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

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

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

#rockbox log for 2010-05-09

00:01:39 Quit bgs000 (Remote host closed the connection)
00:03:19 Quit BHSPitMonkey (Quit: Ex-Chat)
00:03:20 Quit bmbl (Ping timeout: 252 seconds)
00:03:25 Join M3DLG [0] (
00:08:11 Join S_a_i_n_t [0] (S_a_i_n_t@
00:17:05 Quit linuxstb (Ping timeout: 252 seconds)
00:25:01 Join geertvdijk [0] (
00:30:56saratogawhy doesn't the clipv2 define PLUGIN_USE_IRAM?
00:32:47 Join linuxstb [0] (~linuxstb@rockbox/developer/linuxstb)
00:32:56 Quit hebz0rl (Ping timeout: 260 seconds)
00:39:02 Quit petur (Remote host closed the connection)
00:44:20 Quit geertvdijk (Ping timeout: 276 seconds)
00:46:02 Join yawny [0] (
00:47:52 Quit S_a_i_n_t (Ping timeout: 260 seconds)
00:49:15 Join kisak_ [0] (
00:53:12 Quit halmi_ (Quit: halmi_)
00:53:54 Quit adnyxo (*.net *.split)
00:53:54 Quit elcan (*.net *.split)
00:53:54 Quit kisak (*.net *.split)
00:59:34 Join GeekShad0w [0] (
00:59:42 Quit mirak (*.net *.split)
00:59:42 Quit mikroflops_ (*.net *.split)
00:59:42 Quit solexx (*.net *.split)
00:59:42 Quit AlexP (*.net *.split)
00:59:42 Quit arun (*.net *.split)
00:59:42 Quit jordan` (*.net *.split)
00:59:42 Quit mapi (*.net *.split)
00:59:55 Quit ender` (Quit: Be wary of strong drink. It can make you shoot at tax collectors and miss. -- Robert A. Heinlein)
01:00:32 Join mirak [0] (
01:00:32 Join mikroflops_ [0] (
01:00:32 Join solexx [0] (
01:00:32 Join AlexP [0] (~ap@rockbox/staff/AlexP)
01:00:32 Join arun [0] (~arun@unaffiliated/sindian)
01:00:32 Join jordan` [0] (
01:00:32 Join mapi [0] (
01:00:37 Quit mirak (*.net *.split)
01:00:37 Quit mikroflops_ (*.net *.split)
01:00:37 Quit solexx (*.net *.split)
01:00:37 Quit AlexP (*.net *.split)
01:00:37 Quit arun (*.net *.split)
01:00:38 Quit jordan` (*.net *.split)
01:00:38 Quit mapi (*.net *.split)
01:00:41 Join mikroflops [0] (
01:00:57 Join adnyxo [0] (
01:00:57 Quit GeekShadow (Read error: Connection reset by peer)
01:01:10 Join mirak [0] (
01:01:10 Join mikroflops_ [0] (
01:01:10 Join solexx [0] (
01:01:10 Join AlexP [0] (~ap@rockbox/staff/AlexP)
01:01:10 Join arun [0] (~arun@unaffiliated/sindian)
01:01:10 Join jordan` [0] (
01:01:10 Join mapi [0] (
01:01:36 Part domonoky1
01:02:20 Nick fxb is now known as fxb__ (
01:03:13 Join anewuser [0] (anewuser@unaffiliated/anewuser)
01:05:29 Quit mikroflops_ (Ping timeout: 276 seconds)
01:05:58 Quit linuxstb (Ping timeout: 246 seconds)
01:06:51 Join m1ss1onto [0] (
01:07:01 Quit m1ss1onto (Client Quit)
01:12:31 Quit kyraxyg (Quit: CGI:IRC)
01:35:47 Quit pamaury (Quit: Page closed)
01:43:19 Quit Schmogel (Ping timeout: 245 seconds)
01:45:24 Quit anewuser (Quit: for SELL 2 by the price of 1 now!)
01:47:25***Saving seen data "./dancer.seen"
01:48:28 Join hebz0rl [0] (
02:03:51 Quit GeekShad0w (Quit: The cake is a lie !)
02:13:50 Quit DerPapst (Quit: Leaving.)
02:23:49 Join bertrik [0] (
02:23:50 Quit bertrik (Changing host)
02:23:50 Join bertrik [0] (~bertrik@rockbox/developer/bertrik)
02:26:04 Join Rob2223 [0] (
02:27:40 Quit hebz0rl (Quit: Ex-Chat)
02:29:44 Quit Rob2222 (Ping timeout: 268 seconds)
02:49:21 Join Llorean [0] (~DarkkOne@rockbox/user/Llorean)
02:53:29Unhelpfulinteresting. i am not aware of anything the jpeg viewer *should* decode that the core loader does not :|
02:54:01soapIF, if (I don't know Macs) he had a "hidden" file extension...?
02:54:12soap(something easy to do in Windows)
02:54:43Unhelpfulbut that should be trouble for both, shouldn't it?
02:54:47LloreanI dunno
02:54:54Lloreancover.jpg.jpg would open fine in jpeg viewer
02:54:57*soap looks at Unhelpful to answer Unhelpful's question
02:55:05soapbut cover.jpg.png?
02:55:21LloreanDon't we have a .png viewer?
02:55:28soappoor example then
02:55:31LloreanSo that'd open in the png viewer but not work as album art
02:55:51Unhelpfulno, great example! cover.jpg.png would open if you view it, but not as album art
02:55:52Lloreancover.jpeg.somethingelse wouldn't do anything in Rockbox I believe
02:56:12LloreanBoth cover.jpg.jpg and cover.jpg.png could present the behavior he's seeing, I think
02:56:21soapthat's what I was uncertain of, Llorean, IF the viewer would open it but the AA search would fail.
02:56:23Unhelpfulhow would cover.jpg.jpg?
02:56:33Unhelpfulohhhh... because AA search fails.
02:56:58Unhelpfulbut wouldn't the RB file browser show the double extension?
02:57:05soapso back to the other implied question. Can Mac OS hide "true" extensions?
02:57:14*Unhelpful thinks trying to load it in the sim is the best idea.
02:57:16LloreanUnhelpful: I would assume so. Can Rockbox be set not to show extensions at all though?
02:57:35saratogahave him do an ls from the terminal
02:58:08LloreanUnhelpful: In "supported" you don't see extensions.
02:58:17UnhelpfulLlorean: if so that's a potential problem - windows "hide common extensions" option has been an exploit target in the past because people don't think it odd that one file happens to show an extension, so they open file.jpg.exe
02:58:24LloreanSo he would see "cover.jpg.jpg" as "cover.jpg" I assume
02:58:51*Unhelpful would say that *something* breaking AA search is the most likely cause here
02:58:53LloreanBut his OSX browser window looks like it's showing extensions just fine.
02:59:45soapLlorean, rockbox has multiple "ways" not to show extensions at all.
02:59:48LloreanI'd also ask him how he synced his music
03:00:00LloreanSince the folder is 'iTunes Music" and he refers to removing the AA in iTunes
03:00:08LloreanIs it possible he used iTunes to sync it?
03:00:15soapI don't see how he synced his music to be an issue.
03:00:26soapIF that directory structure is of his device as I asked.
03:00:31LloreanIt's not
03:00:45Unhelpfula less likely cause imo would be a loader bug... i don't think i've seen one in many many months
03:00:46soapthen we know nothing. ;)
03:00:56LloreanAt least, on the left bar I don't see anything that indicates a device is attached
03:01:06LloreanUnless "iDisk" is his iPod
03:01:35soaplooking at it again I think that is a "shared" device - NAS?
03:01:37LloreanIt looks like a network share or something, since it says "shared folder" but I don't know anything about OSX really
03:03:21soapSimplest answer is that he doesn't have on his iPod what he thinks he has on his iPod.
03:04:04soapand since Llorean rightfully pointed out that the screenshot does not appear to be of the ipod itself we need to verify that fist. Thanks for chewing the fat on the other possibilities, though, Unhelpful.
03:04:33Unhelpful"the user is always wrong" ;
03:04:53Llorean"the user is likely to not realize you need *exactly* what you asked for, not 'approximately'"
03:05:03Unhelpfulor that :)
03:05:06LloreanThough it's good to know other ways AA could be broken but not look like it (hidden extensions is a good one)
03:05:28pixelmalooks like the Itunes content on his Mac which I've seen stored in nice "usual" subfolders on Macs at work
03:06:37 Quit efyx (Remote host closed the connection)
03:14:02soapahh, my post of desperation. "What version of Rockbox are you running?"
03:15:51 Quit Llorean (Quit: Leaving.)
03:15:53JdGordon1has anyone tried the fm skinning patch?
03:18:48 Quit bertrik (Quit: sleep)
03:19:18 Join Llorean [0] (
03:19:26 Quit Llorean (Changing host)
03:19:26 Join Llorean [0] (~DarkkOne@rockbox/user/Llorean)
03:26:13 Join Blue_Dude [0] (~chatzilla@rockbox/developer/Blue-Dude)
03:27:17Blue_DudeAny last words, wishes, comments, complaints, gripes or other assorted feedback on FS #11250.
03:29:29JdGordon1Blue_Dude: hey, something I completly forgot about untill after I DC'ed on friday... the origoinal icon patch should have sat longer and been made more public only because it broke all themes with icons (I know its irrevlvant now but still...)
03:30:19Blue_DudeIt did? The icons were referenced by enum. Confused why this would make themes not work.
03:31:01JdGordon1it could also explain the large bump... the icon system loads a bmp strip and decides how big each is by height/enum_size
03:31:17 Join linuxstb [0] (~linuxstb@rockbox/developer/linuxstb)
03:32:13Blue_DudeHrm. That's odd.
03:32:29JdGordon1so adding an icon wouldnt actually break anything, it would make each icon slightly shorter looking crap
03:33:02Blue_DudeI saw that behavior. It was fixed by reinstalling the new tango bmps.
03:33:18JdGordon1yes, but that would only fix that icon set
03:33:41JdGordon1as for 11250, I think it should also be settablee from a butotn press
03:33:41Blue_DudeWell yeah. A clean build would make the problem go away though.
03:33:52JdGordon1only for themes using that icons set
03:34:35Blue_DudeStill confused. I changed all the icon sets to add the new icon. A fresh install would have fixed it.
03:34:57Blue_DudeUnless there are custom icon sets out there. Can't help those.
03:35:28*Blue_Dude frets about that.
03:35:36Blue_DudeDon't matter now though.
03:36:57Blue_DudeJdGordon1: I'm kinda looking forward to dumping all the menu code. You get the same functionality but the bin size is close to 1000 bytes less.
03:37:55Blue_DudeAnyone reproduce the wav resume bug? It crashes the player.
03:38:17Blue_DudeI'm going to have to report this...
03:38:36JdGordon1It seems to me that one of the biggest benefits to the hotkey stuff is being able to really quyickly change it... having to find the menu sucks
03:39:28LloreanJdGordon1: What are the values for the list you regularly change it between?
03:39:57JdGordon1I've personally never used it...
03:40:12LloreanSo, why do you think it's something users would change frequently then?
03:40:34Blue_DudeBug report is at FS #11255.
03:41:09LloreanBlue_Dude: Which players have you reproduced that on (and I doubt it's a 3.4 bug?)
03:41:53Blue_DudeHey, JdGordon, put the hotkey setting on the quickscreen. }:-)
03:42:14Blue_DudeLlorean: e200. It's very recent, last couple of days.
03:42:20Blue_DudeCodec changes.
03:42:31LloreanBlue_Dude: you marked it as a 3.4 bug in the tracker, so you probably want to change that.
03:42:46JdGordon1not all the playlist options are available tot he hotkeys right?
03:42:56Blue_DudeI've been off the net so I couldn't do a reversion check to pin down the exact revision.
03:43:10LloreanBlue_Dude: Does the size of the .wav file or anything matter? Can I just grab one a few seconds long from somewhere to reproduce with?
03:43:35LloreanOr channels, sample rate, etc.
03:45:00Blue_DudeI don't know. All I know is that it dumps in the middle of trying to set up. You can see the debug info in the sim, and it doesn't complete its load.
03:45:45Blue_DudeWe had resume problems the last time the codec was updated too.
03:45:48LloreanWell the more information you can include in the bug report (for example, that it's reproduceable in the sim) the better.
03:47:17Blue_DudeI'll report the bug and leave it to the experts to nail down. If they can't reproduce, then I'll get more involved.
03:47:27***Saving seen data "./dancer.seen"
03:47:51Blue_DudeI think I left simple enough instructions to reproduce.
03:49:00Blue_DudeJdGordon: Sorry, missed your comment. No, I'm not changing functionality yet. So far only Playlist Insert is coded.
03:49:34LloreanBlue_Dude: It's better to include as much useful information up front, so that when they're around to test it doesn't matter whether you're around or not.
03:49:44 Nick JdGordon1 is now known as JdGordon (
03:49:46LloreanIf they come to test, and need to ask something, they may not be able to do any more until you get back
03:50:35Blue_DudeThat's pretty simple to change though, if you're interested. It would even be in order, since Playlist Insert is listed last in the enum.
03:51:17LloreanBlue_Dude: For example, I can't reproduce on my Gigabeats with a 16-bit 44.1khz WAV file
03:51:28Blue_DudeLlorean: nope, I'm done troubleshooting. If there are questions later, I'll answer them. I'll be very surprised if they can't reproduce it though.
03:51:42LloreanWell, be surprised.
03:51:44LloreanI can't reproduce.
03:51:47Blue_DudeOK, build a e200 sim and check it out.
03:51:49LloreanCurrent SVN as of 3 minutes ago.
03:52:00Blue_DudeUse a nice long one.
03:52:09 Join chjurk [0] (~chjurk@2001:6f8:1c00:1fa::2)
03:52:11LloreanWhy didn't you provide that info in the report then?
03:52:19Blue_DudeI did say e200 sim.
03:52:25LloreanYou marked the report SWCODEC
03:52:30 Part chjurk
03:52:43Blue_DudeYep. Codec issue. SWCODEC. Follows.
03:52:53LloreanNo, you marked the *player type* as sw-codec
03:53:00LloreanThat means you think it occurs on all software codec players
03:53:01Blue_DudeYes, so I did.
03:53:21Blue_DudeI think it's a reasonable probability. I didn't try them all out.
03:53:26LloreanWell, it doesn't.
03:53:40Blue_DudeOK, take ownership and fix it.
03:54:13LloreanBlue_Dude: Why don't you provide more details for how to reproduce it?
03:54:20LloreanNobody can fix it if they can't reproduce it.
03:54:27LloreanYou may have a problem wav file?
03:54:55Blue_DudeThe wav file is a year old. I've been using it for testing for the last nine months.
03:55:10LloreanYes, and maybe something about that type of .wav is broken, rather than all ones
03:55:18LloreanSo knowing it sample rate, channels, bit rate, and other details could help
03:55:19Blue_DudeIt crashes under those circumstances with a clean build.
03:55:32LloreanNot all .wav files are identical. There are many differences between them.
03:55:41Blue_Dude16 bit 44.1 kHz, stereo. Plain vanilla.
03:55:50Blue_DudeI mean, really.
03:56:19LloreanWell, I can't read minds so I can't know thats what you're doing, and you were being oddly reluctant about refusing to state the properties of the file until now.
03:56:35LloreanMaybe attach a short .wav file sample to the task of one that definitely crashes with current SVN to make testing easier?
03:57:28Blue_DudeThis is a total waste of my time. I've had an extremely long day. I'm going to commit my patch and go to bed. Take or leave the bug report. At this point I just don't care.
03:57:38LloreanBlue_Dude: Does it happen on players, or just the sim?
03:57:56LloreanBlue_Dude: Seriously, it's a waste of your time to *investigate a bug only you can currently reproduce*?
03:59:00Blue_DudeYes, it is. It is not my bug. I just was the first to notice it.
03:59:01 Join n1s [0] (~n1s@rockbox/developer/n1s)
03:59:40LloreanBlue_Dude: And so you refuse to answer even basic questions about the conditions?
03:59:52 Quit Blue_Dude (Quit: ChatZilla 0.9.86 [Firefox 3.6.3/20100401080539])
03:59:53saratogausually codec problems impact all players, unless theres endianness or ASM involved
04:00:13Lloreansaratoga: It doesn't happen on my Gigabeat S with current SVN
04:00:22LloreanWhich is why I was trying to find out more. If it's SIM only that could be telling
04:00:25JdGordonit wont usually cross to hwcodec though
04:00:25saratogai doubt the wav codec has any ASM, but it could be endian specific
04:00:32saratogaalthough x86 and arm are the same
04:00:59Lloreansaratoga: Of course, it would also help (as I said) to know the details of the file being used. =/
04:02:07 Join Barahir_ [0] (
04:02:49 Quit xavieran (Ping timeout: 246 seconds)
04:03:22CIA-5New commit by Blue_Dude (r25905): FS #11250: Hotkey setting method changed to menu item vs button pres in context menu. Manuals updated to match.
04:04:56 Quit saratoga (Ping timeout: 252 seconds)
04:05:15 Quit Barahir (Ping timeout: 240 seconds)
04:06:03 Quit amiconn (Disconnected by services)
04:06:04 Join amiconn_ [0] (quassel@rockbox/developer/amiconn)
04:06:05 Quit pixelma (Disconnected by services)
04:06:05 Join pixelma_ [0] (quassel@rockbox/staff/pixelma)
04:06:25 Nick pixelma_ is now known as pixelma (quassel@rockbox/staff/pixelma)
04:06:26 Nick amiconn_ is now known as amiconn (quassel@rockbox/developer/amiconn)
04:07:43 Join saratoga [0] (~9803c6dd@gateway/web/freenode/x-hpopepfvtzdfbnnj)
04:07:45CIA-5New commit by Blue_Dude (r25906): Fix red
04:12:27 Quit linuxstb (Ping timeout: 248 seconds)
04:14:16 Join linuxstb [0] (~linuxstb@rockbox/developer/linuxstb)
04:16:05 Join ncfi1013 [0] (
04:17:19ncfi1013does rockbox work on the ipod nano video 4th gen?
04:18:44 Quit adnyxo (Ping timeout: 240 seconds)
04:20:58JdGordonOH... THAT ipod video nano 4g....
04:21:00JdGordonstill no
04:21:01 Join kramer3d [0] (~kramer@unaffiliated/kramer3d)
04:21:07ncfi1013just in case there is some confusion i found an image of what im talking about
04:21:13Lloreanncfi1013: The list of supported players is on the front page of the website.
04:22:11ncfi1013is there any support for ipod in question in lucid? still running koala.
04:22:21saratogancfi1013: go ask the ubuntu people
04:24:17 Join xavieran [0] (
04:31:17 Join Zarggg_ [0] (~zarggg@2001:0:4137:9e76:0:fbf3:beb1:ba3d)
04:34:04 Join S_a_i_n_t [0] (S_a_i_n_t@
04:34:50 Quit Zarggg (Ping timeout: 248 seconds)
04:42:11 Join Curulan [0] (
04:46:08 Quit Zarggg_ (Ping timeout: 268 seconds)
04:57:02 Join dys` [0] (
04:57:14 Quit xavieran (Ping timeout: 248 seconds)
04:57:44 Quit linuxstb (Ping timeout: 240 seconds)
04:58:12 Join linuxstb [0] (~linuxstb@rockbox/developer/linuxstb)
04:59:26 Quit dys (Ping timeout: 276 seconds)
05:00:49 Quit Barahir_ (Read error: Operation timed out)
05:19:13 Join xavieran [0] (
05:19:35 Quit linuxstb (Ping timeout: 276 seconds)
05:20:05 Join linuxstb [0] (~linuxstb@rockbox/developer/linuxstb)
05:47:30***Saving seen data "./dancer.seen"
05:51:56 Quit fyre^OS (Read error: Connection reset by peer)
06:00:43 Quit linuxstb (Ping timeout: 246 seconds)
06:08:16 Join FlynDice [0] (~FlynDice@
06:15:35 Quit FlynDice (Quit: Gotta go fly!)
06:15:52 Join FlynDice [0] (~FlynDice@
06:32:38 Nick dys` is now known as dys (
06:35:05FlynDiceWould someone explain the significance of having one's copywrite in the rockbox header attached to the beginning of each file. I am not particularly interested in claiming any credit for work I've done but if there is a legal significance for the rockbox project and I should be doing this I would like to know that!
06:37:14S_a_i_n_tFlynDice: I've often wondered about that myself, I *think* it has just become popular convention to put ones stam on something...
06:38:41saratogaFlynDice: i think its just convention, we need to give copyright to someone, and since there is no Rockbox entity, we give it to whowever made the file
06:39:18 Quit kramer3d (Ping timeout: 265 seconds)
06:41:31FlynDicesaratoga: Thanks, So there's no legal issues to be concerned with if I don't bother with it is what I gather.
06:46:17*S_a_i_n_t adds a simple comment to the newest "hey, how can I remove DRM from my crippled Rhapsody files?" guy on the RB forums.
06:53:09*JdGordon thinks the backdrop size calculation is wrong...
06:53:17JdGordon#define LCD_BACKDROP_BYTES (LCD_FBHEIGHT*LCD_FBWIDTH*sizeof(fb_data))
06:53:44S_a_i_n_tand, for the lamens of the group?
06:56:46 Join mianos [0] (
06:59:15saratogaso if buschel's test_mem changes are correct, I get that IRAM on the Clipv2 is about 2x as fast as DRAM when unboosted, and 3x as fast when boosted
06:59:34saratogaclipv1 results don't look right to me, but they claim that DRAM is faster then IRAM
07:02:09 Join Barahir [0] (
07:05:49 Join S_a_i_n_t_ [0] (S_a_i_n_t@
07:07:36 Quit S_a_i_n_t (Ping timeout: 246 seconds)
07:13:36 Nick S_a_i_n_t_ is now known as S_a_i_n_t (S_a_i_n_t@
07:18:02 Quit S_a_i_n_t (Ping timeout: 248 seconds)
07:26:19 Join S_a_i_n_t [0] (S_a_i_n_t@
07:27:14 Quit Horscht (Ping timeout: 240 seconds)
07:28:22 Quit mirak (Quit: Ex-Chat)
07:29:24 Join Horscht [0] (~Horscht2@xbmc/user/horscht)
07:32:54 Quit FlynDice (Read error: Connection reset by peer)
07:39:29 Join funman [0] (~fun@rockbox/developer/funman)
07:47:32***Saving seen data "./dancer.seen"
07:47:32 Join esperegu [0] (~quassel@
07:49:03JdGordonamiconn: does that #define above look correct?
07:54:59amiconnYes it does
07:57:13JdGordontrying to figure out why the skin buffer is apparnelty not big enough on grey targets
07:58:00 Quit esperegu (Ping timeout: 246 seconds)
07:59:03funmanany trick to make the clip+ crash really fast? i'm tweaking the equ settings but it takes 2 minutes or so
08:00:21S_a_i_n_tI would have thought you'd be the best one for advice when it came to crashing sansas ;)
08:00:51funmani would like to
08:00:53ThomasAHfunman: enabling voice menus and voice clips and browsing through menus and songs
08:01:14funmani need to install voice files though ?
08:04:17funmannevermind I found it
08:06:47funmani have a langs/english.voice, language set to english, voice menus enabled (by default anyway) but nothign is spoken
08:07:03 Join robin0800 [0] (
08:07:39funmanhm i got the clipv1 file though, perhaps that matters
08:08:32 Join esperegu [0] (~quassel@
08:09:14 Quit Boldfilter (Quit: Boldfilter)
08:12:32funmanthanks it works!
08:13:28funmanit crashes 3 instructions after returning from set_cpu_frequency(), but loading the content of given address show it's correct
08:16:03funmanThomasAH: what is the biggest number of items you could ever hear before it crashed?
08:16:20S_a_i_n_theh, this is new...instructions on RB-IRC for "how do I *crash* my clip" as opposed to "how can I make it more stable" ;)
08:16:46funmanS_a_i_n_t: it really is "how can i reproduce crashes and test possible bugfix"
08:17:40S_a_i_n_tFrankly, I'm surprised it was so easy to do.
08:18:11S_a_i_n_tyou could have just gone back to few 1000 revisions :)
08:18:58S_a_i_n_tIt was a remark on how greatly the stability has increased of late.
08:19:13*funman doesn't agree
08:19:29S_a_i_n_tyou *don't* think they're more stable?
08:20:08 Quit robin0800 (Remote host closed the connection)
08:22:34funmanfile context menu isn't voiced correctly it seems
08:23:34 Join FlynDice [0] (~FlynDice@
08:25:04JdGordonamiconn: so "LCD_HEIGHT*LCD_WIDTH*LCD_DEPTH/8" is wrong?
08:25:16JdGordonor that will give a less accurate size?
08:26:58amiconnIt is wrong on targets where depth < 8 and the width or height (depending on pixel packing) is not an integer multiple of the pixel block size
08:27:03funmanThomasAH: hm first time I tried it crashes on the 3rd item and now I just listened to at least 30 items and it still works (using the same code)
08:27:10amiconnAfaik we only have two such targets - the twi Minis
08:27:33JdGordonit will give a value too small or too big?
08:27:38amiconnToo small
08:27:55JdGordoncheers, tat could explain things
08:28:19amiconnYou're testing on mini?
08:29:08JdGordonbut the issue is that fm skins just wont load because there isnt enough room, when there should b
08:29:35S_a_i_n_twhat happened to resizable skin-buffer?
08:29:43*S_a_i_n_t looks for kugel.
08:30:44amiconnThe mini should load fm skins?
08:31:44CIA-5New commit by jdgordon (r25907): use a better value for the needed skin buffer size
08:31:59JdGordonthat might make it possible for it to load fonts though
08:32:25amiconnHmm, other targets shouldn't have problems with that backdrop buffer size definition...
08:33:32JdGordonright now its set to grab 2 full depth screens + 1 backdrop per skinnable screen (2 now)
08:33:35amiconnThe old definition was short by 55 bytes on mini (half a byte per line)
08:33:55JdGordonok, so not enough o make a difference
08:34:43JdGordonthe old way used too much on mono screens apparently?
08:36:02JdGordonit bumpbed the buffer from 22KB to 31KB on the mini
08:36:13CIA-5New commit by jdgordon (r25908): fix red
08:37:25amiconnUmm, mono doesn't support backdrops...
08:38:30amiconnSo probably LCD_BACKDROP_BYTES is set to 0 for them
08:39:13amiconnHowever, some mono themes may use a fullscreen pseudo-backdrop. I'm not sure though - my wps'es don't use any bitmaps
08:41:05JdGordonhaha yeah woops
08:43:06CIA-5New commit by jdgordon (r25909): mono targets dont support backdrops so dont use that #define
08:43:20S_a_i_n_tWPS'es without .bmps (or at least one animation for that matter) make S_a_i_n_t cry...
08:44:58amiconni only ever made own wps'es for the archoses
08:45:03 Quit CaptainKewl (Remote host closed the connection)
08:45:21amiconnI use the bitmap archos one on a few other targets, on the others I just use the builtin one
08:45:42amiconn(builtin default, *not* cabbie)
08:45:48S_a_i_n_tI guess I should start branching out into themeing for targets I don't own.
08:45:53*amiconn can't stand cabbie for more than a few minutes
08:47:05S_a_i_n_tnor can I, I made my own Cabbie (guess you'd call it CabbieV5 going by the other " V's " on the themesite which I find tolerable...but I still don't really like it.
08:47:15*S_a_i_n_t isn't a big fan of yellow.
08:47:30 Quit esperegu (Quit: No Ping reply in 180 seconds.)
08:47:35funmanyou have a clip though? ;)
08:47:52JdGordonMy only gripe with cabbie on the smaller screens is too much text together so its hard to see what is what... unless I use my multifont version
08:47:55 Join esperegu [0] (~quassel@
08:48:12S_a_i_n_tJdGordon: Same.
08:48:20amiconnMy biggest gripe with cabbie is that it's bright-text-on-black
08:48:26S_a_i_n_tMy cabbie manages to make it look "decent" IMO
08:48:56S_a_i_n_tfunman: for about a week, I'm *really* not a fan of that screen.
08:49:04S_a_i_n_tI got rid of it pretty quickly.
08:49:06amiconnWell, /me rates functionality >> design
08:49:24S_a_i_n_tsure, but it's possible to have both.
08:49:35S_a_i_n_tespecially on "large screen" targets.
08:49:53saratogathe combination of small and lots of text in cabbie on the clip makes it really hard to use while running in my experience, otherwise i like it
08:49:58S_a_i_n_tOne of the reasons I like themeing the Nano so much is trying to find balance between the two.
08:50:23amiconnI have more relevant information with builtin status bar + builtin wps than with cabbie. And bright-text-on-black makes my eyes bleed after a few minutes
08:52:15CIA-5New commit by jdgordon (r25910): revert those last few...
08:53:29amiconnHigh resolution battery status and high resolution volume (~15 steps each), to name just two
08:54:17S_a_i_n_tit's possible to put "progressbar logic" in battery/volume, the resolution is WAY higher than the builtin whatever.
08:54:19JdGordon15 steps is hardly high res
08:54:32JdGordon15 pixels is hardly high anything
08:54:38*S_a_i_n_t agrees with this.
08:55:28amiconnWell, iirc cabbie only has 4 battery levels
08:55:40JdGordonthat sounds wrong
08:55:41S_a_i_n_tprogressbar style battery/volume looks *so* sexy...and saves a LOT of skin buffer.
08:55:49JdGordonand it should be using the bar style now anyway :)
08:55:54amiconnAnd volume in the builtin status bar changes to numeric during adjustment
08:55:57S_a_i_n_tbest thing for the wps/sbs since...the last really good thing ;)
08:56:40amiconnI also have time (on targets with rtc - not so much to actually get the time, but rather to easily notice when the rtc is off) and disk activity
08:57:16S_a_i_n_tthe built in bar is just so....ugly, for lack of a better word.
08:57:40funmanThomasAH: how does work for you ?
08:57:42S_a_i_n_tI have all those things in my .sbs, but *prettier* ;)
08:57:56amiconnThe builtin bar is completely sufficient on most of my targets.
08:58:54S_a_i_n_tsufficient, sure. Ugly, sure. ;)
08:59:00amiconnMy largets screen targets are 220x176 (H300 and iPod Color) - and there it's a bit small
08:59:03S_a_i_n_ttotally subjective I guess.
08:59:39S_a_i_n_tI *could* make you an .sbs that looks exactly like the built-in bar but larger?
08:59:51S_a_i_n_tand uses better logic for drawing battery/vol etc.
09:00:00*amiconn forgot the beast, but that's not much wider anyway
09:00:01S_a_i_n_tit's an option...and I wouldn;t mind.
09:00:21amiconnIf it is larger... it wouldn't fit on the archos screen
09:00:28amiconnThat's in fact where it comes from
09:00:40S_a_i_n_tI couldn't do it immediately, but I'm sure more people than you would get a kick out of it.
09:00:43n1sthe default bar is *tiny* on the beast
09:00:56JdGordonand miniscule on the mr500 :)
09:01:16*S_a_i_n_t wonders how it looks on the mini2440 ;)
09:02:27amiconnn1s: Tiny yes, but still recognisable
09:02:34*amiconn doesn't use the beast much
09:03:21*Llorean thinks the default bar looks fine on the beast, but wouldn't want it much smaller than that
09:03:27ThomasAHfunman: I'll try it later today ... currently in familiy interrupt
09:03:33amiconnI have 4 targets (among 15) I use most. They're all non-colour, and quite old. I still prefer them for the sound quality, flexibility, and readability without backlight
09:03:49funmanThomasAH: can't you disable interrupts? .. oops that only work with computers .. ;)
09:03:59n1ssure but it's a bit hard to read :)
09:04:01amiconnIriver H180, Archos Recorder, Mini G2, Archos Player (in that order)
09:04:39ThomasAHfunman: works with family, too, but probably triggers a NMI by the police :)
09:05:27n1samiconn: of my 3 targets, the beast sounds the best by far IMO
09:05:38n1s(h300, c200, beast
09:05:49ThomasAHfunman: and away again ... (additionally I have a patch for the clip* manual in preparation)
09:06:04amiconnI'd expect the H300 to sound better than the beast overall... a bit noisier though
09:06:24amiconnThe c200 audio codec is crappy though
09:06:37LloreanThe noise on the H100 is pretty terrible in quiet conditions, if the H300 is the same, I wouldn't put it up against the Beast.
09:06:56LloreanThe c200/e200 shouldn't even really be talked about when discussing sound quality. :)
09:07:23amiconnWell, the UDA1380 has a bit of background hiss. Imo it still sounds better than those WM* codecs overall
09:07:40amiconnThat's subjective of course
09:08:22LloreanWhen I'm in conditions I can't hear the hiss, I love my H100.
09:08:45JdGordonget tinitus and then the codec really wont matter much :)
09:08:54LloreanBut the lack of it gives me the perception that the Gigabeast sounds "cleaner"
09:09:28n1si don't use the h300 at all basically, since it is about twice the size of the beast (and i think the beast sounds better)
09:10:09CIA-5New commit by uchida (r25911): wave/wave64 LIST chunk parser ...
09:11:11n1si like the size and battery time of the c200 though
09:11:43n1show are the clip + or v2 with battery time and sound quality?
09:12:02funmanhum i notice weird battery indicator when charging with the OF on clipv2
09:12:13Lloreann1s: Great sound quality (to me at least)
09:12:13funmann1s: battery time is very good (next to 20 hours)
09:12:23LloreanEr, regular clip
09:12:24funmanthere is some hiss though
09:12:25LloreanNot clip+
09:12:30*Llorean assumes the clip+ is at least comparable.
09:12:41funman(with rockbox)
09:13:01LloreanI didn't notice a hiss like the iRivers have on my Clip at least.
09:13:04funmanclipv1 battery is awful although it just got better (a bit below 8 hours for mp3)
09:13:05*Llorean didn't look for it either, though
09:14:00 Join ucchan [0] (
09:15:07JdGordonucchan: Hi
09:15:45ucchanHi JdGordon
09:16:10JdGordonI had a hard time understanding your reply to the embedded AA task
09:16:49JdGordonis there any reason why you dont try allocating enough buffer for the image, and if the buffer is full then just fail and try next time?
09:17:35saratogaucchan: are you uchida?
09:18:37n1sfunman: both the v2 and the + have card slots, right?
09:18:46JdGordononly the +
09:18:54saratogafuze and + do, not the clip
09:19:03ucchanimage data size is big (10 -32 kB), then there is possible which the buffer is filled.
09:19:28JdGordonyes, but regular AA has that problem also
09:19:29funmanclipv2 is to clipv1 what fuzev2 is to fuzev1, something like what e200v2 is to e200v1
09:19:31ucchanthen I gave up that dthe data does not store in it
09:20:13funmansame case, different shit
09:20:27saratogaucchan: heres the tta patch I made ages ago
09:20:39saratogaignore the change to vorbis, its just junk
09:20:41ucchanAnd, the codec buffer is not used too much effectively.
09:20:50saratogaunfortunately, its not anywhere near working
09:21:00saratogadecodes one frame and then crashes I think
09:22:44ucchansaratoga, Is the performance enough?
09:22:59saratogaucchan: i don't know, but the format is very simple, so I think it should be fast
09:23:29saratogaucchan: ffmpeg took the same decoder source and improved it further:;a=blob;f=libavcodec/tta.c;h=4bdfd73fbd706c28a335a274a7772870177318bd;hb=HEAD
09:23:46 Quit wincent_balin (Quit: Verlassend)
09:23:48saratogaif you want to port TTA, starting with their code instead of that patch might be a good idea :)
09:23:54ucchanI checked my tta codec is slow (base on original tta hardware player's). for iPod, about 104 %
09:24:11saratogayou already ported it?
09:24:12ucchansorry more less 95 -100 %
09:24:45ucchanMy local environment.
09:24:54saratogai'm sure it can be optimized
09:25:14 Join ender` [0] (
09:25:16saratogagcc is likely to make a big mess of the filter
09:25:32saratogarewriting it in assembly will probably help a lot on the ipod
09:26:09ucchanBut I optimized decode logic, then my codec's speed is 130 % - 140%
09:26:32saratogaare you ready to commit it?
09:29:20ucchanI check only iPod. then I will send the tracker
09:29:38saratogasounds good
09:30:05ucchanplease wait for a moment...
09:30:21saratogafor codecs we require that they compile on all targets, but they don't have to work correctly right away (sometimes its months or years before they work on all devices )
09:31:39ucchanThere might be some problems because I am compiling only on iPod.
09:34:22n1sif it works on an arm target it should work fine on all arm and probably cf too
09:34:43n1sunless there are endianess issues of course
09:39:45ucchantta original decoder, because the endian is considered, it might be unquestionable(may be...).
09:39:56CIA-5New commit by funman (r25912): Update current usage for Fuzev1/e200v2/Clipv1
09:40:08 Join solexx_ [0] (
09:41:50funmanclipv2 crashed in 7 minutes playing mp3
09:42:27saratogai had a crash every 20 minutes so with mine today
09:42:46funmani hoped i had the right fix :(
09:43:33saratogasorry not today, friday
09:43:37 Quit solexx (Ping timeout: 276 seconds)
09:44:34funmanyesterday I listened to flac from µSD on Clip+ and it crashed with CPU boosted
09:45:38saratogait took us years before the last disk issues were fixed on the e200v1
09:47:34***Saving seen data "./dancer.seen"
09:48:41JdGordonwere the (random) clip playback issues ever fixed?
09:48:56funmanaccording to the poll i made in the forum no
09:51:01saratogaJdGordon: for plain mp3 with nothign special going on its basically stable, but turn on too many features or skip through playlists and it'll crash
09:51:21saratogaor at least it'll deadlock
09:51:35funmanrecording errors (on fuzev2 at least) happen because we don't read from the fifo fast enough
09:52:06funmanpcm playing + recording at the same time should be impossible afaik, i'm not sure why i didn't use DMA
09:52:56 Join bmbl [0] (~Miranda@unaffiliated/bmbl)
09:55:00 Quit pjm0616 (Read error: Connection reset by peer)
10:02:13funmanah i remember i couldn't get it to work at all
10:02:21 Quit aberet (Ping timeout: 260 seconds)
10:02:30funmanfred bauer helped a lot here
10:06:40 Join stoffel [0] (
10:12:55funmanclip+ crashed too .. sigh
10:13:11 Join komputes_ubuntu [0] (~komputes@ubuntu/member/komputes)
10:16:17 Join Luca_S [0] (
10:20:01 Quit Curulan (Quit: Curulan)
10:20:12 Quit CGL (Quit: Soy spammero ---- \m/ d(>.<)b \m/)
10:23:14 Join Zarggg [0] (
10:25:32 Quit Kitar|st (Ping timeout: 264 seconds)
10:28:58ucchandelay in sending, I send the tta codec the tracker (FS #11256) please confirm it
10:30:23 Join Kitar|st [0] (
10:31:59 Quit rvvs89 (Ping timeout: 246 seconds)
10:32:00 Quit esperegu (Remote host closed the connection)
10:32:37 Join rvvs89 [0] (
10:37:12 Join esperegu [0] (~quassel@
10:37:45 Join petur [0] (~petur@rockbox/developer/petur)
10:40:18 Quit bmbl (Ping timeout: 252 seconds)
10:41:59 Join bmbl [0] (~Miranda@unaffiliated/bmbl)
10:42:39 Join pjm0616 [0] (~user@
10:44:01 Join stoffel_ [0] (
10:44:29 Quit Luca_S (Quit: CGI:IRC)
10:44:33 Quit stoffel (Ping timeout: 246 seconds)
11:02:07 Quit dys (Ping timeout: 276 seconds)
11:05:23 Join bertrik [0] (~bertrik@rockbox/developer/bertrik)
11:08:37 Quit komputes_ubuntu (Ping timeout: 276 seconds)
11:09:23 Join dfkt [0] (~dfkt@unaffiliated/dfkt)
11:12:17 Join Buschel [0] (
11:27:40 Join pamaury [0] (~c2c7a50a@rockbox/developer/pamaury)
11:31:19 Quit petur (Quit: Leaving)
11:31:32Buschelsaratoga: on an iPod 5.5G 24/100 MHz -> DRAM: 12-16/60-70 MB/s IRAM: 50/200. when taking the asm into account the measured IRAM speed is exactly what is expected when IRAM access has zero wait state.
11:35:11 Join kugel [0] (~kugel@rockbox/developer/kugel)
11:38:07 Join komputes_ubuntu [0] (~komputes@ubuntu/member/komputes)
11:46:42 Join Schmogel [0] (
11:47:25funmanhm i have recording working with DMA but it sounds really bad
11:47:36***Saving seen data "./dancer.seen"
11:48:09funmanlooking in audacity I see a bar at X frequency, where X is really close to the frequency of the DMA callback
11:48:30 Quit freddy_ (Ping timeout: 252 seconds)
11:49:09bertrikMaybe some off-by-one error (just guessing).
11:49:27funmanoff-by-one-sample would hardly be hearable i think
11:49:28bertrikCan you post a wav somewhere?
11:49:56funmanyes, what source material could I use?
11:50:00 Quit Buschel (Ping timeout: 260 seconds)
11:50:47bertrikif there is one extra sample per block, it would give a frequency component at the inverse of the block size
11:51:12bertrikfunman, can you sing? :P
11:58:13funmanpatch: flac:
12:00:01funmani only tested fuzev2 though
12:00:31 Join freddy_ [0] (
12:00:44 Quit antil33t (Ping timeout: 260 seconds)
12:02:04funmanfrequency looks alright
12:02:38bertrikyeah, but it's heavily quantized somehow
12:05:56 Join antil33t [0] (
12:06:04 Join domonoky [0] (~Domonoky@rockbox/developer/domonoky)
12:06:50bertrikyes, it looks like there are only 8 levels for the audio signal
12:07:06bertrik(I'm looking at the flac in audacity)
12:08:24funmani2sin is set for 14 bits per sample, but samples should be transferred as 32 bits (16 bits + 16 bits)
12:08:47funmanthere is a 24 bits mode (32bits + 32bits) but i think rockbox expects 16 bits per sample PCM
12:10:00funmani talked about a problem with recording on ABI forum and noted rockbox recording had few levels, so perhaps this problem exists with current recording ?
12:19:05 Join Jaykay [0] (
12:20:09 Quit shai (Quit: Leaving)
12:20:39funmanbertrik: i put the patch on FS #11257
12:20:47 Join DerPapst [0] (
12:22:15funman <- discussion about recording problems in OF and rockbox
12:27:28 Join shai [0] (
12:35:11 Join loveless [0] (
12:35:42 Quit shaggy-h (Ping timeout: 240 seconds)
12:38:39 Join efyx [0] (
12:39:50lovelessanyone around here familiar with the usb code of the e200rpatcher? I am trying to use it for a different platform, but switching from PP502x to PP6100 seems to have broken the host usb code −− despite the fact that it should be completely identical. I can't seem to find out why this could be tha case.
12:43:06lovelessany ideas on how to proceed with debugging this one? I don't want to waste any more time with my brute-force grep -r approach
12:44:12JdGordone200rpatcher is just used to remove the upgrade blocker
12:45:35gevaertsJdGordon: it uploads code, which I assume is the point here
12:45:59lovelessthe problem is: the code builds fine but the host's usb_find_busses() or usb_find_deviced() (have seen any hanging) throw timeouts and i/o errors.
12:47:12lovelessgavaerts: yes, I only use it for loading test code. but I am at a point where it felt right to get the code tree to switch to the correct platform
12:48:52lovelessJdGordon: it worked fine for the pp6100 view as long as I was using the pp520x target.
12:50:06JdGordonah fair enough
12:50:34 Join Buschel [0] (
12:53:34 Join wodz [0] (
12:55:56 Join shaggy-h [0] (
12:57:50ucchanoh, come back jdGordon.
12:59:11JdGordonAlexP: ping?
12:59:16JdGordonpixelma: ping
13:01:15ucchanAs for embeded aa, Is the big size memory stored in the file buffer, and is it safe ? (buffering.c's source is difficult little..)
13:02:28JdGordonucchan: I'm probably not the best person to talk to, I just found it odd that you were relying on the codec buffer. I would think that it should act like regular AA which is just copied onto the buffer if there is enough room
13:02:30 Join linuxstb [0] (~linuxstb@rockbox/developer/linuxstb)
13:02:40JdGordontalking to Unhelpful would probably be best
13:03:10 Quit Rob2223 (Quit: Rob2223)
13:03:20 Join Rob2222 [0] (
13:07:06funmanbertrik: about battery life, i find it strange that the clipv1/fuzev1/e200v2 do not have the same power usage
13:07:29funmanthis is also true between different clips
13:07:45bertrikis there really a big difference?
13:07:47funmanperhaps we forgot to initialize some register and it is left in an undetermined, device-specific stage
13:07:59funmanmy clip doesn't go past 8 hours
13:08:51funmani was thinking of ascodec registers, did we check if we explicitely write to each of them ?
13:09:05bertrikweird, maybe sandisk changed components
13:09:05ucchanjdGordon: understand
13:09:34funmanbertrik: you think there should be a sane reset value ?
13:09:49bertrikalso I think we're not completely sure about the clips' battery capacity anyway, I tried to extrapolate it from other batteries based on the type number
13:10:43bertrikfunman, I think we haven't checked yet if we write them all
13:10:56funmanah you think there would be different capacities to explain the difference between 2 clips?
13:11:28bertrikfunman, yes that would be a possibiltiy (just a guess for now)
13:13:41 Join mikroflops_ [0] (
13:17:35 Quit mikroflops (Ping timeout: 240 seconds)
13:18:00 Quit esperegu (Ping timeout: 240 seconds)
13:20:35 Join esperegu [0] (~quassel@
13:23:29wodzfunman: You made some comment to my ML post about performance problems with MPIO. Could You be more specific about idea with second timer?
13:24:06funmanif the kernel tick is not fast enough you could experience general slowness
13:25:08funmanperhaps an easy measurement would be to modify the HZ variable and see how speed scale (measure with test_codec)
13:25:18funmanthe HZ constant*
13:25:32 Quit esperegu (Ping timeout: 246 seconds)
13:25:47wodzsorry I don't understand
13:26:17funmanthe CPU is interrupted HZ (= 100) times per second
13:26:35funmanuntil the tick has been executed, nothing else is executed
13:26:58funmanif the tick is executing for 0.01 second, then 100 times 0.01 = 1 second, so not other code can run
13:27:09funmanhopefully the tick is executing for much less than 0.01 second
13:27:39funmanbut if it is 0.005 second, then the speed of other code will feel very slow, just like if the CPU was twice as slow
13:28:40funmandoes it make sense?
13:28:47wodzin general yes
13:29:15funmantry changing HZ in kernel.h to 50 for example
13:29:43ucchanI will quit soon. (dinner, etc). tommorow, I wil login.
13:29:59funmangood bye ucchan
13:30:03ucchanPlease write the question to me in the tracker.
13:30:16 Quit ucchan (Quit: Leaving...)
13:30:54funmanwodz: if you see noticeable speed change, next step is find the part which takes a lot of time in the kernel tick
13:31:06wodzyes now I get it
13:33:08wodzkugel: ping
13:34:29kugelwodz: pong
13:35:27 Join anewuser [0] (anewuser@unaffiliated/anewuser)
13:35:58 Join esperegu [0] (~quassel@
13:36:16wodzkugel: I experince errors from ALSA when running sim
13:38:23 Quit funman (Quit: brb, travelling at ++c m.s^-2)
13:40:12mtwodz, kugel : I receive these errors in the sim :
13:40:14mtALSA lib conf.c:2742:(snd_config_hooks) id of field i is not and integer
13:40:15mtALSA lib conf.c:3079:(snd_config_update_r) hooks failed, removing configuration
13:40:27wodzme too
13:40:41kugelsince when?
13:40:50kugelI have no idea what these errors mean
13:42:22mtI'll try to narrow it down to a particular commit (but will be slow because I'm on a VM currently)
13:43:09wodzmt: we can split bisection to find out
13:43:51kugelteamwork! :)
13:46:54topiktoo bad i just missed funman. for the logs sir, please look at
13:46:55mtwodz: Great, thanks. I'll look how slow/fast it goes first though and tell you ..
13:47:27 Part mianos
13:47:33 Nick fxb__ is now known as fxb (
13:47:37***Saving seen data "./dancer.seen"
13:49:23mtwodz: Forgot that I only have svn on my VM :( .. I'll try svn-bisect
13:58:39 Join MethoS- [0] (~clemens@
14:03:37kugellinuxstb: ping
14:10:21pixelmaJdGordon: yes?
14:10:34JdGordondo you remember issues with the fm skin patch?
14:12:30JdGordonI'm resyncing with the intention of commiting soon
14:12:51pixelmanot since you fixed the issue with unknown presets but I haven't played around with it since. And well the difference to the current radio screen, especially the peakmeters/prerecording time on hwcodec
14:14:11JdGordonwell, I dont have a hwcodec+ radio target, so unless someone steps up....
14:15:04pixelmathe "peakmeter on its own line" thing is there in the WPS too
14:15:38mtwodz : I guess we should start bisect from r25700
14:16:30wodzmt: ok I can look at this after dinner :-)
14:16:51JdGordonwhat is it supposed to look like?
14:17:11mtwodz: ok. I'll bisect a little now. :)
14:18:28pixelmaI can do the rest but the ability to put the peakmeter inside conditionals is something that would need to be coded. JdGordon: on hwcodec the fm screen shows a peakmeter which is replaced by the prerecording time (actually counting up) if you have that set
14:19:16JdGordonok, well work around the fact that pm cant be in a conditional for the time being and make the screen fit
14:26:38 Join Blue_Dude [0] (~chatzilla@rockbox/developer/Blue-Dude)
14:26:44 Join pamaury_ [0] (~c2c7a50a@rockbox/developer/pamaury)
14:27:08Blue_DudeUpdated the bug report at FS #11255: wav file resume crashes the player.
14:27:33pixelmaI find the info important though... hopefully there is another place to put it
14:29:25pixelmaBlue_Dude: metadata inpure wav files?
14:29:41 Quit pamaury (Ping timeout: 252 seconds)
14:30:01Blue_DudeThe wav file is old. It certainly didn't crash the player before.
14:30:02pixelmaI also had one and it crashed the player but I know metadata is not standard in PCM wave
14:30:29Blue_DudeSeems to me that codecs ought to tolerate metadata, not crash.
14:30:34n1swodz: was the clock set up taken from the OF?
14:31:10pixelmaI agree a crash is not nice, the least it should do is reject the file
14:31:48pixelmaif it's unable to play it for whatever reason
14:32:09pixelmaBlue_Dude: thanks for the hotkey menu cleanup by the way :)
14:32:32Blue_DudeYou're welcome.
14:33:40*JdGordon thinks the peak meter handling is braindead compared to the newer stuff :p
14:34:39 Join arbingordon [0] (~w@unaffiliated/arbingordon)
14:38:58Blue_DudeYeah, I created a version of the wav with the same audio content but no metadata and it doesn't crash. Something happened in the last week or so to not tolerate metadata.
14:40:10JdGordonhaha sweet! peakmeter working in conditionals was simple
14:40:53JdGordonneeds some proper testing though... any takers?
14:41:31pixelmaBlue_Dude: the one that crashed for me was ages ago but I guess since metadata in wave is not standard that there can be differences depending on how it was tagged exactly
14:41:58 Join TheSeven [0] (~TheSeven@rockbox/developer/TheSeven)
14:42:08pixelmaJdGordon: I would but my laptop is currently busy with music encoding
14:42:19JdGordoncan i give you a build then?
14:42:21Blue_DudeI didn't even know it had metadata in it.
14:42:53Blue_DudeI'm doing a binary compare of the two files to find out what's going on.
14:42:57pixelmaJdGordon: yeah, I think a c200 build would be easiest (as that one is standard)
14:43:34JdGordonyou dont need the whole zip if you are running pretty close to svn
14:43:41JdGordonso do i need to do the whole zip?
14:44:32pixelmayou can do OndioFM too (would be nice if you could do a backlight build - set in the advanced build options)
14:45:17pixelmamaybe the latter is more interesting and I shouldn't need the whole zip there, only the ajbrec.ajz">
14:46:25JdGordonthis is just peakmeters in conditionals, no fm skin patch
14:46:52pixelmaunderstood, I will just set up a simple test.wps
14:46:52n1sBlue_Dude: uchida added some wav metadata parsing recently, could be the cause, the sim is usually easier to debug for this type of bugs
14:47:31 Quit Schmogel (Quit: Miranda IM! Smaller, Faster, Easier.
14:47:48Blue_Duden1s: understood. First I need to find out what's different between the two files, then maybe I can find out what the parser is choking on.
14:48:08 Quit Buschel (Ping timeout: 248 seconds)
14:48:28pixelmamaybe bisecting for the commit that breaks it could also help"> < ondiofm+backlihgt
14:49:23Blue_DudeHm. Two bytes difference... Weird.
14:49:52Blue_DudeI mean, that's really weird.
14:50:30kugelBlue_Dude: how's your mixer patch going?
14:50:34Blue_DudeOff to the hex editor I go.
14:50:37pixelmaJdGordon: downloaded
14:50:54n1sid guess r25898 or r25911
14:50:56Blue_Dudekugel: Delayed by wav bugs and hotkey patches. :)
14:51:08Blue_DudeOther than that, not bad.
14:53:06pixelmaJdGordon: I'm just reminded as I see it again that the statusbar in the early USB screen vanished in recent builds
14:53:25JdGordonput it on FS
14:56:26Blue_DudeJdGordon: is the "old" statusbar supposed to appear for splash screens? I got that in the WPS screen before I took out the hotkey splash.
14:57:16JdGordonoh hmm... yeah it could now if your wps disables the theme
14:57:42Blue_DudeI've been using cabbie, which does disable the status bar.
14:57:57JdGordonno, ignore that
14:58:00pixelmait shouldn't
14:58:32JdGordonBlue_Dude: where do you see this splash?
14:58:32Blue_DudeThe splash is gone so I can't reproduce, but pixelma's comment reminded me of it.
14:58:52JdGordonif it was shown by the menu then it would show it
14:58:58Blue_DudeI got it when the "Hotkey Not Set" splash came up.
14:59:10Blue_DudeIn the WPS.
14:59:22JdGordonwhich was called after the wps reenables the theme, so yeah expetced
14:59:34JdGordonsimple-ish to fix but if its not an issue anymore
14:59:37Blue_DudeOK, just seemed odd.
15:00:01Blue_DudeI'm not sure there are any other splashes in the WPS.
15:00:54pixelmaJdGordon: pm in the conditional seems to work on both - the c200 and the Ondio - tested with a simple test.wps that just shows pm on hold and nothing when unlocked
15:01:13JdGordonthats how i tested on sim also.. ok good
15:02:01CIA-5New commit by jdgordon (r25913): slightly rework peakmeter handling to make it cleaner and be able to be used in conditionals
15:02:03pixelmaone thing to look out for might be scrolling lines elsewhere
15:02:40 Quit MethoS- (Remote host closed the connection)
15:03:55wodznls: clock setup You mean setup of timer to perform system tick?
15:04:58JdGordonpixelma: do you tihnk you can do the hardcoded fms in the next few days now that is fixed?
15:05:29pixelmaI think so, when an updated patch is available
15:07:10JdGordonthe one i uploaded 20min ago works
15:07:20pixelmascrolling lines alongside such a conditional peakmeter work by the way. I just remembered there were some problems before with %pm in conditional viewports for a while before which is why I tested
15:07:44pixelmaok, will have a look
15:07:54JdGordondo you remember the FS# for that bug?
15:08:04JdGordonI remember it haunting me and not being able to figure it out
15:08:11Blue_DudeOK, here's what I have on the wav file. The one that behaves only contains audio chunks. The one that doesn't contains a metadata chunk at the end. Other than the metadata chunk they are identical, except for two bytes in the header. Those two bytes are the low end bytes of a four byte integer describing the total chunk size of the wav file. My guess is that for some reason, the parser... not ignoring the metadata but somehow trying to read it as an audio subchunk.
15:09:01pixelmaJdGordon: the peakmeter in conditional viewports is fixed but I don't remember the FS# for it
15:09:12JdGordonit was fixed? ok cool
15:09:13Blue_DudeIOW, the parser assumes that if a subchunk exists in the total count, it must be audio.
15:09:18lovelessare any arm-elf toolchain wizards awake, yet?
15:10:27pixelmaJdGordon: need to test again but I saw it in one of my WPSs and think it was fixed. I can easily test though
15:11:11 Join GeekShadow [0] (~Antoine@reactos/tester/GeekShadow)
15:12:04lovelessneed some help with bootloader.bin on -mcpu=arm1176jz-s. seems to produce broken boot/installer images
15:14:49pixelmaJdGordon: the WPS that was affected seems to work fine now (r25710 on my M5... so slightly old but I don't think it's broken again)
15:14:52Blue_DudeAnyway, the bug report is updated again, and I have to go to work. Again. Later!
15:14:57 Quit Blue_Dude (Quit: ChatZilla 0.9.86 [Firefox 3.6.3/20100401080539])
15:14:58pixelmathough you never know
15:18:14 Join petur [0] (~petur@rockbox/developer/petur)
15:19:12 Join Llorean1 [0] (
15:19:59 Quit M3DLG (Quit: RAGE QUIT)
15:21:48 Quit ncfi1013 (Quit: Leaving)
15:22:47 Quit antil33t (Read error: Connection reset by peer)
15:22:53 Join antil33t [0] (
15:23:11 Quit Llorean (Ping timeout: 240 seconds)
15:28:22lovelessokay, rechecked with same code, same setup and the only difference being -ma
15:29:22 Quit komputes_ubuntu (Quit: I haven't slept for ten days, because that would be too long.)
15:29:39 Join komputes_ubuntu [0] (~komputes@ubuntu/member/komputes)
15:29:51loveless... with the only difference being -mcpu=arm7tdmi vs. -mcpu=arm1176jz-s
15:30:16 Quit stoffel_ (Remote host closed the connection)
15:30:58mtwodz, kugel : bisect results in r25850 being the commit that caused the error in audio playback in the sim.
15:34:19 Join M3DLG [0] (
15:41:17 Quit kugel (Remote host closed the connection)
15:42:00 Join kugel [0] (~kugel@rockbox/developer/kugel)
15:43:29 Quit petur (Ping timeout: 268 seconds)
15:45:54 Quit arbingordon (Quit: `)
15:46:39 Join stripwax [0] (
15:47:17mtAnyone having audio working fine in the sim with current svn ?
15:47:38***Saving seen data "./dancer.seen"
15:47:52kugelmt: can you play audio at all?
15:49:45 Quit Jaykay (Quit: ChatZilla 0.9.86 [Firefox 3.6.3/20100401080539])
15:50:03mtkugel : you mean outside the sim ?
15:50:13kugelno, in the sim
15:50:22kugelI have no problem playing audio
15:50:50mtThen it could be something in my drivers, I'll check
15:50:53 Quit bluebrother (Disconnected by services)
15:50:55 Join bluebroth3r [0] (~dom@rockbox/developer/bluebrother)
15:52:44kugelmt: I meant: does it refuse to play entirely or does the error appear after some minutes??
15:52:44mtkugel: I can't play audio at all in the sim
15:52:54mtNo entirely
15:53:27mtExample : Once you select a file for playback, it crashes
15:53:49gevaertsmt: various codecs?
15:54:17kugelhm, something is not alright here too. it skips sometimes
15:54:34mtgevaerts: No, I tried one file and plugin alarm clock
15:55:34wodzmt,kugel: my bisecting ends up at r25850 also
15:57:02wodzI didn't updated system for quite some time so this is not system side error
16:05:00kugelwodz, mt: try please (make sure you re-create the Makefile)
16:05:22 Join petur [0] (~petur@rockbox/developer/petur)
16:05:52kugelit solves the skipping for me
16:07:57*wodz compiling
16:09:03mtI'm compiling now too, but will probably be slower than wodz
16:09:42wodzno it didn't help
16:13:27mtkugel: fails here too.
16:14:16 Join arbingordon [0] (~w@unaffiliated/arbingordon)
16:17:37CIA-5New commit by kugel (r25914): Remove -U_GNU_SOURCE again (was added in r25850). It turns out SDL doesn't work properly with it. The reason I added it for doesn't exist anymore (or, ...
16:18:07kugelmt: unfortunately, I don't have the slightest idea what could cause that
16:18:19kugelthe commit didn't really make any code changes
16:19:24mtMaybe the problem is with my system then ... I'll see
16:19:32AlexPJdGordon: Yo
16:20:23 Join Jaykay [0] (
16:24:51*kugel googles the error and finds
16:27:28kugeland also a mailing thread from 2006 where the same problem occurs
16:28:55wodzkugel: but befor r25850 I didn't get this errors and sim worked ok
16:30:24 Join hellelujah [0] (
16:30:30JdGordonAlexP: hey, been a while :) trying to get back into the fms patch and get it commited.. remember any issues? and which target were you testing on that needed a good buffer bump?
16:31:08AlexPI can't remember any specific ones - the H140 needed a large buffer bump (I think principally due to multifont)
16:31:18AlexPAnd yes, it has been a little while :)
16:31:18 Quit hellelujah (Client Quit)
16:35:19kugelwodz: I believe you
16:35:40JdGordonwith cabbie there its 17KB free so as long as the backdrop image is reused that should be plenty, for the time being anyway
16:35:52kugelthe issue came up in past apparently, I wonder if it was fixed or just disappeared
16:35:54JdGordonwe really need to come up with a more flexible buffer stratergy
16:37:29 Join hellelujah [0] (~59a10eb2@gateway/web/freenode/x-krxkundxvdlwyxry)
16:38:22hellelujahHi. Would iPod nano 4th gen ever be supported?
16:38:39bluebroth3rif someone does the work, then yes.
16:38:41wodzhellelujah: You are free to port :-)
16:39:03hellelujahIf I could...
16:39:10kugelBagder: can you remember how you resolved that issue?
16:39:12hellelujahNot enough skill.
16:39:24kugelBagder: I'm refering to
16:39:27 Part hellelujah
16:40:56JdGordonkugel: tried looking at the old x11 code in svn?
16:46:54 Join Xerion [0] (
16:47:22JdGordonnot really expecting this to go anywhere, but how would people feel about chaning the skin language to use [] for paramterised tags? %V[a|b|c...|] ? and also getting rid of "extra stuff" on tags (i.e %Xda would change to %Xd[a])
16:47:42JdGordonI think we could make the parser much simpler if we did that, and it should make it more readbale
16:48:56kugelJdGordon: I had that idea before, and I'm all for it
16:48:58JdGordonthose changes could be done in a script for all existing themes
16:49:21kugelI think I even posted it here ;)
16:50:26bluebroth3rJdGordon: without knowing any details it sounds good :)
16:51:06JdGordonwho is doing the theme editor gsoc?
16:52:31 Quit n1s (Read error: Connection timed out)
16:53:48bluebroth3rJdGordon: bieber is the student, I'm mentoring.
16:54:51bluebroth3rIIRC he was already asked about such a change and considered it a pretty easy change for the editor
17:01:56 Join adnyxo [0] (
17:05:43pixelmaJdGordon: I'd like the too and if it makes the parser simpler then the better
17:06:05wodzmay I ask to look at FS #11254?
17:06:15JdGordonthe ; tag could cause problems
17:06:34JdGordonI wonder if there is any way we could get away from the line based system we have now?
17:06:35pixelmabecause of the %t?
17:07:01JdGordonno, because it is not really a tag
17:07:21JdGordon%t would become %t[number]
17:07:33JdGordoni.e tags with any args would HAVE to have the []'s
17:11:39 Quit esperegu (Ping timeout: 246 seconds)
17:12:21 Join esperegu [0] (~quassel@
17:15:28 Quit TheSeven (Ping timeout: 246 seconds)
17:15:34 Quit komputes_ubuntu (Ping timeout: 240 seconds)
17:17:34kugelJdGordon: I would prefer () though
17:17:41 Quit antil33t (Read error: Connection reset by peer)
17:17:48 Join antil33t [0] (
17:17:51JdGordoneither work for me
17:17:58 Quit esperegu (Read error: Connection reset by peer)
17:18:45 Join esperegu [0] (~quassel@
17:20:22 Join TheSeven [0] (~TheSeven@rockbox/developer/TheSeven)
17:20:49kugelmt, wodz: sound is also fine on my laptop
17:21:54 Join wincent [0] (
17:22:03 Quit wincent (Changing host)
17:22:03 Join wincent [0] (~wincent@rockbox/developer/wincent)
17:23:57bluebroth3rhow about using {}? ;-)
17:25:01wodzkugel: I glanced at r25850 patch and I don't see any reason why it changes anything but apparently it does
17:27:33kugelmt: you say you compile under a vm?
17:27:46kugelVBox? which distro?
17:27:49pixelmabluebroth3r: wouldn't be that bad as I use () and [] sometimes just like that in the WPS or tags, but don't do so with {}
17:28:19pixelmait also gives a tex touch ;)
17:28:47 Quit evilnick (Ping timeout: 245 seconds)
17:29:44*kugel goes trying to repro it in a vm
17:29:46 Quit kugel (Remote host closed the connection)
17:30:23bluebroth3rpixelma: I wasn't completely serious, I don't care much about it. However, having some opening-closing-braces sounds like a good idea to me. IMO a clash with literals shouldn't happen because a tag always starts with a %
17:32:45*gevaerts prefers <>
17:32:57pixelmathought so, but for reading the .wps file {} would look a bit more like "code" to *me*
17:33:09 Join robin0800 [0] (~quassel@
17:33:10pixelmagevaerts: clashes a bit with conditionals
17:33:23gevaertspixelma: then conditionals should be changed!
17:33:46pixelmaof course
17:34:14 Join komputes_ubuntu [0] (~komputes@ubuntu/member/komputes)
17:37:36mtkugel: yeah virtualbox .. ubuntu but a quite old version .. (8.04 I think)
17:38:26mtyeah 8.04
17:39:36wodzmt: that's curiose I use (x)ubuntu 8.04 also
17:39:45 Join kugel [0] (~kugel@rockbox/developer/kugel)
17:40:09mtwodz: I wanted to ask you about your distro but I totally disregarded that
17:41:01wodztomorrow I can check on some debian and newer ubuntu
17:41:31mtkugel: I think it's leaning more & more towards being a system error rather than being a rockbox bug.
17:42:04wodzmt: not quite - sim used to work and at r25850 stopped
17:44:47mtwodz: While searching for the errors, there was some link about a bug with alsa in 9.10 (didn't open it so not quite sure of the content) .. but anyway I just think that the possibility is higher now that it's something with our systems, imho.
17:46:10wodzmt: It *may be* combination patch & system but we should be sure about that and moreover this shouldn't happen :-)
17:46:51CIA-5New commit by alle (r25915): Manual: Bookmarking update (FS #11227 by Magnus Holmgren)
17:47:16wodzand I use this system daily with a few SDL apps and I don't have any problems
17:47:40***Saving seen data "./dancer.seen"
17:51:41 Join wincent_balin [0] (
17:51:44 Quit wincent (Ping timeout: 268 seconds)
17:52:22 Quit wincent_balin (Changing host)
17:52:22 Join wincent_balin [0] (~wincent@rockbox/developer/wincent)
17:54:17 Quit wincent_balin (Client Quit)
17:54:47 Join wincent_balin [0] (
17:54:50 Quit robin0800 (Remote host closed the connection)
17:55:47 Quit wincent_balin (Client Quit)
17:56:16mtwodz: what I meant was that the patch _might_ have exposed a bug in the system, because quickly skimming through it, I can't really find a change that has a direct relation with that error.
17:56:51 Join wincent [0] (
17:57:11 Quit wincent (Changing host)
17:57:11 Join wincent [0] (~wincent@rockbox/developer/wincent)
17:58:52 Join ansuz [0] (
18:00:25CIA-5New commit by alle (r25916): Use typographic quotation marks
18:08:10 Join Schmogel [0] (
18:09:32kugelmt: "success" I get the error in a vm
18:09:46mtSame distro ?
18:10:00kugelno, debian as of sept 2009 or so
18:10:17 Join Buschel [0] (
18:10:45kugelI'll try if updating helps. but I need to backup the old image so that I don't lose my repro system
18:10:47 Join flatrose [0] (
18:19:07 Join domonoky1 [0] (
18:19:54 Quit domonoky (Ping timeout: 246 seconds)
18:24:42 Join n1s [0] (~n1s@rockbox/developer/n1s)
18:31:41 Join CaptainKewl [0] (
18:48:38 Quit komputes_ubuntu (Ping timeout: 248 seconds)
18:50:38 Join komputes_ubuntu [0] (~komputes@ubuntu/member/komputes)
18:52:31flatroseIs there a way to overclock iPod 5.5G's processor to 100mHz(+) using rockbox? I know it will drin more power but the heat would keep my HDD spinning at -30 next winter.
18:53:44kugelflatrose: yes, ask Buschel for details
18:54:45 Quit wincent (Ping timeout: 240 seconds)
18:55:12 Join wincent [0] (
18:55:27soapflatrose, infact, search flyspray for all patches by Buschel and you'll quickly find it.
18:55:46soapthough many iPods appear unstable above rated speed.
18:59:08bieberI was away when the tag syntax topic came up a little while ago, but I'd like to chime in and say that it would make things really immensely simpler to parse, and I'm all for it
19:00:17 Quit komputes_ubuntu (Ping timeout: 264 seconds)
19:01:22 Nick Llorean1 is now known as Llorean (
19:01:26 Quit Llorean (Changing host)
19:01:26 Join Llorean [0] (~DarkkOne@rockbox/user/Llorean)
19:10:59 Quit esperegu (Ping timeout: 240 seconds)
19:11:44 Join komputes_ubuntu [0] (~komputes@ubuntu/member/komputes)
19:11:46 Join hpmaniac [0] (
19:11:48bieberAnd as for ;'s, my suggestion would be to replace them with an actual tag that has multiple parts, like the conditionals, which should also be a lot easier to parse than what we have now
19:12:28bieberBut I don't know how the theme engine handles parsing and rendering, so it may be easier for that the way it is now...
19:12:30hpmaniachi there
19:13:00 Nick kisak_ is now known as kisak (
19:13:20 Quit hpmaniac (Client Quit)
19:13:59 Quit mt (Ping timeout: 240 seconds)
19:18:12flatrosethanks for the answers, guys!
19:20:27 Join mt [0] (~mtee@rockbox/developer/mt)
19:21:25pixelmabieber: can you give more details about this suggestion? JdGordon recently suggested a %t5<1st part|2nd part|etc.> and I didn't like it at all. It's not easier to read and is less flexible than what we have now. With this every subline has the same timeout while you can have different timeouts now
19:21:55pixelmait's even harder to read when there are conditionals in each subline
19:22:40bieberPerhaps enclose each subline in its own tag that can optionally include its own time value, the way HTML has <ul>/<ol> and <li>?
19:23:54Lloreanbieber: I think the current sublines can be pretty readable - you always know that after a semicolon you're on a new on. If you had multiple characters doesn't it just make each transition more cluttery?
19:24:04Llorean*new one
19:24:19pixelmaI don't know. From a user's perspective ; with optional %t are quite readable
19:24:27LloreanAn explicit tag would just mean that instead of semicolons, we have %nL or something
19:24:58bieberI'm really not sure what the best way to approach it is. Adding explicit delimiters for every line will make it less readable, but easier to parse
19:25:24pixelmawith the exception that you need to write %; if you want to just display a ; in your WPS
19:26:17bieberPerhaps allowing newlines between lines would help readability?
19:26:20Lloreanbieber: I think what would make many WPSes easier to parse for the reader is if they used the commented line break (assuming that still works) to put sublines on multiple lines.
19:26:43Lloreanbieber: You used to be able to end a line with a # to "comment out" the newline, so that you could write more of it on a second line.
19:26:52pixelmabieber: you could already do that by "commenting out" the line-ending
19:27:07pixelmanot used much though
19:27:34bieberIf you can use newlines, it shouldn't be bad to read/write
19:28:48LloreanI don't know if that feature's well documented / documented at all, though
19:28:50bieberPerhaps something like %Ms{ %Mi{line|optionaltime} %Mi{line|optionaltime}}, depending on the names/delimiters that get chosen for the tags
19:29:35bieberLlorean: The CustomWPS wiki entry I've been working off of does say that a comment includes its newline, but it doesn't point out that you can use that feature to break up tags into multiple lines. Perhaps I should add that to the page?
19:29:36 Join esperegu [0] (~quassel@
19:30:04 Join saratoga_ [0] (~463f90ed@gateway/web/freenode/x-lzxwguthctscjalg)
19:30:43Lloreanbieber: What tag would you be breaking into multiple lines?
19:30:55LloreanTags can never span sublines, so when you create a subline you're not breaking up a tag.
19:31:42bieberWell, if we enclosed the sublines in a container tag. Although in general, you _could_ split a tag's options across multiple lines with commented newlines, right?
19:32:02pixelmamaybe the timeout tag can be the sign "start a new subline here"? Although that takes away the possibility to put the %t anywhere in the subline and you wouldn't be able to leave it out and let it use the default time (2s IIRC)
19:32:25Buschelflatrose: will set the clock to 100MHz max, 24MHz default, and will boost while GUI activity
19:32:35Lloreanbieber: Why are you back at putting sublines in a container tag again? That seems like it would generate a lot of additional {} and clutter?
19:33:21pixelmait would be nice to put the %t inside conditionals again BTW which was possible before the tokenizer
19:33:59bieberWell, it would add two additional braces ;) It's just an option that would make parsing a little bit easier, whether to include it for readability is another issue
19:34:01Lloreanpixelma: So depending on which condition was true, the line had variable possible display times?
19:34:29 Quit esperegu (Ping timeout: 264 seconds)
19:34:42pixelmayes, you could even surpress it this way. If I recall there is still an example with %t0 somewhere in the Wiki
19:36:19*Llorean thinks that would be very nice to have back
19:37:09saratoga_anyone want to look at the license in FS #11256 and guess if the lib is GPL safe or not?
19:37:48saratoga_theres actually two nearly identical versions of the TTA decoder floating around, one LGPL, one that license, so i guess if not it could be fixed
19:39:25 Join archivator [0] (
19:40:05bieberIn any case, I'm going to build a generic tag-parsing function for libwps, so once it's decided whether to add delimiters, they'll be easy to implement
19:40:21 Join Strife89 [0] (
19:44:28 Join esperegu [0] (~quassel@
19:45:10saratoga_skimming that license, the only real restriction added is that you have to maintain the copyright notice, and that you cannot use the authors name without his permission, which appears to be GPL ok since the GPL allows a copyright notice and doesn't grant you the right to the author's name
19:45:16saratoga_but i have no real idea
19:45:17 Join lpereira [0] (
19:45:31stripwaxBuschel - is that the same as, or different from, the fs# patch?
19:45:33saratoga_why couldn't uchida just use the explicitly gpl source download :(
19:46:07kugelbieber: did you decide whether to reuse the existing code or not already?
19:47:00 Quit archivator (Ping timeout: 246 seconds)
19:47:28kugelsaratoga: that's bsd afaict, should be gpl safe
19:47:30Lloreansaratoga_: If you have to maintain the copyright notice, but you can't use the author's name without permission, doesn't that restrict your ability to use the code (since that includes the copyright notice and thus the name?)
19:47:43***Saving seen data "./dancer.seen"
19:49:08saratoga_"Neither the name of the True Audio Software nor the names of its contributors may be used to endorse or promote products derived from this software without specific prior written permission."
19:49:40saratoga_kugel is right that is the BSD license
19:49:48 Quit esperegu (Ping timeout: 246 seconds)
19:50:11saratoga_or rather "BSD-new"
19:50:16saratoga_according to wikipedia
19:50:48kugelit's possible though that the last clause bites with gpl
19:51:53saratoga_the last clause is strange since trademarks are implicitly reserved anyway, though maybe theres some country out there where they needed this
19:52:18 Join MethoS- [0] (~clemens@
19:52:46saratoga_FSF says its GPL friendly, so good enough for me
19:54:20 Join esperegu [0] (~quassel@
19:54:48 Quit TheSeven (Ping timeout: 265 seconds)
19:59:22 Join TheSeven [0] (~TheSeven@rockbox/developer/TheSeven)
20:02:24 Quit loveless (Quit: loveless)
20:06:04 Quit petur (Quit: *plop*)
20:06:52 Join archivator [0] (
20:08:26 Join kyraxyg [0] (
20:11:40Bagderkugel: iirc, that x11 asound experiments of mine never got anywhere, we dropped it and later proceeded with the SDL way
20:20:26 Join kugel_ [0] (
20:20:26 Quit kugel_ (Client Quit)
20:22:19 Join petur [0] (~petur@rockbox/developer/petur)
20:26:58 Quit esperegu (Ping timeout: 240 seconds)
20:33:33CIA-5New commit by alle (r25917): Use typographic quotation marks −− correct more places
20:33:42 Join esperegu [0] (~quassel@
20:39:45 Quit saratoga_ (Quit: Page closed)
20:40:17 Join CGL [0] (~CGL@
20:40:58 Quit mt (Ping timeout: 240 seconds)
20:44:01Buschelstripwax: no. the patch I've just pastebin'ed is a 100MHz-version of the FS-version. not all iPods run stable >=100MHz.
20:48:55 Quit MethoS- (Remote host closed the connection)
20:48:59 Quit komputes_ubuntu (Ping timeout: 246 seconds)
20:51:45 Join MethoS- [0] (~clemens@
20:53:27 Join mt [0] (~mtee@rockbox/developer/mt)
21:01:50 Join TechnoKat [0] (
21:03:33stripwaxBuschel : gotcha
21:05:20 Quit kyraxyg (Quit: CGI:IRC)
21:06:29Buschelanything speaking against submitting FS #11253 ?
21:07:12wodzit's ok for me
21:08:19wodzspeaking of test_mem it would be nice to post results from different targets on wiki (like for test_disk, batery_bench, test_codec, test_fps...)
21:08:21 Quit wincent (Ping timeout: 240 seconds)
21:10:40n1swe should add ams for coldfire as well, gcc pretty much never generates movem instructions except for stack stuff in pro and epilogs
21:11:24amiconnWhat for, specifically?
21:12:08n1sif we want to get closer to the bandwidth thoroughput of the ram
21:12:33amiconnwodz: Btw, did you see my optimisation for ata-as-coldfire.S? It could be ported to hd200 - I just didn't want to do that as I can't test it
21:13:01amiconnIt doesn't noticeably improve speed - it saves a few instructions though
21:13:15wodzamiconn: yes I saw this
21:13:17amiconnn1s: Does the test use memcpy() / memmove()
21:13:43n1samiconn: no, that is probably a better ide
21:14:04amiconnI didn't check - those functions are asm monsters on coldfire, utilizing movem in as many alignment cases as possible
21:14:16*amiconn wrote them
21:14:38amiconnThey're in iram as well (it's not *that* important though)
21:14:48Buschelamiconn: the plugin test write and read seperately
21:15:25kugelBuschel: the patch looks good if you ask me
21:15:40amiconnWouldn't it make more sense to bench the mem*() functions?
21:15:57amiconnBtw, that's what my old test_mem.rock did (which never made it to svn)
21:15:58n1shmm, that read case looks like gcc could DSE all but one of the loads since they store into the same var in the loop...
21:16:25*amiconn just remembered what it did...
21:16:57wodzamiconn: could You please look at optimized writes in ata-as-hd200.S? I can't track down where is the problem with byte swaps.
21:17:18amiconnIs there still a problem? I thought it would be working now
21:17:35wodzamiconn: could I also ask You to run current test_mem on CF target?
21:18:00wodzamiconn: optimized writes makes test_disk fail
21:18:25 Join ischeria1 [0] (
21:18:40wodzbut I finally squashed bug and greylib is working on HD200 :-)
21:19:23ischeria1hello. are here any developers familiar with the iriver story (ebook reader)?
21:19:37*kugel has a suspicion re sim audio problem
21:21:04amiconnWhenever I'm looking at some arbitrary asm in rockbox, I find something that's not optimal :\
21:21:26amiconnNo matter whether it was originally written by me
21:23:28CIA-5New commit by Buschel (r25918): Submit FS #11253. Rework of test_mem plugin to bench DRAM and IRAM. Also add ARM assembler and change the result format.
21:24:30kugelwodz, mt: try this one: it works for me
21:24:41*Buschel waits for red to come...
21:25:10stripwaxamiconn - any of mine?
21:25:33amiconnNo idea. This time it's coldfire strlen - that's written by n1s
21:26:44n1samiconn: i know about that, i actually have a half finished version somewhere that loads words at a time and use the magic test trick to check the whole word for nullbytes
21:26:52amiconnwodz: Btw, as I see you replaced jeq by beq in ata-as-mpio.S: There are two versions of beq, beq.s and beq.w
21:27:09n1sjust haven't had time to finish and bench it
21:27:17amiconnThe former is shorter (no extension word) than the latter, but it's limited in branch distance
21:27:53AlexPischeria1: I doubt it - this is the first time anyone hs mentioned it
21:28:18 Nick ischeria1 is now known as ischeriad (
21:28:23amiconnIf you write just beq, the assembler will always choose beq.w - that's what the jeq pseudo-instruction is for: it makes the assembler use the shortest possible version for the branch distance
21:28:45n1swell didn't have time before i left it on my desktop at home, but will try to finish it in a few days
21:29:01amiconnn1s: Reading a byte at a time might be elegant - but it's sloooooow when going through dram
21:29:43n1swell, i remember it benched faster than gcc's mess at least but yes, it's hardly optimal
21:30:39ischeriadAlexP: not long ago, iriver released the sources for the story firmware. I do not know if there are any independent developers yet and where to find them
21:30:51n1sas i said, i will try to finish my new version when u get home in a few days
21:31:04n1ss/ u/ i/
21:31:14AlexPischeriad: Not here at any rate :) I'd try to find owners sites or the like
21:31:14 Quit bertrik (Quit: De groeten)
21:33:58 Quit mt (Ping timeout: 240 seconds)
21:34:16 Quit n1s (Quit: Lmnar)
21:34:34ischeriadAlexP: alright
21:36:26 Join mt [0] (~mtee@rockbox/developer/mt)
21:37:34wodzkugel: I get corrupt patch at line 21
21:37:40 Quit GeekShadow (Quit: The cake is a lie !)
21:38:40wodzok newline was missing
21:41:18 Quit mt (Ping timeout: 240 seconds)
21:42:11 Quit TechnoKat (Quit: Leaving)
21:42:57wodzkugel: this patch works for me
21:45:56amiconnwodz: Hmm, I can't spot an obvious problem in the hd200 ata write :\
21:46:17kugelwodz: what system are you on, btw?
21:46:29wodzkugel: (x)ubuntu 8.04
21:46:37amiconnHowever, you can optimize the swaps wherever only 16 bits are needed. No need to fiddle with the mask... fewer instructions
21:46:53kugelwodz: hm, mt too, maybe it's fixed in newer kernels
21:47:12kugelit seems to work for kernels newer than 2.6.30
21:47:18 Join mt [0] (~mtee@rockbox/developer/mt)
21:47:41wodzkugel: I can test this tomorrow on (x)ubuntu 9.04 and on some debian
21:47:45mtkugel: Worked for me !
21:47:47***Saving seen data "./dancer.seen"
21:48:38wodzamiconn: I am aware of this but as this function fail I tried to eliminate as much irregularities as possible to narrow down the problem
21:48:47mtkugel: Are you going to commit it ?
21:48:55 Join evilnick|ipad [0] (
21:48:58kugeltrying to, yes :)
21:49:11kugelthis vm is a pain
21:52:00amiconnwodz: Hmm, asm optimised reading works?
21:52:34wodzamiconn: yes
21:53:24amiconnNow that is odd
21:54:04amiconnTo me it looks like there is a problem in unaligned read. Need to check thoroughly...
21:55:18wodzamiconn: optimised read + general write pass test_disk test
21:56:06 Join Blue_Dude [0] (~chatzilla@rockbox/developer/Blue-Dude)
21:58:22CIA-5New commit by Blue_Dude (r25919): Fix wav metadata bug, also fix typos and some const police
21:58:57 Join ender [0] (
22:00:18 Quit ender` (Ping timeout: 240 seconds)
22:00:53 Quit kugel (Quit: exit(0);)
22:02:53 Nick ender is now known as ender` (
22:03:48amiconnwodz: Yet there *is* a problem in unaligned read if there is no single 16 bit word to be handled after the first byte
22:04:09amiconnIt will write the first byte twice, omitting the actual second byte
22:04:17amiconn(write to memory of course)
22:05:43 Join kugel [0] (~kugel@rockbox/developer/kugel)
22:07:17CIA-5New commit by kugel (r25920): Undo a change of r25850 which broke SDL audio on some (older kernels?) systems.
22:10:41kugelso, hopefully all my latest bugs are squashed out now so I can move on with raaa :)
22:14:42amiconnwodz: What I mean is that %d2 is only pre-swapped when entering longword/line loops if there is an initial byte+word
22:15:22gevaertskugel: you can always move on and fix bugs later when they're found :)
22:15:38mtkugel: Fixing that specific bug will let me go on with my project too. :)
22:15:42amiconnIf there's just a byte and not a word, it's not pre-swapped, hence the first move.l to RAM will have the topmost byte wrong
22:16:09*amiconn wonders whether C writing makes up for that read bug so that test_disk works overall
22:16:35 Quit Blue_Dude (Quit: ChatZilla 0.9.86 [Firefox 3.6.3/20100401080539])
22:17:13amiconnwodz: Can you test with both C reading and writing? Also, can you compare md5sums of audio files calculated using the md5sum plugin with md5sums calculated on PC?
22:17:31amiconnAlso, are there audible glitches when playing music?
22:17:57amiconnI doubt that though since buffering aligns these days
22:18:10amiconnHmm, md5sum will probably do the same
22:20:05 Quit FlynDice (Ping timeout: 264 seconds)
22:20:21amiconnA similar problem exists for the final word in the unaligned read: If there is one, the leftover byte from the last longword is wrong, since it is swapped twice
22:23:44*amiconn wonders why MPIO didn't use the (dead simple) hardware swap like iriver and cowon
22:24:01kugelgevaerts: I'll start the target tree work. The discussion on the ml wasn't quite a discussion but it confirmed my plams
22:25:03gevaertskugel: yes, I think it's clear that people either agree or don't care much, which is good enough
22:25:35 Quit flatrose (Ping timeout: 265 seconds)
22:25:37kugelalright, fine
22:26:11amiconnIt's just that the ata data lines are connected the "wrong" way - that's why iriver & iaudio have their commands & status words defined different than the rest
22:27:21gevaertskugel: to be honest, I prefer this to a long flamewar :)
22:27:47kugelme too :)
22:32:50amiconnHmm. It's strange though. The C versions are looking 100% correct - so I don't understand why optimized reading on MPIO doesn't cause problems, but optimized writing does
22:32:57amiconnI'd expect the opposite
22:33:01 Quit anewuser (Quit: for SELL 2 by the price of 1 now!)
22:35:15 Quit dfkt (Quit: -= SysReset 2.53=- Ph'nglui mglw'nafh Cthulhu R'lyeh wgah'nagl fhtagn.)
22:35:27amiconnAH wait - the writing *does* have problems - even more serious ones
22:36:21amiconnCombining the words before byte-swapping combines the wrong bytes - and that happens throughout in the unaligned case, not just at the beginning and end
22:36:35amiconnwodz ^^
22:37:58amiconneh, or not...
22:38:06*amiconn is confused
22:39:22amiconnThe unaligned read bug does hold though
22:39:42 Join Blue_Dude [0] (~chatzilla@rockbox/developer/Blue-Dude)
22:39:57domonoky1hi archivator, hows you project going ? :-)
22:42:48 Quit Schmogel (Quit: Miranda IM! Smaller, Faster, Easier.
22:43:00 Nick domonoky1 is now known as domonoky (
22:43:17 Quit domonoky (Changing host)
22:43:17 Join domonoky [0] (~Domonoky@rockbox/developer/domonoky)
22:43:23Blue_DudeCan anyone think why mp3_play_data, mp3_play_pause, mp3_play_stop, and mp3_is_playing functions in the plugin API are disallowed for the simulator?
22:43:26 Quit evilnick|ipad (Remote host closed the connection)
22:44:03kugelBlue_Dude: are they for hwcodec?
22:44:32Blue_DudeStrangely enough, no. They go to voice_thread.
22:44:47Blue_DudeBut only for swcodec.
22:46:25archivatordomonoky: well, I *almost* got it to compile for target but I've stopped working on that aspect for the time being (it's a bit beyond my area of expertise) and am doing fixed-point conversion, using the sim. It's going.. but rather slowly (it's a mess, to put it very, very mildly).
22:46:50 Join FlynDice [0] (~FlynDice@
22:47:03domonokyarchivator: would be good ot post patches about that, so others can see it too.
22:47:08kugelit's not expected to be easy :)
22:47:27kugelgevaerts: time for a first status report I'd say
22:48:09*domonoky also would like to remind all students here of the weekly status report in the mailing list...
22:48:10archivatordomonoky: will do, soon. Want to have something more, though.
22:48:40gevaertskugel: yes indeed :)
22:49:50Blue_DudeThe hwcodec version is in mp3_playback.c. I don't see why the sim is specifically carved out.
22:50:21domonokyarchivator: especially when you get stuck, post patches on the tracker, and ask for fresh looks at it here or in the mailinglist..
22:51:44archivatordomonoky: I'm not really "stuck", it's just hard to grasp all that's going on in flite. I've started drawing diagrams (!), even..
22:51:55Blue_DudeI wanted to point plugins to mp3_play_data for sound effects. That's kinda hard to test if the sim doesn't compile them.
22:53:10domonokyarchivator: if you have those diagrams, maybe it would be good to put it into the wiki with descriptions :-)
22:53:59archivatordomonoky: huh, never though of making them public. Might do something proper, then. My handwriting is horrendous :)
22:55:33 Quit freddy_ (Ping timeout: 265 seconds)
22:57:06domonokyyou dont need to make them public, but if you need them to understand flite, maybe it would be helpfull for others too :-)
22:58:05 Join krabador [0] (
22:58:38mtStatus updates now ? I thought we did those after the official start of the coding period
22:59:09 Quit wodz (Quit: Leaving)
22:59:11kugelI don't see a reason to wait with them
23:00:23mtI wanted to send a mail before though .. but thought I should wait. Will probably send something in the next few days then.
23:00:47gevaertsmt: for students with an extended coding period (due to schedule conflicts), I think the extended period is what matters, not the official period
23:03:33CIA-5New commit by Buschel (r25921): Add non-breaking spaces to mA, mAh, MB and GB. Add playertype for iPod Video when describing the different variants.
23:03:55 Quit Blue_Dude (Quit: ChatZilla 0.9.86 [Firefox 3.6.3/20100401080539])
23:08:01 Join freddy_ [0] (
23:08:49 Quit esperegu (Remote host closed the connection)
23:16:15 Quit Jaykay (Quit: ChatZilla 0.9.86 [Firefox 3.6.3/20100401080539])
23:18:26 Quit Buschel (Ping timeout: 246 seconds)
23:18:40 Join hebz0rl [0] (
23:23:30 Quit pamaury_ (Ping timeout: 252 seconds)
23:23:59 Quit FlynDice (Remote host closed the connection)
23:26:55mtsaratoga, linuxstb : The asf parser seems to be parsing wma pro files correctly now .. I'm thinking of committing that, with a wmapro.codec that would just not play for now.
23:27:09 Quit Xerion (Quit: )
23:27:48mtnot much changes at all really .. ASF is being nice till now. :)
23:29:02 Quit archivator (Quit: Leaving)
23:34:13 Quit petur (Quit: Zzzzz)
23:40:08saratogaBagder: could you add this new test file:
23:41:30Bagdersaratoga: done!
23:42:10CIA-5New commit by mt (r25922): nomsg
23:43:47mtShould I revert that and recommit it with a proper message ?
23:44:17 Quit lpereira (Quit: Leaving.)
23:45:42 Quit domonoky (Read error: Connection reset by peer)
23:47:51***Saving seen data "./dancer.seen"
23:49:55gevaertsmt: IIRC Bagder or Zagor can fix that
23:50:14 Quit hebz0rl (Ping timeout: 246 seconds)
23:50:42 Join hebz0rl [0] (
23:50:45mtgevaerts: Thanks .. I really hope so
23:52:38CIA-5New commit by mt (r25923): Fix ffmpeg revision number in libwmapro/README.rockbox
23:54:22saratogai guess I should make some WMA Pro test files too
23:54:29mtBagder: Can I send you the correct commit message for r25922 so you could fix it ?
23:55:02LloreanjhMikeS: Does the auto-speed whatever that the GigabeatS does respect plugins that ask to be boosted?

Previous day | Next day