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 2007-10-19

00:00:07krazykitpreglow, the build will be done in a few moments
00:00:18rasherpreglow: he did
00:00:41amiconnrasher: Iirc gcc 2.95 for m68k was the standard gcc for amiga. Maybe that's why it's still floating around...
00:02:03preglowrasher: who did what? :V
00:02:30markunpreglow: well, I can try again later
00:02:46rasherpreglow: saratoga posted his wma seeking patch
00:02:49preglowrasher: i know :)
00:02:55preglowkrazykit: i'm going to msg you with the url for a test track now
00:03:28safetydanGodEater (for the logs): stereo speex files don't work because I believe they're still using a floating point decoding path
00:03:42preglowkrazykit: listen to that track, it should have clean sound
00:04:01preglowkrazykit: if something seems wrong, please do a wav write with test_codec and compare the sound with the /test.wav that results
00:04:37 Join SirFunk [0] (
00:06:08preglowkrazykit: and in any way, i'd be thrilled if you could in some way provide me with the first thirty or so seconds of test.wav (or test.flac if you can encode that)
00:07:18krazykitsure, no problem
00:08:29 Quit midkay ("Leaving")
00:08:51krazykitpreglow, sounds just fine to me
00:08:55 Quit Pyromancer ("Leaving")
00:10:16 Join iamben [0] (
00:11:12rasherthere are few things more frustrating than a partially translated interface
00:12:20 Quit midgey ()
00:12:40 Quit Rob2222 (Read error: 113 (No route to host))
00:13:05krazykitpreglow, i didn't notice any obvious problems with the song.
00:15:06moosreally really strange, cause the sound distortion here don't need golden ears at all
00:15:24 Quit bluebrother ("leaving")
00:16:05moosbut I will test this heavily this week end and give you all results preglow
00:16:11 Quit MethoS- (Read error: 113 (No route to host))
00:17:04preglowkrazykit: could i have the first thirty secs of it?
00:17:18preglowi'll host it for moos to hear as well, so he can compare
00:17:29 Quit eigma ()
00:18:33krazykitsure. i'll have to install audacity to slice it down. want it in flac?
00:18:44preglowwould be leet
00:19:10linuxstbCan't flac encode a part of a file directly?
00:19:17krazykitcan it?
00:19:30krazykittoo late anyway, audicity is installed
00:19:48linuxstb−−until mm:ss
00:20:16 Quit ompaul (Client Quit)
00:20:39krazykiti'll just do that, then.
00:21:04preglowkrazykit: but the wav and rockbox sounds the same to you?
00:21:48 Quit PaulJam (".")
00:21:57 Join RaZorbacK [0] (
00:23:13 Quit n1s ()
00:24:25krazykitpreglow, yeah, identical
00:24:37RaZorbacKhi, is there any working link for rbutilqt 1.0.2?
00:24:54RaZorbacKthe one on the wiki seems to be broken...
00:26:59moosGood night/day all
00:27:07 Quit moos ("Glory to Rockbox")
00:27:14rasherRaZorbacK: I think it's been uploaded recently - may be a while before the mirrors sync
00:27:40RaZorbacKok thanks.$
00:27:51keanuRaZorbacK, what OS?
00:28:19RaZorbacKthx really
00:28:33keanuheh, no problem
00:31:08preglowthis sounds sweet to me too
00:31:08 Quit lids (Read error: 104 (Connection reset by peer))
00:31:10preglowargh, moos left
00:31:24RaZorbacKanother question: who is in charge of rbutilqt with bluebrother?
00:31:34krazykitwell, that file isn't going anywhere, so the link won't expire or anything
00:31:50preglowjust need to remember showing him it :>
00:33:13rasherRaZorbacK: bluebrother and Domonoky are the people who currently work on rbutilqt
00:33:19rasherBut anyone's free to help out..
00:33:51RaZorbacKrasher: ok, i will help in suggesting stuff :) that's the more i can do :)
00:36:15 Nick fxb is now known as fxb__ (
00:39:14 Join lids [0] (
00:40:08 Join Mouser_X [0] (n=Mouser_X@
00:41:21 Quit RaZorbacK ("CUL")
00:44:24rasherWhy doesn't Rockbox reboot when usb is plugged on Sansa?
00:44:59iambenshould it reboot?
00:45:27 Quit XavierGr ()
00:45:28keanuiamben, IMO, it should until USB support is fully working on the Sansa.
00:45:29 Quit petur ("Zzzzz")
00:45:32Aveafaicr sansa usb support "is not there yet"
00:45:33pixelmait should but it sufffers from the "hanging on usb screen" bug like observed on Ipods in some builds
00:45:59Avesansa has a disk mode similar to ipod?
00:46:20Avelike hw usb host mode
00:46:39pixelmaat least with my c200 - it's just not easy to see because the display is almost pitch black without backlight
00:47:03 Quit scorche|w ("CGI:IRC (Ping timeout)")
00:47:05keanuAve, no, but it has the OF. Since the bootloader loads the OF when the USB cable is attached, if a full reboot was possible, it should then go the the OF
00:47:36 Join miepchen^schlaf [0] (
00:47:52Avesane thing with ipod (nano) the disk mode is without backlight
00:48:55Aveof meaning original firmware?
00:49:14Avethere was some special word for that...
00:51:09Avethat saved the day right there
00:51:46rasherOddly, plugging the usb cable with the sansa off seems to only start charging, not msc connection. For me anyway, so rebooting to get usb connection might not even work
00:52:46pixelmathat works for me everytime except when I tried on a slow usb port
00:53:09rasherPerhaps I'm using a buggy version of the OF
00:54:05iambenrasher: it goes right to usb mode for me if the OF's db doesnt need refreshed
00:54:22iambenif it has to refresh the db, i have to unplug/replug to get to usb mode
00:54:28keanuiamben, same
00:54:52rasherThat might be it, I suppose
00:55:30pixelmathat could be the difference, my c200 refreshes its database everytime even when I just had it connected for charging :\
00:55:34rasherThough hard to test, since it always wants to update the database
00:55:47iambenit annoys me as well
00:55:58iambenunfortunately, not much that can be done from our side
00:57:06iambenwell since i *only* go to OF to transfer files, of course there are new files every time i boot it
01:00:04 Quit lee-qid ("aufwiederbyebientotsayonara")
01:00:19rasherHere it goes "plug cable, turns on, boots OF, updates database, starts charging"
01:00:42 Join midgey [0] (
01:01:08rasherOh wait, perhaps if I remove the µsd
01:01:52rasherYeah, that did it. Obviously
01:02:43iambenanyone ever have an issue where rb tries to "commit the db" more than just the first boot after a db initialization?
01:03:05 Quit ender` (" I went to a restaurant that serves "breakfast at any time." So I ordered French Toast during the Renaissance.")
01:03:19iambensometimes mine will commit the db on the next several reboots, and i end up w/ duplicates in the db
01:03:32 Quit Rob222241 (Read error: 113 (No route to host))
01:04:36 Quit jgarvey ("Leaving")
01:06:50 Join donutman25 [0] (n=chatzill@
01:07:49 Quit obo ("bye")
01:08:09pixelmaiamben: I think I read you saying you have it on "auto update"? This feature works best with dircache which the Sansa doesn't have, I've been told that it *should* work without it too, just terribly slow. But I've never tried and maybe there is something wrong with it?
01:08:31pixelmaif I'm not terribly mistaken...
01:08:51 Join bytie [0] (
01:09:02 Quit midgey ()
01:09:16 Quit bytie (Read error: 104 (Connection reset by peer))
01:09:23 Join midgey [0] (
01:10:00iambenpixelma: i will try without auto-update, it certainly isnt a feature i need
01:10:22iambensince i only add files every few weeks, and will probably just init the db from scratch every time
01:10:23 Join bytie [0] (
01:11:18pixelmamight be interesting to watch it for a while and compare, maybe it's really a bug
01:12:11iambenits not the easiest to reproduce, but usually i can get it to commit the db an extra time if i just reboot a few times after an init
01:12:51***Saving seen data "./dancer.seen"
01:15:13 Join jhulst [0] (n=jhulst@unaffiliated/jhulst)
01:15:17 Quit bytie (Read error: 104 (Connection reset by peer))
01:16:55 Join bytie [0] (
01:18:17zichoHow can i specify Font color when i create a rockbox theme?
01:18:46 Quit bytie (Read error: 104 (Connection reset by peer))
01:19:01 Join bytie [0] (
01:19:06 Quit bytie (Read error: 104 (Connection reset by peer))
01:20:26pixelmayou specify the foreground colour
01:23:47 Part safetydan
01:25:36 Quit Gnu47 ("I'm off to stalk my Pah'Mach'kai")
01:26:21 Join SkinInd95 [0] (
01:26:55pixelmain your config file there should be a line with it (at least that is explained in the appendix of the manual) you could also have a look at the themes that are shipped with rockbox and how it's done there...
01:29:19 Join Rob2222 [0] (
01:30:55 Quit BigBambi (Read error: 113 (No route to host))
01:32:39 Quit SkinInd95 (Remote closed the connection)
01:32:45*zicho is away: sleep
01:35:27 Quit GodEater_ (Read error: 110 (Connection timed out))
01:35:40 Quit kkurbjun_ ("leaving")
01:36:20 Quit donutman25 ("ChatZilla [Firefox]")
01:38:32 Join alienbiker99 [0] (
01:39:14 Join donutman25 [0] (n=dagni@
01:44:17 Quit krazykit ("leaving")
01:44:20 Join krazykit [0] (
01:46:50 Join Pyromancer [0] (
01:48:00 Part toffe82
01:57:37rashereek, downloads patches off the wiki!
01:58:27preglowi thought we had that fixed
01:58:34preglowthat's an absolute no-no anyway
01:58:36rasherDon't think so
01:58:58preglowi'm pretty sure i raised that issue in the past some time
01:59:29preglowif put a virus there, we could have ourselves a whole botnet compiling builds for us, though :P
02:01:16rasherYeah, seems really bad
02:02:11 Quit chandlerc (Remote closed the connection)
02:08:32linuxstbrasher, preglow: We should ask Zagor to move them...
02:08:39 Quit DataGhost (Nick collision from services.)
02:08:45 Join DataGhost [0] (
02:13:48rasherlinuxstb: with any luck, this should be enough
02:16:10 Join JdGordon [0] (
02:18:48JdGordonhey kkurbjun
02:21:18 Quit Pyromancer ("Leaving")
02:23:13 Quit Arathis ("Bye, bye")
02:25:50 Quit kclaf (Read error: 110 (Connection timed out))
02:27:59 Join dxm [0] (
02:28:19dxmHave a PT-BR version of rockbox?
02:29:32 Part pixelma
02:30:06 Join WalkGood [0] (
02:33:50 Join karashata [0] (
02:33:51 Join toffe82 [0] (
02:34:19 Quit Thundercloud (Remote closed the connection)
02:36:35 Join Pyromancer [0] (
02:38:21preglowwoot, from 255% realtime to 276% realtime on my wma testfile now
02:38:37preglowwith an even bigger difference under real use, thanks to less dsp use
02:39:44preglowalmost never boosts :P
02:39:56iambenand thats on coldfire?
02:40:07preglowwith the output buffer in iram, it won't boost
02:40:18preglowthis change will benefit arm too
02:42:38 Quit Toxicity999 ("Leaving")
02:43:45 Quit BHSPitMonkey (Remote closed the connection)
02:44:05preglowand gnight
02:45:16 Join midkay [0] (n=midkay@rockbox/developer/midkay)
02:45:21 Part dxm
02:45:59 Join Toxicity999 [0] (n=bryan@unaffiliated/Toxicity999)
02:46:49 Quit Toxicity999 (Read error: 104 (Connection reset by peer))
02:48:22 Join Toxicity999 [0] (n=bryan@unaffiliated/Toxicity999)
03:00:50 Quit Ebert ()
03:02:50 Join Soap [0] (n=Soap@rockbox/staff/soap)
03:03:07 Nick idnar_ is now known as idnar (i=mithrand@unaffiliated/idnar)
03:05:20 Join Pyromancer_ [0] (
03:05:20 Quit Pyromancer (Read error: 104 (Connection reset by peer))
03:05:40 Join BHSPitMonkey [0] (n=stephen@unaffiliated/bhspitmonkey)
03:07:53 Join XavierGr [0] (
03:08:01 Quit XavierGr (Remote closed the connection)
03:08:28 Join Rob222241 [0] (
03:09:41 Join Chronon_ [0] (
03:12:54***Saving seen data "./dancer.seen"
03:23:50 Part WalkGood
03:25:48 Quit Rob2222 (Read error: 113 (No route to host))
03:32:39kkurbjunhey JdGordon
03:33:08JdGordonhowsit oging>?
03:33:29kkurbjunpretty good, how about yourself?
03:33:32JdGordonim having problems with the touchpad int he regular build.. the irq seems to only happen once! but works fine in the bootloader
03:33:48kkurbjunare the rest of the interrupts working ok?
03:34:05JdGordonthe current_tick keeps increasing so i tinhk so
03:35:02kkurbjunand the failure you're seeing in the main build is?
03:35:44JdGordoni've added a counter in the gio14 irq and splash it in action.c when it changes, it never goes over 1
03:37:09JdGordon is my current diff... that code works 100% fine in the bootloader :(
03:37:36JdGordonhave you had any luck getting the remote to work again?
03:37:56kkurbjunit's working pretty well in the bootloader build, I havn't tried the main build
03:38:06kkurbjunI should fix the keymap so I can actually try it
03:38:21JdGordonwith svn code? or some local changes?
03:38:30kkurbjunwith the svn code
03:38:31 Quit Mouser_X (Read error: 104 (Connection reset by peer))
03:38:52JdGordondamn, svn doesnt work here
03:39:23kkurbjuntry setting that count to volitile
03:39:48kkurbjunyou can't push up and see the address chang?
03:40:02JdGordonhang on, ill try again
03:40:28JdGordonoh, if you have a sec, can you build the svn gbootloader and see if the cursor follows your touch prperly?
03:40:45 Join ramon8 [0] (
03:40:53 Join Mouser_X [0] (n=Mouser_X@
03:41:02kkurbjunyou had done some work on the clock right?
03:41:27kkurbjunoh, nevermind then.. I'll check here
03:42:40JdGordonyep, remote buttons arnt working
03:44:19JdGordonhmm... woops
03:44:21JdGordonno they are
03:44:51kkurbjunwhat was wrong?
03:45:05 Quit Chronon_ ("User pushed the X - because it's Xtra, baby")
03:45:09JdGordonreturn btn; in the button_read_device before the heartbeat :p
03:45:40kkurbjunwas that a change you did?
03:45:40*JdGordon feels stupid
03:46:02JdGordonim sure i tried it without that though
03:46:18kkurbjunno worries, did you see what held me up so long on the ata interface :)?
03:46:39JdGordoni dont remember the details
03:47:35kkurbjunI forgot to add an offset that was defined right above everything else - all that was missing was an x +
03:48:27JdGordonhehe, yeah that sucks :p
03:48:47kkurbjunyep, cursor works great on the bootloader build
03:49:07JdGordonive started on the kyemaps for the remote, im copying the player keymap which is similar
03:49:40kkurbjunso no calibration is needed for the screen I guess? Or are you already doing that in the driver?
03:50:03JdGordonno, hopefully those values can stay hardcoded
03:50:17kkurbjunthey work perfect for me
03:50:53kkurbjunso you say that the gpio interrupt only occurs once?
03:51:20kkurbjunin the main build, what is that interrupt supposed to be triggered by?
03:51:55JdGordontouch screen data ready
03:52:00JdGordonsame as bl
03:52:13kkurbjunand the other chip triggers it?
03:52:44JdGordonI couldnt find anything in the init which was being done for BOOTLOADER and not otherwise
03:54:13kkurbjunspi is polling right now right?
03:54:46kkurbjunwell, I guess that pin triggers it
03:55:28 Quit Nico_P (Remote closed the connection)
03:55:49kkurbjundid you try making that count volitile?
03:56:03JdGordonill do that now
03:56:54 Quit BHSPitMonkey (Remote closed the connection)
03:57:51JdGordonno change
03:58:18 Join XavierGr [0] (
04:00:22 Join psycho_maniac [0] (
04:03:02JdGordonkkurbjun: yeah, remote buttons arnt working in the main build either
04:03:59 Quit XavierGr (Remote closed the connection)
04:04:14JdGordoncould it be the stuff in crt0.S?
04:04:30JdGordona fre defined(BOOTLOADER) blocks
04:05:19kkurbjuncrt0.S.. I don't think do
04:06:06kkurbjunthe stuff that's bootloader specific is small, and it's mainly to copy it to the right location
04:06:47 Join Calcipher [0] (
04:07:25kkurbjunis uart_init and spi_init called in the main build?
04:08:08JdGordonnope !
04:08:38kkurbjunso in main.c I think they belong
04:08:38 Quit Mouser_X (Read error: 104 (Connection reset by peer))
04:08:50kkurbjunyou could call them in a system_init call or something
04:09:34Calcipherso any good stuff happen since the other night?
04:09:55JdGordonre what?
04:10:06 Join Mouser_X [0] (n=Mouser_X@
04:10:15Calcipherany screen rotation on e200's?, USB? more voice menus? any developments in those categories?
04:10:20JdGordonkkurbjun: it works :D
04:10:29kkurbjunjdgordon, where did you add them :)
04:10:30Calcipheroh hey jd
04:10:31JdGordonCalcipher: no, no, no,
04:10:44Calcipheraw haha, thanks for filling me in
04:10:47kkurbjunok, let's stick with that
04:10:48JdGordonafter enable_interrupts()
04:10:59Calcipherthought that one guy was close on the usb the other morning
04:11:10kkurbjunand pull them out of mrobe500
04:11:14Soapclose doesn't equal a few days.
04:11:18kkurbjundo you want to do a quick commit of that?
04:11:21Calcipherwhat are you guys focusing on right now?
04:11:33JdGordonkkurbjun: yeah ok, in svn in a minute
04:12:13kkurbjuncan you navigate the menus now with the remote?
04:12:27Calcipherah, the mrobe, hard to find now that people can run linux on it
04:12:53JdGordonkkurbjun: i tinhk so :) i have some bedug slapshes happening whihc is killing it
04:13:33kkurbjunCalcipher: I just recently bought mine - they are pretty inexpensive, at least when I got it
04:13:53kkurbjunI guess they are a bit rarer though
04:14:17JdGordonbah, bedug :p
04:15:43JdGordonand now... time to test the touch interface for real :D
04:15:56psycho_maniacgood luck
04:16:11JdGordonit freeking works :D
04:16:14Calcipherwow nice you guys got that working
04:16:28kkurbjun:), great, thats awesome
04:16:44JdGordonwe have to put a bigger default font on though
04:16:56JdGordonmy config has it with jackash-16 and its still tiny
04:16:56Calcipherhe he
04:17:23Calcipherso thats running widescreen since thats the original orientation of the mrobe display, correct?
04:17:46kkurbjunit's currently landscape
04:17:57kkurbjunsince that's the way the hardware is setup
04:18:02 Join curtmack [0] (
04:18:11Calcipherdamn thats freakin impressive, I doesn't sound like it took you guys long to get it working
04:18:26kkurbjunthere's still alot to do
04:18:31JdGordonits portarit.. not ladnscape
04:18:46kkurbjunsound is still an unsolved mystery.. oops
04:18:53curtmackSince I'm sure you guys get this alot, I'll cut it short: blah blah blah, rockbox not working, blah blah blah
04:19:11curtmackalso my ipod is apparently stupid and Menu+Select rebooting no longer works
04:19:31psycho_maniaccurtmack: are you serious? how can you expect to get help when you say that.
04:19:40curtmacknever mind
04:19:43curtmackgot it working again
04:19:49curtmackwell, got it in disk mode at least
04:20:14JdGordonkkurbjun: you want me to commit my keymap file which is partly usable?
04:21:00JdGordonits empty except the standard context table
04:21:06curtmackwait a second
04:21:07curtmackI see
04:21:15kkurbjunyeah, definitely, reduces repeat work. I don't see why we shouldn't be committing alot to keep us all on par with the development
04:22:03JdGordonill hopefully have the touch interface code in svn this arvo, or tomorow night
04:22:20kkurbjunjdgordon, if I rotated the screen, how difficult would tat be for you to readapt?
04:22:30JdGordonnot at all
04:22:32 Quit jhulst (Read error: 104 (Connection reset by peer))
04:22:47JdGordonthe touchscreen -> pixel values would need fiddling, but thats it
04:22:54kkurbjunok, I'm going to work on getting that done, I'm tired of tilting my head
04:23:37JdGordoni tinhk its better portaright though
04:23:51kkurbjundo you use the stand much?
04:23:56JdGordononly for usb
04:24:16JdGordonruntime rotating would be amazing to have :)
04:25:06kkurbjunhmm.. I'll try and make it as flexible as possible.. I think that would be quite a bit more work and could potentially slow it down a bit
04:25:29JdGordonits a pipe dream for all the targets
04:25:50JdGordonI tihnk the eaisest thing i we offer 2 builds when we are ready to go
04:26:17Calcipherhm, as you know, this matter interests me he he
04:26:34Calcipherany way to slide a rotation my way heh
04:27:13CalcipherI bet if you work on that, figuring it out for the sansa would be a breeze
04:27:44JdGordonit would be all or none really
04:27:53JdGordonso yeah, once one target gets it they alll will
04:28:06 Join jhulst [0] (n=jhulst@unaffiliated/jhulst)
04:28:06JdGordonthe problem is that the drawing code needs to know the dimensionsat compile time
04:28:43JdGordonok, shower and lunch time i tihnk :p
04:29:01kkurbjunsee ya, did you commit that keymap?
04:29:13kkurbjunoh, yep, I see it
04:29:16kkurbjuntalk to you later
04:29:48Calciphernice, I like
04:30:08 Quit miepchen^schlaf (Read error: 110 (Connection timed out))
04:30:27 Join miepchen^schlaf [0] (
04:31:11curtmack okay
04:31:44curtmack"The device 'Generic Volume' cannot be stopped right now."
04:31:52curtmackShould I just unplug it?
04:41:09 Quit curtmack ()
04:48:49 Quit donutman25 (Remote closed the connection)
04:48:56 Join Rob2222 [0] (
04:57:22SoapOh, he's gone.
04:58:49psycho_maniaci unplug mine all the time when it says that on my ipod. i look for disk activity on the screen and if i see none i pull the plug
05:00:55 Join toffe82_ [0] (
05:01:16jhMikeSif you don't have delay write enabled, you can just pull it
05:02:39psycho_maniaci must not have that enabled.
05:04:37jhMikeSit's not on by default usually which is good. one USB driver turned it on by default and I didn't know and I had to reformat because of it.
05:05:35psycho_maniacwhoops. where you find this setting?
05:06:15 Join BHSPitMonkey [0] (n=stephen@unaffiliated/bhspitmonkey)
05:06:59 Quit toffe82 (Read error: 104 (Connection reset by peer))
05:07:22 Quit Rob222241 (Read error: 110 (Connection timed out))
05:09:12jhMikeSIN the device manager under disk drives. Goto the properties and select the policies tab.
05:10:01jhMikeSthe device must plugged and drive mapped
05:12:56***Saving seen data "./dancer.seen"
05:15:57 Quit bb (Nick collision from services.)
05:16:02 Join bb_ [0] (n=bb@unaffiliated/bb)
05:19:20 Join RoC_MasterMind [0] (
05:24:10 Join XavierGr [0] (
05:27:18JdGordonjhMikeS: im a drongo, we figured out why it wasdnt working :p
05:29:07kkurbjunJdGordon: do the file names show up properly when browsing for you?
05:29:22JdGordonwhich filenames?
05:30:08kkurbjunwhen I go to choose a font I can't see the names, and if I browse the player, there are icons next to where filenames should be, but I can't see the selection
05:30:22kkurbjuntry selecting a font from the theme option menus
05:31:00JdGordonoh! i thought that was my fault with the list changes for the touchpad.. folders are fine, but no, i cant see files at all
05:31:39kkurbjunI wonder if an lcd function is not working properly?
05:32:07kkurbjunhow are file names and the selector drawn in the file browser?
05:32:33JdGordonsame was as the rest of the items
05:32:49JdGordoncould be update_rect()
05:33:21JdGordonthat doesnt make sense that the menus show but their items dont
05:36:45 Quit midgey ()
05:37:01 Join midgey [0] (
05:37:40jhMikeSJdGordon: what?
05:38:07kkurbjunJdGordon: I think it is in lcd_update_rectangle, I think the partial width updates are broken
05:38:18 Quit midkay (Read error: 113 (No route to host))
05:40:37kkurbjunjhMikeS: I think he was referring to the interrupt problems that were happening with the main build for the M:Robe
05:41:08JdGordonjhMikeS: we forgot to init them :p
05:41:40jhMikeSJdGordon: I have found that to be helpful :)
05:41:55JdGordonhaha, yeah
05:46:25 Join Doomed_ [0] (
05:46:25 Quit alienbiker99 (Read error: 104 (Connection reset by peer))
05:46:26JdGordonkkurbjun: we gotta come up with a faster way to boot rockbox!
05:46:33JdGordongoing through the of is painful :p
05:47:08kkurbjunI agree, someday I would like it to be in flash
05:47:08 Quit amiconn (Nick collision from services.)
05:47:11 Join amiconn_ [0] (n=jens@rockbox/developer/amiconn)
05:47:17JdGordonand yeah, the problem is definatly in lc_update_rectangle., i changed it to just call lcd_update and its working fine
05:49:23JdGordonhmm... its not the only problem though
05:49:34JdGordonbrowse fonts still doesnt work, but the file browser does
05:49:43JdGordonand all the settings
05:51:18 Join aliask [0] (
05:59:54 Quit XavierGr ()
06:02:27 Join KeyLime [0] (n=KeyLime@
06:03:10KeyLimeknow of any good games for rockbox
06:03:24KeyLimeall i have are the ones it came with
06:03:39aliaskKeyLime: Gameboy emulator?
06:03:57aliaskAlso, the ones it comes with are pretty much your only option, unless you want to compile yourself
06:08:34jhulstI filed a patch for a bug reported, FS7831, are there any thoughts or comments on it>
06:15:48 Join chris_mt [0] (
06:16:42 Quit KeyLime ()
06:34:39chris_mtcan anyone give me any insight into the timer_register function?
06:34:52Calciphergba emulation on rb?
06:36:42krazykitCalcipher, no, gba won't happen. most of the targets aren't near fast enough, and/or don't have enough buttons.
06:37:03Calcipherso rb is being emulated by the cpu, and then we have another emulation within that, can the players really process all that
06:37:19Calcipherah ok, haha
06:37:23 Quit midgey ()
06:37:37krazykitrockbox is RUN on the cpu
06:38:36chris_mtNES is coming along, but sound is a bit too much on my ipod
06:38:45Calcipherhm, thought I saw someone mention it was emulated
06:38:57chris_mtno, not even the sim is emulated.
06:39:02Calcipherwow, thats gonna be interesting
06:39:07Calcipherdouble dragon baby!
06:39:10chris_mtall cross compiled
06:40:01chris_mtmegaman 3 :D
06:40:07chris_mtmy old fave...
06:41:12Calcipherdon't wanna get kicked in the face for being off topic so I'll stop
06:41:48chris_mtso... timer_register, anyone?
06:42:26 Quit lazka (Remote closed the connection)
06:42:37Calcipherwhat function does that perform?
06:42:55chris_mtregisters a callback function at a regular interval
06:43:25chris_mtIt takes a cycles parameter, but looking at the firmware source, it's hard to tell what unit of time the interval should be.
06:43:40Calcipherlike for screen refresh?
06:44:06Calcipheror just any calculations
06:44:09chris_mtor emulator vsync, buffer refresh, you name it...
06:44:28 Quit karashata (Read error: 110 (Connection timed out))
06:44:39chris_mtIn this case it's for the NES emulator - trying to emulate the NES timing per scanline
06:45:29Calcipherah I get ya
06:45:57chris_mtthe callback function could do anything you want. like update the screen or check the buttons, etc.
06:46:06Calcipherdamn so its probably very likely the unit would depend on the target right?
06:46:23chris_mtI think it might but I can't confirm that.
06:46:41 Join karashata [0] (
06:46:41chris_mtthe metronome plugin uses a constant called TIMER_FREQ
06:46:49chris_mtbut when I use that I get strange results.
06:46:54Calciphersurprised with all the previous dev work that isn't known
06:47:10chris_mtthe emu runs a zillion miles an hour in the sim, but slow on my ipod.
06:47:26chris_mtso I think I'm setting the timer too short
06:47:48chris_mtand the ipod target is not fast enough, so it's taking multiple timer intervals to draw scanlines.
06:48:02chris_mtbut I want the sim to run at the right speed.
06:48:31chris_mtso who knows... I'll probably just tinker with it until I get an answer
06:48:41chris_mtdo you run rockbox?
06:48:50Calcipheryes I do
06:48:53chris_mton what?
06:49:12Calcipheron a Sansa e280r
06:49:36chris_mtaaah. My girlfriend has an e260 and I've messed with it a bit. pretty cool.
06:49:42chris_mta bit slower than my ipod though.
06:49:54chris_mtnot that it matters for music.
06:51:06Calcipherwouldn't it depend on the type of Ipod
06:51:23chris_mtyeah it might because they use different processors.
06:51:28Calcipherwell we're talking about targets that can atleast do video anyway
06:51:45chris_mtyeah, like the video ipod is a dual core proc.
06:51:51chris_mtnot sure what others are?
06:52:04Calcipherdamn, had no idea
06:52:09kkurbjunJdGordon: font browsing still doesn't work?
06:56:59Lloreanchris_mt: The Sansa e200s are the same processor speed as the modern iPods.
06:57:07Mouser_Xchris_mt: If you're working on 2911, I'd like to know if you have any ideas on why it's not working on the Gigabeat, or what I can do to get it running.
06:58:22Lloreanchris_mt: The iPods use PP5020, 5021, and 5022 and the Sansa is PP5024 if I recall the numbers correctly. But we clock them all to the same speed.
06:59:12Calciphernice to know, might make work easier for chris_mt
06:59:56chris_mtmouser_x: you mean the infones thing?
07:00:05chris_mtwhat's it doing?
07:00:16chris_mtor not doing?
07:00:19Mouser_XWhen I compile, it throws lots of warnings.
07:00:27chris_mtyeah, it does that.
07:00:34chris_mtbut does it compile successfully?
07:00:44chris_mtthe port is not perfect.
07:00:48chris_mtbut it's functional.
07:00:50Mouser_XWhen I have it on my Gigabeat (not sim), and I try to play a ROM (all I've tried is Zelda, so far), it says "loading" and then dies.
07:01:16Mouser_XWell, it does make a *.rock for it, so I assume that it completes the compile...
07:01:32chris_mtyeah, it's compiling.
07:01:52Mouser_XI did this with a build from Sunday.
07:02:04chris_mtI'd say try another rom.
07:02:13Mouser_XI should update, and try again, I guess... Or that.
07:02:16chris_mthave you tried the windows build of infones?
07:02:23Mouser_XNo, I haven't.
07:02:31chris_mtif it's the rom, that should behave about the same way.
07:02:56chris_mttiming and sound is defintiely off in the emu because the targets aren't fast enough.
07:02:57Mouser_XWell, it's from the GoodNES set. I know it's a confirmed good ROM, at least.
07:03:10chris_mtbut I'm trying to tweak the source to make it playable
07:04:06Mouser_XI suspect that the Gigabeat is fast enough to play both the game, and sound (though, you'll probably have to "dumb down" the sound, since the NSF codec takes a fair amount of CPU).
07:05:11Mouser_XThe sound in Rockboy isn't that good. However, all things considered, it's pretty good. I mean, I'm playing Gameboy games on a MP3 player, with sound!
07:05:19chris_mtInfoNES has 3 quality settings but only the highest sounds anywhere near decent.
07:05:27chris_mtthe others have some timing problem
07:05:51chris_mtbuffering problems - but I'm really new to rockbox dev
07:06:30chris_mtisn't gigabeat something like 200mhz?
07:06:57Mouser_XI have Zelda, River City Ransom, Rockman X (could be a homebrew), Startropics, and Ultima - Exodus on my Gigabeat. What do you recommend I try?
07:07:08Mouser_XYes, 300 mhz for the Gigabeat F.
07:07:09chris_mtdunno, whatever...
07:07:12Mouser_X533 for the S.
07:07:14chris_mtI haven't tried any of those.
07:07:32psycho_maniacMouser_X: 533mhz for the S?
07:07:32chris_mtI could try zelda on mine if I have it. let me check...
07:08:10chris_mtHA! zelda does the same thing on ipod
07:08:10Mouser_XHey! River City Ransom works.
07:08:20Calcipherso what about the cpu speed on the newer ipods and the sansa playerS?
07:08:34Calcipherfreakin sweet
07:08:38chris_mtI think my 5.5g is only 80-90mhz
07:08:51 Join ptw419 [0] (
07:09:04Calcipherdamn thats pretty efficient then
07:09:07Mouser_Xchris_mt: Perhaps some form of a compatibility list should be made? It would have been nice to know that Zelda would kill it...
07:09:29chris_mtI haven't gotten that far yet.
07:09:40Mouser_XHmmm. I suppose that's true.
07:09:44chris_mtI've only been testing on ipod and not really coordinating with miket
07:10:03chris_mtso I'm not sure how what I'm doing meshes with what he's doing.
07:10:56LloreanCalcipher: All Rockbox supported iPods and the Sansa are 80mhz.
07:11:11LloreanCalcipher: Except 1st through 3rd generation, I'm honestly not sure what we have the PP5002 targets clocked at.
07:11:35Mouser_Xchris_mt: Hmmm. The Gigabeat controls need work...
07:11:55chris_mtyeah, I changed the ipod to work like rockboy
07:12:03chris_mtbut I'm not sure what the controls are like on gigabeat
07:12:11chris_mtI'll have to try it on the sim.
07:12:19LloreanMouser_X: There used to be a patch for Rockboy sound that made it glitchless (I won't say perfect because I didn't compare it to a gameboy, but no pops or noises) but it only ran full speed on the Gigabeat
07:12:20chris_mtsansa was screwed up too
07:12:37LloreanMouser_X: I think the current Rockbox sound was an alternative to that which worked better on other targets too, a compromise.
07:12:56LloreanEr, "Rockbox" was supposed to be "Rockboy" in that last sentence
07:12:57***Saving seen data "./dancer.seen"
07:13:15Mouser_XLlorean: Cool... Too bad it couldn't be implemented based on abilities, or something...
07:13:26Mouser_XThere is a GBS codec, so that's nice.
07:13:34Mouser_X(It needs work though. Quite a bit.)
07:13:38kkurbjunthat patch still exists.. I have a fix that I've not committed yet
07:13:46kkurbjunin some form I guess
07:14:28 Quit kkurbjun ("Leaving.")
07:14:58 Join MethoS- [0] (
07:15:03LloreanMouser_X: It should be easy to check if the define for the Gigabeat is set, it's more a question of whether someone's willing to get it back into shape, etc.
07:16:21 Join Lambuntu [0] (
07:16:29 Join midgey [0] (
07:17:24Mouser_XHmmm. I should learn C or something...
07:18:08chris_mtwouldn't hurt
07:19:22 Quit jhulst (No route to host)
07:19:31Mouser_XUh, how do I quit?
07:19:39Mouser_XI can't get out of it...
07:20:20 Join XavierGr [0] (
07:20:53chris_mtin the build you have you have to reboot
07:21:22Mouser_X(Not really...) So... When/how can I get a fix for that?
07:21:37chris_mtlearn c :D
07:21:53Mouser_XAlso, the graphics got mangled more and more the further I went into the game.
07:21:54chris_mtI set it up to quit with the hold button for ipod
07:22:00chris_mtdunno about that.
07:22:57Mouser_XWell, under the current circumstances, I don't think I'll be using it much at this time...
07:23:17Mouser_XI look forward to it later though.
07:25:23Calcipherhey I forgot to look into this when I noticed it, but I installed rb on my sansa about 3 days ago
07:25:52Calcipherand I just as a test tried the mpeg player plugin, using elphants test videos
07:26:22Calcipherand each one displayed green video garbage
07:26:52CalcipherI did hear sound on one of the videos, and I could tell the aspect ratios for each were correct
07:28:19psycho_maniacwhat did you use to convert the video?
07:29:35 Quit midgey ()
07:30:11CalcipherI just used the ones on under the mpeg player section in extras
07:30:53psycho_maniacbe more specific? maybe try to re encode it
07:31:29CalcipherYeah when I give it more of a try I'll let you guys know, would it probably be a fast process?
07:31:52CalcipherI'm probably going to sleep soon, so if its a quick process I can try now
07:32:16CalcipherI think I read that some folks use vlc to encode the videos
07:32:27CalcipherI use that as my media player anyway
07:34:24 Quit MethoS- (Read error: 113 (No route to host))
07:35:12LloreanCalcipher: And this happened with a build from Today?
07:35:43Calcipherno, about 2 or 3 days ago
07:35:47 Quit ptw419 (Read error: 110 (Connection timed out))
07:36:35LloreanPlease, don't bring things up unless you've confirmed they still happen on a current build.
07:36:47Calcipherhaven't touched my rb set up since I got it working, was afraid to ruin it
07:37:11Calcipherok, I'll remember to do that, sorry
07:37:58LloreanThings change quickly, and it's hard to tell if people are reporting something that was fixed earlier, or the fact that the earlier fix didn't work, or what.
07:38:27Calcipheryeah, or if one fix might have caused something else to break
07:38:53CalcipherI understand, things are definitely very lively when it comes to rb dev
07:44:18LloreanOne thing, mpegplayer may not work properly with voice enabled right now
07:44:36LloreanI'm not sure if that current issue has been addressed, and it could cause problems like you described.
07:45:39 Quit SirFunk (Remote closed the connection)
07:45:53Calcipherah ok, let me try with the option disabled, also if I wanted to update the plugin its a better idea to update the entire rb install instead, to be up to date
07:46:19LloreanYou should always update the entire Rockbox install, which is really just unzipping to your device.
07:46:58Calcipherwill that reset my user settings, like theme and font selections?
07:47:50chris_mtI think if you don't replace the .config files in .rockbox you'll be fine. Anyone else know for sure?
07:48:26karashataunless the configuration file has changed a lot, it won't wipe them out
07:49:13karashataI've only had mine ever wiped out once since I've been using Rockbox, otherwise only occasionally a feature that's been changed will revert to default
07:49:33karashata(like when they changed the line selector to have coloured versions as well)
07:50:13Calciphercool, thanks for the info
07:51:46karashataheh, it should be on the RB site somewhere, that's where I know it from
07:51:56karashatathat, and experience, of course...
07:52:06 Join pabs [0] (
07:53:05*scorche pokes markun
07:54:05Calciphergetting r15197 through rbutil
07:54:55Calciphernow I will re attempt the elephant dream test video playback
07:55:40 Quit BHSPitMonkey (Connection timed out)
07:57:10Calciphervery nice
07:57:22CalcipherLlorean you were correct to have me update the build
07:57:38CalcipherI'm looking at the test video clear as day right now
07:58:48Calcipheruh oh, now I can't exit video playback
07:59:54LloreanCalcipher: Never had that problem before. Sure you're pressing the right button?
07:59:58Calcipherhad to hard reset the player, holding down the power button
08:00:18CalcipherI tried every button
08:00:33 Join dez`` [0] (
08:00:49Calcipherusually pressing power takes me to the previous menu
08:01:02Calcipherbut it was not responding to anything
08:01:13dez``anyone know a good place to get a pacman rom for the 30gb 5.5?
08:01:37Lloreandez``: Rip it from an Arcade Board
08:01:44Lloreankrazykit: Please don't encourage copyright infringement.
08:01:59dez``did they are all protected by somekinda thing
08:02:08dez``arcade board?
08:02:27 Join BHSPitMonkey [0] (n=stephen@unaffiliated/bhspitmonkey)
08:02:40Lloreandez``: The only legal way to get Pacman ROMs is to own the arcade board, and copy the data off of it.
08:02:41krazykitLlorean, no? the wiki sure seems to encourage it. it links to google searches for the files.
08:03:08scorcheLlorean: i would say "google" would be fine
08:03:17Lloreankrazykit: That should be fixed then.
08:03:19dez``hey psst krazykit you know one?
08:03:25 Quit Lambuntu (Read error: 110 (Connection timed out))
08:03:34Lloreandez``: Do not ask how to break copyrights in here.
08:03:48Lloreanscorche: I'd say "telling people downloading it is breaking copyright" should be fine...
08:04:06scorcheLlorean: it *could* mean google for the way to do it, or even searching for a proper cable
08:04:12Calcipherwell I just duplicated the problem
08:04:12scorcheLlorean: well, of course...
08:04:31CalcipherI treid the video again, same non responsive controls issue
08:04:50LloreanCalcipher: Did you check in the manual for what button you're supposed to press?
08:05:24dez``i got a question
08:05:36Calcipherwell I guess I assumed it would still be the same
08:05:48Calcipherbut I tried them all, I'll check the manual now then
08:06:10LloreanIt should still be the same, I don't believe it's changed recently
08:06:21scorchedez``: if you have one, ask it
08:06:31dez``on the wiki, the screenshots of pacbox for the ipod photo and the video are sideways, is it going to be like that on the ipod or is it just the way the screenshot was took or something?
08:06:33LloreanBut when I asked if you were sure you were pressing the right button, you didn't answer in the affirmative.
08:06:55Calcipherok I checked and no those controls did not respond on my sansa e280r, with the latest build
08:07:45 Join ddalton [0] (
08:07:52 Part chris_mt
08:09:17 Part toffe82_
08:09:24 Part Llorean
08:09:32krazykitdez``, it means you're going to be turning your ipod sideways to use it
08:09:42Calcipherok, tried one more time, with the second elephants dream encode on the mpegplayer section of
08:09:42 Join Rob222241 [0] (
08:09:51dez``thanks krazy
08:09:52Calciphersame issue with the controls
08:11:00CalcipherI have to note that voice menus was active when i tried these, so I will test them with the feature turned off
08:13:00 Join [1]psycho_maniac [0] (
08:13:25 Quit psycho_maniac (Nick collision from services.)
08:13:25 Nick [1]psycho_maniac is now known as psycho_maniac (
08:13:47Calcipherah, after disabliing the voice menu features I have sound with the video AND it responds to controls correctly!
08:14:23CalcipherI made sure every voice menu option was disabled before trying this test
08:15:28ddaltonCalcipher: what sound?
08:16:20CalcipherI was testing the mpegplayer plugin
08:16:38CalcipherIt seemed to have issues when the voice menus options were enabled
08:17:03dez``does "PluginRockboy" work with all gameboy roms?
08:18:10 Join GodEater_ [0] (n=bryan@rockbox/staff/GodEater)
08:18:37psycho_maniacit should
08:18:38Calcipheranyone else want to try recreating that issue on theyre set up?
08:18:44Mouser_Xdez``: I think it's worked with all the ROMs I've tried.
08:18:45dez``k coo
08:19:11psycho_maniacwhat player do you have Calcipher?
08:19:26Mouser_XIt's an e2xx
08:19:37Calciphere280r yeah
08:20:08 Join homielowe [0] (
08:20:12ddaltonCalcipher: what about with voice file names and dirs on does it still work as expected?
08:20:53Calcipherwell I can try that now
08:21:41ddaltonok because it would be interesting to know.
08:21:50*ddalton waits
08:22:33dez``will PluginRockboy work on my ipod 5.5?
08:22:35Calcipherok, I enabled directory and file spelling options
08:22:45psycho_maniacdez``: yes
08:22:48ddaltondoes it work?
08:22:48Calcipherand the problem is reoccuring
08:23:08ddaltonhmmm ok so if you disable all talking but talk_menus what happens?
08:23:13Calcipherso, no sound playback, and controls are unresponsive
08:23:23Calcipheror irresponsive, which ever it is
08:23:37Calcipherlets try that now
08:23:55Calciphergotta hard reset the little guy
08:24:25psycho_maniacslam it agaisnt the ground ;)
08:25:32ddaltonfrom looking at the code I would expect it would work with talk menus enabled.
08:25:52Calcipherok, same thing happens, problem reoccurs
08:25:53dez``does anyone know a place where i can the the Rockboy from? i searched it but nothing important came up
08:26:06ddaltonok can you build your own rockbox?
08:26:31CalcipherI could if i knew how
08:26:34psycho_maniaci dont understand your question dez``
08:26:48dez``the rockboy plugin
08:26:50ddaltonok can you open a flyspray task stating your problem?
08:26:58dez``where can i get it from?
08:27:01psycho_maniacplease rephrase it so it is understandable
08:27:01ddaltonand then test a patch I will atempt to write?
08:27:03Mouser_Xdez``: It's part of Rockbox.
08:27:09Mouser_XDownload Rockbox.
08:27:12Mouser_XIt's in there.
08:27:23psycho_maniacrockboy is installed with rockbox
08:27:34dez``dude that sweeeet
08:27:37psycho_maniacwhy is it that evreything thinks its not part of rockbox?
08:27:47psycho_maniaci ment "that everybody "
08:27:49Mouser_XBecause they don't read the manual?
08:28:14 Quit Rob2222 (Read error: 110 (Connection timed out))
08:28:14dez``of course we dont :)
08:28:15psycho_maniaci mean a quick search for plugins or rockboy would tel you this.
08:28:21Mouser_XBecause the site(s) that lead them here say "It even has games and emulators which you can download!"
08:28:24dez``took me a couple of hours to figur out how to get to the in game menu in doom :)
08:28:46psycho_maniacMouser_X: when i sold my h120 with rockbox flashed on it i MADE my cousin read it before i would sell it to him.
08:29:09Mouser_XIt'd be nice if there was a way to view the manual from within Rockbox...
08:29:44psycho_maniacyeah. i was thinking that too. you could put the html manual on your player and compile the html viewer in your rockbox. if that is up to date now.
08:30:53Mouser_XThe wiki viewer patch (not the "dict" patch) converts the wikipedia XML file(s) to something that it can handle better.
08:31:32Mouser_XPersonally, I think this might be the way to go (convert either the HTML, or the PDF, to something more Rockbox friendly).
08:32:09GodEater_like text...
08:33:39Mouser_XWell, it'd be nice if the images worked (though, they'd most likely have to be resized).
08:34:06ddaltonCalcipher: still there?
08:34:15Mouser_XI use PDFs very rarely. This is probably a dumb question, but can they be converted to a text file?
08:34:25*scorche coughs at Mouser_X
08:34:34Mouser_X(I guess it'd make sense for them to be able to.)
08:34:47Mouser_XWell, excluding copy+paste.
08:34:49psycho_maniacMouser_X: ask google ;)
08:35:17Mouser_XI'd rather go to bed.
08:35:39Mouser_XGood night. (I should have left over an hour ago.)
08:35:46Calcipheryeah just reading up on flyspray so I can do this right
08:36:41ddaltondon't worry im opening it
08:41:10Calcipheroh your making the entry?
08:41:17ddaltonCalcipher: FS #7993. Just give me a few minutes to write this patch then could you please test?
08:41:33Calcipherso have you been able to reproduce the problem?
08:41:45psycho_maniacGodEater_: does your gigabeat mark up easily like the ipod does? i have noticable finger prints all over my ipod
08:42:15JdGordonSlasheri: you round?
08:42:45ddaltonit disables talk_menus so I want to test something hang on...
08:43:03GodEater_psycho_maniac: I keep both in rubber skins, so I wouldn't know
08:45:35 Join kubiixaka [0] (
08:45:35 Quit kubiix (Read error: 104 (Connection reset by peer))
08:45:57dez``yay got pacman running
08:46:37dez``but when i fo to "files" on my ipod in rockbox it doesnt show the .gb rom i got in there
08:46:57 Quit Isolinear ()
08:47:02psycho_maniacis it in the root of the player?
08:47:21dez``i made a new folder called gameboy roms
08:47:33psycho_maniacwell go to that folder. it should be in there.
08:47:46dez``nope :'(
08:47:57ddaltonCalcipher: Hmmm I can't test with out talking menus and file or dir names. But with out them enabled do you hear sound?
08:48:00ddalton(I am blind)
08:48:23psycho_maniacgo in the quick menu and set it to show "all" and tell me if it shows up then.
08:48:44dez``one sec brb
08:48:58Calcipherwith the voice items, each and every one of them disabled, the player functions normally
08:49:14Calcipherincluding sound playback and control response
08:49:31ddaltonso in mpeg player you hear sound?
08:49:35ddaltonwith them disabled?
08:50:03Calcipherwhat target are you testing on?
08:50:08ddaltonok the patch I will write will turn them off but not back on if this works. Then we just need to figure out the next part.
08:50:11ddaltongive me 20 mins
08:52:32Calcipherwell dalton, its 2:50 am here in NY and I have a network test tomorrow
08:52:47dez``how do i get to the "quike menu"?
08:52:57CalcipherI'm already messing my night/morning/day up by having stayed up so late
08:53:24scorchedez``: read teh manual
08:53:29 Join PaulJam [0] (
08:53:39dez``thats the thing, i did
08:54:44psycho_maniacor search the wiki
08:56:14 Join Llorean [0] (n=llorean@rockbox/administrator/Llorean)
08:56:47*scorche spots a manual error
08:57:18psycho_maniacso you changed the file view settings and now it shows up?
08:57:44*scorche opens up a tex file to see how easy it is to edit..
08:57:45ddaltonCalcipher: nearly done. Not sure if it will work though. I am using h300. What are you using?
08:57:59Calciphersansa e280r
08:58:05psycho_maniacok that probably means the file isnt supported
08:58:10dez``thanks mr.psycho :)
08:58:42Calcipherddalton did you get that, I have a sansa e200 series
08:58:44JdGordonjhMikeS: you round?
08:58:51ddaltonoh ok.
08:59:03ddaltondo you want me to make you a build?
08:59:09ddaltonor can you apply the patch yourself.
08:59:30Calcipherwhatever is fastest
08:59:39CalcipherI want to get to bed soon
09:00:02scorchefor the logs, if any manual maintainers are around (and i dont wake up and get it first), 5.13 ("Quick Menu") "The QUICK MENU screen
09:00:02scorcheis accessed with Long Submenuand exited with the same button.
09:00:51Calcipherddalton, so what will you do? will you prepare the build?
09:01:34 Join ender` [0] (
09:01:48dez``what counts as the "record" button on a ipod?
09:02:30dez``im stuck in pause in super bust a move then :(
09:03:06psycho_maniacdid you look up the controls in the manual?
09:03:19dez``but doesnt say
09:03:27dez``found it
09:04:37psycho_maniacdez``: this is not a instant message conversation. you dont have to tell us that you will be right back.
09:05:43dez``ok sorry geez for "flooding" or whatever you call it
09:05:54Lloreandez``: Well, there are channel guidelines. You DID read those, didn't you?
09:06:13ddaltonCalcipher: yes but that might take a while on cygwin. Sorry... but I might be able to test. or if you can stay up for another half hour it should be ready. The patch is almost done. But if it will compile is the next question. :-)
09:06:15dez``yup i think i brushed thu thos
09:06:17 Join spiorf [0] (
09:06:26 Join Isolinear [0] (
09:06:36Lloreandez``: Well, they're enforced here. Continually ignoring them is a quick way to get banned.
09:06:53LloreanYou might want to take a moment to read them more carefully.
09:07:07dez``my bad. sorry fellow people of this channel
09:07:15dez``where would i find them?
09:07:29Calcipherddalton, I'm sorry to say we started this project a little late, especially when I have to get up to take a test I paid $200 to take
09:07:33 Nick fxb__ is now known as fxb (
09:07:34jhMikeSJdGordon: yeah
09:07:39 Join petur [0] (n=petur@rockbox/developer/petur)
09:08:12JdGordonjhMikeS: have you got any idea why some lists would display their items and some wont? is something maybe missing in the lcd driver?
09:08:21psycho_maniacdez``: it shows this link when you first come to this channel.
09:08:22Calcipherafter I finish making a comment on a user interface orientation option request I think I'm heading to bed
09:08:28jhMikeSon m:robe?
09:08:29JdGordonthe lists that dont show items do show the scroll bar and icons and title though!
09:08:40JdGordoncrap, brb
09:08:46 Quit aliask ("ChatZilla [Firefox]")
09:08:48jhMikeSno font drawing?
09:08:52LloreanCalcipher: Due to the way a lot of internals work, a 90 degree rotation of the UI is really not a simple change.
09:08:57LloreanOr so I'm told.
09:10:56*ddalton is compiling
09:11:26ddaltonCalcipher: do you want a build or should I just test?
09:11:50Mode"#rockbox +o Llorean " by ChanServ (ChanServ@services.)
09:12:14CalcipherI can try a build if its done
09:12:29JdGordonjhMikeS: i thought it might be scroll, but that doesnt make sense, because it shows no text even if it doesnt need to scroll
09:12:38ddaltonmine is compiling so it will take about 30 minutes for your build
09:12:49JdGordonthe menus display fine, but entering any setting screen doesnt
09:12:59***Saving seen data "./dancer.seen"
09:13:00ddaltonthere are probably going to be errors so maybe tomorrow is best.
09:13:13dez``thank you Llorean. I have now finished reading the IRCGuidelines that you have sent me the link to. I will no longer ask "can i ask a question?" from now on. I promise to not use vulger this channel. So please dont ban me ;)
09:13:23CalcipherLlorean It sounds like the same principals used in the gigabeat patch can be applied to the players that have the sansa e200 series players screen sizes, after some ui positioning adjustments
09:13:58LloreanCalcipher: The Gigabeat patch requires that you keep two different versions of Rockbox on your player, a 4:3 and a 3:4 version, doesn't it?
09:14:12jhMikeSJdGordon: framebuffer copies should work it looks like
09:14:50Calciphersome people mentioned the fucntion being implemented in runtime, which is not even what I'm asking for, I know thats asking for alot more, a pre boot selection of orientation mode is just as good for me
09:15:34LloreanCalcipher: Unfortunately, it's incredibly unlikely a method that requires two different builds be made will be accepted in SVN.
09:16:24jhMikeSJdGordon: sounds like some corruption of something like FRAME gets changed or something
09:16:58jhMikeSI can only imagine without a device to poke
09:17:12CalcipherI would be happy with the option to use a separate 3:4 build if need be
09:17:38ddaltonCalcipher: what os are you on?
09:18:10LloreanCalcipher: Yes, my point was that it's unlikely to be included in the official version. The patch tracker is for things working toward inclusion.
09:18:57Calcipherah I see
09:19:18ddaltonCalcipher: it compiled...
09:19:25ddaltonwhat player I will make you a build
09:19:40 Quit dez`` ("Pull the pin and count to what?")
09:19:42CalcipherSansa e200 series
09:20:21JdGordonwell this just sucks :p
09:20:28Calcipherddalton how do i get your patch?
09:20:29JdGordonand im outta time, have a good weekend all
09:20:52ddaltonI will upload a build there if I can as well as the patch give me a few minutes
09:20:58 Join Zagor_ [0] (
09:21:05*ddalton Last comment is probably not true with cygwin :-)
09:23:13 Quit Zagor (Remote closed the connection)
09:23:21 Nick Zagor_ is now known as Zagor (
09:23:50*amiconn_ likes speedups, even if they're tiny
09:23:55amiconn_They tend to add up :)
09:24:01 Nick amiconn_ is now known as amiconn (n=jens@rockbox/developer/amiconn)
09:24:50*Llorean likes them too.
09:25:23amiconnI squeezed 3..4% more speed out of the filtering stuff for APE on coldfire
09:25:47amiconnThat is with the predictor still untouched (but in the works...)
09:26:04LloreanSo you're confident 3000 will reach realtime?
09:26:26amiconnVery probably, but it might struggle even with an optimised predictor
09:26:38amiconnI'm at 97% realtime now for -c3000
09:27:12LloreanVery close then
09:27:51LloreanWell, close to pure realtime decoding. Did we determine an approximate number for actual non-skipping playback?
09:27:56amiconnYeah, but you need around 120% to get no skipping even when buffering etc
09:28:36LloreanYeah, and lossless spends more time buffering than lossy to boot.
09:28:49amiconn-c2000 is now 135% realtime. Still a lot of boost, but plays nicely
09:29:02ddaltonCalcipher: Still a wake? Nearly done....
09:30:12Calcipherddalton yeah
09:31:18 Join pixelma [0] (i=pixelma@rockbox/staff/pixelma)
09:35:39ddaltonjust uploading give it about 10 seconds
09:36:56ddaltonok get it from the last comment. re the mpeg player stuff.
09:37:33Calcipherddalton thanls
09:38:05ddaltontry it out... I doubt it will work though. Since hardly anything I do works the first time :-)
09:39:11ddaltonCalcipher: could you send me one of your video clips so I could test it out? I don't watch video.
09:39:42Lloreanddalton: He's testing with the sample clip we make available, I believe.
09:39:54Calcipherthats correct
09:39:57ddaltonwhat is the "sample clip "?
09:40:01CalcipherI'll get him a link
09:40:15ddaltontry that build... because I had problems to
09:40:34JdGordondont bother
09:40:40JdGordonyouve got an infinite loop there
09:41:28ddaltonhis using one of them videos?
09:41:35JdGordonyou should have got a warning do_shutup will always be non 0
09:41:58ddaltonwait I will remove that and it has a statement return 0 doesn't it?
09:42:48ddaltonmaybe I did...
09:42:55ddaltonsee line 326 in talk.c
09:43:33CalcipherOk ddalton, I tried again, with only the voice menu feature enabled, and no other voice features enabled, and the no sound and no controls issue persists
09:44:40 Quit JdGordon ("Konversation terminated!")
09:44:49ddaltonCalcipher: what about just voice file?
09:45:35Calcipherlets see
09:45:56ddaltoncan you turn off the video once it doesn't play?
09:46:07 Nick fxb is now known as fxb__ (
09:46:11ddaltonso hit select on your mpeg file then stop and see if it talks
09:46:22 Join pondlife [0] (n=Steve@rockbox/developer/pondlife)
09:48:08Calcipherddalton do you mean the voice filenames that can be set to number or spell?
09:48:35 Join kclaf [0] (i=kclaf@
09:49:50PaulJamI was wondering if some dev who uses the randon folder feature could have a look at FS #7077 ?
09:50:12ddaltonCalcipher: yes
09:50:34ddaltonpondlife: you closed my task. The one you said doesn't look right to me.
09:50:37ddaltonsee my comment
09:51:41Calcipherddalton the problem persists even then
09:51:55ddaltonwell the patch pondlife said won't fix it.
09:52:21ddaltonCalcipher: so what happens if you go into mpeg player with my build ghen exit?
09:52:28ddaltonis voice turned off.
09:52:43CalcipherI see sound playback conflict making sense, but why do the controls not respond due to that option being enabled, very strange
09:53:52Calcipherddalton I tried once with voice menus enabled, then once with voice menus disabled, but voice file names set to spell
09:54:09Calcipherand both times the problem was recreated
09:54:33ddaltonso you basically need voice off?
09:54:51Calcipheryes, completely it seems
09:55:40 Quit psycho_maniac (" I love my HydraIRC -> <-")
09:55:54amiconnjhMikeS: I'm not sure whether this has to do with the dualcore stuff, but if you run test_codec on PP5020 and don't let it run undisturbed (e.g. you press buttons to light up the backlight, plug the charger or similar stuff) there is a high chance that test_codec will crash (got prefetch aborts several times)
09:56:04ddaltonsorry... I was trying to start my screenreader
09:56:12amiconnAlso, it loves to lock up the device after quitting test_codec
09:56:19amiconnDoesn't happen on single core...
09:56:33Calcipherddalton I just tested it yet again with no voice features enabled and the playback has sound and reacts to controls normally again
09:56:35ddaltonwait... if I execute my function at the start of mpeg player it should hopefully work.
09:57:58Calcipherddalton so will exiting the player reinitialize the voice features?
09:58:39Calcipherhas anyone else here been able to recreate this problem!? I need a substitute for tonight
09:58:57amiconnHmm, wait. Could be that this was just due to voice being enabled...
09:59:12amiconntest_codec should really disable it like mpegplayer does....
09:59:15ddaltonfrom my code it should 1. Start mpeg player two turn all of voice off 3 exit when you exit it.
09:59:23jhMikeSamiconn: well, I can make a single core build of anything and see what's up though I didn't have unusual trouble with it myself on dc.
09:59:28ddaltonthis is just a test remember
10:00:56ddaltonCalcipher: just compiling... will post an update in a minute.
10:01:21pixelmaCalcipher: the green blocks in mpegplayer with voice enabled is known (and happens on other targets as well), unresponsive buttons is new to me
10:01:52Calcipherhm, only when voices are enabled
10:01:59pondlifeddalton: Your patch wouldn't help much.
10:02:08ddaltonpixelma: Im trying to fix that. My fix is to just turn all of voice off before we do anything in mpeg player.
10:02:09CalcipherI can play normally with the voice features disabled
10:02:13ddaltonbut I doubt it will work.
10:02:26Lloreanpixelma: Maybe with the changes to how IRAM is given to mpegplayer it's a new symptom of the same problem? I seem to recall voice being related to IRAM anyway?
10:02:29pondlifeChanging the global_settings does the same thing as talk_disable_menus(), but in a more horrible way
10:02:50ddaltonif it works then we can just add a call to a function that turns voice on before the return
10:03:10ddaltonok well lets see what happens...
10:03:16pondlifeddalton: If you change global_settings to disable voice, then someone resets their player during mpegplayer, they will not have voicing when it reboots!
10:03:29pondlifeYou need to *understand the problem* before you type any code.
10:03:34pixelmahmm... no
10:03:47ddaltonpondlife: ok im just trying to get it to work. then I will fix the little problems like that.
10:04:15pixelmaCalcipher is right, green blocks are gone (revision 15194) but I can't quit on my sansa c200
10:04:18pondlifeYou don't get my point.... don't code that way... first, find out what the problem is
10:04:25pixelmawith voice enabled
10:04:42ddaltonI know but we have to disable all voice including files and dirs.
10:05:06pondlifeWhat causes this? Is it IRAM use?
10:05:11ddaltonjust let me try this. if it works then I will not use global_settings just the standed functions.
10:05:24pondlifeBut the standard functions are already called!
10:05:32ddaltonhe said, however that it occurs with just voice file names on
10:05:44pondlifeFS #7939 puts them in the right place
10:05:59pondlifetalk_menus_disabled() should also disable filename voicing
10:06:07Calcipherpixelma can you help out and test this issue or give feed back, I really really gotta get some sleep
10:06:13pondlifeIt's not named very well - it should be talk_disable() really
10:06:24Calcipherhave a test I paid for tomorrow
10:06:33Calcipherand now its 4am
10:06:39pondlifeSleep well!
10:06:54GodEater_are we still waiting on some specific features in 7738 to be implemented before it gets commited ?
10:07:10LloreanCalcipher: Feel free to just go. It's a problem on all players, anyone can test, really.
10:07:11pondlifeddalton: With FS #7939, do you still get a problem?
10:07:17ddaltonpondlife: wait a minute...
10:07:20pondlifeI don't.
10:07:24pondlifeOn H340
10:07:29ddaltonok I give up...
10:07:39ddaltonI will just poste this version and try it myself.
10:08:00LloreanGodEater_: I believe we're waiting on people to test it on various iPods and pick good defaults for the settings. And decide about configurability.
10:08:07pixelmaand also no sound, I got one video that just stays black... Calcipher can just give feedback
10:08:21Lloreanpixelma: Dual core or single core target?
10:08:25pondlifeSorry, I'm not meaning to put you off, just to say you need to be sure that you know what it is that's broken before you attempt a fix.
10:08:34GodEater_Llorean: it looks like it needs syncing again at the moment
10:08:41pixelmaLlorean: that was on the c200, will try M5 too
10:09:03pondlifeAh, is it only a problem on dual-core?
10:09:12Lloreanpondlife: We'll find out. :)
10:09:18CalcipherOk, so atleast some other people have recreated the same problem
10:09:31ddaltonim just testing now...
10:09:35GodEater_ah no - I take it back - it just won't patch against Nico's MoB tree
10:09:56*ddalton goes for dinner
10:10:05Calcipherwell good luck on fixing that, see you all later, thanks for the help ddalton and all of you
10:10:21LloreanGodEater_: One thing is that I think a semi-general feeling is that it should not be "configurable", but having it configurable would help greatly in narrowing down good hard-coded settings.
10:10:51GodEater_personally, I've not fiddled with the config for it at all
10:11:02GodEater_I've left it at whatever it uses for default - seems to be perfect as is.
10:11:24LloreanSame here on my Nano
10:11:33LloreanAt least, with v12, haven't tested the latest.
10:11:38amiconnpondlife: Then it should probably be renamed as well, to talk_disable();
10:11:44 Join Tavnos [0] (
10:11:54pondlifeIt should, and it should also work properly :)
10:11:56pixelmapondlife, Llorean: the issue is there on my M5 too, even worse - Elephants Dream stays completely black while on the c200 I got video at least
10:11:57pondlifeI'm fixing now
10:13:01 Join Benoitb [0] (
10:13:16pixelmaand it needs a hard reset to get out of it (that's with r. 15194 too)
10:13:21GodEater_Llorean: do we know what ipods it's been tested on ?
10:13:26amiconnBtw, I wonder why we have talk_disable() and talk_enable(), instead of the more rockbox like talk_enable(bool enabled)
10:13:32TavnosHello, I'm working on a Remote WPS for my Iaudio X5. I found an error in the documentation, how does it works to correct it ?
10:13:36Lloreanpixelma: That would make sense, since video is handled on the other core, right?
10:13:41pondlifeamiconn: You are correct
10:13:42jhMikeSamiconn: seems test_codec wasn't updated with the extra check?
10:13:48LloreanGodEater_: Nano and Video? :)
10:14:41BenoitbLlorean: I'm snk4ever from the forum, you told me my message was in an innapropriate place, could you advice me on where it should be ? I don't want to cause trouble and disorder
10:14:46LloreanGodEater_: I suspect that it's the physical size of the wheel that makes the biggest difference.
10:15:24LloreanBenoitb: You posed the message as a "Here's something for you guy. Feel free to improve it." Rather than "I'm trying to make a video script, but I can't make it do this, any help."
10:15:31LloreanIt very much reads like you're just posting information.
10:15:57 Part Chronon ("Leaving")
10:16:01BenoitbLlorean: so I leave it where it is but change the content ?
10:16:18LloreanJust leave it.
10:16:20GodEater_Llorean: but if it's been tested on those - which have the two extremes of wheel size as I understand it, then surely it's now fine ;)
10:16:38LloreanGodEater_: Well, it has one default for "Nano" and one for "All other iPods" I *think* (I haven't verified this)
10:16:49GodEater_oh really - didn't realise
10:16:54LloreanBenoitb: But in the future, if you're posting a question, try to actually ask a question.
10:17:19TavnosDoes someone read me ?
10:17:21LloreanGodEater_: - use WHEEL_SENSITIVITY = 6 for iPod nano - use WHEEL_SENSITIVITY = 4 for all other iPods
10:17:31GodEater_ah ha
10:17:31LloreanTavnos: Submit a patch against the manual.
10:17:46LloreanGodEater_: Though I believe most of the wheels are *fairly* close to the same size?
10:17:57TavnosLlorean the error is on the wiki...
10:18:15GodEater_Llorean: I think you're right...
10:18:20LloreanTavnos: Then you just edit the wiki page. That's the whole point of it being a wiki.
10:18:30LloreanGodEater_: In which case we're once more back at "It's probably fine". :)
10:18:55LloreanGodEater_: My preference is that all configurability be hidden away behind the Debug menu so that people can still test different settings, but it's clear they're not for keeping.
10:19:02BenoitbLlorean: I'm not actually asking a question, I thought "It is something handy, I should make others benefit from it and as well some knowledgeable people might give me good advices to improve it for the best of everybody" I did not know it was a wrong way of doing things
10:19:09TavnosLlorean I don't have edit rights on the Wiki
10:19:25LloreanTavnos: It should tell you what to do to get edit rights, though.
10:19:33BenoitbMy idea was to post it on the wiki if it eventually came to a cleaner aspect
10:19:44Benoitbbut I thought it was not yet ready
10:19:59LloreanBenoitb: Then again, you have a question. "I would like to post this on the wiki, but it seems to have a limitation. Does anyone know how to fix these problems?"
10:20:15TavnosLlorean : I'd register, and now it says... Ask for rights on IRC chan, that's what I do :)
10:20:15Benoitbthanks for the formulation
10:20:21GodEater_Llorean: I've just looked, and to my eyes the ipod color's wheel is ever so slightly larger than that on the video
10:20:25LloreanTavnos: Alright, what's your wikiname?
10:20:29GodEater_which sort of surprised me - but the difference is tiny
10:20:35TavnosLlorean : NicolasPerrin
10:20:50LloreanGodEater_: Yeah, *probably* not a concern then. :)
10:20:56GodEater_I can't imagine it
10:21:07GodEater_I'd test on the ipod in question, but I think the owner would shoot me :)
10:21:28LloreanTavnos: Done
10:21:47TavnosThanks :)
10:23:44TavnosLlorean : I have another question... Why did you check that I read the documentation for the rights ?
10:24:36jhMikeSamiconn: I just had it mess up playback after using...single core build on H10. test_codec needs the state check added.
10:25:05jhMikeSdo you use voice at the same time?
10:25:23LloreanTavnos: Because you didn't simply say "I tried to register for the wiki and it told me to ask in here." Checking that you've read at least suggests that you know to use your real name, and so on.
10:25:37amiconnI often have voice enabled. Right now I disabled it for doing benchmarks on my H10
10:25:42Mode"#rockbox -o Llorean " by Llorean (n=llorean@rockbox/administrator/Llorean)
10:26:04amiconnWith voice enabled, I got those prefetch aborts. With voice disabled I can do benchmarks
10:26:57TavnosLlorean : ok no problems, I was just affraid to fault on something
10:28:39*jhMikeS tests with state check enables
10:30:09pondlifeI'm just knocking up a patch that makes the talk_disable(bool) interface generic, to disable all talking, internally.
10:30:28pondlifeJust testing it now, will pop on Flyspray in a short while
10:31:18pondlifeI wonder if PLUGIN_IRAM_STEAL (or somewhere similar) should auto-disable talking?
10:31:30pondlifeWe'd need a PLUGIN_IRAM_RELEASE to match though
10:37:54ddaltonpondlife: that would mess up the talking plugins when that is commited.
10:38:10pondlifeOnly those that use IRAM
10:38:25ddaltonwhat's iram
10:38:51pondlifeThat is, assuming IRAM use isn't compatible with talking... I don't know if that's always the case (now or going forward).
10:38:59jhMikeSamiconn: yes, much more reliable with that. np going into test_codec with music playing.
10:40:30jhMikeSguess I'll switch it back to dual and evaluate again
10:46:18 Quit karashata ("Leaving.")
10:49:35GodEater_is there some reason why the debug menu seems not to be using Jd's "new" menu code ?
10:49:56ddaltonamiconn: can you turn rec off when in the recording screen?
10:50:02jhMikeSsince when isn't it? he converted it.
10:50:10ddaltonso if I write some code to turn it off when a button is pressed?
10:50:13ddaltona certain one.
10:50:19ddaltonso it can talk...
10:50:21amiconnGodEater: Larger code. He tried it.
10:50:47amiconnddalton: No you can't. We already told you. Turning off recording would break prerecording.
10:51:42ddaltonI know but for a certain button. just for that one time your in the rec screen. So -enter ec screen -hit long rec for example stop recording -press something else start again
10:52:14 Part Benoitb ("Kopete 0.12.5 :")
10:52:45*jhMikeS thinks some distinctive beep system would be nice there.
10:53:11amiconnOn hwcodec there is
10:53:39amiconn(well, more like a burp, but it's the best we can do with the MAS while in recording mode)
10:53:46jhMikeSit's possible for swcodec. if some use the hp jack, I guess the piezo needs to work.
10:54:51amiconnThe MAS has quite good protection against plops. The only way I found to make noises in recording mode is to quickly toggle DAC power on & off...
10:55:18jhMikeSusing full voice during recording would be utterly impractical and imo dangerous to data
10:55:22amiconnAnd it's not-so-quickly because it requires i2c communication, and the MAS does a lot of clock stretching...
10:56:17 Join bluebrother [0] (i=aFNji7zx@rockbox/staff/bluebrother)
10:56:20amiconnRecording uses long & short "beeps" on hwcodec - a long one to signal start recording, a short one for stop and pause
10:56:56jhMikeSironically one target that doesn't record can do beeps with a timer output tied to the audio codec
10:57:29jhMikeSgigabeat - one timer out line is connected to an analogue input
10:57:31daurnimatorhas anyone expressed an interest in getting rockbox on the iphone?
10:57:50amiconnWell, on ipod we could use the piezo
10:58:06amiconnThat wouldn't require any fiddling with the pcm input/output
10:58:15jhMikeSthat what I was just saying, and required when the headphone jack is also an input
10:58:53amiconnBut what could we do on e.g. H10?
10:58:54jhMikeSactually, I think HW beeps should be preferred where they can be used instead of pcm buffer mixing
10:59:00amiconnOr H1x0/H300...
10:59:02linuxstbdaurnimator: I don't think anyone has seriously mentioned it - i.e. no-one is working on it.
10:59:13jhMikeSI can spit samples out on TX while recording
10:59:31daurnimatorwell, is anyone interested in it (thats a question for you alll... now)
10:59:44jhMikeSe200 does that already that for WYHIWYG monitoring :)
11:00:11amiconnCould you do that for coldfire too?
11:00:16Lloreandaurnimator: It doesn't make too much sense as Rockbox currently stands.
11:00:30jhMikeSsure, just change the audio output source and play some samples
11:00:42daurnimatorLlorean: true
11:00:55linuxstbdaurnimator: Can't the iphone run third-party apps already?
11:01:04 Quit atsea-22 (Read error: 104 (Connection reset by peer))
11:01:04Lloreanlinuxstb: They announced an SDK now, I believe.
11:01:04jhMikeSI never could find a way to do i2s loopback on PP
11:01:06 Nick fxb__ is now known as fxb (
11:01:17Lloreanlinuxstb: But I believe there's already third-party ways to get third-party apps on too
11:01:30daurnimatorSDK is web apps only AFAIK
11:01:53linuxstbYou mean code that runs on a server, not on the iphone?
11:02:21Lloreandaurnimator: The new one they announced, you mean?
11:02:38daurnimatoruh. how new?
11:03:14LloreanWell, it's not out yet, and there aren't many details, so I'm confused as how you know it's web-apps only
11:03:19 Part pondlife ("Gone")
11:03:56LloreanApple announced it's due in Feb, iirc
11:05:04 Join atsea-22 [0] (i=atsea-@gateway/tor/x-017a3ab149816930)
11:05:24LloreanEither way, official SDK or third-party method, it's still really more suited for Rockbox-as-app, so any work should/could probably start there anyway
11:07:39jhMikeSamiconn: I can't seem to hurt playback at all now after test_codec with the state check. I think a more reliable interface needs doing since now it's used in a regular user plugin however.
11:08:36 Join pondlife [0] (n=Steve@rockbox/developer/pondlife)
11:11:30pondlifeHmm, revises the use of talk_disable_menus() etc. to disable all talking, with a simpler API. But it doesn't fix mpegplayer...
11:12:38Lloreanpondlife: Is disabling talking the very first thing done in mpegplayer?
11:12:49pondlifeIt is with that patch
11:13:00pondlifei.e. before any splash() calls
11:13:02***Saving seen data "./dancer.seen"
11:13:30jhMikeStest_codec should probably steal iram as well
11:13:39pondlifei.e. FS #7994 incorporates FS #7939
11:17:30 Join courtc_ [0] (
11:17:55 Quit courtc (Read error: 104 (Connection reset by peer))
11:21:08 Join mertd [0] (
11:22:46mertdhi... how can i make my lcd off while my ipod video is idle? because there is a big battery problem.
11:23:30pondlifemertd: Poweroff idle timer?
11:23:53mertdbut i want my ipod to continue playing the songs..
11:24:14pondlifeAh, not idle then...
11:24:22mertdoo yee soryy.. :)
11:24:31mertdi mean while playing.
11:24:38pondlifeYou can only make the backlight go off AFAIK
11:25:13mertdbut why there is not such a way to turn the lcd off while playing..
11:25:28pondlifeIt's thought that the LCD isn't a major reason for the battery draining...
11:25:33pixelmaafaik the lcd doesn't need too much power and on the Video the display is actually quite readable without backlight
11:26:47mertdok i understand... ok then.. thanks for your helps..
11:27:47 Part mertd
11:29:12 Nick fxb is now known as fxb__ (
11:29:39Lloreanpondlife: It's a little more than simply "thought". There's pretty strong evidence that the LCD isn't a significant factor in the power drain problems on the PortalPlayer devices. :)
11:30:35LloreanIn fact, I *think* there's an option to power it down on the Sansa even, though I'm not certain on that one
11:31:08jhMikeSyes, e200 has a standby
11:31:33markunscorche: you were poking me?
11:32:08LloreanjhMikeS: Is there any disadvantage to it on the e200, making the assumption that the screen is unreadable while backlight is off?
11:32:54jhMikeSIt's not very readable even with sun on it imo
11:33:18LloreanI agree, I was just wondering if powering down the LCD caused a delay or anything when the backlight came back.
11:33:36jhMikeSyes, much more from standby that simple display off
11:34:10 Quit ddalton ("leaving")
11:34:24LloreanI've been trying to keep my eyes out for "options that can be removed", such as the disk power down.
11:34:39jhMikeSthough, one value could be specified to force it on always. one person has found it readable enough.
11:35:14LloreanjhMikeS: I certainly don't find it visible enough to be usable, but that's me.
11:35:22LloreanI can "see" it, but I wouldn't *want* to see it.
11:35:40jhMikeSit wouldn't be a new option anyways, just a new value
11:36:29 Nick webmind_ is now known as webmind (
11:36:50LloreanI've just been trying to determine if the menus, in general, could be shortened and simplified in some areas.
11:36:52 Quit webmind ("Reconnecting")
11:36:55 Join webmind [0] (
11:37:56LloreanFor example, I'm wondering how many people have replaygain tags, but disable replaygain, or among those that use it how many use an option other than "track gain if shuffle" (or if the on/off option could just be incorporated so that the options are off, track, album, or track if shuffle)
11:38:14jhMikeSI don't doubt they could or at least all the backlight ones
11:38:38pondlifepixelma: Could you perhaps try - version 3 fixes mpegplayer for me
11:38:43pondlifeOn H340 that is
11:38:52jhMikeScan't we just dump crossfade? :)
11:39:13pondlifeNo! :)
11:39:34pondlifeIt's my #1 favourite feature
11:39:44jhMikeSargh! see a counselor. :)
11:40:19LloreanOr the Equalizer On/Off: My understanding was that bands set at 0db don't affect CPU usage anyway, so is it really necessary to be able to turn it off when you can just "clear" the EQ?
11:40:29pixelmapondlife: will do
11:40:31LloreanAssuming a quick method to clear it were apparent that is.
11:40:43linuxstbpondlife: I think I mentioned a project in IRC a while ago which proposes new metadata tags to improve crossfading. Did you see it?
11:40:45jhMikeSLlorean: and a quick method to restore it?
11:41:06pondlifelinuxstb: Yes, but I'm generally happy with our current simple implementation
11:41:24LloreanjhMikeS: Well, I think a lot of users just use presets as it is. But what I was thinking is that, were I not paying attention I'd go to the EQ, and set a value, and be confused as to why my sound wasn't different because I wouldn't expect to also have to enable it.
11:41:38linuxstbpondlife: OK ;) I guess you won't be the person adding it to Rockbox then...
11:41:41jhMikeSI use that option but as long as another way exists to perform it as quickly, I'd be happy.
11:42:02LloreanI use it too, but I'm trying to think about the general complaint that some of Rockbox is "Too complicated"
11:42:13pondlifeGreat work on H300 mpegplayer guys, it's usable at last
11:42:53jhMikeSLlorean: well, at least the uses can't claim we patronize them
11:43:00LloreanjhMikeS: To me, "equalizer" seems like one of those options (like volume, or bass/treble) where you just expect to change it, not turn it on and off too.
11:43:25LloreanThere's not an on/off option for "Bass/Treble filters" :)
11:43:39jhMikeSno, but I wanted one :)
11:43:50LloreanAh, I see you may be the wrong person to discuss this with, then. ;)
11:43:56 Quit spiorf (Remote closed the connection)
11:43:57jhMikeSwhen debugging stuff it quite nice to have a quick switch
11:44:01LloreanVery true.
11:44:15LloreanBut there's also a "Debug" menu. Heh.
11:44:21*linuxstb mentions .cfg files
11:44:25*pondlife would prefer to scrap the EQ on/off too...
11:44:45pondlifeAs long as there's no penalty when all bands are at 0dB
11:44:47LloreanAnyway, what I'm trying to do is figure out where settings can be merged, or assumptions can be made, so that there are less "options" but the same amount of "functionality"
11:44:54Lloreanpondlife: iiuc, there shouldn't be.
11:45:07pondlifeLlorean: I applaud your aims. Good luck.
11:45:22Lloreanpondlife: Of course, the important part is to do so without making it *more* confusing. :)
11:45:40pondlifeIndeed, and to enable a new Settings menu structure...?
11:45:46pondlifeOR rather layout
11:45:56LloreanYes, a secondary goal would be to reorganize them
11:46:07*pondlife will be very happy when "General Settings" has gone
11:46:17LloreanPlayback vs Sound Settings needs a little work too.
11:47:00jhMikeSI'd want more quick switches actually...conviently accessible. I don't need the option though.
11:47:25pondlifejhMikeS: .cfg files are great for testing
11:47:53LloreanjhMikeS: Seen this:
11:47:54pondlifeHmm, mpegplayer works fine from start, but resume still locks..
11:48:02jhMikeSThose always get messed up somehow
11:48:21*jhMikeS wonders if he should :)
11:48:27LloreanjhMikeS: It seems like the option to quickly switch to the default value. If it also stored the previous value, so that you could quickly toggle to default, or toggle to previous...
11:48:45jhMikeSLike "last_ch" on a TV remote?
11:48:54LloreanIn my mind, yes.
11:49:12LloreanYou could then toggle *any* feature off/on easily.
11:49:12pondlifeWe need to suppress config.cfg from the Browse .CFG files option too (or save user presets and config.cfg in different places).
11:49:20LloreanWell "off" being "default" and "on" being "your setting"
11:49:37Lloreanpondlife: I'm confused when you say that.
11:49:54LloreanI "run" config.cfg frequently (every time I accidentally clear my settings on my gigabeat, for example)
11:50:01pondlifeWell, if I want to select a configuration, config.cfg is not much use.
11:50:01jhMikeSSort of more general form for talk_disable(_menus)
11:50:06rasherZagor: actually, why not put the gcc patches in svn and download them from viewvcs?
11:50:29pondlifeLlorean: Doesn't config.cfg contain the current (ie.. cleared) settings?
11:50:50Lloreanpondlife: It contains the "last saved" settings. Immediately after a cleared boot, it contains the settings from your last shutdown still, until you trigger a save.
11:50:56pondlifeAh, ok
11:51:10LloreanVery handy for a klutz like me who pressed the menu button too soon during boot.
11:51:29linuxstbrasher: Sounds a nice idea.
11:51:30pondlifeFair enough.. It's just an entry I always skip. I assumed it should really be cleared by Clear Settings too!
11:51:56LloreanI don't think there's any harm in it not being cleared by clear settings, though
11:52:12LloreanIt's actually more convenient that it's not I think, as it'll be cleared as soon as they change anything anyway, or shut down, I believe.
11:52:21TavnosHi, Is there a way to emulate Remote control in Rockbox Simulator ?
11:52:41jhMikeSDoes the settings clear trigger a save of them on the next ata idle callback?
11:52:51pondlifeYes, I think so
11:53:13pondlifeIn which case it may not be the most reliable way to restore your settings.
11:53:24pondlifeI always have some user .cfgs anyway
11:53:36LloreanYeah, I do too.
11:53:37jhMikeSthe clear shouldn't mark the settings as dirty then
11:54:24amiconnlinuxstb: around?
11:55:15LloreanAnyway, I think I saw somewhere that someone had actually written out the rockbox menu layout. Has anyone else seen that as I'm not finding where it was mentioned any more, so I may just be crazy.
11:55:36pondlifeNo, I saw it too, so we're both crazy
11:55:47pondlifeNot sure where... rasher's?
11:56:09jhMikeSLlorean: Did it look a bit disturbing?
11:56:16LloreanI think it was one of the people working on one of the myriad menu customization patches.
11:56:16 Quit Mouser_X (Connection timed out)
11:56:22LloreanjhMikeS: Never looked at it, just saw it mentioned
11:56:36LloreanTrying to decide if it's a better use of my time to hunt it down, or just boot up one of my players and start typing.
11:57:04jhMikeSprobably #2. I don't thinks it's a lengthy undertaking.
11:57:06 Join Tanuva [0] (n=tanuva@
11:57:09LloreanProbably not
11:57:29LloreanI do like things being done for me, though. :)
11:57:53linuxstbamiconn: Yes
11:57:59jhMikeSprobably would take 10min and you'd spend 30 just finding it more
11:58:20LloreanYeah, already started on the typing myself
11:58:27 Join underleave [0] (
11:58:34underleaveThe sound quality of sansa e200 with rockbox is good as iaudio/iriver ?
11:58:34 Join Nico_P [0] (n=nicolas@rockbox/developer/NicoP)
11:58:38amiconnlinuxstb: Am I correct that PredictorX and PredictorY are the same code, except every Y replaced by X ?
11:59:03 Join obo [0] (n=obo@rockbox/developer/obo)
11:59:11jhMikeSunderleave: e200 sounds very good. an x5 is a tough one to beat though.
11:59:15LloreanNico_P: Question: is the buffering API available to plugins?
11:59:26Nico_PLlorean: not yet
11:59:44LloreanJust curious was all, thanks.
11:59:52 Join ramon8 [0] (
11:59:53underleaverockbox improves the sound quality in sansa?
12:00:00LloreanjhMikeS: The e200 has several noise issues though, I think
12:00:06pondlifeHmm, I appear to be stuck in the recording screen!
12:00:17pondlifeMeters are working, as is menu
12:00:25jhMikeSunderleave: I think it sounds better than their stuff, yes. It's kind of striking actually.
12:00:30linuxstbamiconn: It's basically the same, yes. It was originally one function, called twice in each decoding stage, once per channel.
12:00:42pondlifeOnly way out is to shut down
12:00:46jhMikeSLLorean: say what? I never hear them.
12:00:58jhMikeSOnly for mic recording.
12:01:10 Join MethoS- [0] (
12:01:17 Join Rondom [0] (
12:01:45LloreanjhMikeS: You haven't heard the noise during buffering, or when the wheel light is on?
12:02:04pondlifeHmm, it happens if I (a) enter recording using LONG REC and (b) have done a short recording
12:02:08LloreanOn an unrelated note, what is "Custom" under "Channel Configuration"?
12:02:38amiconnlinuxstb: Nice, that allows a quick advance now in predictor-cf.S :)
12:02:49jhMikeSLlorean: I can hear very, very quiet noise as long as _no_ audio is playing right at buffer start.
12:03:08jhMikeSLlorean: It a stereo width control
12:03:22pixelmapondlife: the patch helps on the M5 (but there's still a quirk when I manually "stop", which should bring up the menu but I'm just presented with a blank screen and then it return to the filebrowser) - and there is no change on the c200
12:03:24 Quit MethoS- (Read error: 104 (Connection reset by peer))
12:03:44underleaveso, if i want sound quality, i better pass sansa e200 ?
12:03:58jhMikeSLlorean: But I'll admit I didn't connect Custom and Width right off the bat. The arrangement is totally wrong.
12:04:02LloreanjhMikeS: I can hear noise any time audio rebuffers *if* my audio is quiet enough. As well, the ring light used to make a noise, but I haven't checked on that in a while. What kind of headphones are you using?
12:04:08pondlifepixelma: Did you try several times on the c200? I think there's still an intermittant problem..
12:04:25LloreanjhMikeS: I think "Custom" and "Width" seems odd, since I would expect stereo width to work as long as my audio is set to "Stereo"
12:04:27jhMikeSunderleave: not imo but maybe it affects newer models more.
12:04:38LloreanjhMikeS: Mine's one of the oldest.
12:04:48Lloreanunderleave: Despite what I said, I'd still recommend it for quite good sound quality
12:05:15jhMikeSI gues mine's pretty darn quiet.
12:06:00LloreanjhMikeS: I think it's dependent on your headphones more than anything else. Some people complain of it a lot, some people claim it doesn't exist. I can hear it if I look for it, or could last time I did.
12:06:13underleavewhere can i check what languages supported
12:06:16underleavein e200
12:06:39jhMikeSHaving such long analog lines with a builtin codec is just stupid. The codec the should be right by the HP jack, mic and other analog lines.
12:06:40pixelmawith just mp3s (or wav) I get the best sound quality with the hwcodec Ondio :D
12:06:43linuxstbunderleave: You mean which languages Rockbox supports?
12:06:47Lloreanunderleave: Rockbox supports the same languages on every player.
12:07:09jhMikeSLlorean: I think 16-ohm phones? x5 gets more clicking with them. Tried it myself.
12:07:40 Nick fxb__ is now known as fxb (
12:07:41LloreanjhMikeS: Mine are 16-ohm, yes.
12:07:50linuxstbrasher: Is having problems? It's extremely slow for me...
12:08:44rasherlinuxstb: I don't know much more than you, but yes, it seems slow
12:09:04 Join MethoS- [0] (
12:09:05Lloreanrasher: That will be quite handy, thanks.
12:09:06pixelmapondlife: I tried ED 3 times (always start at the beginning) and other videos which stay completely black, but that might broke for a different reason don't know
12:09:07linuxstbrasher: OK. The front page isn't loading at all for me...
12:09:24pondlifepixelma: OK, thanks
12:09:36pondlifeLlorean: Did you see rasher's link there?
12:10:51Lloreanpondlife: Yup
12:10:58LloreanAlready downloaded a local copy to play with and everything
12:11:27 Join moos [0] (
12:13:22LloreanjhMikeS: Anyway, so iiuc, Stereo Width is only used if you choose "Custom" for channel configuration?
12:13:32jhMikeSLlorean: yes
12:14:22LloreanAny reason why "Custom" can't be removed, and Stereo Width only matter if you choose "Stereo"?
12:14:27underleaveso, where can i check about the languages supporteD?
12:14:53jhMikeSThough arranging the DSP to simply output stereo and shortcircuit width control when in stereo mode would be not too difficult
12:15:07pixelmaLlorean: amiconn likes it to have it preset and just switch it on and off.. ;)
12:15:14Lloreanrasher:"> is prettier though, isn't it?
12:15:31 Join barrywardell [0] (
12:15:39jhMikeSLlorean: another preset w/switch user is present too I guess :)
12:15:51Lloreanpixelma: That'd be a handy case for the "Previous Channel"-style Default selection I mentioned to jhMikeS then
12:15:57amiconnLlorean: 'Custom' needs sligthy more processing power even with width == 100
12:16:14amiconn(on swcodec only)
12:16:30Lloreanamiconn: It seems one ought to be able to bypass when it's at 100% though.
12:16:31jhMikeSwe could change that so width=100 nulls that stage
12:17:27amiconnIt might also confuse people who accidentally set the width to 0, and then wonder why Stereo sounds like Mono
12:18:13jhMikeSthe two settings right now don't look related at all though.
12:18:22LloreanWell, "Custom" needs to be named "Custom Width" then, I think
12:18:35LloreanRight now, I fear people will set a custom width and not know that 'Stereo' overrides it
12:18:43jhMikeSUse Stereo Width
12:19:11LloreanjhMikeS: Well the option for the width is called "Stereo Width" which is why I'd expect it to affect the "Stereo" option under "Channel Configuration" rather than the "Custom" option there
12:19:27pixelmabut the current implementation often confuses me, I often forget to enable it and wonder why it doesn't take effect... exactly
12:19:27rasherLlorean: even I did that..
12:19:35LloreanI think either the "Custom" option there needs to be removed, or renamed to "Custom Width" or even "Custom Stereo Width" to go with the fact that the option to set the width is so called.
12:20:24jhMikeSWhen I first used rockbox, I selected "Custom" and wondered why I didn't hear any change so set it back to "Stereo". The I went to another option called "Stereo Width" and tried changing that and didn't know why I didn't hear it. For awhile I then just assumed the software was broken. :)
12:21:02LloreanSo, either a name change, or (imho slightly better) assume that if set to Stereo, and width is no 100%, they want a custom width.
12:21:15jhMikeSI have a way of doing this sort of thing...menus that are also setting values
12:24:15pixelmasame with the selector bar colour(s) by the way
12:24:46 Nick Tanuva is now known as Tanuva|reboot (n=tanuva@
12:27:45Lloreanpixelma: What about the colours?
12:29:17*jhMikeS sees that idea got no op
12:29:29 Quit MethoS- (Read error: 113 (No route to host))
12:29:46pixelmaLlorean: well you can set the colour(s) but you also have to set the "Line Selector Type" to get a coloured selector bar
12:30:08LloreanYes, I'm not sure how to fix that, since there's still "inverse" as an option, isn't there?
12:30:28pixelmayes, and the pointer
12:30:50jhMikeSguess what I mean is, you select the item "Custom" then custom is enabled. Perform some action on that same item actually takes you to the setting of the value.
12:31:03amiconnlinuxstb: Btw, with both filters and predictor disabled, decoding runs at ~283% realtime on coldfire, not varying with -c level
12:31:37LloreanjhMikeS: Makes sense to me. It's just an extension of the existing menus, sorta. Some of the settings are "options" but one is just another "menu", at least from a user perspective
12:32:13jhMikeSalso, I'm there any such thing as an "audio icon"? If not, let's invent it.
12:32:16linuxstbamiconn: I would hope that stage could be improved a lot...
12:32:18pixelmajhMikeS: that's what I thought when the colour selector bar got added
12:32:22LloreanjhMikeS: "Audio icon"?
12:32:41LloreanjhMikeS: I think there's an icon by default for Sound Settings that's usually a note
12:32:43amiconnAnd I'm about to test predictor-cf.S :)
12:32:44LloreanOr a speaker
12:32:51jhMikeSSome tone or pattern associated with the voice clip to tell what it is.
12:33:19LloreanAudio cues to tell you what's going on without words?
12:33:34linuxstbamiconn: I will be impressed if that works first time... I struggled with it on ARM.
12:33:40jhMikeSheck, I'd probably use it
12:33:46amiconnIt sounds wrong...
12:34:07amiconnNo static, but volume is too low and goes up & down
12:34:26Lloreanpixelma: I'm sure there's *some* way to simplify the selector colour options, I'll think on it too and see if I can come up with something
12:34:39amiconnHmm, maybe just the emac mode...
12:34:53pixelmaLlorean: good luck... :)
12:35:30Lloreanpixelma: Well, my goal is to make the menus intuitive enough that it drastically reduces the amount of requests for custom menus. I've already taken on an impossible task, so a "fairly difficult" one doesn't make it too much worse. :)
12:36:36amiconnlinuxstb: If I get that working without significant overhead, it makes for a nice speedup
12:36:37LloreanjhMikeS: What situations would you use these cues in?
12:36:44amiconn177%->209% for -c1000
12:37:04LloreanjhMikeS: I can see them for handy things like where splashes are. A "confirm", "cancel", "wait", and "done waiting" cue at least
12:37:25linuxstbamiconn: Nice. What about -c3000 ?
12:37:37amiconnTesting right now
12:38:04 Join lee-qid [0] (
12:38:16linuxstbamiconn: BTW, this improvement is probably worth mentioning in the MajorChanges page when you're finished.
12:39:13amiconnMy code follows the arm version very closely wherever possible
12:39:38amiconnSometimes I need more instructions, because cf is strictly 2-operand
12:39:58amiconnSometimes the sophisticated addressing modes help saving instruction
12:40:22amiconnLike the loop count decrease - it's a single instruction that acts directly on the stack value
12:40:51amiconnarm needs 3 instructions for that due to load-store
12:41:07jhMikeSLlorean: To tell whether the item is a menu, setting or a combo menu/setting. That way key consistency is slightly less important. You know what the widget is by sound and hence how to operate it.
12:41:33amiconnHmm, -c3000 ends up at 103%
12:42:32LloreanjhMikeS: We already *kinda* do that with visual icons too.
12:43:07jhMikeSLlorean: right. but blind users don't see those. they just hear strings, right?
12:43:13pondlifeHmm, can anyone decode !!QUEUE_LEVEL from talk.c ?
12:43:29LloreanSo what you do is allow talk clips to be associated in the same way as image icons.
12:43:37jhMikeSif (QUEUE_LEVEL != false)
12:43:43pondlifeDoes that mean (QUEUE_LEVEL != 0)
12:43:49LloreanSo an audio file gets the audio talk clip immediately before its personal one, if there's an talk clip for the "audio file" icon
12:43:51pondlifei.e. an int to bool
12:44:41amiconnHmmmmm, actually playing -c2000 or higher yields static, and after that, even -c1000 do the same until the next reboot
12:45:06linuxstbamiconn: Yes, I didn't get huge speedups on ARM either, but I put that down to the C code already being well optimised.
12:45:07jhMikeSLlorean: the icon could be it's own voice clip followed by the string clip.
12:45:18LloreanjhMikeS: That's kinda what I tried to say. :)
12:45:48LloreanNeedless to say, I like the idea
12:46:04amiconnlinuxstb: -c3000 and higher are obviously dominated by the filtering. For lower modes, doing the predictor in asm is helpful, but the entropy decoding is the larger part
12:46:05jhMikeSThey just need to be nice sounds, not annoying beeps
12:46:36 Quit PaulJam (".")
12:46:54linuxstbamiconn: Yes, that's what I concluded. I don't think there's much more we can squeeze out of the predictor/filter stages.
12:46:56jhMikeSI guess they'd be customizeable too but I suppose a standard set would be needed to combine into the voice build
12:46:57 Join B4gder [0] (i=db8e3061@gateway/web/cgi-irc/
12:47:19LloreanjhMikeS: Probably so, yes.
12:47:27linuxstbB4gder: Hello China.
12:47:34B4gderhey ho
12:47:54amiconnlinuxstb: My estimation for current SVN is that the predictor needs ~37% and the entropy decoder needs ~63% of the decoding time on coldfire (for -c1000, which does no filtering)
12:48:08B4gderfake ipods everywhere in this country :-)
12:48:20linuxstbB4gder: Are you bringing some back?
12:48:21amiconnAnd it looks like asm will make the predictor about twice as fast, similar to the filter
12:48:50 Join desowin [0] (
12:48:53B4gderI've seen at least two different iphone fake models too
12:49:06jhMikeSthe MaoPhone
12:49:14B4gderI'm gonna try and see if I find any nice players to bring home tomorrow
12:49:26B4gderI'm not too interested in the fake ipod ones
12:50:05amiconnHmm, maybe I shouldn't fiddle with the parameters on stack?
12:50:12amiconn(and rather copy them)
12:53:42 Quit Rick (Read error: 104 (Connection reset by peer))
12:54:10 Join Rick [0] (
12:54:27 Quit nerochiaro (Remote closed the connection)
12:54:32 Join nerochiaro [0] (
12:58:54preglowLlorean: agree with the "custom", it's confusing, and needlessly so
12:59:05 Join Thundercloud [0] (
13:00:21 Nick Tanuva|reboot is now known as Tanuva (n=tanuva@
13:00:56Tanuvabeim booten schwarzer bildschirm, kein xserver, nichtmal ein terminal ist zu bekommen...
13:01:09LloreanTanuva: This is an English channel, please.
13:01:18LloreanIt's also an on-topic channel.
13:01:56Tanuvaoh, sorry, this is btw the wrong channel... the message was supposed to appear in kubuntu-de... :/
13:09:22 Quit B4gder ("CGI:IRC (EOF)")
13:09:30 Join midgey [0] (
13:10:52 Nick Tanuva is now known as Tanuva|frickel (n=tanuva@
13:11:32 Quit midgey (Client Quit)
13:13:03***Saving seen data "./dancer.seen"
13:15:25 Quit barrywardell (Read error: 110 (Connection timed out))
13:19:40 Quit iamben (Read error: 104 (Connection reset by peer))
13:20:51 Quit lee-qid ("aufwiederbyebientotsayonara")
13:22:44 Join Arathis [0] (
13:24:42 Join einhirn [0] (
13:29:52 Quit einhirn ("Miranda IM! Smaller, Faster, Easier.")
13:31:11 Quit tedrock (Client Quit)
13:33:22 Join Gnu47 [0] (
13:35:14 Join tedrock [0] (
13:40:54 Nick Tanuva|frickel is now known as Tanuva (n=tanuva@
13:48:30 Quit marcosource (Remote closed the connection)
13:48:34 Join marcosource [0] (
13:51:50 Join Dark_Apostrophe [0] (n=darkapos@supporter/monthlybyte/DarkApostrophe)
13:52:05Dark_ApostropheHas the ogg playback problem been solved?
13:52:35markunwhich problem exaclty?
13:53:49preglowmarkun: ah, btw, did you check out the icode/text stuff for vorbis yesterday?
13:54:37Dark_Apostrophemarkun: A few svn revisions ago, your ogg playback stopped working
13:54:50Dark_ApostropheSo I didn't upgrade my rockbox since it worked on a prior version
13:55:42markunpreglow: I see I've changed .icode to .text in my own tree
13:57:56preglowwell, whatever has it working
13:58:03markunI believe I had some trouble after including a header file in imdct_arm.S to be able to use .text in case of the S3C2440
13:58:10preglowmoos: here?
13:58:36LloreanDark_Apostrophe: Which SVN revision did this start with?
13:58:46LloreanAnd was there a bug report associated with it in the tracker?
13:58:50Dark_ApostropheI don't remember, I just saw you discussing it in IRC
13:58:50preglowLlorean: i'm pretty sure he's talking about the same thing i am
13:58:54Dark_ApostropheSo I chose not to upgrade
13:59:02Dark_ApostropheBut my version didn't have the issue
13:59:05Dark_ApostropheSo I didn't upgrade.
13:59:25 Nick Tanuva is now known as Tanuva|frickel (n=tanuva@
13:59:26preglowlatest rockbox doesn't have the issue, but it's not resolved properly
13:59:44Dark_ApostropheSo I should not upgrade?
14:00:09pondlifeAny voice users around? I'd really like to commit but it could do with some testing...
14:00:28pondlifeDoesn't do much in theory, but might save space and tidies stuff up a bit.
14:00:50pondlifeCould do with a MAS test
14:01:38moospreglow: yes, hello
14:02:17markunDark_Apostrophe: are you talking about this?
14:03:14markunand did you read this too?
14:03:39LloreanThat's also hardly "a few" SVN revisions ago. Several hundred...
14:04:08LloreanFixed nearly 3 weeks ago...
14:04:24Dark_ApostropheAny other issues I should know about, or is it safe to upgrade to a newer rockbox?
14:04:40markunDark_Apostrophe: it's always extremely dangerous! :)
14:04:50preglowmoos: that's a gigabeat f wav write of the the first test track, i assume you hear nothing wrong with it?
14:04:59LloreanDark_Apostrophe: This is why it helps if you file actual bug reports, or verify that one has been filed for a problem you know about
14:05:04LloreanThen you can just check if it's closed.
14:05:16moospreglow: let's listen...
14:05:21Dark_ApostropheLlorean: Like I said, I never had the problem - I didn't upgrade to the affected svn revisions
14:05:54LloreanDark_Apostrophe: And because you never had it, you decided it was better not to care if there was a proper bug report, and instead just wait around and ask people "Is it fixed?" until someone said "Yes"?
14:06:03markunand now it's fixed so all is good
14:06:47jhMikeSpreglow: need ears?
14:06:50LloreanDark_Apostrophe: It's somewhat fortunate you found someone who knew what bug you were talking about, instead of someone who thought this was a recent problem, as suggested by your "a few revisions ago"
14:06:59moospreglow: nope, that's not what I have with musepack on gigabeat unfortunatly :(
14:07:10preglowjhMikeS: not really, no
14:07:17Dark_ApostropheLlorean: I'm a newb, sorry.
14:07:36preglowmoos: then i have no idea what's wrong with your playback of musepack :/ please do do a test_codec wav write when you can
14:08:25moospreglow: I'll now quicly if I can. Have a link to point me, for what do I need to do?
14:08:44preglowdo you do your own builds?
14:08:44markunDark_Apostrophe: just upgrade and have fun :)
14:08:50Dark_ApostropheOk. :)
14:09:12moospreglow: yup
14:09:27markunmoos: do you get it with an official build as well?
14:09:30preglowmoos: ok, open apps/plugins/SOURCES, and insert "test_codec.c" (without quotes, of course) on a line
14:09:58preglowmoos: then build normally, copy the build over, use "open with" on the test file, then choose test_codec, then write to wav
14:10:04moosmarkun: yeah, that is with svn builds, and the sound problem came after one musepack optimisation commit
14:10:21preglowmoos: that will create test.wav in the root of your player, listen to that and tell me if that has the same distortion
14:10:21moospreglow: ok, let me 2 min
14:10:26 Join Mouser_X [0] (n=Mouser_X@
14:11:00 Nick Tanuva|frickel is now known as Tanuva (n=tanuva@
14:11:11linuxstbDark_Apostrophe: You can always just make a backup of your .rockbox folder, and if you have problems with a current build, go back to it.
14:11:54jhMikeSpreglow: just asking...thought you might be fishing opinions
14:12:05Dark_ApostropheMy current version is r014811-070921
14:12:09Zagorbertrik: if you want to help more concretely, it would be great if you can write the host application for bulk testing. not only does it save me time, but it will probably find more bugs than if I write it myself.
14:12:13Dark_ApostropheNo wait
14:17:29 Join Honkboy [0] (
14:20:03TavnosHi, can someone help me with WPS conditional tags ?
14:20:32 Part Honkboy ("Leaving.")
14:22:03TavnosLlorean can you help me ?
14:22:07 Nick Tanuva is now known as Tanuva|off (n=tanuva@
14:24:45TavnosDark_Apostrophe can you help me with WPS ?
14:25:04Dark_ApostropheNo, I'm a noob here as well. :(
14:25:42TavnosNo pronlems
14:25:54TavnosZagor can you help me with WPS ?
14:26:45ZagorTavnos: what about it?
14:27:19TavnosI want to display battery level as images
14:27:38TavnosI made images, and I loaded them
14:29:22 Join Tavbis [0] (
14:29:22 Quit Tavnos (Read error: 104 (Connection reset by peer))
14:29:36 Nick Tavbis is now known as Tavnos (
14:29:48Tavnossorry Zagor I was disconnected...
14:30:04linuxstbTavnos: Why don't you try and find another WPS that does what you want to do, and copy the code?
14:30:10LloreanTavnos: Please don't just start selecting names and sending them messages. For many people that causes their computer to make a noise, and it's considered impolite unless it's something you know in advance they're involved or interested in.
14:30:38preglowlinuxstb: frame length is constant for wma, yes?
14:30:57linuxstbIf you mean total number of samples per frame, then no.
14:31:35preglowlinuxstb: i thought block length changed within a frame, but a frame always had the same amount of samples
14:31:51 Quit Nico_P (Remote closed the connection)
14:32:12linuxstbpreglow: I may be wrong, but I don't think that's true. You could try adding some DEBUGFs in the sim to test.
14:32:37Tavnoslinuxstb: I want to understand how it works... I think I should use things like that : %?bl<%xdA|%xdA|%xdB|%xdC|%xdD|%xdE|%xdE>
14:33:07Tavnosbut How rockbox knows the battery thresholds...
14:33:17preglowi've got a hack going that lets me access output samples from the internal array of libwma, but it kinda depends on frame_len not changing
14:33:17linuxstbTavnos: I can't help, all I can do is point you to the documentation - - and to other WPSs as examples.
14:33:29preglowlinuxstb: s->frame_len is only ever set in init(), it seems
14:33:34Tavnoslinuxstb : I'm just on this page :)
14:34:33 Join webguest54 [0] (i=4ccb89b9@gateway/web/cgi-irc/
14:34:58 Join Crash91 [0] (n=evil91@
14:35:59 Quit Rondom ("Ex-Chat")
14:41:21*preglow wonders how to handle errors in wmas
14:41:28preglowmoos: any news?
14:43:44 Join mf0102 [0] (n=michi@
14:45:28linuxstbamiconn: Any luck with the predictors?
14:46:09 Join jgarvey [0] (
14:46:54amiconnFound 2 mistakes, but I still get static :(
14:47:25preglowthat good old libffmpegFLAC/coldfire.S feeling
14:47:31*amiconn finished the PP502x -c5000 speed test in parallel
14:48:03amiconnSomeone should perform a fresh test on gigabeat
14:48:43*markun points at linuxstb
14:52:15*linuxstb feels pointed at
14:55:20preglowlinuxstb: would it be hard to provide resync support in the asf stuff?
14:57:52 Join nicktastic [0] (n=nick@unaffiliated/nicktastic)
14:58:18linuxstbpreglow: I think it would be... Looking at the parsing code, there doesn't seem to be any significant magic at the start of a packet.
14:58:27moospreglow: sorry, testing...
14:59:34linuxstbpreglow: Why the sudden interest in error recovery?
15:01:35preglowlinuxstb: because ffmpeg deals with it by default, and my one file i'm interested in doesn't decode smoothly
15:02:24preglowamiconn: do you think there's a stall after mac.w as well?
15:02:50linuxstbpreglow: So it's definitely an error in the stream, rather than a bug in Rockbox?
15:02:57preglowlinuxstb: ffmpeg says error
15:03:00moospreglow: the .mpc file, and the .wav file doesn't sound the same at all :)
15:03:13preglowlinuxstb: i think it decodes fine in windows
15:03:18preglowlinuxstb: might be an ffmpeg bug
15:03:48preglowmoos: and you're absolutely certain you have tried resetting all your settings? this pretty much rules out that the error is in our musepack codec, even with the new optimiazations
15:04:05amiconnThere's no stall after mac.w nor mac.l, but there is a stall when you fetch the %accN too early after a mac streak
15:04:12preglowamiconn: that's what i mean
15:04:27amiconnI don't think that depends on whether it's .w or .l
15:04:34preglowamiconn: mac.w has about a fourth less work to do than mac.l
15:04:50preglowluckily, freescale doesn't exactly spoil us with consistent information on this
15:05:10preglowamiconn: ok, then i see no point in using mac.w instead of mac.l in the 16 bit part of libffmpegFLAC/coldfire.S
15:05:33amiconnWorth a test...
15:05:40preglowi doubt i'd be able to measure it
15:05:54amiconnNot in the real thing, but in a test loop
15:06:44amiconnAn empty loop with 45158400 iterations should take 3 seconds
15:07:00amiconn(when not boosted)
15:07:24 Quit underleave ()
15:07:25amiconnDisable interrupts to be sure only that loop runs, put a puts/lcd_update before and after, and measure with a stop watch
15:07:27 Join ataxic [0] (
15:08:07amiconnThen put your mac.w/movlcr.l sequence in it, recompile, run & measure again...
15:08:20*amiconn puts such test loops into the debug menu
15:09:00moospreglow:wow, I reseted all seetings, and now all is fine, very cool. But after to reset I saved my seeting(with all dsp options of) and the distortion came again. I will see what could in my seeting cause this if this not an dsp option
15:09:36preglowmoos: post your cfg file
15:09:41moospreglow: sure
15:09:49preglowi think i know what's wrong...
15:12:31preglowtry turning off replaygain
15:12:54moosyup just saw is in reviwing the config, let's try
15:12:55LloreanBad replaygain tags could certainly cause clipping, couldn't they?
15:13:07***Saving seen data "./dancer.seen"
15:13:31linuxstbamiconn: I've updated with the current Gigabeat performance (-c5000 is running now)
15:13:36LloreanAlso, a replaygain preamp of 12?!
15:13:39moosmaybe the high preamp
15:13:44mooshehe :)
15:14:47amiconnlinuxstb: Ah, nice. So the predictor helps a bit on gigabeat, but the higher modes don't see much speedup because of the filtering
15:15:03 Quit Crash91 (Read error: 104 (Connection reset by peer))
15:15:16moosthta definitively the high preamp values
15:15:21amiconnI had a quick look at the arm filtering y'day and it looks like gcc already produces pretty much optimal code
15:15:38amiconnPerhaps some arm guru could spot an optimisation...
15:15:42ataxicIs it possible to create a module that can play an mp3 (or sets of mp3s) and then mute at certain points and play at certain times. those times would be specified in a txt file or something. Or play 2 mp3s at the same time perhaps?
15:15:43 Join Nico_P [0] (n=nicolas@rockbox/developer/NicoP)
15:16:09moospreglow: and musepack seems to be the only codec that cause those stranginess, and absolutely not on coldfire
15:16:25ataxictrying to create something for my running where i normally add beeps in my trance mixes
15:16:58preglowmoos: well
15:17:07preglowmoos: you probably don't have replaygain tags on the other files
15:17:23preglowmoos: coldfire probably clips, but it should sound like shit anyway
15:17:33preglowmoos: not as bad as overflow like you're probably hearing
15:17:36preglowbut still shit
15:18:22moospreglow: on coldfire, with reasonable volume no more hearing clips
15:18:34mooseven with high premp values
15:18:45preglowstandard arm doesn't have saturating add instructions :/
15:18:49 Join n1s [0] (
15:19:25 Quit webguest54 ("CGI:IRC (EOF)")
15:19:40linuxstbpreglow: We'll have it soon... (armv5 and armv6 targets on the horizon)
15:19:46moosohh, I understand then :)
15:19:56mooslinuxstb: nice
15:20:12preglowmoos: well, at least your problem is resolved
15:20:23*moos still prefers coldfire targets for now :)
15:20:27*preglow too
15:20:48*preglow hugs his h120
15:21:10LloreanDon't most replaygain scanners let you pick a level to.. I guess "normalize" is the closest word, at?
15:21:12*linuxstb wonders if amiconn can squeeze 30% better performance out of APE on Coldfire for -c3000...
15:21:24preglowlinuxstb: what setting is the badassest?
15:21:30moospreglow: yeah thanks, and now I know that those actual ARM targets doesn't like much preamp replaygain :)
15:21:44Lloreanpreglow: -c5000
15:21:46LloreanI think
15:21:48linuxstbpreglow: See
15:21:54preglowand that is still way out of our league, yea?
15:22:06preglowhole shiznit
15:22:10preglowholy too
15:22:13linuxstb1.7% realtime ;)
15:22:36*preglow doesn't like ape
15:22:38amiconnColdfire was 0.85% realtime with -c5000 before my filter optimisations
15:22:50mooswow :)
15:22:54amiconn8 hours for decoding a 4 minute track
15:23:08linuxstbpreglow: Are you suggesting a 1280 order filter is a bad idea? ;)
15:23:16preglowsweet fecking lord
15:23:24preglowthey use that and STILL achieve lossless playback?
15:23:46 Join Genre9mp3 [0] (
15:23:50linuxstbTo be fair, it's called "insane" for a reason, and I'm not sure many people use it for playback
15:24:21amiconnI think that coldfire could be ~10% realtime with the insane buffer in iram
15:24:33linuxstbAt least it woul dbe something to keep the Gigabeat S's CPU amused with.
15:24:35moospreglow: about those saturations with the replaygain preamp on mpc, do you think at one reason why before the big Buschman optimisations there aren't ?
15:24:38amiconn(which would be only possible for X5/M5)
15:24:54preglowmoos: they shouldn't affect it at all
15:25:04preglowmoos: and i'd be very, very surprised if they do
15:25:15amiconnI will do that test at some point, but not now
15:25:32Genre9mp3Did anyone else got a mail from the author of WinFF asking for permission to include Rockbox presets?
15:25:47preglowbut it sounds like a good idea to me
15:25:53LloreanThey need to be edited
15:26:00LloreanNewer versions of ffmpeg are incompatible with the current presets
15:26:09moospreglow: oh, I don't remenber touched my RG config, and those distortions came when I updated with this build (I was in vacations this time)
15:26:13LloreanDue to the name changes for some of the codecs
15:26:22amiconnAnd there would be a long list of rockbox presets...
15:26:30Lloreanamiconn: I've made most of them already
15:26:44moospreglow: but when I find time, I could perform a test with and without this patch
15:26:54LloreanThough they're for pre-name-change ffmpeg.
15:27:08amiconnPluginMpegplayer is missing a couple of necessary resolutions these days...
15:27:12preglowSoundCodecs could really need some cleaning up
15:27:12markunwhere is ippiDCT8x8Inv_Video_16s8u_C1R defined? (used in mpegplayer)
15:27:31preglowmoos: it's easy to disable the optimizations
15:27:37LloreanGenre9mp3: Considering I created the presets.xml on our wiki, I release all rights to it into the public domain (for what tiny amount that may be worth). :-P
15:28:00moospreglow: yeah? (maybe easy for a dsp an like you ;)
15:28:34*markun thinks he has a clean svn tree, but gets an error while compiling :(
15:28:38preglowmoos: in synth_filer.c, just find the "elif defined(CPU_ARM)" part and comment out the #define OPTIMIZE_FOPR_SPEED below it
15:28:41preglowmoos: that's that
15:29:01moosoh ok, let's try...
15:29:17preglowSoundCodecs needs cleaning :/
15:29:22 Join barrywardell [0] (
15:29:55amiconnLlorean: H10 5/6GB 1:1 is missing, and the Sansa c200 is missing completely
15:30:49amiconn(and yes, the 1:1 *does* make sense on that small display)
15:31:24preglowlinuxstb: but packet size is the same for all packets, yes? (wma)
15:31:26Lloreanamiconn: Now that we can actually do any even resolution, isn't 4:3 possible on that screen?
15:31:38preglowlinuxstb: more like asf, actually
15:31:42*Llorean should remake the presets for more accurate 16:9 resolutions.
15:32:09amiconnBtw, 4:3 would *not* cover a 160x128 screen
15:32:17amiconn160x128 is 5:4
15:32:39LloreanBut 120 wasn't an option previously
15:32:43amiconnI think we should have fullscreen, 4:3 and 16:9 for each target
15:32:44LloreanWe had to use multiples of 16
15:33:08LloreanBut now it's multiples of 2, so I can do "Fullscreen" "4:3" and "16:9" for each target
15:33:18LloreanOr at least, much much closer approximations
15:33:45amiconnI'm not sure whether we want the greyscale versions right now or not yet
15:33:53pondlifeGenre9mp3: If you're conversing with the author of WinFF, maybe point out that it's Rockbox, not Rock Box ...:)
15:34:08Lloreanamiconn: No harm in including presets for every target that builds mpegplayer, I think.
15:34:28amiconnIn fact they would just add one screen size to the list - the mini
15:34:41amiconnAll other greyscale targets are 160x128
15:34:52preglowamiconn: btw, does libmpeg only decode luma for our grayscale targets?
15:35:15Genre9mp3pondlife: aha, so I wasn't the only one who got the mail
15:35:19Lloreanamiconn: Right now I've named the video presets by target name
15:35:20 Join aliask [0] (
15:35:25pondlifeNo, you were...!
15:35:33amiconnpreglow: I don't know. I asked linuxstb whether libmpeg2 can o that, but he also doesn't know
15:35:37Lloreanamiconn: Partially because, for example, the iPod Video should be handed a much lower framerate video than the Gigabeat, etc.
15:35:39pondlifeBut the site refers to Rock Box
15:35:48Lloreanpreglow: Can mpeg2 encode a video with just luma?
15:36:03Lloreanpreglow: If so, I could add that for grayscale target presets.
15:36:07Genre9mp3pondlife: I see... I'll mention this in my reply
15:36:11amiconnHmm, would make sense to already encode them as grey
15:36:14preglowamiconn: but we don't right now, no?
15:36:31amiconnWould save some decoding time even when we could ignore chroma
15:36:47amiconnThe chroma would still need to be skipped, meaning some minimal decoding
15:37:21amiconnLlorean: Hopefully that'll change a bit for the better soon...
15:37:51amiconnpreglow: I have no idea...
15:38:08Lloreanamiconn: Still, I've pretty much tried to hand pick values target by target. For example, I encode audio for portalplayer targets at higher bitrate than for the others on the assumption that it won't really hurt (other than filesize)
15:38:41amiconnAudio bitrate hits PP more than coldfire
15:38:49LloreanIn video?
15:38:54LloreanIsn't audio handled on the other core?
15:39:02amiconnAh, yes
15:39:24amiconnBut on greyscale PP it matters. Much.
15:39:59LloreanI think with this next set of presets I'll try to get feedback from an owner of a target covered by each preset then, to be sure
15:40:06amiconnMpegplayer is currently *somewhat* usable on H1x0 (too low framerate, but that will hopefully be solved soonish as well)
15:40:20amiconn..but on greyscale PP, it's nothing but experimental
15:40:45amiconn(H1x0 statement also applies to M5)
15:41:42amiconnOn greyscale iPods, audio playback in mpegplayer stutters. This is due to the cpu both decoding audio and running the graylib
15:42:51Nico_Ppondlife: ping
15:42:53LloreanAlright, so lower bitrate audio for grayscale iPods than for colour iPods.
15:42:57amiconnUnless we get our libmad much more efficient for arm, this problem cannot be solved
15:43:09Nico_Ppondlife: have you tried mob on your H300 ?
15:43:17pondlifeYes, a little
15:43:32pondlifeNot much changed, I'm afraid.
15:43:43amiconnThe graylib needs around 50% cpu power due to its isr, and the mono lcd bridge doesn't allow going faster
15:43:50pondlifeSkipping during buffering caused locks
15:43:58pondlifeThat's the main issue really
15:43:59Lloreanamiconn: Care to hazard a guess at a bitrate I should pick for mp3 audio then?
15:44:22pondlifeI've been looking into another problem in SVN today, so not on MoB at the mo.
15:44:33amiconnNot sure at all. It would be helpful if mpegplayer would support lower sampling rates than 44.1kHz...
15:45:15amiconnThat'd require sample rate configuration on PP, which we can't do yet
15:45:40bluebrotherhmm, seems like Display FPS isn't implemented on h100?
15:45:42jhMikeSamiconn: we could run it through DSP np. I don't know why I haven't just done that.
15:45:49amiconnbluebrother: It can't
15:45:57Nico_Ppondlife: you haven't found a reliable way to cause the lockups ?
15:46:31amiconnbluebrother: Or rather, it could, but not for fullscreen video
15:46:53barrywardellbluebrother: do you need a rbutil build for osx?
15:46:56linuxstbOr we could render it into the yuv buffer before blitting...
15:47:13Lloreanamiconn: Okay then, at some point when I have test videos I'll track down someone with a gray iPod and see which ones stutter
15:47:24pondlifeNico_P: Just keep skipping backwards
15:47:39bluebrotherbarrywardell: would be great
15:47:55amiconnLlorean: It will be model dependent. Larger LCD == worse, so the minis will be better, with the 2nd gen worse than the 1st
15:47:57barrywardellok, i'll build it now...
15:47:58Nico_Ppondlife: with the playlist, as before ?
15:48:05LloreanjhMikeS: Should I assume that resampling will happen in mpegplayer and create presets on that assumption?
15:48:12pondlifeI was just using the left button.
15:48:16amiconn(LCD hookup in mini 2nd gen is serial so it's solower than the others)
15:48:27pondlifeMind you, I had SVN lock up this morning...
15:48:39Lloreanamiconn: Okay, then I'll make more presets and try to get lots of testing.
15:48:41Nico_Phehe, SVN isn't bugfree either ;)
15:48:44pondlifeSo maybe something in the H300 hardware code..
15:48:58pondlifeThis was just navigating through settiings
15:48:59jhMikeSLlorean: no, in the core DSP. Lowering the codec samplerate is a bigger deal than resampling anyway.
15:49:22amiconnLlorean: 1st..3rd gen ipods are still worse - slower because their broken cache *and* still running single core
15:49:33LloreanjhMikeS: Well, I rather meant, "should I assume that mpegplayer will be able to handle lower samplerate audio at some point soonish"?
15:49:37jhMikeS1) decode a frame 2) process it 3) put it in the mpegplayer pcm buffer
15:50:05linuxstbjhMikeS: Is your SPC multi-core codec in SVN yet?
15:50:21Lloreanamiconn: So I should assume video probably won't be "playable" on those for the time being?
15:50:29LloreanAt least, not watchable
15:50:33jhMikeSLlorean: I guess I could do a few hrs and get it to work
15:50:46amiconnI think so
15:51:18LloreanjhMikeS: Well, I'm leaving town for the weekend in a few hours, so I won't be testing the presets until I get back anyway.
15:51:22jhMikeSlinuxstb: I can't do that without making it swappable. No multithread codec can be swapped, single or dual yet. The threads will run again if they're not blocked before the swapout happens.
15:51:22LloreanAt the earliest.
15:51:47LloreanWell, rather, I won't be testing them on any target I don't own.
15:52:15linuxstbjhMikeS: OK. I'm thinking about trying the same thing for the APE decoder, but I doubt I'll be doing it in the immediate future anyway.
15:52:50linuxstbI think it probably needs more single-core optimisation as well, before it's worth trying dual-core.
15:52:50jhMikeSLlorean: ok, it probably time I did that little bit of work anway...was planned for a long time now.
15:53:05markunpreglow: looks like with mdct_arm.S I go from 617.6 to 617.8 % realtime on my test file, I had expected much more
15:53:51jhMikeSThe worse SPC's will play on dual-core with echo at about 85% boost if that gives any perspective. Some arm asm was put in there but doesn't have the profound effect emac does on cf.
15:54:24 Quit XavierGr ("One firmware to rule them all!")
15:54:43jhMikeSThat is with full parallel processing. EMU on COP. PCM/DSP on CPU.
15:55:01linuxstbjhMikeS: The files that really need to be realtime for the codec to be useful are the -c3000 encodings - and they're only 37% realtime at the moment on PP....
15:55:33Lloreanlinuxstb: At what compression level does ape start routinely coming out smaller than the best flac can do?
15:55:44 Join scorche|w [0] (
15:55:51LloreanOr "the best any of our other lossless codecs can do" even
15:55:53jhMikeSIf decoding and output can be split cleanly, it shouldn't be difficult.
15:55:58amiconnWell, -c2000 is called "normal"
15:56:35linuxstbamiconn: -c5000 has finished on my gigabeat - 17%
15:56:53Lloreanamiconn: How many people do you think willingly use a very cpu intensive codec that's not widely supported and just use "normal"? :)
15:57:08 Quit scorche|w (Client Quit)
15:57:52preglowmarkun: how much cache does gigabeat have?
15:58:03linuxstbLlorean: In my unscientific test of 1 track, even -c1000 is smaller than flac -8
15:58:09jhMikeShow much more compression does APE offer anyway?
15:58:16amiconnlinuxstb: Like before, as I would have expected because the filtering didn't change
15:58:19markunpreglow: 16KB data and 16KB code
15:59:13 Quit ramon8 (Read error: 110 (Connection timed out))
15:59:14*jhMikeS sees Llorean asked that Q too essentially
15:59:22LloreanjhMikeS: The hydrogenaudio wiki suggests FLAC is "58.7%" and Ape is "55.5"
15:59:23amiconnAnd filtering is the major cpu drain in the higher modes
15:59:35 Join scorche|w [0] (
15:59:41LloreanIt doesn't say what compression level is used for those, though
15:59:55LloreanIt also says Monkey's doesn't support replaygain.
16:00:12LloreanI would assume that's more a matter of whether the tagging supports it.
16:00:20linuxstbHere is my test track with all 5 APE levels, compared with flac -8 -
16:00:56jhMikeSUsually expect 60% for FLAC at highest on most music. Exhaustive searching doesn't do a heck of alot better for it.
16:01:31Lloreanlinuxstb: Looks like 2000 is probably acceptable to "settle" for at least
16:01:54 Nick parafin|away is now known as parafin (
16:02:04jhMikeSfor that tiny difference APE sucks all the processor power? hmmm.
16:02:27linuxstbjhMikeS: Exactly... But some people prefer maximum possible compression.
16:02:44amiconnjhMikeS: I have several tracks where flac only compresses to around 75%
16:02:49*jhMikeS rolls eyes
16:02:51amiconnflac -8, that is
16:02:55jhMikeSamiconn: white noise?
16:03:28n1sThat new TAK codec seems interesting, comparable compression to APE and seems quite fast, at least on computers
16:03:55jhMikeSI average better than that. FLAC seem to do better than 60% 1/2 the time and worse the other half. Average over thousands of tracks was about 60%.
16:03:57LloreanjhMikeS: I have to admit, my average with flac seems to be in the mid to high 60s.
16:04:01linuxstbn1s: Yes, if the author ever gets around to releasing the source....
16:04:29n1slinuxstb: yeah, and he want's to rewrite the whole thing in c first too iirc...
16:04:38*linuxstb never understands when people say "the source is in a bad state, but here's a binary..."
16:04:40jhMikeSn1s: as opposed to doing it by hand? :P
16:05:33n1sjhMikeS: yup, by hand encoding has proven quite slow indeed :-D
16:06:09n1si mean I read some comparison of encoding and decoding speed compared to flac on a computer and tak was faster
16:06:09markunlinuxstb: shame?
16:06:11jhMikeSRPN calculators speed it up a bit though
16:07:30linuxstbmarkun: But would you think less of someone who's invented a lossless codec just because their implementation wasn't perfect? I just think people care about the wrong things...
16:08:55 Quit RoC_MasterMind (Connection timed out)
16:09:08scorche|wmarkun: some decent pictures here too:
16:09:15 Join RoC_MasterMind [0] (
16:10:03scorche|wincluding one at the pizza party showing me with a wet bottom :)
16:11:36Nico_Pscorche|w: in, isn't that a rockbox guy holding a beer? :p
16:12:04Nico_Poh and he does seem to have a wet bottom :D
16:12:17pondlifeHope no beer was wasted
16:12:41scorche|wNico_P: you are going to have to host it on a personal site...websense blocks it..
16:13:09markunscorche|w: I even found myself in one of the pictures :)
16:13:33scorche|wmarkun: the one looking up at the google sign has all 3 of was perfect
16:13:39preglowjhMikeS: got to thinking, when we get realtime dsp, we won't have the benefit of iram buffers coming straight from the codec anymore, that sucks :/
16:14:12markunscorche|w: at least 3 pictures with me so far
16:14:45markunscorche|w: which is the one with the google sign?
16:15:09scorche|wmarkun: a bit earlier up...(i cant see it because it is blocked
16:15:16scorche|wactually, let me try something...
16:15:17jhMikeSpreglow: indeed...something I had though about. it's got some serious difficulties.
16:15:47markunscorche|w: ah yes, I see us now :)
16:16:44markunNico_P: who else are wearing baby blue shirts :)
16:17:09jhMikeSpreglow: Perhaps with the proper threading, only the buffer length of the codec would be needed anyway.
16:17:24scorche|wNico_P: yup...that is me, and kkurbjun is on the right
16:17:45preglowjhMikeS: that's another little shitty thing with preemptive multithreading, we can't just share iram around when we know it's not used by other people
16:17:52Nico_Pscorche|w: he's the one reaching for pizza ?
16:17:59preglowjhMikeS: well, we can if we mutex them ;)
16:18:10jhMikeSwhy not? if you "know" it's not used.
16:18:27preglowjhMikeS: if the codec is always running in a thread of its own, preemptively, how would we know?
16:18:46jhMikeSyou ask it or queue it
16:20:04 Quit Zagor ("Client exiting")
16:20:46Nico_PjhMikeS: what's the current state of multicore usage ?
16:21:07jhMikeSusage? just mpegplayer for useage.
16:21:24Nico_Pare some threads moved to the COP?
16:22:01jhMikeSnot yet
16:22:22preglowjhMikeS: well, we need a way to solve that anyway, buffer in ram will absolutely crush dsp performance, and that needs to be tight if it's going to be realtime
16:23:29preglowouch, illinstr
16:24:22preglowat 10002000, wtf
16:24:44jhMikeSbuffer in ram won't crush DSP on an ARM processor with a good data cache at all
16:25:09preglowin the middle of crossfeed_data...
16:25:15preglowjhMikeS: talking coldfire
16:26:54preglowlinuxstb: do asf packets point to the next packet?
16:29:41linuxstbpreglow: Not as far as I know.
16:30:36Nico_Ppondlife: I've managed to make the sim segfault, but it hasn't been very informative so far
16:30:47 Join saratoga [0] (i=9803c6dd@gateway/web/cgi-irc/
16:31:02Nico_Pyes, but the backtrace was less useful than I hoped
16:31:55pondlifeMaybe you could make the timing super bad by deliberately slowing down the buffering? Read 1 byte at a time or something?
16:32:17Nico_PI've slowed it down already yeah
16:32:27Nico_Pnot as badly but enough I think :)
16:32:30 Join toffe82 [0] (
16:33:11Nico_PI think I need to protect the linked list anyway, so I'll add mutexes and see if they improve the situation by any chance
16:34:01Nico_Pthe backtrace *seems* to indicate the linked list could have been in an inconsistent state
16:36:42markunwould it be useful to have a atrac3 decoder in rockbox?
16:37:18linuxstbI used to have one atrac3 file (Realaudio), but can't find it any more...
16:37:56linuxstbYou can't have too many audio codecs though...
16:38:09barrywardellbluebrother: also, it is expected that updating an already up-to-date bootloader fails on h10?
16:38:21*preglow wants a regular atrac decoder
16:39:06linuxstbpreglow: Do any software encoders exist?
16:39:30preglownone i know of
16:39:46preglowin a similar case to wma, i also have one minidisc release i want to be able to decode perfectly :P
16:40:05linuxstbSo you can rip the atrac data from the minidisc?
16:40:45preglowi assume it's possible..
16:40:55linuxstbafaik it's never been possible...
16:41:22preglowreally? that sucks
16:42:14linuxstbafaik, the best you can do is to grab the uncompressed output via s/pdif.
16:42:28preglowahh, elite
16:42:35preglowi see it's now available as flac on
16:44:04 Join Zap [0] (
16:44:09Zaphi to all :D
16:44:26saratogathere was once a directshow atrac3 encoder
16:44:34saratogait might have been able to decode
16:46:08saratogaobviously not gpled though
16:46:45Zapno news for ipod nano 2 gn and rockbox ? :(
16:47:06pondlifeHmm, I just IllInst-ed in mpegplayer on H340
16:47:07GodEater_no news because no-one's working on it
16:47:33pondlifeAll I did was leave the video paused for 5 mins, then pressed STOP
16:48:42Zapuh ok :(
16:49:23scorche|wwhoops...that would be my armrest saying hi...
16:50:55markunsaratoga: we could use the ffmpeg atrac3 code
16:51:25markunprobably more open than the directshow filter :)
16:52:47saratogai didn't realize there was one in ffmpeg
16:52:54markunI thought so :)
16:52:58saratogamight be fun
16:53:00markunalthough I've never tried it
16:53:40markunlinuxstb: together with the cook codec we should be able to support quite some real-audio files then, right?
16:55:30bluebrotherbarrywardell: no, I don't think that should fail.
16:56:07barrywardellbluebrother: it only happens when doing a complete install. I'll look for the problem and let you know if I can find it
16:57:03bluebrotherDomonoky did the bootloader check stuff so you might want to ping him too.
16:58:26 Quit mf0102 (Read error: 110 (Connection timed out))
16:58:58 Join mf0102 [0] (n=michi@
16:58:59 Quit aliask ("ChatZilla [Firefox]")
16:59:36 Quit petur ("connection reset by beer")
17:07:30barrywardellak, thanks
17:12:01preglowwoot, my file decodes almost to the end
17:13:11***Saving seen data "./dancer.seen"
17:16:13Calciphermorning/afternoon/evening all
17:16:42nicktasticDoes rockbox provide a way to increase loudness beyond the standard volume control - e.g., preamp?
17:17:11 Quit Zap ("Sto andando via")
17:17:27preglowwoot, my file decodes to the end :P
17:18:40 Join Crash91 [0] (n=evil91@
17:18:44markunpreglow: did you fix a bug in the decoder?
17:18:50markunhi Crash91
17:19:07preglowmarkun: nah, i just started ignoring errors in the stream a number of times
17:19:13Crash91just wanted to ask, how the USB driver going?
17:20:15markunCrash91: last thing I heard, Zagor was making good progress with the ums driver
17:20:21 Join Addi2 [0] (
17:20:44*scorche|w wonders if he should be making a rule about no more asking
17:21:02*Crash91 thinks you should put it in the topic
17:21:18scorche|wwell, yeah, i would..
17:21:28Crash91markun: UMS is the same as MTP?
17:21:34scorche|wno no no
17:21:47scorche|wMSC, you are thinking of
17:21:53Crash91or was it UMD? something like that...
17:22:16Crash91i remember there is MTP and MSC and theyre known as two others....=P
17:22:29scorche|wmsc = ums
17:22:38Crash91so UMS=MSC and erm ?UMD?=MTP?
17:23:03scorche|wunicorns of mass destruction?
17:23:04saratogaUMDs are optical disks
17:23:11saratogathe psp uses them i think
17:23:33*Crash91 is very confused
17:23:38 Quit barrywardell ()
17:23:51saratogaonce ums is working, we could implement MTP though, if someone wanted
17:24:08saratogamight be interesting to do the apple and ms protocols
17:24:16saratogahave a player that syncs with everything
17:24:22Crash91well i guess the only advantage of MTP over MSC in the OF is that MTP doesnt need a database refresh
17:24:38saratogathis has nothign to do with the OF
17:24:59saratogathis would be rockbox's USB mode, which doesn't need to refresh the database up front
17:25:14scorche|wsaratoga: why would we want to do MTP?...cant everything be done with UMS?
17:25:28Crash91yeh, but MTP is stupid...MSC can serve the exact same purpose..with the advantage of letting you use files off disk
17:25:43saratogawell theres probably someone out there who wants to sync to wmp, and if he submits a patch i don't see any reason to reject it
17:25:52Crash91MTP is just for transferring music and AFAIK, for WMP
17:26:04Crash91well WMP does the exact same thing with MSC
17:26:07scorche|wsaratoga: but cant you do that with a UMS device?
17:26:13Crash91it detects it either way
17:26:25Crash91scorche|w: yes, you can
17:26:30saratogawith MTP WMP could transfer play counts, song rating, etc I guess
17:26:40saratogai think it allows more integration
17:26:47saratogathough i haven't looked at WMP in years
17:26:55 Quit FOAD ("I'll be back")
17:27:02Crash91i still use it because of the auto AA downloader
17:27:11Crash91but otherwise...VLC
17:27:26saratogaalso, you might need mtp to do AA, though i'm just BSing here
17:27:49Crash91no i dont think so
17:27:52Crash91no, you dont
17:28:11Crash91i transferred over MSC recently and i did get AA
17:28:16Crash91in OF of cosue...
17:29:16Calcipherwell I have better luck transfering music to the players music directories automatically, under winamp 5, but only when my e200 series is set on mtp mode
17:29:18 Quit Gnu47 (Nick collision from services.)
17:29:25 Join Gnu47 [0] (
17:29:54Calcipherotherwise in msc mode winamp just sends over the music to the root, treating the player as a mass storage device than a player
17:30:20Calcipheralso doesnt see it as e280r when in msc mode, only as drive letter
17:30:25Crash91well...differenent players i guess
17:31:09*Crash91 is off to play games
17:31:14 Quit Crash91 ("Bye Bye!")
17:31:26Calcipherthe integration is flawless when the device is set to mtp mode, seems more programs like mtp mode
17:31:57scorche|wwindows programs, at least
17:32:25saratogai'd rather have itunes support, then i could sync to foobar w/ foo_dop
17:32:29Calciphermeans I don't have to worry about manually organizing and reordering things on the player
17:33:02saratogause it to upload my logs
17:33:38*pondlife hopes for his first green delta...
17:33:40 Join FOAD [0] (
17:34:31Calcipherso was there any progess with that mpegplayer / voice menus issue we were talking about a few hours ago?
17:34:57Calcipherjust curious, I'm not that hung up about it personally
17:35:06 Join lazka [0] (
17:37:58markunCalcipher: if it's just a few hours ago you could check the IRC logs
17:38:46CalcipherI can give it a look, I'm new to this community, and all its neat info
17:39:05 Join linuxstb_ [0] (n=linuxstb@rockbox/developer/linuxstb)
17:39:18amiconnLooks like I'm unable to spot the bug in my coldfire predictor asm :(
17:39:35pondlifeCalcipher: I thought I'd fixed it, but no. not yet
17:39:54pondlifeYou may want to try the new build and let us know though
17:40:11pondlifeIt now works some of the time...
17:40:16pondlifeWooh, green delta !
17:40:17Calciphersure, so the latest currect build?
17:40:41Calcipherok, I'll let you know, so what targets was it tested on so far??
17:40:46 Quit linuxstb (Read error: 113 (No route to host))
17:40:48pondlifeI only have an H300
17:41:05pixelmapondlife: but a little bit of yellow ;)
17:41:13Calcipherah ok, I'm jealous, I always wanted one of those but was too broke
17:41:29Calcipherinstead bought an 80 dollar used e280r
17:41:56 Join Domonoky [0] (
17:44:03pondlifeIt's always the MAS sims....
17:46:18n1slinuxstb_: thought you might be interested in news on frontpage stating that they changed the license of faad2 so that the fsf stated that it now os gpl compatible
17:49:39preglowwhy, that rocks
17:49:43preglowtoo bad it's still shite :)
17:50:51Calcipherpondlife should I run rbutil and do a full install
17:50:57Nico_PjhMikeS: would it be sane to have one mutex per handle to prevent closing or moving a handle that is being buffered ?
17:50:59pondlifeThat should do, yes
17:51:05Calcipheror just a rockbox install without the bootloader update
17:51:07Calcipheroh ok
17:51:16pondlifeNo need for bootloader update
17:51:33saratogai like how faad claims to be only "reasonibly fast"
17:51:42Calcipherok, I thought I saw someone mention it was good to up that too
17:51:47preglowit certainly isn't more than that
17:51:58pondlifeCalcipher: If you haven't updated recently, then yes.
17:51:59preglowi want the helix one :/
17:52:16jhMikeSNico_P: well, there's one spinlock per thread...fine grained locking can be beneficial...but it's tricky to make sure all heirarchy is correct.
17:52:38saratogayes i'd rather the helix people become gplv2 compatible
17:53:06pondlifepixelma: Better now? I think the other yellow was already slopping around...
17:53:21Nico_PjhMikeS: "there's one spinlock per thread" => what does that imply ? and so do you think it's a reasonably good idea ?
17:53:50Calcipherfirst I'll use mpeg player with all voice options disabled
17:53:52pixelmapondlife: yes, I know. Only meant the yellow for hwcodec recording sims
17:54:03pondlifeJust checking
17:54:58jhMikeSI'm talking about the dual-core scheduler because it's analagous. as long as the lock is used on both ends and actually provides freer access to data it is. if times spent in a lock are really short, possibly not.
17:56:05Nico_PI'm having a hard time followong you :)
17:56:16pondlifejhMikeS: I believe Nico_P's having a problem on a single core, is this also relevant there?
17:56:58jhMikeSIf the critical section is finished quickly, it's probably overkill to provide a mutex for each handle but better to use one.
17:56:59Calcipherwow, very nice pondlife
17:57:06pondlifeIt's not fixed
17:57:16pondlifeIt's become intermittent, I think.
17:57:27CalcipherI just tested the elephants dream with voice directories enabled, and theres no issues
17:57:52Calcipheroh boy, thats never good when problems become "wacky"
17:57:54jhMikeSFor dual-core, these things are done to prevent one core from interfereing with another unless they have some sort of sending a message or waiting on the same object.
17:57:55Nico_PjhMikeS: the problem is that one mutex would probably be enough, but I need to know which handle locked it
17:58:40pondlifeNico_P: Which threads are contending? The buffering thread and a.n.other API-calling thread?
17:58:43jhMikeSLock the mutex, save the handle. Just before unlock, NULL the locker.
17:58:49 Join linuxstb__ [0] (
17:58:50Nico_PI think my problem is that currently a handle can be closed or moved while being buffered. I need to control that
17:59:02 Part Llorean
17:59:32pondlifeNico_P: Can you make it so that only the buffering thread accesses the data structures?
17:59:39Nico_Ppondlife: yes, buffering is on the buffering thread (duh!), and bufclose can be called on the handle that is being buffered
17:59:56Nico_Ppondlife: not really
18:00:03pondlifeOK, so bufclose just sends a message to the buffering thread to tell it to close it?
18:00:07Calcipherits back now
18:00:16amiconnpreglow, linuxstb: Any idea how to debug this beast short of staring at it again and again, comparing implementations?
18:00:17Nico_Ppondlife: yeah, maybe
18:00:28Calcipherwhen I disabled directory voice, and activated menu voice
18:00:55pondlifeCalcipher: It's intermittent, not based on a particular setting. Do you find it locks hard when it fails?
18:01:06pondlifei.e. locks up with a black screen.
18:01:10Calcipherbut a strange thing i noticed, I was initially able to use the controls when the video started, because I saw the wheel light up on input
18:01:17 Join linuxstb [0] (
18:01:39pondlifeI suspect it's got stuck waiting for the voice thread.
18:01:47jhMikeSNico_P: you need a handle usage count really.
18:01:57Nico_PjhMikeS: per handle ?
18:02:20Calcipherbut after a second or two it stopped responding to input and the wheel light was stuck on, and now I will be hard resetting, the video is playing but with no sound or response to user input
18:02:27linuxstbn1s: Unfortunately Nero have still not made faad2 GPL compatible.. I mentioned it here a few days ago.
18:02:35jhMikeSprotect them all with the same sync object. though if a handle is no longer needed, should buffering continue on it?
18:02:45Nico_PI had thought of that, but only for multiple bufopens on the same file... currently one bufopen == one handle id. why do I need an usage count ?
18:03:08n1slinuxstb: too bad :-/
18:03:10Nico_PjhMikeS: the idea is that if a handle isn't needed it's closed
18:03:14 Join ender1 [0] (
18:03:16 Join kugel [0] (n=kugel@unaffiliated/kugel)
18:03:30preglowamiconn: i'm afraid no. the only way i can drastically improve my chances of finding a bug is not looking at the code for a couple of days
18:03:47preglowamiconn: i've got time to take a small look if you want
18:03:59 Quit ender` (Read error: 104 (Connection reset by peer))
18:04:16advcomp2019pondlife, i decided to try it myself... the video stays going but they controls does not work until you do a hard reset
18:04:26 Join Domonoky_ [0] (
18:04:33Nico_PjhMikeS: Ideally closing a handle would interrupt its buffering... maybe making buclose send a message to the queue is the best thing to do, as the buffering loop checks for messages
18:04:37 Quit Rob222241 ()
18:04:37pondlifeadvcomp2019: Which device?
18:05:17advcomp2019the sansa e200 series
18:05:20 Join Rob2222 [0] (
18:05:26jhMikeSNico_P: so you do have a race condition. I suspect buffering sould enter the mutex when it's doing something with the data. When a handle is being changed, that mutex should be used to exclude the buffering thread?
18:05:38Calcipherdoes it play with sound on yours advcomp2019
18:05:39linuxstbn1s: If you're interested, see starting at 11.57.48
18:05:46pondlifeNico_P: The API routines themselves shouldn't access the internal data at all, not even to attempt sanity checks
18:06:01jhMikeSQueues are a nice cheap easy way to sync stuff of course.
18:06:17advcomp2019Calcipher, i can test it again
18:06:47Nico_Pturning all API calls into queue senders seems like a big complication to me. some taks multiple args
18:07:09Calcipherok, and also please see if this happens when you disable voice menus, but enable voice directories, or voice files
18:07:12pondlifeadvcomp2019: Calcipher: Do we know which revision started crashing?
18:07:24pondlifeProbably worth working backwards to find outr
18:07:36nicktasticDoes rockbox provide a way to increase loudness beyond the standard volume control - e.g., preamp?
18:07:38Calcipher15179 atleast thats the first one I noticed it on
18:08:06pondlifeCalcipher: If you could try 15178 and work back one at a time, that would be best
18:08:18nicktasticIf not, maybe replay gain could be used to effectively increase loudness?
18:09:03CalcipherSure, I'll be glad to test the builds, but I can't do that at the moment, later today when I return
18:09:52Nico_PjhMikeS: I'm not sure full mutual exclusion between buffering and the rest is necessary. things are pretty well contained IMHO
18:10:18Nico_PI think bufclose is the only real problem left
18:10:42Nico_P(now that I've protected the linked list)
18:10:47amiconnpreglow: I know that there is still a bit of optimisation potential (reordering instructions, mostly). That's on purpose - it's as close to the arm version as it can atm.
18:10:47advcomp2019pondlife, Calcipher is right no sound too
18:11:45jhMikeSNico_P: if it's not need then of couse don't
18:11:48 Quit linuxstb_ (Read error: 110 (Connection timed out))
18:12:14n1snicktastic: the RG preamp can probably do what you want but it will most likely introdue clipping
18:12:29jhMikeS_but is it safe for both COP and CPU to use at the same time?
18:13:02Nico_PI think so
18:13:23jhMikeSok, cool
18:13:40preglowamiconn: what size is the offset part of the cf indexed addresing mode?
18:13:45preglowamiconn: signed eight bits, no?
18:14:04amiconnI don't use indexed addressing at all
18:14:10preglowno, im just wondering
18:14:16nicktasticn1s, Hmm, so is there no other approach to increase loudness, that you know of? (I don't want to noramlize, but after hours of googling, RG seems to be the only option)
18:14:23amiconnJust offset addressing, and that is 16 bit signed
18:14:30Nico_PjhMikeS: each handle has its own reading and writing end. the reading is managed by the calling threads, and the writing is managed by the buffering thread
18:14:34preglowyesyes, but that's not why i'm asking
18:14:41Avewould it be possible to use the graphic eq to implement a dolby nr kind of pre-emphasis to outputted audio?
18:14:48 Quit ender1 (Read error: 104 (Connection reset by peer))
18:14:51Avefex when using a cassette-adapter for old car player
18:14:58jhMikeSNico_P: but what about the access of the handle data itself?
18:15:17amiconnpreglow: Indexed addressing only has 8 bit signed offsets
18:15:50 Join ender` [0] (
18:15:51Nico_Pwriting is done only by the buffering thread, reading is done only by the others
18:16:00n1snicktastic: the dac in your dap is probably the limit, increasing the signal in the digital domain will introduce clippign quite fast in most cases so if you really need higher volume a separate amp is better
18:16:43Nico_PjhMikeS: well actually bufalloc is a bit of an exception... it can be used by a caller thread to write data to a handle
18:17:25Nico_Pit opens a handle then copies data over to it... all done in the calling thread
18:18:33Nico_PI fail to see how it can be a problem though... bufalloc is only used for metadata. data is copied once and then the handle is basically read-only
18:18:40jhMikeSbut how many calling threads are there?
18:18:45 Join chandlerc [0] (n=chandler@
18:18:49pondlifeLots, in theory
18:19:26Nico_Pcurrently only the audio thread an the codec thread. audio thread does the bufopen and bufclose and codec thread reads the data
18:19:49nicktasticn1s, Well, I'm asking in regard to two albums in particular, which are unusually quiet. My DAP can get quite loud, but these albums are just very quiet. I was hoping there would be 'room' to boost the signal without clipping, but I have little understanding of DSP. ;)
18:19:58Nico_Pthe wps thread is also a user of the metadata handles I think
18:20:09nicktasticn1s, thanks for the advice
18:20:23n1snicktastic: that sounds exactly what RG was made for
18:20:31pondlifeUltimately video could use the buffering too
18:20:49pondlifeBest to assume that your API routines shouldn't do anything unsafe
18:21:22markunpondlife: can other plugins also use the buffering to get some extra memory (like the jpeg viewer)
18:21:51 Quit Domonoky (Read error: 110 (Connection timed out))
18:21:55Nico_Ptrue, but are the calls unsafe if they each have a well defined role (which I think is the case, except for bufalloc and once I'll have fixed bufclose)
18:22:17pondlifemarkun: Anything that needs to read a list of files...
18:22:44Nico_Pmarkun: hopefully yes but they'd have to negotiate with the playback code to get it to free some space
18:22:45pondlifeNico_P: Depends mainly on what data they access.
18:23:03jhMikeSNico_P: so the codec thread never closes anything?
18:23:11Nico_PjhMikeS: no
18:23:21pondlifeNico_P: Actually, the buffering code would be allocating memory anyway, it's all one big buffe
18:24:05jhMikeSbut could the codec thread be using it when the audio thread closes it?
18:24:45Nico_PjhMikeS: I'm pretty sure it couldn't
18:24:57 Join hcs [0] (n=agashlin@rockbox/contributor/hcs)
18:25:03Nico_Pthe audio thread closes handles the codec doesn't need anymore
18:25:25jhMikeShmmm...pretty sure doesn't cut it :) esp. on two cores. the logic should mutually exclude it.
18:25:40Nico_Ppondlife: I'm not sure what you mean, but bufopen fails when there's no space left. it won't forcefully close handles to create free space
18:25:57jhMikeShow does it know the codec thread doesn't need it btw?
18:26:34pondlifeI would work on the basis that all API calls should be thread-safe
18:26:49*jhMikeS does need a nice going over to understand the code so be able to comment better
18:26:58Nico_PjhMikeS: that's through track_ridx. tracks[track_ridx] tells us which handles the codec needs
18:27:12preglowamiconn: can't see anything obvious
18:27:36Nico_PjhMikeS: track_ridx changes when the codec switches tracks
18:28:21Nico_PI've kept most of the architecture in playback.c in that respect
18:28:37jhMikeSeek...I thought that would be history to not have a track limit
18:29:00pondlifeOne thing at a time... :)
18:29:14pondlifeThjis is about buffering.c, not playback.c
18:29:35Nico_PI do need to store handle IDs somewhere, otherwise playback.c needs access to the memory handles
18:29:49jhMikeSwell, you don't want to write buffering.c to accomodate playback.c...well not if it's not easily revised anyway
18:30:30pondlifeI'm ok with playback.c having a static list of buffer handles, the structure is small
18:30:47pondlifeWe can sort that out later if needed
18:31:00preglowis there any way of making rockbox play a sequence of tracks randomly without having to do shuffled playlists?
18:31:07jhMikeSdoes it really need to? why can't it access the memory handles?
18:31:39pondlifeThe buffer API could have a FindFirst/FindNext added... later.
18:31:56pondlifeBut for now, let's fix up the race conditions
18:32:18pixelmapreglow: you mean without setting it to shuffle in the settings?
18:32:18pondlifeGet something stable to commit and build upon
18:32:31jhMikeSI'm wondering if the that isn't a critical part of doing so in some way.
18:32:35n1spreglow: insert shuffled?
18:32:40preglowpixelma: i think i should just read the manual...
18:32:48 Quit Mouser_X (Read error: 104 (Connection reset by peer))
18:32:50pixelmapreglow: "insert shuffled" on a folder or song
18:33:05pixelmaI'm not sure if that is in the manual...
18:33:06Nico_PjhMikeS: well there's nothing saying playback.c really can't access the linked list of handles, but then we'd need to ensure it never writes them. I think a level of abstraction is a good thing
18:33:10 Join BigBambi [0] (n=Alex@rockbox/staff/BigBambi)
18:33:12preglowi don't want the tracks to be played in the same order over and over again
18:33:30pregloweven if the order is shuffled
18:33:56Nico_Pthe point was to hide the buffering process behind a set of well-defined (and possibly thread-safe) API calls
18:34:07preglowi think what i really want is of-style random playback
18:34:10n1spreglow: write a plugin that keeps shuffeling :-)
18:34:19pondlifepossibly? definitely thread-safe
18:34:25*linuxstb notices that all codecs apart from AAC appearto fit in the iFP's 128KB codec buffer
18:34:31pixelmapreglow: "repeat shuffled" though I'm still not sure I understand fully
18:34:44amiconnpreglow: Set your repeat mode to random
18:34:55amiconnThen it reshuffles at the end of the list
18:35:13preglowamiconn: but it still plays all the files in a list through before one can reappear, yes?
18:35:31preglowamiconn: i don't want that :)
18:35:36preglowlinuxstb: including bbs?
18:36:09linuxstbpreglow: Yes, the whole codec buffer is 128KB IIUC
18:36:19linuxstbThat's assuming current SVN compiles on the iFP..
18:36:55linuxstbRockbox currently gives 1MB to codecs on most targets (512KB buffer + malloc buffer)
18:37:04preglowan abominaiton, it is
18:37:07preglowthat malloc buffer...
18:37:13amiconnIt's a pity that the iFP port is basically a one-man port, and tomal isn't around often :|
18:37:42saratogawhat codecs actually need a malloc buffer that big?
18:37:44jhMikeScan't say I'd ever want one myself
18:37:47linuxstbpreglow: I agree. At the very least, we should try and remove that, and make codecs malloc out of the initial 512KB
18:38:27amiconnpreglow: Damn :( Thanks for looking anyway
18:38:44*amiconn goes to print the whole thing, along with the C version for reference
18:38:55Nico_PjhMikeS: I assume it would help if I wrote some kind of doc outlining the responsibilities of each call so that you can help me make them thread safe ?
18:39:13 Quit obo ("bye")
18:40:38jhMikeSwhat might help more is a playback.c without large amounts of unneeded code #ifdef 0 'ed. just to have functioning stuff only there.
18:40:53preglowwhy the ifdefs?
18:41:01Nico_Pyeah I'll take that code out soon
18:42:05Nico_Ppreglow: I disabled some chunks of code to have them at hand if needed
18:42:42 Quit lazka (Remote closed the connection)
18:43:10 Join muesli- [0] (n=muesli-@
18:44:17 Join darkapostrophe [0] (
18:44:35linuxstbHmm, iFP port fails in debug_menu.c...
18:46:19 Quit mf0102 (Read error: 110 (Connection timed out))
18:46:51 Join mf0102 [0] (n=michi@
18:47:42 Quit jgarvey ("Leaving")
18:48:48 Join muesli-_ [0] (n=muesli-@
18:49:08linuxstbSeems musepack, wma, APE, SID, NSF and SPC all need more than a 128KB codec buffer...
18:49:13 Part muesli-_
18:49:16Nico_Ppondlife: do you have your H300 at hand ?
18:49:59Nico_PI haven't been able to crash my gigabeat with my usual recipe yet
18:50:09preglowlinuxstb: yeah, it sounded too good to be true, that
18:50:25preglowvorbis fitting also surprises me
18:50:32preglowthat should be pretty small
18:52:22linuxstbThe linker complains about both IRAM and RAM being full...
18:54:24linuxstbIf I can read maps, the BSS for musepack is 150KB by itself...
18:54:26preglowspeex also has to be larger than that
18:55:07linuxstbSpeex fits...
18:55:09preglowjhMikeS: oy, is there anything that needs doing for h120 spc?
18:55:25preglowlinuxstb: that's only because it uses the malloc buffer
18:55:37jhMikeSpreglow: like?
18:55:47preglowjhMikeS: wouldn't know, which is why i'm asking :P
18:56:13preglowlistened to spcs on ipod yesterday, and that doesn't even do echo...
18:56:30jhMikeSThe DSP is totally emac. If something can be trimmed, go ahead. I did a retest of msac instead of neg and mac and it's didn't kill it this time. I don't know what was going on there.
18:56:50preglowresults were equal this time?
18:56:52preglowthat's good to know
18:57:14n1spreglow: gaussian interpolation is disabled for cf for performance reasons I guess
18:57:34jhMikeSbut something else seems troubling, it seemed fractionally faster with a trapf between msac and mac...I really need to do test_codec runs on it.
18:57:35preglowbummer, then that needs doing
18:58:31 Quit Dark_Apostrophe (Connection timed out)
18:58:31 Nick darkapostrophe is now known as Dark_Apostrophe (
18:59:13jhMikeSIn any case, I got 78% boost running SW tone controls with just msac...same as the neg/mac
18:59:16n1spreglow: about the ringing in midi playback do you think it is because of the samples being 44kHz someone in the forum thought it was this and called the effect "aliasing" and playing at 44kHz in the sim I can't hear it
18:59:17 Join BigMac [0] (
18:59:52jhMikeSmy usual dkkjd2 test for that
19:00:35amiconnjhMikeS: On cf it's normal that slight changes which change function sizes or move functions around can change performance somewhat, even if one wouldn't expect it. That's due to cache aliasing
19:00:48 Quit Addi2 ("Leaving")
19:00:59preglown1s: riiiiight, it isn't playing at 44.1khz?
19:01:39n1snope, not on cf but i intend to try that out as soon as i hunt down a bug
19:01:48jhMikeSamiconn: I figured some movement of code could do that. I don't think it's limited to CF either.
19:02:11preglown1s: you will get more aliasing at 22khz or whatever you use, yes, but it didn't sound like that was the problem
19:02:22 Quit BigMac (Client Quit)
19:02:31amiconnWell, in theory it can affect all caches. But the cf instruction cache is direct mapped, while the PP cache is 4-way set associative
19:04:04 Join spiorf [0] (
19:04:56 Quit pondlife ("Read error: 110 (Connection slimed out)")
19:05:16 Quit muesli- (Connection timed out)
19:06:48 Quit nicktastic ("Leaving")
19:13:13***Saving seen data "./dancer.seen"
19:13:50 Join Morey [0] (
19:15:44 Join ilgufo [0] (
19:17:18 Quit Guile`` ("KVIrc 3.2.0 'Realia'")
19:18:21MoreyjhMikeS: It would be alright for you to change the queue'ing stuff in mpegplayer now if you are still interested.
19:19:02 Join bertrik [0] (
19:21:21linuxstbMorey: Hi. Was there a reason you changed mpegplayer so it fills the whole buffer before starting to play, rather than just filling to the low watermark?
19:23:17jhMikeSMorey: I'm interested since that what it should be based on. Don't know if I'm up for "now" though. :)
19:24:16MoreyjhMikeS: now = anytime, with little hinder to our efforts
19:24:40 Join barrywardell [0] (
19:25:12Moreylinuxstb: That was changed by Gwynne for debugging, but it seemed to have little effect on speed so it was left. I suppose we could bench mark it.
19:25:38linuxstbMorey: What target(s) are you normally testing on?
19:26:10Nico_PjhMikeS: things look much more stable to me now that I've fixed bufclose
19:26:27Nico_Ppity pondlife is gone though
19:26:53 Quit mf0102 ("Verlassend")
19:26:58linuxstbMorey: Only buffering a small amount means playback starts almost instantly on my ipod Color, compared to waiting a few seconds with the current code.
19:27:19MoreyWe are still working on a solution to include seeking after the movie has started. The issue we have now is that if a person exits the seek without selecting a seek position, the original place in the mpeg is lost and it's hard to resume playback at exactly that spot. Which would be weird to the user.
19:28:12markunMorey: I would prefer it if "seeking after the movie has started" would replace the current seeking
19:28:15linuxstbHow can a person not select a seek position? Audio seeking doesn't work like that.
19:29:10MoreyEnter the menu - look at the seeking thumbnails - and press the power but (sansa) which exits the menu without making a selection.
19:29:58linuxstbWhy not just map seeking to long press on fwd/rew, like audio, instead of making it a menu?
19:30:12Moreylinuxstb: If changing the buffer fill back is better I don't feel it will cause any problems.
19:30:31 Quit homielowe (Read error: 110 (Connection timed out))
19:31:16Moreylinuxstb: The actual seek would take about 4 seconds and the screen would splash or clear for that time. ppl might not like that.
19:31:43jhMikeSwhy should the mpeg keys work differently than playback anyway?
19:32:43MoreyjhMikeS: I don't understand your comment.
19:33:09MoreyIt's the delay ppl would not like.
19:33:19jhMikeSI think linxstb said it better
19:33:47MoreyIt can be done, it's just slow. :)
19:33:48 Join darkapostrophe [0] (
19:33:50jhMikeSMorey: you're still buffering all data from point to point? I can't see any reason not to even display video frames while seeking rapidly. MPEG allows that.
19:35:23MoreyjhMikeS: It's the aud/vid resync that makes the delay.
19:35:38linuxstbMorey: I disagree. The problem seems to be that we need to work to reduce that 4s, rather than work around that issue.
19:36:14jhMikeSseek by the video, then find the audio frame that matches closely enough
19:36:24bertrikhow about seeking with the volume control keys while the movie is put on pause (e.g. with select keys)
19:36:30linuxstbAnd as long as a "Seeking..." splash is displayed, I think users could live with the delay.
19:36:51rasherThere's a delay in audio seeking as well in some situations
19:37:24 Quit Dark_Apostrophe (Read error: 104 (Connection reset by peer))
19:37:24 Nick darkapostrophe is now known as Dark_Apostrophe (
19:37:48jhMikeSI just can't buy that a noticeable delay should have to be there. The decoders can handle error as well, in fact that's meant for that.
19:40:07MoreyIf I get something working I'll start a patch thread and maybe some of you can help me resolve the delay issues.
19:40:57MoreyI also like the idea of supporting the avi container.
19:41:45markunMorey: because it has a jumptable at the end?
19:41:48jhMikeSI think better threading control could help too
19:41:58 Quit spiorf (Read error: 110 (Connection timed out))
19:42:00linuxstbMorey: Have you investigated whether AVI has any advantages over MPEG-PS?
19:42:16 Join spiorf [0] (
19:42:42markunlinuxstb: one advantage is that we can then say "yes" when people ask "does rockbox play avi?" ;)
19:42:59Moreylinuxstb: Sorry, been too busy. It's on my todo list though. :)
19:43:06BigBambiI redid the MPEGPlayer fps rates on the gigabeat last night, and even on that the pause before playback starting was annoying
19:43:18linuxstbmarkun: I'll leave you to explain why it doesn't play 99% of AVIs then... ;)
19:43:35markunlinuxstb: and I'll leave that to llorean :)
19:43:48linuxstbmarkun: Then we're all happy ;)
19:43:57jhMikeSwell, he can just tell everyone to encode their AVIs in MPEG streams first
19:44:08markunlinuxstb: btw, when's the next pubcon?
19:44:36linuxstbNothing planned - there was one due yesterday, but no-one could make it in the end...
19:44:48markunif I can find a cheap ticket I would like to visit you guys
19:45:08 Quit Morey ("Ninja IRC v1.5.8.1(#1) exiting after 31m36s of use")
19:45:09linuxstbI'm sure we'll become available...
19:51:10 Join Hukt [0] (
19:52:28 Join pepie34 [0] (
19:54:46 Quit zicho (Read error: 104 (Connection reset by peer))
19:55:01n1sstevenm: for the logs: your "quick & dirty" pitch bend depth committed to svn in r15112 causes artifacts in at least one of my test files.
20:00:34 Join zicho [0] (
20:04:54 Join spiffytech [0] (
20:06:12 Quit saratoga ("CGI:IRC (EOF)")
20:08:20preglowspeex currently has some test programs commited to svn, i should just go ahead and delete those, right?
20:08:43 Quit Hukt0nWax (Read error: 110 (Connection timed out))
20:08:43 Nick Hukt is now known as Hukt0nWax (
20:08:57*n1s chants kill! kill!
20:09:38 Quit barrywardell ()
20:13:55 Quit ilgufo (Nick collision from services.)
20:13:58 Join gufo [0] (
20:15:55 Join lee-qid [0] (
20:16:24amiconnHmpf, looks like this coldfire asm is fubar
20:16:57amiconnI can't spot a single mistake even by comparing it with the C code, and still it only produces static :(
20:17:25*amiconn did a full line-by-line comparison
20:19:23 Quit zicho (Remote closed the connection)
20:19:32 Join obo [0] (n=obo@rockbox/developer/obo)
20:20:15preglowamiconn: i'll bet there's something hiding around :>
20:22:22*linuxstb has similar memories of debugging the ARM version
20:22:48 Join zicho [0] (
20:24:25 Part pixelma
20:24:49*amiconn gives up
20:25:14 Join Zagor [0] (
20:25:21 Join MethoS- [0] (n=clemens@
20:25:51 Join merbanan [0] (n=banan@
20:31:35 Quit moos (Read error: 110 (Connection timed out))
20:32:52*preglow has similar memories of abandoning the new libmad imdct for coldfire :P
20:33:15 Quit MethoS- (Remote closed the connection)
20:34:48Rob2222IPOD Settings reset: "You can reset the settings with turing player off and back on with hold button oN" ....
20:35:00 Join hannesd [0] (
20:35:02Nico_PGodEater_, GodEater: ping
20:35:08Rob2222how can i start the player with hold activated?
20:35:19 Join XavierGr [0] (
20:35:37n1sRob2222: flick the hold on when it's booting
20:35:48n1sbut if you do it too fat you'll boot the OF
20:36:03 Quit spiorf (Remote closed the connection)
20:36:38Rob2222ah ok, thats the problem
20:37:39amiconnpreglow: The only thing that might help would be tracing a run, to see where it goes wrong, but I have no idea how that could be done on target ...
20:37:44 Join PaulJam [0] (
20:37:45 Join ilgufo [0] (
20:39:30n1samiconn: would comparing it to a disassembly of the c version help? else you could always code up a quick emulator :->
20:40:16 Quit merbanan (Remote closed the connection)
20:40:25amiconnThe disassembly of the C version is of course very different. Typical gcc crap for m68k/cf
20:40:33Nico_Pamiconn: then you haven't abandoned after all ? :)
20:40:35preglowunreadable too
20:41:36amiconnNico_P: I hate wasting time, but I hate even more when that time was wasted for nothing :(
20:42:34Nico_Pyeah I know the feeling ;)
20:44:36 Quit hcs ("Leaving.")
20:44:52 Join pondlife [0] (n=Steve@rockbox/developer/pondlife)
20:44:54 Part pondlife ("Gone")
20:48:47preglowcan't one use I*_ATTR for data inside a function?
20:49:06amiconnOf course not
20:49:18amiconnauto vars live on the stack, wherever that is
20:49:26preglowi'll just declare these bitches static
20:49:28preglowthat should solve it
20:49:41amiconnYes, static should work, also within functions
20:50:06preglowthe new speex way of handling its overridable functions generates tons of warnings
20:50:20amiconnBut depending on where you're using it, you might not want to do that. The stack might already be in iram
20:50:26amiconnE.g. the codec stack is
20:50:29preglowamiconn: it's for const tables
20:50:43preglowamiconn: speex has them in a function, and i want to change as little as possible
20:50:53preglowsyncing is boring enough as it is
20:50:55 Join MethoS- [0] (n=clemens@
20:51:00amiconnWeird way of coding, really...
20:51:13preglowif they're local to the function, then why not
20:51:21preglowextra semantics never hurt anyone
20:51:33amiconnpreglow: Then I would *always* make them static
20:51:46preglowamiconn: why?
20:51:49amiconnNot making them static means they're not truely const
20:51:58preglownot in the rom way, at least
20:52:01preglowwell, i've made them static now
20:52:03amiconnThey will be constructed on stack *at runtime*
20:52:12 Join midgey [0] (
20:52:17kubiixakai have a weird problem with compiling clean source
20:52:26amiconnThat means larger and slower code for no reason
20:52:30preglowbut now i get tons of errors for each bloody overridden function i don't use
20:53:53preglowi take it there's no way to make gcc not complain about unused static functions in headers without introducing compiler switches...
20:53:55amiconnlinuxstb: Btw, is there a reason why the predictor is supposed to return an int? It is never checked...
20:54:14 Quit gufo (Read error: 110 (Connection timed out))
20:54:49amiconnpreglow: Not that I know of. It only works if those functions are static inline
20:56:21linuxstbamiconn: No, they should be void...
20:58:11amiconnpreglow: Putting code which results in actual binary output into header files should be banished, imo
20:58:42preglowamiconn: i'll inline them, then
20:58:52preglowamiconn: not my fault, the speex author suddenly went for this scheme
20:59:50amiconnYes, I thought so. Still, it's broken design
21:00:32amiconnBut if we want to use that codec... :/
21:00:45 Join jhulst [0] (n=jhulst@unaffiliated/jhulst)
21:00:46amiconnYou could put some #if 0 around the unused stuff
21:01:22kubiixakai have a weird problem with compiling clean source
21:01:22kubiixakait look that in lang.h aren't generared all needed lang_xxx variables
21:01:22kubiixakaand later in compiling i got errors like:
21:01:22DBUGEnqueued KICK kubiixaka
21:01:22kubiixakamenus/theme_menu.c:142: error: initializer element is not constant
21:01:22kubiixakamenus/theme_menu.c:142: error: (near initialization for `set_lse_col_.desc')
21:01:23***Alert Mode level 1
21:01:23kubiixakamenus/theme_menu.c:144: error: `LANG_SELECTOR_TEXT_COLOR' undeclared here (not in a function)
21:01:51n1skubiixaka: make clean?
21:02:08kubiixakai have fresh cygwin with latest revision
21:02:59kubiixakai thing it is same issue as here:
21:03:37amiconnCurrent SVN definitely builds, as you can see in the build table.
21:04:11 Join chris_mt [0] (
21:04:21kubiixakai know, but where can be the problem
21:04:39bertrikwhat gcc version do you use?
21:04:44 Join merbanan [0] (n=banan@
21:06:57 Join linuxstb_ [0] (
21:07:09 Quit linuxstb (Nick collision from services.)
21:07:10 Nick linuxstb_ is now known as linuxstb (
21:07:33bertrikI'm no expert yet, but I use gcc 3.4.4, arm-elf-gcc 4.1.1 and perl 5.8.8 with cygwin without problems so far
21:07:54 Join Frazz [0] (
21:08:39 Join linuxstb_ [0] (n=linuxstb@rockbox/developer/linuxstb)
21:08:58kubiixakai will check all versions
21:09:33 Quit PaulJam ("Verlassend")
21:11:24***Alert Mode OFF
21:11:54 Join jgarvey [0] (
21:12:52kubiixakawhere can i get arm-elf-gcc 4.1.1 ?
21:13:17***Saving seen data "./dancer.seen"
21:14:04bertrikIt's not officially supported I think. I already had an arm-elf-gcc toolchain installed before rockbox. I got it from
21:14:11amiconnThe recommended gcc for arm targets is 4.0.3
21:14:39amiconnYou can always build your own manually, it's described in the wiki
21:15:15kubiixakaso i have proper gcc
21:15:44amiconnYou can also use and it will build the crosscompilers for you, and for cygwin we provide precompiled gcc packages
21:16:05amiconn(mostly because compilinmg gcc on cygwin takes, ahem, a "while")
21:16:08kubiixakai have compiled succesfully rockbox in past many times, but I have reinstalled windows + cygwin
21:17:39Nico_Pare there any volunteer testers for mob on a slowish target (coldfire would be nice) ?
21:18:00n1sNico_P: sure
21:18:17n1sshould I install git now?
21:18:32amiconnColdfire isn't slow. Try playing ape -c2000 on PP :P
21:19:03Nico_Pn1s: no need
21:19:28n1sso will you make me a build or should I get a snapshot?
21:19:35Nico_Pn1s:;h=mob, click snapshot
21:20:05Nico_Pamiconn: "slowish" was a figure of speech for something slower than a gigabeat :)
21:20:06 Quit linuxstb__ (Read error: 113 (No route to host))
21:20:33amiconnAh, compared to that insane target it's slower, yes...
21:21:03Nico_Phehe, if the F is insane, what do you call the S ?
21:25:03 Quit _pill ("changing servers")
21:25:14Nico_Pn1s: I'm particularly interested in potential lockups caused by repeated skipping
21:25:29n1sNico_P: building
21:25:34Nico_Pthanks :)
21:26:09 Join ennorico [0] (i=5a8013c1@gateway/web/cgi-irc/
21:26:21preglowamiconn: not possible, the .h is included in nearly every .c and what is used is different from time to time, luckily, the functions are so smalled they might as well be inlined
21:27:50amiconnIf you compile such stuff with -O0 you'll end up with tons of dead code in the final binary
21:28:11amiconnI mean when they're not declared inline
21:28:20amiconnstatic inline, even
21:28:33preglowluckily, we're not doing that
21:28:38preglowthanks anyway, it worked fine :)
21:29:29CtcpIgnored 1 channel CTCP requests in 0 seconds at the last flood
21:29:29*amiconn hopes his ugly hack will work...
21:30:38preglowhack for what?
21:31:19amiconnJust the brute-force approach: I hacked some stuff into predictor_decode_stereo() that writes all 3 important arrays to files before and after doing its magic
21:31:33amiconnThat is, it does this on the very first call
21:31:57amiconnThen I'll hack in the asm version, and compare
21:32:15preglowwell, at least that should work
21:32:17preglowbrb again
21:32:21n1sNico_P: yup skipping like a madman makes it lock with disk spinning and backlight on :-/
21:32:41 Join Lambuntu [0] (
21:33:13 Join jac0b-work [0] (
21:33:28 Quit Arathis (Remote closed the connection)
21:34:15 Join barrywardell [0] (
21:35:12 Join Arathis [0] (
21:35:40 Nick chris_mt is now known as criznach (
21:36:38 Quit animeloe ("This computer has gone to sleep")
21:38:05jac0b-workI am interested in learning C does anyone have a suggestion of what book I should read?
21:38:45 Join rotator [0] (n=e@rockbox/developer/rotator)
21:38:50n1sNico_P: also inserting a couple of tracks while playing locked up
21:39:40Nico_Pn1s: inserting them where ?
21:40:17n1sin the playlist, first I did a "insert next" which worked then regular "Insert" which locked up
21:42:16Nico_Pok I'll see if I can reproduce
21:42:51 Quit MethoS- (Remote closed the connection)
21:43:47 Quit ennorico ("CGI:IRC")
21:44:20Nico_Phmm inserting isn't a problem here... is the lockup consistent ?
21:45:56n1spretty much, if I insert while it's still buffering it locks up after the first one otherwise after the second
21:45:57 Join MethoS- [0] (n=clemens@
21:46:13n1sso it seems like insert while buffering is broken...
21:47:55n1shmm, now it doesn't happen anymore...
21:48:28linuxstbjac0b-work: This is the C bible - But you might also want to just go to a local bookstore and browse, and choose something that suits you and your prior knowledge.
21:49:19 Join pixelma [0] (i=pixelma@rockbox/staff/pixelma)
21:49:30 Quit kubiixaka ("Miranda IM! Smaller, Faster, Easier.")
21:49:52n1sNico_P: now inserting is rock solid, the skip-like-a-madman lockup is still there though :-/
21:50:28Nico_Pn1s: are you skipping forward or backwards ? and what kind of tracks do you have ?
21:50:55n1sforwards, almost only ogg/vorbis, wnat me to try something else?
21:51:04 Join Asteriskk_ [0] (n=herb@
21:52:03Nico_Pn1s: hmm maybe seeing if you get the same thing in the sim could be good if it's not too much of a hassle
21:52:16Asteriskk_so I got a microsd card for my sansac200, it shows up in the computer as an additional drive, but nothing turns up in the player itself. is this normal?
21:53:08maxkelleyin rockbox, nothing will show right now
21:55:51Asteriskk_anything I can do about it?
21:56:39 Join ompaul [0] (n=ompaul@gnewsense/friend/ompaul)
21:56:55 Quit ilgufo ("So Long, and Thanks For All the Fish -")
22:00:05 Quit jac0b-work ("ChatZilla [Firefox]")
22:03:33n1sNico_P: nope, skipping like a madman works fine both forwards and backwards in the sim
22:04:01 Join donutman25 [0] (n=chatzill@
22:04:48 Quit barrywardell ()
22:04:51 Quit BigBambi (Read error: 113 (No route to host))
22:05:03 Quit desowin ("use linux")
22:07:26 Join barrywardell [0] (
22:08:23 Quit pepie34 ("Ex-Chat")
22:10:47Zagorbertrik: did you see my message earlier in the log?
22:11:10 Join moos [0] (
22:11:35bertrikno didn't see it yet, i'll have a look at it
22:15:17bertrikthis is what I used as a host test application:
22:15:55 Quit bertrik ("bye")
22:16:51 Quit amiconn (Nick collision from services.)
22:16:58 Join amiconn [0] (n=jens@rockbox/developer/amiconn)
22:18:32 Part spiffytech
22:18:45 Join midgey_ [0] (
22:20:02 Quit midgey (Read error: 104 (Connection reset by peer))
22:20:36barrywardellZagor: any chance you could put this up on the download server:">
22:21:10barrywardellthanks. it goes in the rbutil/macosx dir
22:22:16Zagorbarrywardell: done
22:26:14 Quit Calcipher (Read error: 110 (Connection timed out))
22:27:45*n1s curses macro hell in combination with people "fixing" 64 bit warnings by doing some incorrect cast...
22:29:32 Part BrianHV
22:29:35 Quit Frazz ("Leaving")
22:32:15 Quit scorche|w ("CGI:IRC (Session timeout)")
22:33:51 Join scorche|w [0] (
22:34:30 Join darkless [0] (
22:36:12 Quit jhulst ("Konversation terminated!")
22:38:29n1sanyone wants to do a quick test on a 32 bit sim?
22:40:46amiconnHmm, the bug is most likely in the filter B calculation....
22:42:42preglowlooking forward to seeing the bug :P
22:43:18amiconnAfter one single loop iteration, YcoeffsB[2..4] and XcoeffsB[2..4] are wrong. The decoded buffers are yet unaffected, but if I continue running with the C version, I get complete static for a few seconds
22:43:39amiconn(yes, my test hack calls the asm version for one single loop iteration)
22:44:16preglowouch, speex is 112% realtime for a 44.1khz file
22:44:34preglow300% for 19kbps 16khz wb
22:44:40preglownano, even slower on coldfire...
22:44:55preglowi think time has come for some gentle speex massaging
22:51:20 Part criznach
22:51:53preglow86% and 228% for the same files...
22:52:08 Join webguest51 [0] (i=50d81efc@gateway/web/cgi-irc/
22:53:17 Quit merbanan (Remote closed the connection)
22:54:06webguest51Zagor: I think I can run the usbcv if you want.. I have a e200
22:55:06Zagormy wife has a windows box I can use too. thanks anyway.
22:55:55Zagorunfortunately I'm not yet at the stage where I can run usbcv.
23:01:00 Join darkapostrophe [0] (
23:01:37 Quit Dark_Apostrophe (Nick collision from services.)
23:01:39 Nick darkapostrophe is now known as Dark_Apostrophe (
23:03:15 Quit bluebrother ("leaving")
23:05:29amiconnLooks like a simple, freaking letter B was missing
23:06:10amiconn...and gas didn't complain about the undefined macro - simply because it doesn't see that macro
23:06:17preglowthe best kind of bug :P
23:06:25 Join zehsound [0] (
23:06:25preglowgas really isn't made for humans
23:06:25amiconnI had XADAPTCOEFFS instead of XADAPTCOEFFSB
23:06:38 Part Asteriskk_
23:06:38preglowit's the xml of assemblers
23:06:38amiconn(and the same for Y)
23:07:11 Join sound [0] (
23:07:32 Quit sound (Client Quit)
23:07:35amiconngas is not to blame here - these are preprocessor macros
23:08:03TavnosHi there :)
23:08:27amiconnSo if I have (XADAPTCOEFFS-16,%a5) and the macro is undefined, it collapses to (-16,%a5) for gas, which is perfectly valid
23:10:46Nico_Pamiconn: what are you writing btw ?
23:10:55 Quit zehsound (Client Quit)
23:11:07 Join BigBambi [0] (n=alex@rockbox/staff/BigBambi)
23:11:13 Join sounddude [0] (
23:11:42amiconnIt plays! No static! \o/
23:11:50 Quit sounddude (Read error: 104 (Connection reset by peer))
23:12:07 Join sounddude [0] (
23:12:16Tavnos\o/ my first WPS near finished :)
23:12:41Nico_Pamiconn: APE optimisation ?
23:13:14 Quit sounddude (Read error: 104 (Connection reset by peer))
23:13:20***Saving seen data "./dancer.seen"
23:14:09amiconnI should probably commit this now, and try to squeeze it a bit after wards
23:14:30amiconn-c1000 is now 208%, -c2000 is 146%
23:14:30 Nick parafin is now known as parafin|away (
23:14:42amiconn-c3000 is barely realtime, but not usable
23:14:51Nico_Pamiconn: are there any notable arch differences between m68k and arm that could cause crashes on m68k and not on arm (I'm thinking things like alignment)
23:15:16amiconnThe opposite is more likely
23:15:49 Join sounddude [0] (
23:15:49amiconnm68k/coldfire doesn't care about alignment (except that it's slower). arm does
23:15:50Nico_Pok :(
23:16:01 Quit sounddude (Read error: 104 (Connection reset by peer))
23:16:09Nico_Pand there's nothing else apart from alignment ?
23:16:43amiconnIf you access non-existing memory locations (reading data), you'll get a hung bus on coldfire
23:16:59amiconnThe visible effect would be a complete freeze
23:17:16amiconnDon't know what arm does when accessing non-existing locations
23:17:30Nico_Phmm that could be something
23:17:43*amiconn suspects dangling pointers
23:17:45Nico_Pwhat would the sim do in that case ?
23:17:56 Join sounddude [0] (
23:18:24amiconnOn coldfire, the 'catch memory accesses' debug feature might help
23:19:08preglowgeh, new warning
23:19:27Nico_Pif only I had a colfire target at hand
23:20:24Tavnosi have a question Nico_P, where are you from ?
23:20:42Nico_PTavnos: France. I'm from Paris and study in Toulouse
23:21:05TavnosNico_P : Salut cousin francais :)
23:21:18Nico_Psalut :)
23:22:28 Join grapeshit [0] (
23:22:31Nico_PTavnos: can you PM ?
23:22:48grapeshitHow's the fm tuner on sansa e200 + rockbox ?
23:23:57Zagorgrapeshit: it works
23:25:43grapeshitis it good ?
23:26:15Zagorit's not very sophisticated, but it gets the job done
23:26:57grapeshitand how's the overall sound quality, better than original fw ?
23:27:36preglowno, it sounds like a bag of shit :(
23:27:48ZagorI should think the radio quality is very similar
23:28:09Zagorbut my OF doesn't have radio, so I can't compare...
23:28:12preglowwe just don't get quality statements like that anymore
23:28:29scorche|wi think rockbox's was a little better, but that is from a month or so ago
23:28:37scorche|wthe OF's was shit
23:29:14scorche|wi could get channels with rockbox that i couldnt with the OF
23:29:30scorche|w(or at least hear them clearly...)
23:31:54amiconnpreglow: Imho the old workaround was better (using unsigned j;)
23:32:12preglowamiconn: that would probably break the other for loops it's used in
23:33:57grapeshitscorche: now u can hear all clear ?
23:35:02scorche|wgrapeshit: well, i dont have the device...i woudlnt say it was all clear, but it was certainly better
23:35:29 Join jhulst [0] (n=jhulst@unaffiliated/jhulst)
23:35:33 Quit webguest51 ("CGI:IRC (EOF)")
23:36:08grapeshitI need a very good fm reception device with rockbox
23:36:26grapeshitis there's one ?
23:36:40 Quit MethoS- (Remote closed the connection)
23:37:05 Join MethoS- [0] (n=clemens@
23:39:45 Quit Tavnos ()
23:41:57preglowthe h120 is decent, i think
23:42:02Nico_Pwow... wps_offset is -14 !
23:42:05krazykitgrapeshit, the sansa is great, but FM reception is a little spotty
23:44:49preglowwould any of you guys happen to know if the 8 bit paletted modes are slower than the rgb modes these days? on ordinary computers, this is
23:45:12grapeshitspotty ?
23:45:49grapeshithmm i currently have iaudio, and the fm reception isnt good enough
23:45:53 Join bluebrother [0] (i=Yw6AUPnx@rockbox/staff/bluebrother)
23:46:12Zagorpreglow: for 2d or 3d?
23:46:33 Part krazykit ("Connection reset by beer")
23:46:40 Join krazykit [0] (
23:46:51preglowZagor: 2d? i didn't even know 3d did that kind of thing
23:48:52preglowperhaps for textures, *shrug*
23:49:53Zagorwell benchmarks usually show lower performance for 32-bit than 16-bit graphics, so I would assume 8-bit would be even faster. but as you say I haven't seen any tests with 8-bit graphics.
23:50:36 Quit midgey_ (Client Quit)
23:51:42preglowi think the palette stuff might be emulated in some way these days
23:51:51preglowif it wasn't for the lookup, there'd be no reason for it to be slower
23:52:53Zagorotoh the lookup ought to be done in local ram, so should carry a rather small penalty compared to the data transfer
23:54:03*Domonoky_ just commited a fix to the talkfile generation in rbutil... it looks like nobody has used this function in reallife, otherwise the bug would have been noticed .. :-)
23:57:20barrywardellDomonoky_: I came across a bug in rbutil today
23:57:36barrywardellcomplete install fails to install the bootloader on my H10
23:58:36 Join darkapostrophe [0] (
23:58:46Domonoky_barrywardell: any error message ?

Previous day | Next day