroand as for the timeline what do i consider the dates that i give to the project to be till ... i mean the end date only ...
02:54:23saratogagive an estimate for how long you think you it'll take to do each part of the project
02:55:21RoronoaZorothe refactoring of the original code will take me about 1-2 weeks
02:56:54RoronoaZorofor the porting i am assuming about 3-4 weeks .... not much sure as i have only seen the documentation not the code
02:57:05saratogahave you had a chance to look at the mpegplayer code ?
02:57:09RoronoaZoroyes
02:59:05 Join RadicalR [0] (~radicalr@c-69-255-49-110.hsd1.va.comcast.net)
02:59:43saratogahow much of it is just the mpeg1/2 decoder and how much is actually the player part?
03:00
03:01:04***Saving seen data "./dancer.seen"
03:01:29RoronoaZorompeg2dec is the decoder files
03:01:43 Join adnyxo [0] (~aaron@adsl-065-013-002-216.sip.asm.bellsouth.net)
03:01:55 Quit adnyxo (Remote host closed the connection)
03:02:42RoronoaZoroand also the stream.c and .h they check the format of the file
03:03:35 Quit totonono (Quit: Quitte)
03:03:47RoronoaZoroso i think about 1/2 is the part that is mpeg1/2 specific
03:06:14RoronoaZoroi am still having a problem i am not able to get the cross compiler working so i am not able to build rockbox yet
03:07:35saratogawhat is your problem?
03:09:46 Join anewuser [0] (anewuser@unaffiliated/anewuser)
03:12:24RoronoaZorohttp://pastebin.com/fN81v1L7
03:13:05RoronoaZorothis is what happens when i run rockboxdev.sh
03:14:32 Quit robin0800 (Remote host closed the connection)
03:15:29saratogaRoronoaZoro: what kind of computer is that?
03:16:23saratogayou could also try deleting those temp files and redownloading them in case they really are corrupt
03:17:58 Quit JohannesSM64 (Max SendQ exceeded)
03:21:59 Join Blue_Dude [0] (~chatzilla@adsl-235-206-131.mco.bellsouth.net)
03:22:14Blue_DudeKeymaps! I gotcher keymaps right here!
03:24:26CIA-5New commit by 03Blue_Dude (r25433): Switch on hotkey on several targets. Stole the WPS Show Track Info key for most of them. Made up a free key for the rest.
03:24:36RoronoaZorowhere do the temp files get stored .... i am not able to find...
03:27:17saratogayour logs mention /tmp/rbdev-dl
03:27:26saratogai would check there
03:27:36saratogafailing that, open up the script and see where it stores things
03:29:36 Join JdGordon [0] (~jonno@rockbox/developer/JdGordon)
03:30:13JdGordondircache was the cause of the slow boots then?
03:37:53*RoronoaZoro proxy really troubles
03:38:25 Join CGL [0] (~CGL@190.207.232.114)
03:39:19S_a_i_n_t_does anything useful actually get stored in /tmp/rbdev-dl ? I just ended up deleting it...
03:39:22S_a_i_n_t_Oooops.
03:41:16 Join sustineo [0] (~42116ef5@giant.haxx.se)
03:41:51RoronoaZorostill the same problem http://pastebin.com/RXpkEsWb
03:41:53CIA-5New commit by 03Blue_Dude (r25434): Fix someone elses yellow and red
03:42:28sustineo−−−−unrelated−−−− hey, i was making a wps, and i was wondering if there is a way to put images / text info over the album art. is there?
03:42:41*RoronoaZoro says this time rockboxdev.sh downloaded the files
03:42:54S_a_i_n_t_sustineo: not reliably...
03:43:04 Join Tomis2 [0] (~Tomis@70.134.100.53)
03:43:13sustineo@saint: it crashes if you do?
03:43:52S_a_i_n_t_No, it just doesn't ever seem to look right, there is little controll (actually none) over what gets drawn first.
03:43:59sustineooh
03:44:10S_a_i_n_t_It all gets drawn in one big batch at the end.
03:44:33sustineoalternatively, can you rotate text data? (like 90 degrees)
03:44:56S_a_i_n_t_not without getting *really* creative with viewports
03:45:38sustineo>_>
03:45:40sustineothanks
03:45:46 Quit CGL (Ping timeout: 245 seconds)
03:45:49 Quit sustineo (Client Quit)
03:45:59 Join Beta2K [0] (~Beta2K@d24-36-126-101.home1.cgocable.net)
03:46:13 Join n1s [0] (~n1s@rockbox/developer/n1s)
03:46:36 Quit Tomis (Ping timeout: 252 seconds)
03:46:36 Nick Tomis2 is now known as Tomis (~Tomis@70.134.100.53)
03:48:29JdGordon22:36:49 gevaerts Was the WPS loaded at boot back in the pre-sbs days? <- No, it was a "recent" change
03:49:17 Part froggyman
03:49:55 Part RoronoaZoro ("Ex-Chat")
03:50:14 Quit Blue_Dude (Quit: ChatZilla 0.9.86 [Firefox 3.6.2/20100316074819])
04:00
04:03:41 Join Barahir [0] (~jonathan@gssn-5f757aa7.pool.mediaWays.net)
04:05:20 Quit komputes (Quit: I haven't slept for ten days, because that would be too long.)
04:07:12 Quit Barahir_ (Ping timeout: 264 seconds)
04:16:24 Join JohannesSM64 [0] (~johannes@cm-84.215.75.42.getinternet.no)
04:20:36 Quit linuxguy3 (Read error: Operation timed out)
04:32:45 Join Rob2222 [0] (~Miranda@p4FDCB716.dip.t-dialin.net)
04:33:09 Quit Adubb ()
04:35:43 Join xiainx [0] (~xiainx@modemcable195.238-202-24.mc.videotron.ca)
04:35:54 Nick S_a_i_n_t_ is now known as S_a_i_n_t (S_a_i_n_t@203.184.1.208)
04:36:35 Quit Rob2223 (Ping timeout: 265 seconds)
04:45:29 Join angelwolf71885 [0] (~chatzilla@cpe-173-171-133-36.tampabay.res.rr.com)
04:46:55 Quit angelwolf71885 (Client Quit)
04:53:04 Join TillW [0] (~Till@CPE0018395fb63b-CM00195ee38f2c.cpe.net.cable.rogers.com)
04:58:49 Join saratoga_lab [0] (~9803c20d@gateway/web/freenode/x-tmegqaqlgrgwjthu)
05:00
05:01:08***Saving seen data "./dancer.seen"
05:07:51 Join jhMikeS [0] (~jethead71@rockbox/developer/jhMikeS)
05:28:07 Join phanboy4 [0] (~benji@c-174-49-112-244.hsd1.ga.comcast.net)
05:32:24UnhelpfulS_a_i_n_t: i don't see how messing w/ viewports could let you rotate text...
05:34:48 Join blairb [0] (~blairb@121-73-216-35.broadband.telstraclear.net)
05:35:42Unhelpfulsoap: realistically this needs the *rest* of an AVC decoder to matter. isn't codec video player on the GSoC list for this year?
05:36:04soapdid you mean to highlight soap, Unhelpful?
05:36:53Unhelpfulsoap: i might have, you thought this might be worth moving, i wanted to give you an idea what else we need for the IPU filter to help
05:37:03soapfor I was speaking in much more general terms, not at all attempting to be specific to the task at hand (as it were)
05:37:19Unhelpfulit could optionally be used to clean up mpeg-1/2 video now, if somebody can figure it out
05:39:28saratoga_labwhat does it actually do? resize or more stuff like iDCT?
05:39:36Unhelpfulthat's re: jhMikeS's question also i suppose. you don't need deblock, it's not anywhere in the spec, but running it on the decoded frames before display can pretty up your low-bitrate stream a good bit.
05:40:02S_a_i_n_tUnhelpful: well, not "automatically" rotate *any* text, but I've proven (with much viewport fuckery) that its possible to get text to scroll/display top to bottom/bottom to top.
05:40:24S_a_i_n_tI thought that was more the effect he was after, not rotating a specific block of text.
05:41:29 Join jeffp [0] (~jeffp@barmen.interhost.no)
05:41:37saratoga_labjhMikeS: as long as you're here, how difficult will it be to separate the mpeg1/2 stuff from the core of mpegplayer?
05:41:43 Part jeffp
05:42:04Unhelpfulsaratoga_lab: it's a smoothing filter intended to blend together macroblock boundaries. it's helpful with low-bitrate mpeg-1/2 streams, and a specific deblocking filter is part of the decode process for AVC
05:42:21saratoga_laboh its just the h264 deblocking filter?
05:42:40Unhelpfula good chunk of AVC's quality/bitrate improvement probably comes from making deblock mandatory and doing it before frames are used as references
05:43:03jhMikeSsaratoga_lab: (sorry, wasn't watching this closely and thought I was in this channel actually) I planted some seeds there, but didn't go full out. It would take some thought but its not a from-scratch job.
05:43:10Unhelpfulsaratoga_lab: the IPU has a postproc filter that at least claims to be able to filter per the AVC spec.
05:43:40*jhMikeS slowly recovers his bearings
05:43:52saratoga_labjhMikeS: is the basic app (buffering, audio thread, seeking) able to work with any format or would any parts need to be reimplemented?
05:44:25saratoga_labi don't know much about video decoding beyond the high level transform stuff, so its hard for me to judge what one of these projects involves
05:45:41jhMikeSsaratoga_lab: the video/audio threads probabaly need better abstraction. they're rather tied to the codecs themselves. seeking, buffering and parsing are pretty much abstracted each other.
05:45:56saratoga_labdid you email that student?
05:46:08jhMikeSsaratoga_lab: yes I did
05:46:20saratoga_labgreat, i felt bad since no one could answer any of his questions
05:47:15 Join Finixlives [0] (~4a22104a@giant.haxx.se)
05:49:03jhMikeSsaratoga_lab: np. I hope they aren't _too_ esoteric about the codecs' nitty-gritty. Well, whatever, I'll give it my best.
05:49:24saratoga_labi don't think hes a codec person at all, so probably not
05:50:35saratoga_labi suggested some combination of AC3 support, MPEG4 video and implementing resizing
05:50:45saratoga_labso expect questions about what that would involve
05:52:11 Quit JohannesSM64 (Quit: WeeChat 0.3.2-dev)
05:55:10jhMikeSsaratoga_lab: I guess its one of those things where once you get the multi-format going, the rest would be easier.
06:00
06:00:37 Quit katyl (Quit: Ex-Chat)
06:02:33 Quit panni_ (Read error: Connection reset by peer)
06:07:33 Quit Finixlives (Quit: CGI:IRC (EOF))
06:08:55 Quit CaptainKewl (Remote host closed the connection)
06:20:05 Join kramer3d [0] (~kramer@unaffiliated/kramer3d)
06:30:51 Quit anewuser ()
06:35:56 Join Adubb [0] (~aldubuc@67.201.160.144)
06:46:03 Quit xiainx (Ping timeout: 246 seconds)
06:49:51 Join jeffp [0] (~jeffp@barmen.interhost.no)
06:51:28 Part jeffp
07:00
07:01:10***Saving seen data "./dancer.seen"
07:17:34 Quit leavittx__ (Ping timeout: 258 seconds)
07:17:34 Quit leavittx_ (Ping timeout: 258 seconds)
07:17:34 Quit leavittx (Ping timeout: 258 seconds)
07:26:36 Join CGL [0] (~CGL@190.79.148.166)
07:29:39 Join leavittx [0] (~leavittx@89.221.199.187)
07:30:07 Join leavittx__ [0] (~leavittx@89.221.199.187)
07:30:09 Join leavittx_ [0] (~leavittx@89.221.199.187)
07:33:20 Quit Horscht (Ping timeout: 264 seconds)
07:33:40 Join Horscht [0] (~Horscht2@xbmc/user/horscht)
08:00
08:02:19 Join mitk [0] (~mitk@195.117.162.130)
08:03:07 Part Tomis
08:04:58 Join kramer3d_ [0] (~kramer@unaffiliated/kramer3d)
08:06:18 Quit kramer3d (Ping timeout: 258 seconds)
08:11:43 Quit BHSPitMonkey (Remote host closed the connection)
08:28:13 Quit CGL (Quit: Saliendo)
08:30:02 Join funman [0] (~fun@rockbox/developer/funman)
08:31:15 Join ender` [0] (krneki@foo.eternallybored.org)
08:43:09funman12h40 runtime on Clip+ and running, battery level 17%
08:43:59 Join froggyman [0] (~me@unaffiliated/froggyman)
08:44:27 Join moos [0] (moos@rockbox/staff/moos)
08:44:36 Join webguest67 [0] (~5edf559a@giant.haxx.se)
08:45:21 Quit webguest67 (Client Quit)
08:47:39 Quit dionoea (Changing host)
08:47:39 Join dionoea [0] (~dionoea@videolan/developer/dionoea)
08:47:52mitkfunman: What file you are running/listening on your Clip+ actually?
08:48:18funmanmp3 stereo 44.1kHz @192kbps
08:49:27mitkSo I think you almost beat Fuze v1 actually. Congratulations.
08:51:47 Join efyx [0] (~efyx@lap34-1-82-225-185-146.fbx.proxad.net)
09:00
09:00:23 Quit dmb (Quit: Leaving)
09:00:56 Quit froggyman (Ping timeout: 265 seconds)
09:01:12***Saving seen data "./dancer.seen"
09:01:12 Join froggyman [0] (~me@unaffiliated/froggyman)
09:01:18 Quit dmb_ (Quit: Leaving)
09:01:59saratoga_labrated battery life is only 15 hours, so getting 13 now is really impressive
09:03:05funmanthat makes me think we forgot something really stupid in the AMSv1
09:03:16 Quit ender` (Quit: Smoking is one of the leading causes of statistics.)
09:03:31funmanit would be nice to dump the as3517 registers when the OF is running and playing to compare
09:16:34 Join ender` [0] (krneki@foo.eternallybored.org)
09:21:22 Quit leavittx_ (Ping timeout: 258 seconds)
09:21:31 Quit leavittx__ (Ping timeout: 240 seconds)
09:21:45 Quit leavittx (Ping timeout: 258 seconds)
09:24:04JdGordonfunman: did that patch I did for you work?
09:24:18 Join bmbl [0] (~Miranda@unaffiliated/bmbl)
09:24:41funmandidn't test
09:25:05funmando you have a target with FM ?
09:26:26JdGordonnot for another week or so when my boxes get here
09:27:28JdGordonit used to be about an extra hour battery life if you leave it in the menus yeah?
09:27:32funmani thought kugel might be more aware than me of what/how needs to be done
09:27:37JdGordon(general target independant rule)
09:27:42funmanhm?
09:27:53JdGordonrandomly asking the channel...
09:28:19 Join Luca_S [0] (~5d3fc54b@giant.haxx.se)
09:28:56 Join stooo [0] (~sto@g227066084.adsl.alicedsl.de)
09:29:36 Quit stooo (Client Quit)
09:30:46funmanhm that can't be right, GPIOB_PIN1 should be the fm i2c scl, but it is the home button on fuzev2 ..
09:30:53 Quit togetic (Ping timeout: 245 seconds)
09:33:05amiconngevaerts, Torne: The strange thing is that I observe this boot slowdown on archos as well, which doesn't have dircache at all
09:33:44amiconnBoot charting is in svn now, but that doesn't help when going back...
09:35:19amiconnTorne: If it has no ill side effects, we could probably turn on LBA48 for all ata targets except the archoses
09:35:45 Join pamaury [0] (~pamaury@rockbox/developer/pamaury)
09:36:00amiconnThe latter is due to the fact that you can't use >128GiB on those anyway without taking additional measures, since the usb-ata bridge doesn't support lba48 either
09:36:34amiconn(You'll need a multivolume build and two partitions, having access to the 2nd partition in rockbox *only*)
09:38:14 Quit linuxstb (Ping timeout: 258 seconds)
09:40:50*JdGordon starts a batt bench in the menu to compare with wps usage
09:40:58JdGordonhave a good weekend all
09:41:03 Quit JdGordon (Quit: Leaving.)
09:46:17Luca_Sfunman: when looking at debug ports, pressing the home button does not change the value of any of the GPIO indicator
09:46:21 Join einhirn [0] (~Miranda@p54858DB8.dip0.t-ipconnect.de)
09:46:46funmandoes home button work anyway?
09:47:03 Join einhirn_ [0] (Miranda@vpn10.rz.tu-clausthal.de)
09:47:04Luca_S"hotkey not set" in the file browser
09:47:14Luca_Sso, it does something
09:47:22Luca_Sin the wps, it takes me to the playlist
09:48:00funmanbtw about the date I did that: change in rockbox, reboot to OF, change in the OF => and then both were synced
09:48:40Luca_Sok, trying right now (damn me, today I forgot the sansa cable at home)
09:49:22funmanmc2739: does this ring you a bell about RTC on sansas ?
09:50:31 Quit einhirn (Ping timeout: 240 seconds)
09:51:23Luca_Sok, works for me too
09:52:07 Join flydutch [0] (~flydutch@host225-165-dynamic.15-87-r.retail.telecomitalia.it)
09:52:19Luca_Sdid you see the report about i2s panic I had last night?
09:52:33 Join togetic [0] (~togetic@unaffiliated/ibuffy)
09:53:00funmanyeah
09:53:13Luca_SI was messing with recording etc
09:53:23Luca_Sbut I only had it once
09:53:31 Join linuxstb [0] (~linuxstb@rockbox/developer/linuxstb)
09:53:54 Join stoffel [0] (~quassel@p57B4C277.dip.t-dialin.net)
09:53:56 Quit einhirn_ (Quit: Miranda IM! Smaller, Faster, Easier. http://miranda-im.org)
09:54:16 Join einhirn [0] (~Miranda@p54858DB8.dip0.t-ipconnect.de)
09:55:20Luca_Show strange.. the wps just disappeared o_O
09:55:30Luca_SI activated random folder advance
09:55:36Luca_Sthen pressed next while playing
09:55:50Luca_Sand the wps backdrop disappeared
09:56:06funmanah so i wasn't dreaming yesterday =)
09:56:07Luca_Sreappeared by opening the menu and back
09:56:54Luca_Salso: the volume can't be set to mute (other samsa have this glitch, isn't it?)
09:57:06funmanyes i think it's written in SansaAMS wiki
09:57:13Luca_Smoving the scrollwheel does not turn on the backlight
09:57:32funmanother buttons do?
09:57:44Luca_S(in the wps screen)
09:57:44Luca_Syes
09:58:03 Join JanDo [0] (~JanDo@212.44.150.75)
09:58:34Luca_Swell, not exactly true
09:59:02Luca_Sit makes it brighter when its dim, but does not turn it on when it's off
09:59:11Luca_Smaybe it's by design?
09:59:59 Join mischasworld [0] (~quassel@p5B215158.dip.t-dialin.net)
10:00
10:00:14funmanno change in brightness for me
10:00:24*pamaury has got a shinny new Clip+ ! :)
10:00:37funmanbut yeah it doesn't turn the backlight on
10:00:46funmanpamaury: great! quick write the USB driver ! :)
10:00:50Luca_S:D
10:00:58pamauryyeah, do we have some doc ?
10:00:59Luca_Sthe power button is on 2nd bit from the left of D2, confirmed (on the wiki page it says Power?)
10:01:08ThomasAHpamaury: how boring ... you want a rockboxed used clip+ :)
10:01:25funmanpamaury: give me your email and i send you the linux patches for as352x/as353x
10:01:26*pamaury first need to install rockbox on it, and even before, need to open the box
10:01:31Luca_S(replace D2 with GPIOD)
10:01:56funmanLuca_S: you have a wiki account,
10:01:58funman?*
10:02:21Luca_Slife is beautiful: jewels works fine! :D now my boss has to threaten me to get me back to work :D
10:03:14Luca_Sfunman: maybe, long time ago I sent a wps - will need to check if i can recall the credentials
10:04:00saratoga_labpamaury: we have a lot of info on the USB controller, see the USB note here: http://www.rockbox.org/wiki/SansaAMS#Port_Status
10:04:14funmansaratoga_lab: it might be a little different from as3525 though
10:04:35saratoga_labah good point
10:04:47funmanbtw I tried to write some code for AMSv1 but i still don't understand completely although I had some feedback in both the OS & the device (sometimes)
10:06:30Luca_Sfunman: I'm not on the wikiusers page so no, I don't have an account
10:06:44funmancreate one and we'll approve you
10:07:09Luca_Sthough I'm on the UserList
10:07:25Luca_SLucaLeonardoScorcia
10:09:54funmanyou're in WikiUsersGroup so it should be ok i guess
10:11:02 Join kramer3d [0] (~kramer@unaffiliated/kramer3d)
10:13:37 Quit kramer3d_ (Ping timeout: 246 seconds)
10:17:35funmanLuca_S: i don't know why backlight isn't shown when scrollwheel is turned, perhaps kugel will know
10:18:25Luca_SI start to suspect there's something wrong with the debug io/ports screen
10:18:38funmanwhy ?
10:18:57Luca_Sthe hold button doesn't show any change, but it gets detected
10:19:18Luca_S(i.e. it works)
10:19:29Luca_Ssame for the home button
10:19:49funmanif i read correctly, one has to write on some pins before reading buttons
10:20:02funmanso you won't necessarily see it in IO ports
10:20:11 Quit shaggy-h (Ping timeout: 240 seconds)
10:20:43 Quit einhirn (Ping timeout: 260 seconds)
10:21:07Luca_Sah, that would explain it - I won't change the wiki page then. maybe a dedicated item in the debug screen "test button readings" would be useful
10:21:22funmanif they work in rockbox, no need to test =)
10:21:39Luca_Sstrange that the power button gets detected all the time as a gpio change however
10:22:19 Join merbanan [0] (~banan@c-62-220-165-110.cust.bredband2.com)
10:23:41linuxstbjhMikeS: Do you think you'll be around Rockbox this summer and able to help out with any mpegplayer SoC project?
10:25:29n1sjhMikeS: do you have plans for the voltage/freq scaling on the beast+
10:25:36n1ss/+/?
10:28:15 Join anubisfremen [0] (~c18cf902@gateway/web/freenode/x-paxbhnyyubyobmee)
10:28:44 Nick anubisfremen is now known as Guest12483 (~c18cf902@gateway/web/freenode/x-paxbhnyyubyobmee)
10:29:14gevaertspamaury: you say you find the storage_* integration awkward. Do you mean something else than that the current patch is done half in ata.c and half in storage.c in a random looking way?
10:31:32 Quit togetic (Ping timeout: 260 seconds)
10:31:33pamauryno I mean that :) Perhaps there should be a cleaner interface between storage.c and ata.c/mmc.c/... regarding this point but perhaps I'm wrong, I know nothing about the storage, it's juste that I find it bizarre :)
10:32:20gevaertsOK, everything is fine then :) That's the bit that needs work
10:33:04saratoga_labGuest12483: which codec were you interested in?
10:33:34gevaertsAs to whether it makes sense for flash, I'd say yes. While latency is of course much lower, these background threads are still using up valuable bandwidth that's needed by user-visible things like actually buffering
10:33:35 Quit Guest12483 (Quit: Page closed)
10:35:39 Quit phanboy4 (Ping timeout: 265 seconds)
10:36:02 Join emrecelikten [0] (~c18cf902@gateway/web/freenode/x-qabfmrgrnpjvdavj)
10:36:06emreceliktenOK, I am back
10:36:13emreceliktenWell I am interested in AAC
10:36:20emreceliktenThe difference seems quite big
10:36:40emreceliktenI think I may be able to improve the performance of that one.
10:37:22saratoga_labemrecelikten: thats a good project
10:37:27saratoga_labi think a lot could be improved there
10:37:43saratoga_labhave you done any codec stuff before
10:38:13emreceliktenSadly, no. But I am generally interested in signal processing, I think I can handle it
10:38:41saratoga_labi take it you've done a DSP course before?
10:39:24emreceliktenWell, sort of. Not officially though, yet
10:39:30emreceliktenWent to the lectures
10:40:26emreceliktenI plan to go for a Ph. D. in speech synthesis
10:40:48saratoga_labhave you taken a class in fourier series or transforms or any related math?
10:42:15 Quit Kitr88 (Ping timeout: 258 seconds)
10:43:02 Join Kitr88 [0] (Kitr88@BSN-182-4-208.dial-up.dsl.siol.net)
10:43:32emreceliktenNot officially, as I have said. But I am sure that I can handle it :)
10:44:19saratoga_labthats ok
10:44:38saratoga_labits not required that you have done all the calculus, though the project may be a little harder if you have not
10:44:43linuxstbemrecelikten: Do you have any C and/or assembly language experience?
10:45:09emreceliktenYes, I do. I'm a third year computer engineering student.
10:45:20emreceliktenAlso with a minor on electronics and communications engineering
10:46:07saratoga_labhave you done any ARM programming before?
10:46:10emreceliktenI can program a 8051 with assembler
10:46:13emreceliktenNo, I haven't
10:46:28saratoga_labMIPS?
10:46:46saratoga_labthats what we did when i was in undergrad anyway
10:47:00linuxstbemrecelikten: I don't think that's a problem though - ARM is very easy to pick up if you're comfortable with low-level programming.
10:47:33emreceliktenI think I may be able to learn that easily, they're almost all the same
10:47:50linuxstbemrecelikten: Do you own any Rockbox devices?
10:48:24emreceliktenNo, I won't be able to test the code on a real device
10:48:35emreceliktenBut I remember that you had emulators for that?
10:48:40gevaertsno
10:48:46emreceliktenI can buy one as well
10:48:56gevaertsWe have simulators, not actual emulators
10:49:00linuxstbYes, you really need to get at least one if you're going to work on Rokcbox.
10:49:07emreceliktenOk, will do
10:49:16linuxstbThe simulator is the Rockbox code compiled to run on a PC - so not useful for benchmarking etc.
10:49:35saratoga_labwe can develop codecs on PC, but actual optimization requires hardware
10:49:48emreceliktenTrue
10:49:59gevaertsIf you'd get accepted, you get enough money to buy one cheap second hand player :)
10:50:00saratoga_labso usually its something like verify algorithmic correctness on PC, then move to the device for optimization
10:50:23emreceliktenI was thinking about algorithmic optimizations first
10:50:35saratoga_labyes thats something that can begin on PC
10:50:49saratoga_labcertainly in the case of libfaad since we'll probably want to rewrite large parts of it
10:50:51emreceliktenBut I will buy one then
10:51:27saratoga_labdepending on where you are, sandisk players can be very cheap and are arm9 based
10:51:49saratoga_labthat reminds me, are you familar with the idea of fixed point calculation?
10:52:00saratoga_labi don't expect that anyone who hasn't done codecs has actual experience with it
10:53:30emreceliktenWell, you are right :)
10:53:43saratoga_labok so thats another thing to look at
10:54:00saratoga_labbasically its just doing decimal operations approximated with integers rather then floats
10:54:11saratoga_lablots and lots of that involved in this project
10:54:38saratoga_labsince we think the ffmpeg versions of most of the algorithms involved in AAC-LC (and probably all of the AAC-HE) are better then the ones we have
10:54:42saratoga_labbut they're all floating point
10:54:48saratoga_labwhich we can't use
10:54:59emreceliktenSo we have to convert them to fixed-point, right?
10:55:05saratoga_labyeah
10:55:25emreceliktenOkay
10:55:42saratoga_labso all the * operations become function calls
10:55:49emreceliktenI have found some Sansa players, is there a specific one that you recommend?
10:56:15emreceliktenLike C240 for example
10:56:15saratoga_labthe e200, clip, or fuze
10:56:24saratoga_labthe c240 will work as well
10:56:30emreceliktenOr will any E2x0 will do?
10:56:40saratoga_labthe e series is good too
10:56:42gevaertswell, c240v1 will work. c240v2 might not
10:56:53saratoga_labgood point
10:56:53gevaertsthe e series all work
10:57:09saratoga_labwhich time zone are you in?
10:57:14emreceliktenGMT+2
10:57:40emreceliktenI will get an E then
10:58:22saratoga_labin the US at least, the clips are the cheapest by far
10:58:55 Quit bluebrother (Disconnected by services)
10:58:55 Join bluebroth3r [0] (~dom@rockbox/developer/bluebrother)
10:59:13 Nick bluebroth3r is now known as bluebrother (~dom@rockbox/developer/bluebrother)
10:59:42n1sclip may not be much fun for video though
11:00
11:00:00gevaertsn1s: this is audio I think
11:00:05n1sah, wron project, sorry
11:00:20 Quit kramer3d (Quit: Leaving)
11:01:16***Saving seen data "./dancer.seen"
11:02:45emreceliktenHmm OK then
11:02:50emreceliktenAnything else that I should know?
11:03:21funmanc200v2 works quite well, some patching might be involved though but we can provide help on this
11:03:51saratoga_labyour proposal should try and show that you're motivated, aware of what the project will involve, etc
11:04:43funmanClip+ 15h runtime, still running \o/ (at 60MHz, will try 24MHz when this test is finished)
11:04:47saratoga_labbut looking into how the aac spec actually works is probably impractical given the complexity of it
11:04:55saratoga_labwow
11:04:59saratoga_labbeating the OF already
11:05:16linuxstbemrecelikten: My advice would be that very few potential students get involved with Rockbox before SoC. If you do that, and show you know what you're doing, you would increase your chances.
11:05:30funmanLuca_S: could you bench the battery on your fuzev2 ?
11:05:39S_a_i_n_twow...nice runtime for the clip+
11:05:43S_a_i_n_tcongrats.
11:06:49emreceliktenOkay, thanks for the tip. I know I haven't done any work on this field before but this one seems quite interesting for me, that's why I'm here
11:06:59emreceliktenI will try to progress before SoC
11:07:20emreceliktenSoC is not that important actually, I want to spend my summer working on something
11:07:37emreceliktenAnd SoC would force me to do it
11:07:38saratoga_labwell even if you don't end up doing SoC you're certainly welcome to hang out here
11:07:44emreceliktenThanks
11:09:27emreceliktenI will try to learn about fixed point arithmetic in my spare time now, then
11:09:35emreceliktenThat's almost the first think I need to do anyway
11:09:40emreceliktenThanks for the tips, I have to go now
11:09:40saratoga_labthat should be pretty straight forward
11:09:50emrecelikten*thing
11:09:51saratoga_labits mostly just annoying more then anything
11:10:28saratoga_labi told someone else who asked about the codec projects to look at how jpeg works on wikipedia
11:10:54saratoga_labsince its basically the simplist codec possible, and most of the stuff in jpeg is in all codecs for audio or video
11:11:14Kohlrabijpeg is based on DCT?
11:11:37saratoga_labcorrect
11:11:41Kohlrabi:)
11:11:45Kohlrabinot jpeg2000, though
11:11:48Kohlrabithat uses wavelets
11:11:53Kohlrabino wonder it fails :D
11:13:50Unhelpfulfails how? it's smaller at the same and has a number of interesting features... it's just too slow and complex :/
11:14:05KohlrabiI think it's shown that wavelets don't work well with image
11:14:11Kohlrabiat least the recent implementations
11:14:18funmanis wavelet used with audio signals?
11:14:42Unhelpfulfunman: it's been tried by a number of forgotten proprietary codecs ;)
11:15:30Unhelpfulit's of some benefit with handling transients but never seems to beat mdct overall
11:15:56Luca_Sfunman: not today unluckily: it's not fully charged and i forgot the cable at home. from a subjective estimate the consumption seems in line with the OF, the battery indicator also seems well calibrated. I'll charge this evening and bench it tonight.
11:16:28 Quit emrecelikten (Ping timeout: 252 seconds)
11:16:51saratoga_labfunman: no, the DCT family seems to work better then wavelets for audio
11:17:02funmanLuca_S: i plugged the output to my soundcard and record PCM (needs loads of free space), and I look at the recorded data to see at which time it stopped
11:17:13saratoga_labi don't think theres ever been a wavelet audio coder outside of scientific literature
11:17:38Kohlrabisome (biased?) view on wavelets in video coding: http://x264dev.multimedia.cx/?p=317
11:18:22saratoga_labi think the reason the dct is so hard to beat is that our ears are basically fourier transformers, so its hard to beat a fourier transform based coder
11:18:33saratoga_labits sort of the "natural" choice
11:19:38 Join togetic [0] (~togetic@unaffiliated/ibuffy)
11:19:50 Quit saratoga_lab (Quit: Page closed)
11:24:00Luca_Shm.. after a panic, rockbox rebooted and as soon as it booted, the 'shutting down' message appeared and shutted down the player
11:24:33funmani've seen that also
11:31:29UnhelpfulKohlrabi: my experience has been that jpeg2k looks better at high compression, but as one of the commenters points out, a wavelet codec can effectively encode the image at a lower resolution to achieve high compression ratios
11:32:00KohlrabiI'd guess that it also highly depends on the source
11:32:24KohlrabiSinc efrom my understanding wavlets can effectively code global patterns
11:32:29KohlrabiSince from*
11:32:55KohlrabiI never dug into coding, though :(
11:33:36funmanclip+ stopped at 15h30
11:33:47 Quit mischasworld (Ping timeout: 245 seconds)
11:33:53S_a_i_n_tnice...
11:34:10Kohlrabi:O
11:34:15Unhelpfulafaik the most useful thing done for audio with anything quite like wavelets is transform-based lossless (useful since the transform approximates mdct and can scale to lossy)
11:34:31S_a_i_n_tOF+30, that's not to be sneezed at, in this stage of development.
11:35:13Kohlrabiwow
11:35:22KohlrabiI didn't chekc rb for some time
11:35:33KohlrabiHow did the Clip+/v2 port progress that fast? :O
11:35:41S_a_i_n_tfunman?
11:35:45S_a_i_n_t;)
11:35:46Kohlrabiamazing
11:35:55funmanmost of Clipv1 code could be reused as-is, some other parts needed only little modifications
11:36:09S_a_i_n_tmodest...I like it ;)
11:36:22KohlrabiNow I wished I didn't give my Clipv2 away
11:36:28Kohlrabioh well
11:36:54Kohlrabi"Externalk Storage": "Yes" means that you can also read the micro-SD?
11:37:13Kohlrabifunman: Congratulations on your work
11:37:26funmanKohlrabi: right
11:37:31Kohlrabioh my
11:37:47KohlrabiThat's so awesome, after 1.5 years I wil use rb again :D
11:38:50Kohlrabifunman: any caveats before I put rb on my Clip+?
11:39:42 Quit n1s (Ping timeout: 252 seconds)
11:40:13funmaneverything is written on SansaAMS page of the wiki but basically: writing doesn't work, expect some crashes, and volume can't be set very low
11:40:40 Join dfkt [0] (dfkt@unaffiliated/dfkt)
11:41:49funmanOF is advertised as running 15 hours, so we already do better
11:41:59Kohlrabigreat
11:42:14KohlrabiThe battery lifetime with vorbis q5 already starts to annoy me a bit
11:42:46Kohlrabiwait
11:42:59KohlrabiI hope the SansaAMS is no april fool's page? :D
11:43:25funmani didn't try vorbis though
11:44:58JanDofunman: Hello. My Clip+ runtime 15h40min and still running. Played FLAC v8 75%vol(-13) in KOSS the PLUG
11:45:11funmanJanDo: nice !
11:45:39funmani'm just recharging it and will try a build with reduced CPU speed
11:46:01funmancan you compile your own rockbox build?
11:46:40funmanJanDo: btw which revision did you run ?
11:46:41JanDoyes, yesterday i co SVN code and setup Cygwin.
11:47:17JanDonow testing at r25423
11:47:21funmanhttp://pastie.org/900094 <- here is the patch i'm going to try
11:50:28Kohlrabifunman: do you know about the pitch hifts in the orignal firmware? Is rockbox affected by that, too?
11:51:41 Join n1s [0] (~n1s@rockbox/developer/n1s)
11:55:46Luca_SI have my music organized as /Artist/Album/Track.mp3. is there any way to say "play everything" without using the database?
11:56:25*S_a_i_n_t suspects to see better than 15:30 going down to 24MHz
11:56:40Luca_SKohlrabi: http://www.rockbox.org/tracker/task/10906
11:56:48S_a_i_n_tthe previous test was at 60MHz, no? Or was it at 30?
11:57:22S_a_i_n_tLuca_S: make a playlist?
11:57:31Luca_Ssaratoga:
11:57:33Luca_Sops
11:58:42Kohlrabithanks
11:59:15Luca_Ssorry, i meant S_a_i_n_t: fuzev2 doesn't have write support - I got close with random folder advance, but with it the shuffle setting is useless since it shuffles in the current folder
12:00
12:00:06S_a_i_n_tmake playlist manually, on a PC?
12:00:36S_a_i_n_twow...apparently I miss words at random now...ooops :/
12:01:38Luca_Soh, I didnt think of that - is the file format documented anywhere? i can't seem to find it in the wiki
12:02:30S_a_i_n_thaving not ever used a playlist once, I can't say sorry...perhaps the manual?
12:02:40S_a_i_n_t(note: not a RTFM comment)
12:02:58 Join Multiplex [0] (~5773e1e8@giant.haxx.se)
12:04:27Multiplexis the only way of getting the source to use SVN? I mean it used to be possile to get a tarball of the source
12:04:50Luca_SGoogle will help me, thank you. RTFM would have been perfectly fine anyway :D
12:06:49S_a_i_n_tSource Tarball used to be on the Daily Build page, I take it that isn;t the case anymore?
12:08:50 Quit Adubb (Read error: Connection reset by peer)
12:09:31MultiplexNo, - I followd the link from currnet builds page that said that "Daily builds, voices, fonts and source" but I can't see the source or the voices there
12:10:03 Join einhirn [0] (~Miranda@p548584F9.dip0.t-ipconnect.de)
12:11:20funmanS_a_i_n_t: 60MHz yes
12:12:18funmanKohlrabi: i didn't measure the real pitch, I assume there would be no difference between Clipv1 & Clipv2/+ but feel free to verify
12:12:22 Join liar [0] (~liar@clnet-p09-185.ikbnet.co.at)
12:13:03funmanhm sorry, since we run the PLL at a different frequency it will be different
12:13:09Kohlrabiah
12:13:17Kohlrabigood
12:13:42 Join BlackSwan [0] (~kvirc@80.93.115.29)
12:13:58BlackSwanHi
12:15:19BlackSwanWhere can I found DominikWenger?
12:16:18funmanhe's domonoky on irc, he's often here on the evenings but he'll read the logs
12:16:30 Join leavittx [0] (~leavittx@89.221.199.187)
12:16:32 Join leavittx_ [0] (~leavittx@89.221.199.187)
12:18:07 Quit Rob2222 (Ping timeout: 265 seconds)
12:18:12BlackSwanfunman: O...
12:18:23mtBlackSwan: Also, ask your question to the channel, maybe someone else could answer you.
12:19:11bluebrotherBlackSwan: evening means evening in european time
12:19:12BlackSwanI want speek about his themes on summerOfCode2010
12:19:51bluebrotherBlackSwan: there are not "his" themes. He's volunteered as potential mentor for some project ideas.
12:20:32BlackSwanHm... And what that means?
12:20:42mtBlackSwan: The mentor for this project has not been decided yet.
12:21:17 Quit Multiplex (Quit: CGI:IRC (EOF))
12:21:24*bluebrother wonders what the project is BlackSwan is talking about
12:21:31 Quit liar (Ping timeout: 258 seconds)
12:21:55BlackSwanBut if I want to try help You with web-site... With him i must speak?
12:21:57 Join liar [0] (~liar@clnet-p09-185.ikbnet.co.at)
12:21:57bluebrotherbesides, as a student you should generally consider the community your primary mentor. The assigned mentor is responsible for some paperwork stuff
12:22:05mttheme editor ? Although now that I think of it, maybe I jumped to a conclusion too quickly
12:22:11bluebrotherBlackSwan: talk to the community.
12:22:27*bluebrother 's guess was the "web services" stuff
12:22:48 Join Rob2222 [0] (~Miranda@p4FDCB716.dip.t-dialin.net)
12:23:43BlackSwanI can code on PHP,HTML, some JavaScript... Works with mySQL...
12:24:09bluebrotherBlackSwan: if you have questions just ask. This channel is logged, so people can read the logs. If they don't someone can point them to the logs :)
12:25:39BlackSwanQuestion is - how and with who I nedd comunicate, to help with web services?
12:25:49BlackSwan*need
12:28:17bluebrotherhow is easy: the same way all other developmen communication happens. Most happens on IRC, but the mailing list is also a channel (though noticably less used than IRC for development talk)
12:29:53BlackSwanOh... I dont think that posting here my e-mail but... grafmailgraf@mail.ru
12:29:59bluebrotherwho is a bit more complicated as a couple of people did work in the web services we already have. domonoky is the one who was recently working on the theme site.
12:30:11BlackSwanI hope there are no spammers ^)
12:30:15bluebrotherBlackSwan: err ... I was not saying anything about posting an email address.
12:30:16BlackSwan:)
12:30:30Torneamiconn: well, if you extract the bootchart patch you can apply it to older revisions, and just throw away the bits that don't apply ;)
12:30:44bluebrotherI was saying that the mailing lists are used as well for development talk.
12:31:31Tornedo we have somewhere i can upload some test builds to, or shall i just stick tehm on my own httpd
12:31:45bluebrotherI assume you've read the "Information for students" section on the SummerOfCode2010 wiki page?
12:32:05BlackSwanbluebrother: O, it was my idea :)
12:32:14 Quit funman (Quit: free(random());)
12:32:28gevaertsTorne: I don't think we have
12:32:37gevaertsI usually just host test builds myself
12:32:54bluebrotherBlackSwan: your idea as in that you didn't pick it out of the project ideas on the SummerOfCode2010 page?
12:33:39BlackSwanbluebrother: my idea was posting e-mail...
12:34:36Tornegevaerts: the list of people having problems of one kind or another that might plausibly be caused by the startup workaround thingy is getting rather long, so I'm gonna do a big pile of builds with the clearing-iram fix in instead
12:36:35S_a_i_n_tTorne: I've forgotten to ask, how's your ipod going since that patch?
12:37:11BlackSwanI must go away... I hope some body from developers send to my E-mail...
12:37:20TorneS_a_i_n_t: still works perfectly fine
12:37:40TorneS_a_i_n_t: but the instant i started bisecting the other day and went to an old revision, my ipod immediately locked up and needed resetting after shutdown ;)
12:37:55TorneS_a_i_n_t: which kinda proves that my player still does it. Just not if you clear iram. So i think we win :)
12:38:01 Quit BlackSwan (Quit: When two people dream the same dream, it ceases to be an illusion. KVIrc 3.4.3 Shiny(svn-3438) http://www.kvirc.net)
12:38:32soapis there a FS task where I can follow this?
12:39:00soap11149, sorry for asking before looking
12:39:30 Join kugel [0] (~kugel@rockbox/developer/kugel)
12:40:45Torneanyway, when i'm actually properly awake I am going to do a test build with that change for all the relevant models based on current and 3.5
12:40:59Tornein the hope that *some* of the people affected will test ;)
12:42:22S_a_i_n_tOr give "shutdown troll" something new to complain about ;)
12:44:04linuxstbTorne: Any idea how long iram will keep its content?
12:44:44Tornelinuxstb: eternally, as far as i understand
12:44:55Tornewell, yaknow. until the battery runs out.
12:45:04TorneI think it's kept powered when the pcf goes to deep standby
12:45:18Tornei could be wrong
12:45:23linuxstbOK, so the time the ipod is "off" shouldn't affect the reliability of power-on?
12:45:32Tornedon't think so.
12:45:46TorneI have previously had the poweron issue come up immediately after powering off, or when powering on after it being off all night
12:46:08Tornetbh i am wildly speculating here
12:46:19Tornethe clearing iram thing is basically a big guess with no "scientific" basis ;)
12:46:26Tornebut it appears to work
12:46:53linuxstbWell, the Apple bootloader checks iram for things, so it's a plausible theory.
12:47:24linuxstbWhat if you write the "diskmode" magic to iram, then power-off? Does it always then start in disk mode?
12:47:29 Join Geert_van_Dijk [0] (~chatzilla@cc412026-a.zwoll1.ov.home.nl)
12:47:30TorneOoh
12:47:35TorneThat sounds like science :)
12:47:39Tornei'll try that in a few
12:47:40*linuxstb apologises
12:47:53 Nick Geert_van_Dijk is now known as geertvdijk (~chatzilla@cc412026-a.zwoll1.ov.home.nl)
12:48:09Tornethe parts of the bootloader i have disassembled basically mean that when it boots the OF it writes some magic string to iram
12:48:20Torneand if it sees that magic string *again* on next boot it assumes the battery is dead
12:48:37Tornebecause it assumes that spinning up the hard disk caused the voltage to drop too low to run the system
12:48:37linuxstbThat also sounds like science...
12:49:05Tornewhich is a fair-enough thing to do
12:49:26Torneit avoids the Beast reboot loop situation, i suspect ;)
12:49:43Torneso i think we are maybe triggering *some* of this logic by accident
12:49:58Torne(probably not all of it, there's more than just the one thing in iram, it has several flags that i've not figured out)
12:50:15Torneso we're sending the bootrom down a path that it shouldn't ever see and it's doing half of the "no battery" logic then falling over
12:50:55Tornebut logically the bootrom should always work if IRAM is empty, because that's what happens after the battery *does* die completely and then gets recharged, and probably also on hard-reset
12:52:11bluebrotherwhat would happen if the IRAM would get cleared on poweroff? Shouldn't that be similar to when the battery got ran out completely?
12:52:28*bluebrother finds that error message on shutdown pretty annoying
12:54:40Tornebluebrother: yes, that's what I'm doing
12:54:45Tornebluebrother: FS #11149
12:54:58Tornebluebrother: i've tested this for a week or so on my player and not had the poweron issue.
12:55:05Tornei'm gonna put up test builds in a bit
12:55:59S_a_i_n_tI've failed to see the "batt low" icon on shutdown once, the only time I see that icon it usually means "white screen of death" on the nanos
12:56:16bluebrothernice. Got to try this (though my mini was rarely affected by the shutdown problem anyway)
12:56:29 Quit leavittx (Ping timeout: 240 seconds)
12:56:37TorneS_a_i_n_t: it appears that on some players it is either relly hard to see or actually not visible
12:56:47TorneS_a_i_n_t: our shutdown process has already turned off the backlight *and* the lcd where possible
12:56:58TorneS_a_i_n_t: but on at least some models/firmwares the OF turns the lcd back on
12:57:12Torne(though not the backlight)
12:57:33 Quit leavittx_ (Ping timeout: 258 seconds)
12:57:38S_a_i_n_tOn my nanos at least, if the backlight is off, and the idle shutdown occurs it actually turns the backlight *on* for shutdown.
12:57:51S_a_i_n_tnot sure if thats some setting of mine or not though.
12:58:05linuxstbBoth generations?
12:58:16S_a_i_n_tpretty sure.
12:58:22Tornethe backlight coming on when idle shutdown kicks in is normal..
12:58:24Tornei think.
12:58:28S_a_i_n_tI'll check, one sec.
12:58:30Tornethe "idle shutdown" event counts as a UI interaction
12:58:37Torneso it makes the backlight come on ;)
12:58:44TorneThe shutdown process later does turn the backlight off.
12:58:53S_a_i_n_tAh, I thought it was a bit weird...thats all.
12:59:42CIA-5New commit by 03bluebrother (r25435): Warn when selecting system proxy settings with invalid values. ...
13:00
13:01:18***Saving seen data "./dancer.seen"
13:08:35 Join leavittx [0] (~leavittx@89.221.199.187)
13:08:47 Join leavittx_ [0] (~leavittx@89.221.199.187)
13:11:25amiconnTorne: Yes I can do that (and there is no ;) ), but that would have been a reason to leave it as a patch
13:11:42 Quit stoffel (Ping timeout: 246 seconds)
13:13:22 Join robin0800 [0] (~robin0800@cpc2-brig8-0-0-cust964.brig.cable.ntl.com)
13:14:58Torneamiconn: well, it's not just for looking into the past..
13:20:25 Part JanDo (" www.qip.im ")
13:24:14 Join Casainho [0] (~chatzilla@87-196-52-105.net.novis.pt)
13:25:32 Join JanDo [0] (~JanDo@212.44.150.75)
13:39:15Torneah, blue fixed my yellow
13:39:24Tornei didn't notice because the buildserver was broken ;)
13:40:34gevaertsmoos: you might want to try FS #11167
14:00
14:05:07 Join funman [0] (~fun@rockbox/developer/funman)
14:05:36 Join jeffp [0] (~jeffp@barmen.interhost.no)
14:06:57 Part jeffp
14:07:01funmanshould a battery charging graph look logarithmic ? (charges slower and slower)
14:10:09pixelmabluebrother, AlexP_, others: I cleaned up goban.tex in two parts - the button table and used [item] for the submenus as well - (a) I'm not a 100% sure if we agreed on the latter, does someone remember? (b) is it worth it to split a commit into two parts or not?
14:10:34 Join stripwax [0] (~Miranda@87-194-34-169.bethere.co.uk)
14:11:24 Quit geertvdijk (Read error: Connection reset by peer)
14:13:32 Join webguest44 [0] (~5edf559a@giant.haxx.se)
14:13:41 Quit webguest44 (Client Quit)
14:13:50 Join stoffel [0] (~quassel@p57B4C277.dip.t-dialin.net)
14:14:29 Quit einhirn (Quit: Miranda IM! Smaller, Faster, Easier. http://miranda-im.org)
14:17:17 Quit stripwax (Read error: Connection reset by peer)
14:20:06 Join stripwax [0] (~Miranda@87-194-34-169.bethere.co.uk)
14:20:32funmanI opened a poll for frequency of crashes on Clipv1 : http://forums.rockbox.org/index.php?topic=24375.0
14:21:01gevaertsfunman: is "Official test builds" really the place for that?
14:21:10funmanI figured "Official Test Builds" was the good forum category since you need the Official (Test) Current build
14:21:26gevaertsThe current build is not a test build
14:21:42funmanNo ?
14:22:22gevaertsno. That forum is for modified builds by developers who want to get more widespread testing of a patch
14:22:57funmanWhere shall I put this forum poll ?
14:23:32funmanAudio Playback perhaps
14:24:03bluebrotherpixelma: I'm not aware of an agreement on using items for submenus or not (but maybe I simply didn't notice) but IMO it's a good idea.
14:24:12gevaertsI'd say general discussion. That's where Torne put the ipod shutdown poll, so there's a precedent. Audio Playback might be good as well
14:24:30 Nick fxb__ is now known as fxb (~felixbrun@h1252615.stratoserver.net)
14:25:02 Join panni_ [0] (hannes@ip-95-222-52-93.unitymediagroup.de)
14:25:10bluebrotherpixelma: no idea if it's worth splitting the patch in two commits. Won't hurt but I don't think it's necessary.
14:25:22bluebrotheras the changes rely to the same thing
14:25:51 Nick fxb is now known as fxb__ (~felixbrun@h1252615.stratoserver.net)
14:28:42 Quit JanDo (Ping timeout: 246 seconds)
14:28:42pixelmabluebrother: I just saw that there are even more changes in this file. In the introduction there are a few \\ to force newlines which I removed
14:29:51 Quit mitk (Quit: Leaving)
14:29:56bluebrotherpixelma: there was some talk about forcing newlines with \* a while ago. It might be good to discuss if paragraphs should use indents or blank lines
14:37:52pixelmaok, so leave the change out of the commit for now?
14:38:08pixelmajust referring to the newline one
14:38:40bluebrotherpamaury: regarding your question about translations the other day: I'm not aware of any policy for that. If you update a translation and feel confident enough about it just commit it.
14:39:15bluebrotherat least I can just commit contributed translations, and I haven't seen anyone reviewing them before. It would be good to have, though.
14:39:19pamauryok, I'm not sure about consistency mainly
14:39:30bluebrotherit's just that someone needs to understand the language for reviewing :)
14:39:42pamaurybut go on, I'll continue this week-end or tonight
14:39:50pamauryI guess it's better than nothing
14:39:51 Join steve|m [0] (~steve@p4FD47C96.dip.t-dialin.net)
14:39:59bluebrotherpixelma: well, \\ forces a newline while \* a new paragraph so those are a bit different.
14:41:42Kohlrabifunman: I get some error messages about playlists not found when trying to playback vorbis files on Clip+
14:42:36Luca_SKohlrabi: if playback starts fine after the message, then it's alright: it's just complaining that it can't write a playlist file to disk
14:42:45Kohlrabiah
14:42:50KohlrabiIt starts fine
14:43:01Kohlrabithough the text is a bit garbled on the display
14:43:18KohlrabiBut all in all it works very well already.
14:45:04pamauryyeah, rockbox on Clip+ !!
14:45:51arbingordonit didn't use to run on the clip+ right? (new port?)
14:46:12pamauryit's unstable
14:46:36arbingordonso is the gigabeat S port. but that's what I'm using and its flawless :D
14:47:13Tornearbingordon: only for some people ;)
14:48:06pixelmabluebrother: is that a yes or no - or shall I replace \\ with \* ?
14:48:39gevaertsarbingordon: just wait until that thing decides to format its disk right when you wanted music :)
14:48:46pamauryfunman: I can confirm that the volume doesn't get very low, if you still had a doubt :)
14:49:00bluebrotherpixelma: I would remove the \\ if it makes sense.
14:49:08bluebrotheri.e. go with what you currently have :)
14:49:34 Quit xavieran (Quit: ➤➤➤➤➤➤ UniCode shall rein forever! ➤➤➤➤➤➤)
14:49:53pixelmaI should probably do some test compiles which would also tell me how it looks in the PDF... ;)
14:50:22 Join xavieran [0] (~xavieran@ppp118-209-242-153.lns20.mel6.internode.on.net)
14:54:32 Join JanDo [0] (~JanDo@212.44.150.75)
14:57:36 Quit stripwax (Quit: http://miranda-im.org)
15:00
15:01:19***Saving seen data "./dancer.seen"
15:05:26 Join domonoky [0] (~Domonoky@rockbox/developer/domonoky)
15:08:00 Part JanDo (" www.qip.im ")
15:10:26 Join Schmogel [0] (~Miranda@p3EE21B8F.dip0.t-ipconnect.de)
15:13:35 Join JanDo [0] (~JanDo@212.44.150.75)
15:14:12Tornehttp://forums.rockbox.org/index.php?topic=24376.0 <- does this make sense
15:14:17Torne(test build for startup issue)
15:14:39gevaertsdomonoky: I noticed on DevConEuro2010 that your name is on the list of people who'll attend DevCon, but not on the list of people who want a bed
15:15:21domonokyuh, someone must have removed that..
15:16:59 Quit FlynDice_ (Remote host closed the connection)
15:33:35*Torne tries to remember all teh things he told people he would do/investigate so he can make a proper todo list, and fails
15:37:04funmanhm too bad the Fuzev2 crashed while testing battery
15:37:52 Quit Casainho (Quit: ChatZilla 0.9.86 [Firefox 3.6.2/20100316060223])
15:38:29funmanand the Clip+ stops charging at 63% ?
15:42:58 Quit pamaury (Quit: Quitte)
15:47:10 Join xiainx [0] (~xiainx@modemcable195.238-202-24.mc.videotron.ca)
15:50:57 Join xou [0] (niko@eris.feh.name)
15:51:12 Quit funman (Quit: free(random());)
15:52:12 Join phanboy4 [0] (~benji@c-174-49-112-244.hsd1.ga.comcast.net)
15:55:34Luca_Sfunman: which capacity is your fuze?
15:55:46linuxstbTorne: Did you try writing the diskmode magic to iram before power-off?
15:55:52Tornelinuxstb: not yet
15:55:59Tornegonna test that in a few
15:56:54Luca_Smine is 8gb (+8gb uSD) tonight i'll try to fill it up to see if it reads correctly on the entire space
16:00
16:01:20xouhi, i'm thinking about applying for a GSoC-project at Rockbox
16:01:50xouthe "Rockbox as an App" caught my attention
16:02:26xouis that project still available? :)
16:03:27linuxstbxou: All projects are "available" until a student is selected.
16:04:25kugelyes, the fuze has some problems. for example, sometimes playback just stops instead of continuing (I suspect the rebuffering fails)
16:04:52kugellinuxstb: or two :)
16:05:12kugelwe might have 2 students working on RaaA (independently) since we want it so badly
16:05:31kugelbut that seems unlikely
16:05:32 Quit n1s (Ping timeout: 276 seconds)
16:06:04linuxstbkugel: Anything is possible. But it's not worth thinking about until we have all the applications, and know how many student slots google give us.
16:06:06Luca_Sbtw, I noticed that my fuze seems to get quickly hot when running the fft plugin
16:06:22Luca_Snow i stopped it to let it cool a bit
16:06:39Luca_Si will try in a few minutes to exclude the "warm hands" cause :P
16:06:46xouok... im really interested in having this feature beacause my iPod is probably breaking down soon and i already thought about making an app out of rockbox a few weeks ago
16:07:33xouand im probably going to buy some sort of android-based device as an replacement for my ipod
16:08:04kugelyou will need a decent application, we appear to have multiple applicants for RaaA
16:08:15xouok
16:09:37Tornehey wait what
16:09:46Tornei just updated my regular build and now it *doesn't* take six seconds to load the font any more
16:10:01Luca_Snah, I probably just left it on the laptop, sorry for the noise about the fft plugin..
16:10:27xouwhat are the requirements for working on this project? i did a lot of C/C++ coding and some Java for University and some smaller projects, but mostly commandline-stuff, nothing with GUIs
16:10:41linuxstbTorne: Congratulations - you fixed it!
16:10:48Tornelinuxstb: haha
16:10:52linuxstb(unless it now takes 10 seconds...)
16:10:58Tornelinuxstb: no, it's practically instant now
16:11:05Torneeven if i delete glyphcache
16:11:12Tornebut i haven't changed anything!
16:12:14Tornei didn't even catch up very far in svn, I was already only a few days out of date
16:12:23xouwould i be required to have the knowledge on building GUI/Android apps already, or do you think it is possible to learn all that stuff while working on the project?
16:12:42 Quit Strife89 (Quit: Leaving.)
16:13:26kugelxou: there are no "requirements", but you should expect that any experience in the items listed under skills involved helps a lot
16:13:48linuxstbxou: In my view, the main requirement is that you're motivated and willing to put the hours into the project over the summer. But you should also realise that (in my view at least) a lot of the project won't be dealing with Android at all, it will be changing lots of parts of Rockbox to separate the "application" part from the "firmware/OS" part, and preparing an infrastructure for porting to something like Android.
16:15:07xouok, thank you!
16:15:20Tornelinuxstb: no, it doesn't start diskmode if you write the string out before shutdown
16:15:33Tornelinuxstb: so that doesn't really prove anything either way :)
16:15:35linuxstbTorne: I'm not sure what that proves though...
16:15:39linuxstb;)
16:17:18TorneOops.
16:17:26TorneBootchart has broken the ondio backlight mod :)
16:17:32Tornei've stolen the letter "b" in configure from it by accident
16:19:25CIA-5New commit by 03torne (r25436): Change bootchart to be "c" in configure, instead of "b" which clashes with the ondio backlight mod
16:26:59 Join pamaury [0] (~c2c7a50a@rockbox/developer/pamaury)
16:27:56kugelgevaerts: why do you need the info about the thread for fs#11167
16:27:57kugel?
16:29:26gevaertskugel: because the thing only cares about disk accesses if they were done by another thread
16:30:37 Join mitk [0] (~mitk@chello089078013146.chello.pl)
16:31:47 Quit pamaury (Ping timeout: 252 seconds)
16:33:33 Quit stoffel (Remote host closed the connection)
16:36:35linuxstbsaratoga: I was reading http://forums.rockbox.org/index.php?topic=13883.30 and wondered if you were aware that "VOB" is an an mpeg program stream? i.e. the container mpegplayer already supports. Or do you mean something else?
16:45:01 Join plauclair [0] (~p-l_aucla@dsl7-141.express.oricom.ca)
16:51:44 Quit linuxstb (Ping timeout: 258 seconds)
16:53:33kugelthis is my current draft if someone wants to look at it: http://pastie.org/900367
16:53:40kugelcomments are welcome
16:56:58 Quit xiainx (Ping timeout: 260 seconds)
17:00
17:01:21***Saving seen data "./dancer.seen"
17:03:57Luca_Skugel: http://www.rockbox.org/irc/log-20100402#09:57:13
17:05:07 Part plauclair
17:05:15kugelLuca_S: yes, I know about this issue, but I have no idea where the problem is
17:05:43Luca_Sah, ok, sorry
17:07:02 Join linuxstb [0] (~linuxstb@rockbox/developer/linuxstb)
17:07:50 Join xiainx [0] (xiainx@wpa062030.Wireless.McGill.CA)
17:08:51kugelLuca_S: it's very strange
17:10:44Luca_Salso, i set the wheel light to always on. but when backlight goes off, it goes off the wheel light too
17:11:40 Join stripwax [0] (~Miranda@87-194-34-169.bethere.co.uk)
17:12:49 Join jgarvey [0] (~jgarvey@cpe-065-190-066-089.nc.res.rr.com)
17:13:04kugelthat's possible
17:13:05 Quit stripwax (Client Quit)
17:14:15 Join stripwax [0] (~Miranda@87-194-34-169.bethere.co.uk)
17:14:26kugelI'm not sure if we can fix that
17:16:48 Join dockimble [0] (~dockimble@77.227.1.24)
17:17:09 Join anewuser [0] (anewuser@unaffiliated/anewuser)
17:18:48 Join stoffel [0] (~quassel@p57B4C277.dip.t-dialin.net)
17:18:53 Nick fxb__ is now known as fxb (~felixbrun@h1252615.stratoserver.net)
17:18:59Luca_Sthe brightness setting is not always coherent: often (but not always) if I move slowly from 1 to 2, the brightness stays the same. if I cycle between 1 and 2 repeatedly, it sometimes changes brightness.
17:22:37kugelwe'll look into it once we have fixed the real problems :)
17:23:28Torneaha, i figured out why font_load takes 6 seconds for cabbie on my ipod
17:23:37Tornethe 15pt version of Helvetica is *way* bigger than the other sizes
17:24:05Tornecabbie only uses 15pt on 320x240 or 240x320 screens, i.e. ipodvideo/beast
17:24:17 Join CGL [0] (~CGL@190.79.138.162)
17:25:43Tornei guess the 15pt version contains more characters than the others
17:25:56Torneit still shouldn't really take 6 seconds :(
17:26:07Tornebut if i load a smaller font it takes <0.5s
17:26:07 Quit merbanan (Ping timeout: 276 seconds)
17:26:15 Join merbanan [0] (~banan@c-62-220-165-110.cust.bredband2.com)
17:26:33merbananswitched to a wireless card that works
17:26:52kugelTorne: profile it!
17:27:59 Quit stripwax (Read error: Connection reset by peer)
17:28:06kugelsaratoga: ping
17:29:13topikyou're pitching yourself as a student for the RaaP GSoC project, kugel?
17:29:23topikRaaA even
17:30:50 Join stripwax [0] (~Miranda@87-194-34-169.bethere.co.uk)
17:30:57Tornekugel: aha, it'sb ecause it's too big to load the whole font
17:31:07Tornekugel: if the font's over 60000 bytes it has to use the glyph cache stuff
17:31:14Torneif it's less than that it just loads the whole font file in one go
17:31:17Tornewhich is way more efficient
17:31:34Torneit seems like a bad idea to have the font used by the default theme be one that provokes this behaviour ;)
17:31:46Tornesince for me it makes font loading over ten times slower
17:31:58Torne(might just be the 5.5G's crappy disk)
17:32:11linuxstbTorne: But it's also good for the default font to support lots of characters....
17:32:19Tornelinuxstb: yes, but it doesn't on every other device
17:32:29Tornedo ls -l *Helvetica* on the fonts
17:32:38Tornethe 15pt one is *way* bigger than all the others
17:32:46Torneand onlt 320x240 devices use 15pt
17:32:56Torneother devices use smaller (or bigger) versions, which presumably have less characters
17:33:02Torneso if that's the motivation, we're not applying it very well ;)
17:33:25linuxstbThe "fix" to that problem isn't the one you want though...
17:33:30Torneheh, indeed
17:33:55Tornehas someone got a non-ipodvideo device they can test on, with the bootchart?
17:34:06Torneand see how long it takes to load a font that fits vs one that doesn't?
17:34:25Torne14-Adobe-Helvetica vs 15-Adobe-Helvetica
17:35:29Luca_Splayback stuck on fuzev2 (not a panic, just stopped playing while still responsitive, pausing and resuming does not help): would it be useful to tell you the values in the debug > buffer screen?
17:35:40 Join jennifur [0] (~bunnytaur@cpe-72-224-19-1.nycap.res.rr.com)
17:36:47gevaertsTorne: I'll test in a minute
17:36:56Tornegevaerts: thanks
17:37:16kugeltogetic: yes
17:37:34 Join Adubb [0] (~aldubuc@67.201.160.144)
17:37:56kugeltopik: ^
17:38:25kugelI mean, I apply, but I don't make the decision whom to pick
17:38:37topikyou bring quite the resume
17:39:04topikRaaA is an exciting direction
17:39:20Tornegevaerts: i'm not sure how the font cache works entirely but it *looks* like it will result in font loading being dozens/hundreds of read() calls
17:39:30Tornegevaerts: wheras a font that fits without caching is *one* read() call
17:39:43Torneso that's a horribly bad case in general, but *especially* on the 5.5G with the sector emulation
17:39:44gevaertsTorne: I seem to vaguely remember some discussion about that a while ago
17:39:54 Quit merbanan (Remote host closed the connection)
17:39:56Tornewhich makes small reads way worse than on other devices
17:40:12kugelI thought the glphcache is a simple dump of the font buffer
17:40:18 Join stripwax_ [0] (~Miranda@87-194-34-169.bethere.co.uk)
17:40:30Tornekugel: i mean the mechanism to load/render fonts larger than the buffer
17:40:34Tornenot just loading the cache specifically
17:40:37 Quit stripwax_ (Client Quit)
17:41:12*linuxstb wonders how many people make use of the 345 Ethiopic glyphs in 15-Adobe-Helvetica
17:41:13Torneand no, the cache only saves the indexs of the glyphs, it looks like
17:41:22TorneCrap, yes. That's what it does.
17:41:25Tornesee glyph_cache_load
17:41:28 Join merbanan [0] (~banan@c-62-220-165-110.cust.bredband2.com)
17:41:33Torneit reads 2 bytes at a time from .glyphcache
17:41:40Tornethen goes and reads taht *one* glyph from the font
17:41:47Tornehundreds/thousands of tiny reads
17:41:57gevaertsTorne: 25 ticks to load 15-Adobe-Helvetica on gigabeat F
17:43:49kugelTorne: I mean I thought the glyphcache would be read with a single()
17:43:54Tornekugel: nope.
17:44:05Tornekugel: and even if it was, that wouldn't help, because it reads *each glyph* from the font seperately
17:44:15Tornethe cache is, sadly, not a copy of the font buffer
17:44:20Torneit is just a list of what characters are in there.
17:44:34Tornegevaerts: hm, that's not very slow ;(
17:45:35Tornegevaerts: but what is it for 14?
17:46:51kugelTorne: maybe it should be a changed so that it's a simple dump
17:47:13Tornekugel: maybe. it wouldbreak if someone changed the font file, then
17:47:16kugelbut I assume you might get out of sync with the font if it changes between boots
17:47:19gevaertsTorne: 7 ticks
17:47:39Tornegevaerts: right. and i think it will get *slower* the more you use the cache, also
17:47:48Tornegevaerts: since if your cache just mentions the ascii chars it won't load anything else..
17:47:53gevaertsYes, probably.
17:48:00Tornebut if your cache also mentions, say, a bunch of CJK glyphs it will take *longer*
17:48:05Torneup to the limit of the size of the cache
17:48:19gevaertsSo just changing the font would clear it, and then it would slowly getting slower
17:48:29Torneso yeah. it's 4 times slower for you
17:48:49Tornebut it's more like 20 times slower for me, probably because of the sector emulation
17:49:17Tornesmall random reads are bad in our fat implementation anyway but they are *much worse* when emulating a different sector size ;)
17:51:33Tornefor me it's 22 ticks to load 14, ~600 ticks to load 15
17:51:40Torneso yeah, over 25 times slower :)
17:52:04Tornesucks to be me
17:52:13jennifurI've put rockbox on my FuzeV2, and I keep getting a Panic error about 1 to 2 seconds in when I play a file from the uSD card.
17:52:14Tornefortunately i don't actually use cabbie normally
17:52:20Torneso this doesn't affect my actual use of the player
17:52:31Torneit just affects the boot time every time i try and test something with the default settings ;)
17:53:25Tornei'm gonna raise a FS# for this so other people can see more easily
17:54:15linuxstbTorne: The metadata parsers may also have that problem (lots of small reads). So the slow database times on the 5.5g may not only be because they normally hold lots of files.
17:54:28Tornelinuxstb: yeah
17:54:52Tornei'm primarily concerned about this font thing because it's affecting the boot time with default settings, though
17:54:59Tornewhich affects *all* users who haven't picked another theme yet :0
17:55:12Tornei guess it's not so bad on the other devices if it only takes 250ms to load
17:55:18Tornebut even 250ms is not quick
17:55:23Torneand 6 seconds is just crazy
17:57:49stripwaxthis is only on the 80GB 5.5g, right? 30GB 5.5g unaffected?
17:58:10stripwaxlikewise 5g
17:59:41Tornestripwax: i assume it's only the 1024-byte-sector disks, yes
17:59:57Tornebut even on a disk without that problem it's still4-5 times slower
18:00
18:00:12Torne250ms is still not very nice when it could be 70ms
18:00:21AlexP_bluebrother, pixelma: I have a patch that removes all the \* and the \\ etc and changes the latex settings to have new paragraph have a blank line rather than an ident (I *much* prefer this, I think it is much easier to read.) This also means that there is e.g. a blank line after a note without having to force it. I need to sync it and clean it up, then I'll stick it on flyspray
18:01:47Tornehttp://www.rockbox.org/tracker/task/11168 <- bug raised
18:02:21 Part JanDo (" www.qip.im ")
18:02:25bluebrotherAlexP_: nice.
18:03:18gevaertsTorne: there's another option. Increase the font buffer on ipod video :)
18:03:23Tornegevaerts: This is true
18:03:32Tornethe font buffer is 60000 bytes on devices with >2mb ram
18:03:45Tornewe could go higher on devices with >16, or similar
18:04:45 Join funman [0] (~fun@rockbox/developer/funman)
18:05:03Tornethe font in question is 227KB though
18:05:10Torneso we'd ahve to make it 256KB or whatever
18:05:24Torneis losing 200KB of ram to make fat fonts load faster worth it? :)
18:06:12gevaertsIf this were an ideal world, all ipods with the affected disk would have 64mb ram
18:06:15Torne35 of the 73 fonts we provide currently fit in 60000 bytes
18:06:40Torneif we increased it to 256kb then 61 of them would fit
18:06:53Tornegevaerts: i'm still not sure if they do or not
18:07:30funmankugel: if your application is read by google, s/OF/OF (Original Firmware)/g
18:07:57 Quit stoffel (Remote host closed the connection)
18:09:22gevaertsThis will probably be even worse if you use multifont
18:09:25TorneYup
18:09:29Torneor is it?
18:09:35Tornedoes it load the extra fonts to the font buffer?
18:09:39Torneor does it put them in the skin buffer?
18:09:54gevaertsIt puts them in the skin buffer, with I think 10K per font
18:10:07Torne10K?
18:10:17Tornethat's.. way too small :)
18:10:17gevaertsI'm not sure
18:11:25Torneanwyay, i dunno what the best thing to do is
18:11:29Tornebut i think we should do *something* :)
18:12:14Tornegevaerts: anyway, now that i've solved that mystery i'm gonna test your patch
18:12:17gevaertsThe best thing to do is of course to make the glyph cache code less seek-happy
18:12:20Tornegevaerts: and see if i can get my boot time down any lower :)
18:12:26Tornegevaerts: i'm wondering if the cache is even in order
18:12:29Torneif not that would be *punishingly* bad
18:12:48gevaertsIt probably isn't
18:12:54Tornethat would explain it ;)
18:13:52Torneyeah, it saves them by doing lru_traverse
18:14:21Torneso i would guess we're probably doing several hundred seeks at minimum
18:14:51TorneFAT will cache different 512-byte sectors, even though the cache is useless
18:15:02Torneand then the ATA code will cache different 1024-byte sectors too, even though that cache is also useless
18:15:07funmankugel: your application looks good (i noted a few typos in #rockbox-community)
18:15:08Torneso, huge amounts of seeking and copying all round.
18:17:37 Quit mitk (Quit: Leaving)
18:19:24funmanLuca_S: i have a 8gb fuzev2 , crashes are expected (I experienced them the most when loading several plugins in a row on Clip+), I believe it is due to SD driver but test_disk will help to confirm once writing works
18:20:29jennifurfunman: I've been experiencing some crashes too on my fuzev2, but only when I try to play a track from the uSD card.
18:20:40 Join tomers [0] (~chatzilla@bzq-84-109-85-100.red.bezeqint.net)
18:20:56 Quit dockimble (Quit: WeeChat 0.2.6.3)
18:21:15Luca_Syeah. internal storage seems more stable (it's slower IIRC?)
18:21:32tomersIn viewer.c:viewer_draw(), what exactly is the size of the arrays scratch_buffer and utf8_buffer?!
18:21:51kugelfunman: md5 can "verify" as well, can't it?
18:22:04tomersunsigned char scratch_buffer[max_columns + 1]; - and max_columns is a global variable... Any ideas?
18:22:30bluebrothertomers: that's C99
18:23:31tomersbluebrother: does rockbox is written to c99?
18:23:35 Join bertrik [0] (~bertrik@rockbox/developer/bertrik)
18:24:19bluebrotherthere are a number of places that use C99 features. I don't think we *try* to use all features of C99 though (don't see a reason to do so anyway).
18:26:19tomersthank so the explanation. does all our code compile with C99 features (but we may choose not to use them), or does those specific places in the code are configured by makefiles to compile with C99 features enabled?
18:26:56stripwaxwhen scrolling through art in pictureflow, on ipodvideo, what is the limiting factor (it's pretty slow) - disk reads (from the image cache) or lcd updates?
18:27:12stripwaxor even the rendering
18:27:47bluebrothertomers: gcc defaults to using C99, so if you don't want it you explicitly need to turn it off. I don't think its done anywhere in Rockbox' Makefiles
18:28:34bluebrotherI don't see much point in not using C99 features unless you want the code to compile with other compilers that don't support C99.
18:28:48tomersbluebrother: can i use this C99 feature, 'variable-length arrays', say, in firmware/drivers/lcd-bitmap-common.c firmware/font.c ?
18:29:17bluebrothersure, it's a language feature. It's a different question if it makes sense using it.
18:29:17funmanjennifur: yes it seems microSD works less fine since we switched to dynamic CPU frequency
18:29:20tomersI mean, will it be accepted if I use it? It would really help me with FS #11095
18:29:45funmankugel: md5 could work yes, we just need to patch md5 plugin to not write its result on storage but rather on screen
18:30:15*funman slaps dionoea for that
18:30:18Tornetomers: yah, sure. we generally use whatever gcc will accept
18:30:26bluebrothertomers: I'm not aware of any policy to not use language features in firmware code (or any other place).
18:30:32Tornetomers: too much of it is highly useful to ignore ;)
18:30:49tomersTorne, bluebrother: great, thanks a lot
18:30:54bluebrotherC99 has quite a bunch of nice and useful features.
18:31:06*tomers agrees
18:31:12Tornewe use gnu extensions also, so yeah, knock yourself out :)
18:31:20bluebrotherunfortunately the numbers of C compilers supporting C99 is pretty small. Even MSVC doesn't support C99.
18:31:24 Quit Luca_S (Quit: CGI:IRC)
18:31:38bluebrother(which is the major reason why you can't build Rockbox Utility using VS)
18:31:46tomersbluebrother: as long as we can use it, I don't care :-)
18:32:05funmanrockbox is quite dependant on gcc anyway
18:32:17bluebrotherwell, we do want a free toolchain so MinGW is the way to go anyway
18:33:03bluebrothertomers: we're also pretty much tied to gcc because of inline assembly. There's no language standard for that, so other compilers do it different. Same goes for the assembly code itself.
18:34:21funmanI had tried to use RVCT armcc and the first showstopper was dependencies generation
18:35:35tomersbluebrother: thanks for explaining :-)
18:42:01 Join captainkewllllll [0] (~2669ecc2@gateway/web/freenode/x-lbxmmpcnkobmrhca)
18:43:22 Quit jennifur (Quit: jennifur)
18:44:16 Quit sevard (Ping timeout: 252 seconds)
18:45:38CIA-5New commit by 03tomers (r25437): Manual: Document new 'alignment' option in text viewer, and remove now obsolete limitation
18:47:13 Quit anewuser ()
18:49:38Tornehttp://www.rockbox.org/tracker/task/10756 has this been committed already? (freeing init code)
18:50:01funmanyes
18:50:15 Nick fxb is now known as fxb__ (~felixbrun@h1252615.stratoserver.net)
18:50:51funmanr25013
18:59:42amiconnTorne: Read access through the sector cache shouldn't cause noticeable slowdown. Writing single 512 byte sectors is what suffers a lot
19:00
19:01:24***Saving seen data "./dancer.seen"
19:01:51 Join sevard [0] (sev@216.164.6.24)
19:03:24amiconnIn fact it should be slightly faster
19:03:48amiconnHmm, well, unless it's jumping back & forth between files, in which case it will be slower (but not much)
19:04:39gevaertsI've made an attempt at cleaning FS #11167 up. It still might need to be generalised to be usable by non-dircache
19:09:37amiconntomers, bluebrother: We do use C99 features in various places. It's just that we don't want *some* of those features, by convetion
19:10:23amiconnTwo examples are C++ style line comments ( // ), and mixed declaration and code
19:10:56funman// comments and mixed declaration and code area already present in several places?
19:11:12 Quit sevard (Read error: Operation timed out)
19:11:50amiconnIf they are, it's a violation of our coding style
19:12:43gevaertsfunman: we also don't want bugs, but we still have some
19:12:53kugelmixed declaration and code is against our coding style? I didn't know that
19:12:59funmanamiconn: I always mix declaration and use of variables
19:13:08funmanand // comments although less than the above
19:13:21bluebrotheramiconn: our coding style says to not use mixed declarations and code?
19:13:49funmanalso in my understanding coding style is left to the desire of the developer creating the file, as long as individual files respect the same style
19:14:22funmanit's true that CONTRIBUTING specifically mentions /* */ over //
19:14:27*linuxstb remembers the same as amiconn, but it's not in docs/CONTRIBUTING....
19:14:56amiconnfunman: Brace placement style is left to the desire of the developer
19:15:24amiconnIf CONTRIBUTING does not mention not to mix declarations and code, that should be added
19:15:32 Join sevard [0] (sev@216.164.6.24)
19:15:45funmanno
19:16:06funman% grep -Ie "[^:]"// *|wc -l
19:16:06funman30219
19:16:33funmanIMO mixing declaration and code is a good idea: declare variables when they are used, not 80+ lines over when they are used
19:16:36bluebrotherif we don't want to mix declarations and code it should really be added. I'm doing it all the time (though in C++ code). But why should we not want to use that feature? The only reason I can think of is C89 compatibility
19:16:37kugelthe mix rule is new to me
19:16:49linuxstbfunman: If it's 80 lines away, your functions are too big...
19:17:14*bluebrother wants a monitor that can show 80 LOC
19:17:28funmantrue, but really I learned that gcc -pedantic (-std=c89?) produced warning for declarations made inside code
19:17:37funman+ when I joined rockbox community
19:17:45*AlexP_ spots jhMikeS on the mailing list
19:17:50 Nick AlexP_ is now known as AlexP (~ap@rockbox/staff/AlexP)
19:17:58amiconnNot mixing declarations and code helps improving code quality, imo
19:18:11kugelAlexP: ?
19:18:24AlexPkugel: What?
19:18:42kugel"* AlexP_ spots jhMikeS on the mailing list"
19:18:50AlexPI don't know how else to explain that
19:18:53amiconnI'm *not* saying that all declarations have to be at the beginning of each function, just at the beginning of a block
19:19:13kugelare you talking about the guy that want to apply for gsoc?
19:19:14linuxstbDoes mixing declarations and code count as a "C++ism"? (I don't know C++)
19:19:16amiconn(wherever it is allowed in C89)
19:19:32funmandeclaring loop counters just before the for() is better for me
19:20:32AlexPkugel: No, jhMikeS - Mike Sevakis, the Rockbox developer
19:20:35*bluebrother wouldn't consider mixing declarations and code as C++ism, though imagines that some people would
19:20:53funmanlinuxstb: at least it's ok with -pedantic -std=c99
19:20:58*bluebrother points to IrcNicks
19:21:21bluebrotherfunman: it's a C99 feature, so it should be ok with that options ;-)
19:22:29funmani don't know to which point gcc is laxist, but mixing code and declarations only produce a *warning* with *-pedantic* for C89
19:23:17bluebrotherfunman: you sure?
19:23:59funman101%
19:24:38funmangcc version 4.4.3 (Ubuntu 4.4.3-4ubuntu5)
19:24:51bluebrotherfunman: not for me
19:25:17bluebrother4.4.3 on F12: test.c:9: warning: ISO C90 forbids mixed declarations and code
19:25:33bluebrotherthat's with gcc -o test -std=c89 -pedantic test.c
19:26:05bluebrotherwithout -pedantic it compiles
19:26:28funmanhttp://pastie.org/900568
19:27:04funman(cc points to gcc)
19:27:26*linuxstb recalls Bagder and LinusN being strongly against mixing code and declararations, and commits being criticised for that. But things must have become slacker...
19:27:56bluebrotherwell, that's the same I'm seeing here (and which is correct): -pedantic -std=c89 errors out.
19:28:12bluebrotherthough I'm not sure if I consider it good behaviour to not error out without -pedantic
19:28:17funmanonly a warning
19:29:07bluebrotherah, overlooked that. True. IMO it should error out in c89 mode
19:29:24*bluebrother assumes that you can make it do so with the correct command line option
19:30:00funman-Werror
19:31:59funmanrockbox doesn't build with -std=gnu99 -pedantic
19:35:31CIA-5New commit by 03funman (r25438): Only build alarm_menu.c if needed
19:37:06ThomasAHfunman: it seems I can't charge my clip+ while running rockbox, too. I had it connected to a Motorola mobile phone charger for more than 5h today and now it is at 77% ... I think it was about the same level when I started
19:37:31funmanThomasAH: it charged from 0% up to 63% for me, then I had to use the OF
19:37:41ThomasAHfunman: but the charging icon was displayed, even when starting that high
19:37:48funmanyeah
19:38:31ThomasAHfunman: getting below 63% isn't that easy if I connect it regularly with OF to copy new rockbox builds, voice file, talk clips, ... :)
19:38:50*funman bets ThomasAH doesn't listen to his Clip+ 15 hours in a row
19:39:13ThomasAHfunman: right ... only about 1-2h each day
19:40:01ThomasAHfunman: though it might be more these days, I bought a short connector to connect it to the kitchen radio :)
19:40:06 Quit kugel (Remote host closed the connection)
19:40:17funmanamiconn, linuxstb: if you think mixing declarations and code in rockbox (gnu)C(99) code should be forbidden, please bring it on the mailing list
19:42:49ThomasAH(oh, and I copied two jpegs of smurfs to it yesterday (harmony and a generic smurf) :)
19:43:31funmanno smurfette? you crazy Oo
19:44:22ThomasAHfunman: thought about it, but she wasn't in the mood to be copied to a smurf smurf :)
19:44:50ThomasAHok, of for dinner ...
19:47:35 Quit linuxstb (Ping timeout: 276 seconds)
19:50:50 Join linuxstb [0] (~linuxstb@rockbox/developer/linuxstb)
19:52:38 Quit antil33t (Read error: Connection reset by peer)
19:52:44 Join antil33t [0] (~Mudkips@203-184-54-232.callplus.net.nz)
19:53:03 Quit funman (Quit: free(random());)
19:58:37 Quit linuxstb (Ping timeout: 240 seconds)
19:59:23gevaertsThe 170ms it takes to load the built-in sbs with cabbiev2 on my gigabeat f seems to be entirely spent on loading the backdrop. I guess that means the sbs change itself does indeed have no effect on boot time
20:00
20:03:11 Quit sevard (Ping timeout: 276 seconds)
20:03:16 Join linuxstb [0] (~linuxstb@rockbox/developer/linuxstb)
20:07:02 Quit linuxstb (Read error: Operation timed out)
20:09:38 Join sevard [0] (sev@216.164.6.24)
20:12:12 Quit tomers (Ping timeout: 265 seconds)
20:12:30 Quit TillW (Quit: This now concludes our broatcast day.)
20:22:02 Quit xiainx (Ping timeout: 276 seconds)
20:22:52 Join linuxstb [0] (~linuxstb@rockbox/developer/linuxstb)
20:25:30saratogalinuxstb: no i didn't realize they were identical
20:27:01linuxstbObviously most VOBs contain AC-3 audio, and that's stored differently to mpeg audio.
20:28:39linuxstbVOBs are just a restricted form (e.g. certain maximum buffer sizes, fixed-size (2048 byte) packets) of mpeg-2 program streams. They also have extra things (e.g. subtitle streams, navigation info) as private streams within the program stream. Plus even more extra stuff on DVDs as IFO files.
20:33:45 Quit saratoga (Quit: Page closed)
20:36:24Lloreanbluebrother: Another key point (to me) is that if characters aren't showing, a user is likely to ask "how do I fix this" or check the manual. If boot is slow, they're more likely to just think "oh, that's how things are, I guess."
20:37:09bluebrotherLlorean: right.
20:37:45bluebrotherin any way, making boot slower is definitely the worst option.
20:37:49LloreanAgreed
20:38:22bluebrothergiven that my mini needs around 6 seconds already. I did even flash my h100 to make booting faster (which is really great)
20:38:31LloreanBetween gevaerts's dircache fix and getting the right combination of font size and buffer, the iPod 5G should have a pretty snappy boot again.
20:38:40Lloreanbluebrother: Is your mini using dircache?
20:38:52 Quit flydutch (Quit: /* empty */)
20:39:09bluebrotheryes. Wanted to try gevaerts dircache fix today, didn't get around till now.
20:39:26bluebrotherthough I wonder how much of a difference it will make −− my mini is cf modded.
20:44:51 Join pamaury [0] (~c2c7a50a@rockbox/developer/pamaury)
20:45:32 Join Zagor [0] (~bjst@rockbox/developer/Zagor)
20:46:14 Join killan_ [0] (~nnscript@c-38fd70d5.06-397-67626721.cust.bredbandsbolaget.se)
20:46:28 Quit killan (Read error: Connection reset by peer)
20:50:05stripwaxthe dircache patch sounds great - presumably would prevent disk thrashing too, and hence speed up buffering also?
20:50:10bluebrotherLlorean: ok, on my mini the difference is not noticable, at least not when timing it with an ordinary stopwatch. I guess it's related to the CF mod.
20:50:25gevaertsstripwax: yes, that's the point
20:50:33stripwaxbut what is "the known bug that Slasheri told me would fix" ?
20:51:01gevaertsIt might be known, but not by me :)
20:51:04stripwaxgevaerts - wasn't sure if the point was to prevent disk thrashing, or just speed up the boot to main menu. everyone's a winner though.
20:51:55pamaurygevaerts, stripwax : I don't have time right now but I'll explain it to you later ;)
20:52:32LloreanCan't you summarize?
20:52:34gevaertsstripwax: basically the boot is slow because of thrashing. It's possible to fix that by putting the dircache init later in the sequence, but then the thrashing potentially interferes with buffering. The patch basically inserts a sleep in the dircache thread at the place it yields anyway in case another thread has accessed the disk recently
20:53:58 Join jennifur [0] (~bunnytaur@cpe-72-224-19-1.nycap.res.rr.com)
20:54:39gevaertsIt might be useful to do this for database updates as well, but then we have to think hard about making sure that dircache and tagcache don't wait for each other
20:56:20 Join xiainx [0] (~xiainx@modemcable195.238-202-24.mc.videotron.ca)
20:57:14stripwaxcan we record the thread id of the thread that most recently accessed the disk - and then tagcache waits unless the thread id is dircache? (or vice versa)
20:59:03gevaertsthat's one way to do it
21:00
21:00:06gevaertsThe way I thought of yesterday was to just make the time intervals for "recent" different enough, which would give one of the two priority as well
21:01:28***Saving seen data "./dancer.seen"
21:04:07 Quit sevard (Ping timeout: 240 seconds)
21:05:31niekieHrm. Seems RockBox boots *much* faster than the OF here :)
21:11:17pamaurygevaerts, stripwax, Llorean : bug description follows
21:11:21 Join sevard [0] (sev@216.164.6.24)
21:12:37pamauryThe known bug is quite technical bug one consequence is for example that if you open a file while dircache is on, then dircache got deactivated and then is renable, the file is not more considered opened by dircache, so there is an inconsistency
21:13:41pamauryAnd any operation that modifies the metadata of a file during that dircache rebuild might lead to inconsistent dircache state (where the dircache data is not the real one)
21:14:33gevaertsThe deactivate/activate thing won't be impacted by boot changes I think
21:14:53gevaertsThe other one, hm. What about the playlist control file?
21:15:14pamauryplaylist control is the mainly affected file
21:15:55pamaurygevaerts: the bug will be more visible because the rebuild will take longer so during that time, more files can be opened/modified
21:16:14gevaertsIt will take a bit longer. Not really much though
21:16:16stripwaxwhen dircache is enabled, don't you need to reboot for it to take effect? i'm not sure I understand how dircache can be enabled while a file is already opened.
21:16:48pamaurywhen dircache rebuilds, it's disabled, it passes though to dir_uncached
21:17:07pamaury(basically)
21:17:09stripwaxand dircache can't see/query file open status?
21:17:31pamauryit's not that simple
21:20:06pamauryThis bug is not super hard to fix but (a) Slasheri said he'll do it but I think it hasn't got time (b) it's a bit tricky, need to go into details, think hard ;)
21:21:41pamauryAnd furthermore, whathever the implementation you choose, you can end up in a situation where you have no choice but to redo the rebuild because you can't keep track of all the changes that happened during that time in a constant space
21:22:09gevaertsmaybe, but that's better than an inconsistent state
21:22:23LloreanWould it be possible to make the filesystem read-only during the background dircache scan?
21:22:36CIA-5New commit by 03bluebrother (r25439): Implement system proxy values retrieval on OS X.
21:22:41LloreanI mean, there's a lot less likelihood someone will want to write to their filesystem in the first three seconds after boot.
21:23:02gevaertsLlorean: the problem isn't really the "someone" I think, it's the "something", i.e. playback
21:23:10gevaertsFor me it's 12 seconds
21:23:15pamauryLlorean: that also happens during tagcache rebuild
21:23:17Lloreangevaerts: Well, deny it writing until dircache is complete?
21:23:40pamauryread-only is clearly not a solution to me
21:23:57Lloreanpamaury: Why?
21:24:16gevaertstoo much impact elsewhere
21:24:28LloreanRockbox can do almost everything on our read only targets as it is.
21:24:37LloreanDuring the tagcache build, what's it preventing that will be a problem?
21:24:46gevaertsYou suddenly have to make *everything* that writes to disk able to handle intermittent write failures
21:25:14gevaertsYou don't know, these are background things. Anything could be running
21:25:26 Join stripwax_ [0] (~Miranda@87-194-34-169.bethere.co.uk)
21:25:28pamauryand that's a mess, for example if it happend during playback, playlist control file will be read-only, I think it will trigger an error message
21:26:07gevaertsYou could handle one or two instances, but all of them including plugins? What about e.g. recording?
21:27:11*gevaerts is assuming that dircache can have the same issues durin partial rebuilds
21:27:25pamauryMy point is that here is a solution to this problem, it's just that there has to be some code for this problem and that it's not very fun to write it, but I should write it
21:32:26*pamaury also notes that to make this run with only a small memory overhead, it requires a bit of trickery, otherwise, it will eat several KiB of memory
21:33:40gevaertspamaury: you recently made it eat around 30K less, you have some margin :)
21:35:35pamaurygevaerts: true but that would be stupid. I mean, if you keep track of every modified file during the rebuild, you have to save the name => 250 bytes/file. It can be done if much less (I think it can be done in 4 bytes/file :)) but there are some corner cases that have be stupied. The bigger problem is deletion I believe
21:38:16LloreanCan some features just be disabled during dircache background scan? If someone tries to delete, splash "Dircache is committing. Please wait for disk use to end and try again."
21:38:58LloreanI don't see that the dircache scan needs to be 100% invisible to the user, just have a "minimal" impact.
21:39:43 Quit stripwax (Read error: Connection reset by peer)
21:39:55pamauryWell, for me a cache has to be transparent, but that's my opinion.
21:40:12gevaertsLlorean: again, it might not be this easy. File operations are not only triggered directly by users through the GUI
21:40:28gevaertsYou'd have to track down all cases and fix them
21:40:39 Join carl_sagan [0] (~steve@adsl-68-88-123-124.dsl.lgvwtx.swbell.net)
21:40:51pamauryit's easier to modify *just* dircache than all the rest :)
21:41:50carl_saganok question. i installed rockbox on an ipod and it installed fine, but when i shut down the ipod and restarted it, it said the bootloader could not be found and to boot with disk mode
21:42:12carl_saganso i did that, and tried to reformat the ipod. however, under disk mode it says the ipod is write only
21:42:46carl_saganbut i discovered that it would boot with the old ipod firmware if the hold switch was on when it started
21:42:55carl_saganso i reformatted with no problem from the old firmware
21:43:02pamauryAnd by the way, dircache code is already using magic tricks, so one more will not make the code horrible ^^
21:43:27carl_saganso obviously the old firmware is intact, but every time i start the ipod it still says rockbox bootloader could not be found and says to boot with disk mode
21:43:47bluebrothercarl_sagan: Rockbox doesn't tell you that the bootloader is missing. That's the Rockbox bootloader telling you the main build is missing.
21:43:58carl_saganok...
21:44:04bluebrotherwhich means you installed only parts of it. How did you install?
21:44:05carl_saganbut if i reformatted it how is the bootloader still there?
21:44:14LloreanIt's in the same place as the original firmware.
21:44:16bluebrotherI.e. did you use "complete install" or the "Installation" tab?
21:44:36carl_sagani used the complete install
21:44:45carl_saganit was fully functional with rockbox before i turned it off
21:45:03bluebrotherthe Ipod uses a firmware partition. Rockbox modifies that to boot the Rockbox bootloader first. Then it can boot either Rockbox (which is placed on the main data partition) or the Apple firmware, which is also located in that firmware partition
21:45:17gevaertsHow did you reformat?
21:45:35bluebrotheralso, what host are you using? Mac?
21:45:36 Quit bmbl (Quit: Bye!)
21:45:46carl_sagani turned the ipod on with the hold switch on so that the apple firmware would load
21:45:48gevaertsAnd which exact ipod is it?
21:45:51carl_saganthen i opened the ipod with itunes
21:45:55carl_sagannano version 2
21:46:08carl_saganreset it with itunes, and then used windows reformat to reformat it
21:46:29*bluebrother has this strange "declaration doesn't declare anything" error again :(
21:47:08bluebrotheryou don't want to format an Ipod using the window format
21:47:38carl_sagani reformatted it several times. i also used some that were referenced for use with ipods
21:47:46carl_saganlike i think its the HP one? or something
21:48:14carl_saganbut every time, it reformats fine, but when i restart it in disk mode, the computer detects it as unformated
21:48:25LloreanIt sounds like you have a bad disk.
21:48:25carl_saganand if i try to format it with disk mode on, it says it is write only
21:48:36carl_saganor various forms of "the format failed"
21:48:36LloreanWhich would also explain why rockbox.ipod vanished / became corrupted.
21:48:53carl_saganthats...possible i suppose
21:49:05carl_saganbut why was the ipod functional with apple firmware before i installed rockbox?
21:49:17bluebrotherwell, if you want to reformat you should use the restore functionality of itunes, not windows format. Windows is keen on formatting everything as NTFS, which doesn't work on the ipod
21:49:33bluebrotherdid you reformat it before installing Rockbox?
21:49:44carl_saganwith itunes, yes
21:50:11gevaertsrestoring it using itunes doesn't remove the rockbox bootloader? That sounds strange
21:50:29bluebrotherit does.
21:50:38bluebrotherat least it did for me (on a Mac though)
21:51:09gevaertscarl_sagan: if you tell itunes to restore it, and do *nothing* else, what happens?
21:51:37carl_saganit does just like it would with a normal ipod "the reformat is complete, restarting ipod in 5 etc."
21:51:58carl_saganand the the ipod restarts and comes up with the bootloader could not load rockbox.ipod
21:52:40gevaertsDoes itunes make a difference between "reformat" and "restore"?
21:53:00carl_saganit says "restore" but that is the only option listed
21:53:36*gevaerts has neither an ipod nano nor itunes, so he can only work from hearsay, but that definitely does not sound normal
21:54:14bluebrotheritunes only knows about restore
21:54:40carl_sagan?
21:55:03bluebrothersounds like the flash is bad. How old is the ipod?
21:55:37carl_saganit is definately old
21:55:42gevaertsIsn't there a way to reset the FTL, or does the restore do that?
21:55:57carl_saganftl?
21:56:47gevaertscarl_sagan: I think your best bet is to come back later. Our main nano2g specialist (TheSeven) isn't here right now
21:57:00carl_saganalright i'll check back
21:57:04carl_saganthanks for the help guys
21:57:35 Quit carl_sagan (Quit: Leaving)
21:59:30*bluebrother feels like the compiler error is a naming clash with OS X APIs again :(
22:00
22:02:07 Quit sevard (Ping timeout: 240 seconds)
22:04:37 Join Blue_Dude [0] (~chatzilla@adsl-235-206-131.mco.bellsouth.net)
22:05:10Blue_DudeUpdating the manual to reflect new keymaps. I probably missed some things.
22:06:56S_a_i_n_trestore (iTunes restore) *should* reset the FTL, unless something is going drastically wrong somewhere...
22:07:18 Join jeffp [0] (~jeffp@barmen.interhost.no)
22:10:02 Part jeffp
22:10:17S_a_i_n_tthough, definitely *definitely* come back & talk to TheSeven...
22:11:13CIA-5New commit by 03Blue_Dude (r25440): Manual update for keymaps, hotkeys
22:12:26 Quit stripwax_ (Quit: http://miranda-im.org)
22:16:14 Quit Zagor (Quit: Leaving)
22:17:25 Join stripwax [0] (~Miranda@87-194-34-169.bethere.co.uk)
22:21:35 Quit pamaury (Quit: Page closed)
22:22:36 Quit robin0800 (Remote host closed the connection)
22:23:13 Quit pixelma (Disconnected by services)
22:23:13 Join pixelma_ [0] (quassel@rockbox/staff/pixelma)
22:23:16 Quit amiconn (Disconnected by services)
22:23:18 Join amiconn_ [0] (quassel@rockbox/developer/amiconn)
22:23:33 Nick pixelma_ is now known as pixelma (quassel@rockbox/staff/pixelma)
22:23:40 Nick amiconn_ is now known as amiconn (quassel@rockbox/developer/amiconn)
22:25:08 Join sevard [0] (sev@216.164.6.24)
22:29:50 Join tvelocity [0] (~tony@weg38-1-82-237-37-150.fbx.proxad.net)
22:34:13liarS_a_i_n_t: itunes does not reset the ftl, but we have tools to do that
22:38:03 Quit dfkt (Quit: -= SysReset 2.53=- Ph'nglui mglw'nafh Cthulhu R'lyeh wgah'nagl fhtagn.)
22:38:39 Join geertvdijk [0] (~chatzilla@cc412026-a.zwoll1.ov.home.nl)
22:39:11pixelmaAlexP: does that mean I would break your patch if I committed the removal in goban.tex already? Besides... I have to find out first why my patch is breaking the (or at least some manuals)...
22:39:48AlexPpixelma: No, go ahead - the patch is well out of date and needs significant fiddling anyway :)
22:40:12 Join dfkt [0] (dfkt@unaffiliated/dfkt)
22:41:04 Quit geertvdijk (Read error: Connection reset by peer)
22:43:08 Join tomers [0] (~chatzilla@bzq-84-109-85-100.red.bezeqint.net)
22:44:45 Quit liar (Ping timeout: 258 seconds)
22:45:21pixelmaAlexP: alright, I hope I'll find the error soonish, tex is not being helpful :\
22:46:00 Join geertvdijk [0] (~chatzilla@cc412026-a.zwoll1.ov.home.nl)
23:00
23:00:39 Join GeekShadow [0] (~Antoine@reactos/tester/GeekShadow)
23:01:29***Saving seen data "./dancer.seen"
23:04:09 Join S_a_i_n_t_ [0] (S_a_i_n_t@203.184.2.219)
23:04:45 Join JohannesSM64 [0] (~johannes@cm-84.215.116.196.getinternet.no)
23:05:02 Quit S_a_i_n_t (Ping timeout: 260 seconds)
23:10:56 Quit Blue_Dude (Quit: ChatZilla 0.9.86 [Firefox 3.6.2/20100316074819])
23:15:30 Join Smith [0] (~4b519c19@gateway/web/flash/gateway.tiramisu.in/x-ydzwsqwqvsovgabn)
23:16:35 Part Smith
23:21:16 Quit jgarvey (Quit: Leaving)
23:22:54 Quit captainkewllllll (Quit: Page closed)
23:24:22CIA-5New commit by 03bluebrother (r25441): Move utils.cpp functions into separate class and split it up. ...
23:24:38*pixelma reads months old dev list emails and notices that the 23rd of March went by silently
23:26:06pixelmadid the release cycle "reset" after the last release which was postponed, so that the 3 months time to the next release count from the actual release date?
23:26:53gevaertsI think it would be a bad idea to do a two-month cycle suddenly.
23:27:14 Quit xiainx (Ping timeout: 264 seconds)
23:27:37gevaertsI personally also think that a four month cycle is slightly better
23:27:45pixelmanot a "cycle", just getting back on track. But I always thought releasing every 3 months was a but much
23:28:17pixelmabit too
23:28:18bluebrotherIMO we shouldn't get into a release cycle with a release around christmas
23:28:26gevaertsthat too
23:29:02LloreanI'm not sure the time between releases matters too much.
23:29:09LloreanOne way or another it always seems to take people by surprise.
23:29:27LloreanAnd of course there's the usual "I want to squeeze this in before the freeze" mentality that, I think, doesn't benefit our users much.
23:29:47*Llorean thinks a "oh, the freeze is coming, I may as well just let my big change happen after the release so there's more time to test" benefits users more.
23:30:15LloreanThe freeze should mostly be about clearing up known bugs, not trying to track them down in last-minute added big new code.
23:30:51pixelmaI haven't seen this now... no-one seemed to even remember that 23rd of March would have been the usual date
23:31:26AlexPI'm still convinced that the last release was a couple of weeks ago :)
23:31:31AlexPIt seems like it
23:31:34LloreanI think we may want to reconsider what the release is. Not just the cycle, but come up with a solid reason why we're doing it and *how* we're going to do it, because I think different people see it as far too different of a thing.
23:32:36LloreanFor some it's a time "to get things done by" when it's remembered. Others (or at least, me) thing things should generally start slowing down before the release even if there's not an explicit freeze, to try to make sure stability is ramping up or at least not getting worse as the freeze approaches.
23:32:51 Join xiainx [0] (~xiainx@modemcable195.238-202-24.mc.videotron.ca)
23:32:53CIA-5New commit by 03bluebrother (r25442): System Info: display OS X version number and architecture too.
23:33:12*gevaerts wants a four month cycle, but he also thinks that we should release soon because there seem to be no large changes right now...
23:34:13pixelmaLlorean: I agree with the "don't commit right before release (or freeze)" and I am aware that there was this big discussion last time which I just skimmed again, but didn't get the feeling that this was a "usual" attitude
23:34:15*Llorean wouldn't mind a four month cycle, and is more just concerned about general release awareness and care to try to avoid delays.
23:34:27Lloreanpixelma: It's not a "usual" attitude over everyone.
23:34:34LloreanIt's a specific attitude among some people though.
23:34:36linuxstbLast release was 2nd Feb, so 4 months would be two days before devcon.
23:35:44gevaertsthat could be good or bad
23:35:52linuxstbYes, I'm not sure which...
23:36:15gevaertsDevcon as "let's write new code, it doesn't have to work", or "Let's fix bugs!"
23:36:22bluebrotherit might make sense to pick a dedicated release manager for each release.
23:36:37 Quit tomers (Ping timeout: 265 seconds)
23:37:41linuxstbgevaerts: If we've branched, we can do both...
23:37:56linuxstbi.e. release on the final day of devcon.
23:38:12pixelmaLlorean: I'm just against generalisation and would want to try avoiding bad blood beforehand
23:39:53Lloreanpixelma: I'm just suggesting that we decide whether we want to say "please try to avoid significant changes to functionality for (some time period) before the freeze to help ensure the manual is up to date and stability is at its best" or just say "the freeze is the only period where commits matter, even one day before it is fair game"
23:40:08LloreanIf you have a policy in place in advance, it helps avoid arguing about it later.
23:41:40gevaertsWell, basically, "to help ensure the manual is up to date and stability is at its best" is what the freeze itself is for, but yes, *too* intrusive changes should stop a bit sooner
23:41:42pixelmaok, thanks for the clarification. That's probably worth talking about...
23:42:03Lloreangevaerts: Well that's why I said "significant" rather than "any"
23:42:10LloreanI guess definition for "significant" would also be important
23:42:13 Quit xavieran (Ping timeout: 276 seconds)
23:42:26gevaertsOne basic problem we have is that there just aren't enough testers for some of the features
23:43:04LloreanI think it's fine to make the user base also our tester base. It's just ideal to do so right *after* the branch so that there's a recent stable version for them to use too.
23:43:22LloreanThe longer something that needs heavy testing has before it's in a release, the better.
23:50:18 Quit ender` (Quit: There is no reason anyone would want a computer in their home.-- Ken Olson, president, chairman and founder of DEC)
23:51:37 Quit Status (*.net *.split)
23:51:37 Quit preglow_ (*.net *.split)
23:51:37 Quit blithe_ (*.net *.split)
23:51:37 Quit Farthen (*.net *.split)
23:51:37 Quit stavrob_ (*.net *.split)
23:51:48 Join Status [0] (Status@gentoo.lonis.org)
23:52:52 Join preglow_ [0] (thomj@tvilling2.pvv.ntnu.no)
23:52:52 Join blithe_ [0] (~blithe@72.14.176.144)
23:52:52 Join Farthen [0] (~Farthen@static.225.178.40.188.clients.your-server.de)
23:52:52 Join stavrob_ [0] (~sam@78-105-125-218.zone3.bethere.co.uk)
23:54:37 Join xavieran [0] (~xavieran@ppp118-209-62-101.lns20.mel4.internode.on.net)
23:55:23 Part jennifur
23:55:25 Join temp1234 [0] (~8c718113@gateway/web/freenode/x-wksmgufhfnftbpry)
23:57:50 Quit merbanan (Read error: Operation timed out)
23:58:17gevaertsLlorean: the big problem is that I don't really like rockbox not working with some systems if there is a known way to make it work...

Previous day | Next day