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 2005-09-01

00:02:09***Saving seen data "./dancer.seen"
00:04:16WettI see lines like CONFIG_HWCODEC == MASNONE, which HWCODEC corresponds to the iriver ?
00:05:23linuxstbUpdate to the latest CVS. It was changed recently. Or search for "pcm_"
00:06:54Wettyeah, have some pcm_play_data (my package is just a few days old)
00:07:38Wettbut pcm is uncompressed data, I guess i'll have to use a codec, right ?
00:17:25 Join [IDC]Dragon [0] (
00:18:28linuxstbWett: Yes. You can either attempt to modify Rockbox to give plugins access to the codec API, or just link your plugin against one of the codec libraries (e.g. libmad), and do your own decoding "internally".
00:19:54Wettok... i don't know at all how i'll be able to do that but... Oo I guess i may understand better if I look at some .h files
00:20:06linuxstbIn the short term, it's going to be easier to just link against libmad and copy code from mpa.c to do the decoding.
00:20:30Wettwhat is mpa.c ?
00:21:09linuxstbIt's the MPEG codec - i.e. the code that sits between libmad and Rockbox.
00:21:43Wett(and I guess libmad is the for mp3 :)
00:21:47linuxstbYou basically just need to find out how to use libmad - you feed it compressed MPEG frames, and it gives you back PCM data.
00:21:59linuxstbYes - and MP1/2
00:22:10Wettok, should be not as hard as I thought
00:22:41linuxstbThe .c files in apps/codecs are relatively small, and should be relatively easy to read.
00:23:08Wettokayy, thanks a lot ! i'll see what I can do from now on :)
00:23:48Moosis it still for video plugin port for irivers? :)
00:24:04linuxstbYou could also look at the old "mpa2wav.c" plugin - I wrote this before Rockbox could play audio on the iriver in order to test libmad. As the name suggests, it decodes an MPEG audio file (e.g. mp3) to WAV.
00:24:09Wettyep !
00:24:18linuxstbIt's in CVS, but I think it's been "removed".
00:24:20Wettalready managed to play the video :)
00:24:44linuxstbDoes your plugin play the video format used on the Archos?
00:24:57Wetti just port it
00:25:19MoosWett: your man is amiconn or IDC Dragon
00:25:41Wettamiconn already helped me yesterday on iriver lcd ^^
00:26:14Wettlinuxstb -> how can i find it if it's been removed ?
00:26:17webguest63The feature-thaw is looking to be eventful
00:26:37linuxstbWett: Check the "Attic" when browsing CVS:
00:26:45Moosisn't you IDC Drangon, who ported this video things in archos?
00:27:39 Quit ashridah ("Leaving")
00:28:15[IDC]Dragonnot ported, but made
00:28:23MoosHello hehe :)
00:28:32Wetthello !
00:29:50[IDC]Dragondos it play sound, too?
00:30:01WettI'm working on it
00:30:26[IDC]Dragondid you extend the size?
00:30:43[IDC]Dragonor playing backwards-compatible?
00:30:44Wettbtw, i guess it's possible to modify tools in order to produces 160x128 video ?
00:31:04[IDC]Dragonwhich are you using?
00:31:05Wettlol, it's compatible for now, i'm just playing little videos
00:31:16Wettfor now, the easy way
00:31:16 Quit LameD ("CGI:IRC (EOF)")
00:31:35[IDC]Dragonwhich is?
00:31:39Wetti took a little look at rvf_mux.c, it looked possible to change
00:32:01linuxstbIs it possible to transcode MPEG-2 to the Rockbox format (under Linux)?
00:32:01Wettusing RockVideo.exe
00:32:08[IDC]Dragonthe DirectShow one is nicer, but not open source
00:32:17Wettand a directshow plugin
00:32:57Wetthm, so i'll definetly have to modify rvf_mux
00:33:17[IDC]Dragonor I'll have to modify the DirectShow filter
00:33:51WettI didn't find the source, and I'm not sure at all i'll be able to do that
00:33:53 Join gromit` [0] (n=gromit`
00:34:18WettI'll begin with the other solution ^^
00:34:23[IDC]DragonI haven't disclosed the sources for that
00:34:39[IDC]Dragontoo much overlap with my job
00:35:14[IDC]Dragonbe prepared for huuuge video files then
00:35:52Wetthm... how can I find the source for the directshow filtrer then ?
00:36:25[IDC]Dragonyou can't, I'm not releasing them
00:36:47Wettoh, okok I misread sorry
00:37:02[IDC]Dragonyou probably won't be able to compile that anyway
00:37:03linuxstbDoes anyone know if the "halftone" program work with the yuv4mpeg format used by mjpegtools?
00:37:14Wettyes, i guess Oo
00:37:58[IDC]Dragon(needs MSVC and the DirectX SDK)
00:40:07Wett(I have ! lol, it doesn't matter)
00:40:36[IDC]Dragonah, ok
00:41:25 Join TCK [0] (
00:48:09Rickiriver = gapless support? (friend is asking)
00:48:19RickI don't recall if it was ever implemented
00:48:36 Quit ender` (Read error: 113 (No route to host))
00:48:44webguest63 It is. There's a very very minor glitch
00:50:09Rickwho're you anywawy? :)
00:51:09webguest63just a user
00:51:27webguest63hm, just checked.. couldn't here any transition at all
00:51:28Rickoh, ok :P
00:51:59OnkelJonashey... do any of you have experience with "the godfather"? (tagging program)
00:52:29 Quit paugh ("Leaving")
00:55:22linuxstbRick: iriver is 100% gapless for everything apart from MP3s
00:55:53linuxstbFor MP3s, I think that gapless is now working (using lame to encode the tracks), but there may be a few minor problems remaining.
00:57:06webguest63This was from an mp3 I had split with a program that reportedly splits at frame-boundries
00:58:27linuxstbThat should work fine then. It will be decoded as if it was a single long file.
00:58:48linuxstbThe problem comes when tracks are encoded individually.
01:03:15 Join Febs [0] (
01:07:45 Quit Sucka ("a bird in the bush is worth two in your house")
01:11:28Moosgood night @ all !
01:11:38 Quit Moos (" HydraIRC -> <- Try something fresh")
01:14:01 Quit bluebrother^ ("Leaving")
01:15:19 Quit webguest63 ("CGI:IRC")
01:19:55 Quit DangerousDan (Read error: 110 (Connection timed out))
01:26:52 Quit hicks ("Too lazy to change my quit message")
01:31:00 Join solex_ [0] (
01:31:29Wett[IDC]Dragon : Can I know how are stored audio blocks in the rvf format ? Are they composed of just one frame, or more like little mp3 files ?
01:32:06[IDC]Dragonthey are whatever fits
01:32:14[IDC]Dragonnot frame aligned
01:32:34[IDC]Dragonhowever, there is a tiny header in each block
01:32:59[IDC]Dragontelling you about frame starts
01:33:25[IDC]Dragonthe mp3 data is bitswapped
01:33:50Wettand don't know what it means ^^
01:33:52[IDC]Dragonbut the file format is prepared for other audio formats
01:34:27[IDC]Dragonbitswapped like the voice files
01:34:31Wetti'm not very familiar with audio format anyway
01:34:51[IDC]Dragonno, this is a crappy Archos workaround
01:35:10[IDC]Dragonthe serial shifts the bits out in wrong order
01:35:32[IDC]Dragonso we're better off if the data is already bitswapped
01:35:52Wettthe only thing i understand here is : it's better so
01:36:36 Quit matsl (Remote closed the connection)
01:37:05Wettif a read well, the codec decode frame per frame. How big a frame can be ? Is there only one frame in an audio block ?
01:37:09[IDC]Dragonthis format is compensating a hardware shortcoming
01:37:32[IDC]Dragonno, frames are unrelated to blocks
01:38:35[IDC]Dragonbut for seek convenience, my audio header tells you if and where a frame starts within this block
01:39:30Wettoh ! so there's just a stream all along the file, wich I have to feed with blocks when i run out of data, right ?
01:40:11[IDC]Dragonminus the headers, yes
01:40:48Wettgot it. thanks !
01:41:23[IDC]Dragonand for the special case of Archos mp3 audio, you need to twist the bytes
01:41:44[IDC]Dragonlike done in talk.c, too
01:42:05[IDC]Dragonwhich also deals with pre-bitswapped mp3 data
01:42:18Wettbut, will I have to kinda twist them back ?
01:42:30Wettif the mp3 is already bitswapped
01:42:56[IDC]Dragonyes, this is what I'm trying to say
01:43:25Wettarg sorry i'm a little slow :p
01:45:30Wettso, what does it concretely say ? that bits are reversed ?
01:45:48[IDC]Dragonyes, MSB becomes LSB
01:45:59WettxD google is my friend
01:46:30[IDC]Dragontalk.c is your better friend
01:46:35 Quit solex (Connection timed out)
01:46:55[IDC]Dragonsuch is done there, already
01:47:32Wettyes ! I'll just have to steal some code, exactly what I do from the beginning
01:48:18 Join Mxm`Pas`Bien [0] (
01:48:24[IDC]Dragonsuch a thief!
01:49:14Wettlol kinda :D
02:02:12***Saving seen data "./dancer.seen"
02:05:54 Quit Maxime (Read error: 110 (Connection timed out))
02:13:44 Quit preglow ("leaving")
02:18:04Wettgoodnight, I'm going to sleep in a while, too
02:18:08[IDC]Dragonand happy ROMBoxing
02:18:24[IDC]Dragonto any Archos users out there
02:18:38 Quit [IDC]Dragon ()
02:23:17Wettgn all :)
02:23:21 Quit Wett ()
02:28:54 Quit AdNauseam ("leaving")
02:48:54 Quit linuxstb (Read error: 113 (No route to host))
04:02:13***Saving seen data "./dancer.seen"
04:46:51pikerockbox;iriver can you lock remote and mainunit independently yet ?
05:24:59 Join yeft [0] (
05:25:35yeft!seen midk
05:25:47yeftno .. no
05:26:31 Quit yeft (Client Quit)
06:02:15***Saving seen data "./dancer.seen"
06:46:26 Join midk [0] (
06:47:20 Join LinusN [0] (
06:48:59 Join webguest65 [0] (
06:59:50 Quit webguest65 ("CGI:IRC (EOF)")
07:08:32LinusNbleh, the line-selector uproar is getting louder each day
07:11:54 Join kurzhaarrocker [0] (
07:21:40kurzhaarrockerIn convbdf there's a variable which is declared in the middle of a function instead of the begin of the block. That causes problems with old compliers. Do we care about that?
07:24:06kurzhaarrockerThat variable is int ofr = 0;
07:24:08*kurzhaarrocker must go
07:24:21 Quit kurzhaarrocker ()
07:24:21midkkurzy! :)
07:27:10LinusNhe obviously didn't check the cvs log, i fixed that yesterday
07:28:07 Join ender` [0] (
07:34:26 Quit TCK (Read error: 104 (Connection reset by peer))
07:36:13 Join tvelocity [0] (
07:47:14CoCoLUSthe sony psp launched today all over europe... and i was thinking
07:47:21CoCoLUSit -would- be a viable plattform for rockbox :)
07:47:27 Quit midk ("Leaving")
07:53:38 Quit ender` (Read error: 110 (Connection timed out))
07:54:31 Join DangerousDan [0] (
08:02:17***Saving seen data "./dancer.seen"
08:17:37 Join kurzhaarrocker [0] (
08:19:51 Quit solex_ ("leaving")
08:25:53 Quit DangerousDan ("Miranda IM! Smaller, Faster, Easier.")
08:28:45 Join solex [0] (
08:42:53 Nick Mxm`Pas`Bien is now known as Mxm`PasLa (
08:43:35kurzhaarrockerIn recording.c the variable path_buffer is defined in line 247 and in line 307. I am convinced that the definition in 307 can be omitted as both variables are merely used as parameters and not for storage.
08:44:30kurzhaarrockerhm. Or we could move the global definition in 247 to a local variable of trigger_listener.
08:46:26kurzhaarrocker(which might save MAX_PATH Bytes)
08:49:39kurzhaarrockerStrange. It doesn't seem to affect the size of rockbox. Are static variables allocated on a heap?
08:51:14LinusNin the .bss segment
08:51:23LinusNallocated at startup
08:51:41LinusNso it won't affect the size of the rockbox image
08:51:47LinusNunless it's initialized
08:52:00LinusNthen the init values have to be stored in the image
08:52:12kurzhaarrockerno it isn'n initialized
08:53:04amiconnLinusN: I found another strong argument for the line-selector pointer removal: The arrow didn't scale so it worked very poor with larger fonts. The inverse bar doesn't have that problem
08:53:40LinusNyeah, but that would have been easy to fix
08:53:50LinusNi'm having second thoughts
08:54:41kurzhaarrockerYou don't want to put the arrow back, do you?
08:55:51LinusNpersonally, i don't care, more than i like the reduced code complexity because of the way it was imlemented
08:56:32LinusNwe could make it a little simpler by centralizing the cursor display
08:56:40LinusNand reintroduce the arrow
08:56:54 Join cheriff [0] (
08:57:10LinusNso maybe we should revert the change and rework the cursor code after the release
08:57:42LinusNthe fm code is small enough to fit anyway
08:58:17kurzhaarrockerBtw: When you have "resume on startup" enabled "Show recording screen on startup" doesn't work. Bug or feature?
08:58:38LinusNi'd say bug
08:58:45kurzhaarrockerme too :)
09:01:44kurzhaarrockerShould I write a bug report or add that in wiki to ReleaseTodo?
09:01:57amiconnLinusN: I think it's about time to centralise the listbrowser code
09:02:07LinusNamiconn: absolutely
09:02:31LinusNthe tree and playlist browsers would be a lot simpler
09:02:37amiconnA related thing is that we should have a request() in addition to our splash()
09:02:41LinusNand the lcd remote display would too
09:02:47LinusNamiconn: yeah
09:02:49amiconnTree, playlist, menu
09:04:36amiconnI am thinking about real *quick* menus, but I'm undecided whether this would be a good idea
09:04:56amiconnThis results from my double-click discussion with Jörg
09:05:05 Join linuxstb [0] (
09:06:17amiconnWhat about a menu that pops up with a short click of one button, then stays for a short time, allowing to select n different options with n buttons. If nothing is selected in say, one second, it would disappear again
09:07:20kurzhaarrockerHard to tell without trying the feel - But I think I'd prefer the quick menues as they are now.
09:08:49amiconnYes, but the current quick-menus have some problems
09:08:55kurzhaarrockerOr do you mean you still could use those quick menus as key combos and they would disappear immediately?
09:09:05amiconn(1) Only the recorders have 2 of them, and the iriver has one
09:09:28kurzhaarrocker(It's that key combo thing I'm after)
09:09:34amiconn(2) Due to the way they work, the quick-operation mode doesn't work for many buttons
09:09:57amiconnThat's why they use only 3 (they could use 4), and even one of the 3 doesn't work when flipped
09:12:38LinusNhow would this menu work, i mean how are the items selected?
09:14:38amiconnThe items would be selected by a click of a button
09:14:46LinusNas today?
09:15:10amiconnIt's basically an "extended double-click", the second click may be a different button than the first
09:15:46LinusNlike the nokia phones?
09:16:09amiconnHmm, I don't know the current nokia gui at all
09:17:25LinusNok, so the only real difference would be that it disappears automatically?
09:19:31amiconnDifferences against current quick-screens: They wouldn't operate as button combos, in order to block less buttons. They would allow more buttons within the menu because of that. They would disappear automatically
09:22:23amiconnThey would disappear either if a button is selected or they timeout
09:22:52kurzhaarrockerapropos button combos: the recorder is off. I hold the f2 button and then switch it on with the on button. It turns on but doesn't boot. Is there a hidden feature or another bug?
09:25:48LinusNi wouldn't like them to disappear when a button is pressed, that would make it a pain to cycle the repeat mode setting
09:26:02LinusNkurzhaarrocker: a hidden feature
09:27:57 Join B4gder [0] (
09:29:52LinusNkurzhaarrocker: hmmm, f2 *should* boot normally, if i understand it correctly
09:31:13kurzhaarrockerer, sorry, I meant F3
09:32:29LinusNf3 starts the MiniMon
09:32:50LinusNa way of restoring a corrupt flash via the serial port
09:37:39linuxstbLinusN: Obviously I'm in favour of reversing the cursor patch. I'm also more than happy to work on cleaning that part of the code - unless others wish to do it.
09:39:23 Join paugh [0] (n=pete@2001:5c0:8fff:ffff:8000:0:3e03:6822)
09:39:47LinusNlinuxstb: let me think about it
09:40:10linuxstbThanks - I can't ask more than that.
09:43:06kurzhaarrockerSomething else I failed to implement: In recording screen I'd like to make the on-button switch to wps and play the last recorded file immdiately.
09:44:25LinusNkurzhaarrocker: submit a patch
09:44:37kurzhaarrockerUnfortunately I failed
09:44:53LinusNkurzhaarrocker: :-)
09:44:56kurzhaarrockerafter recording it switches back to the menu
09:45:30kurzhaarrockerThat is quite generic code where I don't know how to hook in that it should return to wps instead.
09:45:39 Quit linuxstb ("Leaving")
09:46:33 Nick cheriff is now known as cheriff_AWAY (
09:47:09kurzhaarrockerI'm talking about main_menu.c , rec_menu.
09:57:57LinusNkurzhaarrocker: that isn't trivial, i know
10:00:05 Quit webguest61 ("CGI:IRC (Ping timeout)")
10:01:10kurzhaarrockerI thought: maybe store a hint in some static variable in recording_menu, abuse SYS_USB_CONNECTED to quit the menu at all, and than check and process that static hint within the caller of main_menu. But that feels so ugly.
10:02:21***Saving seen data "./dancer.seen"
10:03:18*B4gder buckles up
10:03:26*Rick steals B4gder's buckle.
10:04:57LinusNok, now the arrow cursor is back
10:09:40kurzhaarrockerDecreasing CONFIG_BLOCK_VERSION is funny.
10:10:10LinusNdoesn't matter in this case
10:12:38 Join [IDC]Dragon [0] (
10:14:56LinusNok, finally we can have conditional images in the wps
10:15:07 Join thomjoha [0] (
10:16:00 Nick thomjoha is now known as preglow (
10:16:06preglowso, someone whined the cursor back? ;)
10:16:42solexLinusN: conditional images are cool. I didn't know they didn'twork unitl now, but I already thought about using them! :)
10:16:58LinusNpreglow: yes :-)
10:16:59[IDC]Dragonwhat a pity, we're too weak to clean out the feature clutter... ;-)
10:17:39LinusNwhat is rockbox, if not features?
10:17:55preglowLinusN: change the default to bar, then we can remove the cursor again in six months or so, heh
10:18:10[IDC]DragonI want the arrow on the right, and non-solid, can I have an extra option, please?
10:18:29LinusNsubmit a patch :-)
10:19:49solexis there a big performance/memory penalty associated with conditional images
10:19:57solexsay, for play status?
10:20:15*solex thinks about implementing his own status bar
10:20:39solexso all images are kept in memory all the time?
10:34:30pikerockbox;iriver can you lock remote and mainunit independently yet ?
10:36:06LinusNit doesn't work?
10:36:30pikereason I ask is, I feel myself unable to use the remote, when I take up the mainunit I have somehow managed to go into the settings menu
10:36:43 Join darkskiez [0] (n=darkskie@
10:36:55pikeif you say the 2 locks are independant, I'll do more testing
10:37:26B4gderwe don't have remote support yet, afaicr
10:37:34LinusNthe remote keys work
10:37:39B4gderthey do?
10:37:43darkskiezmy iriver has died :( power light comes on when you plug it in to charge, but doesnt ever finish, and it doesnt power up. Anyone have any idea if its a common fault that might be repairable ?
10:38:48LinusNdarkskiez: did you do anything to it?
10:39:09LinusNlike connect it to a third-party charger?
10:39:24darkskiezi did yeh, it charged fine once on it though
10:39:28darkskiezand the spec matched
10:40:04LinusNand the polarity was ok too?
10:40:30darkskiezyup, it successfully charged it one night, polarity and voltage matched
10:40:50LinusNthe symptoms show that it is killed beyond repair
10:40:52darkskiezwent to turn it on a few days later, thought battery was flat, plugged it in to charge and it didnt stop
10:41:24LinusNyou could try to replace the battery, but i doubt that it would help
10:41:31amiconnboooooooooo for the line selector reversal! :-((
10:41:32darkskiezjust as rockbox was maturing too :(
10:42:15darkskiezis the drive pinout the same as a laptop hdd ?
10:42:20*[IDC]Dragon joins the boooooooo
10:42:23darkskiezconnector wise, if I want the data
10:42:40LinusNdarkskiez: don't remember
10:42:50pikewhat does "Dir Buffer is Full" mean, I get that when I enter some of my folders
10:43:23LinusNpike: it means that the directory buffer is too small to hold all file/dir names
10:43:31LinusNhow many files are there?
10:43:55pikewould need to check
10:44:04pikecurrently trying locking
10:44:24LinusNpike: either organize your files better, or increase the buffer size
10:44:49LinusNgeneral settings->system->limits
10:44:51CoCoLUSthe latter one being the quicker solution
10:48:58pike546 files
10:49:40LinusNthe default buffer size is 400 files
10:51:50 Join leftright [0] (
10:52:26leftrightperhaps the default buffer should be increased for those that dont read manuals and phone support :)
10:52:37pikehow does rockbox handle this: player is off, charger is connected. leave it during night, later you power it on while charger is still connected. now indicator shows that it's charging
10:53:41B4gderleftright: no
10:53:45preglowboost ratio isn't a very good performance indicator :)
10:54:16LinusNpike: rockbox doesn't yet detect that the battery isn't being charged, only that the charger is connected
10:54:55leftrighthmm, why does the charge light extinguish after charging the batt ?
10:55:04leftrightwith charger still connected
10:55:08pikeah I see, so all the charge indicator shows, is that external power is used ?
10:55:19pikebut it indicated charging !? :)
10:55:26pikegreen diode was off
10:56:03[IDC]Dragonhehe, Creative delivered ~4000 Zen player with a virus on the disk
10:56:41LinusNpike: rockbox doesn't yet detect that the battery isn't being charged
10:57:32pike2x negative declaration, too confusing
10:58:58LinusNon the iriver, rockbox only detects that the charger is attached, it doesn't yet detect the battery full condition
10:59:18darkskiezpike: i think with the iriver is off and u charge it rockbox doesnt do anything.
10:59:34leftrightfor info, I have been playing mp3 files since the 27th (with 27th's build) and R-Box hasnt crashed once
10:59:54LinusNdarkskiez: true
11:03:42darkskiezif I was buying a new mp3 player now my beautiful iriver has died, it it worth getting an H300 series ?
11:04:27B4gderwhat decides if it is "worth it" then?
11:04:47B4gderit is probable that Rockbox will run on it in a future
11:05:22paughdarkskiez, i've heard that iriver are pretty good ad fixing or replacing broken units regardless of the circumstances. maybe try sending it back?
11:10:51pikeloaded with rockbox ? ;)
11:11:19amiconnLinusN: Now we're almost back to square one with the code size issue :(
11:12:31[IDC]Dragonhow do I get gcc 3.3.6 ina convenient way? (cygwin complied)
11:13:03[IDC]Dragonis BC up to that?
11:13:03 Quit leftright ("CGI:IRC (EOF)")
11:13:27paughpike: "regardless of the circumstances"
11:13:47amiconn[IDC]Dragon: I've just built it myself.
11:14:14amiconnThe good thing is that if you have an up-to-date cygwin installation, you don't need the newlib workaround any more
11:14:33amiconnThat reduced gcc build time from >2 hours to ~30 minutes on my laptoü
11:15:26[IDC]DragonI don't know if my cygwin is up to date
11:15:41[IDC]Dragondo you mind sending me the binary?
11:15:50amiconnI am updating my cygwin installation more or less regularly
11:16:07amiconnThe cygwin installer handles this very well
11:16:24darkskiezpaugh: out of warranty :(
11:16:30[IDC]DragonI'm a fan of "if it aint broken, don't fix it"
11:17:14amiconn[IDC]Dragon: My binaries are built to run from /opt/sh1. If you also want it that way, I could zip this folder.
11:17:15[IDC]Dragonafter overcoming being an adolescent version junkie
11:17:31*[IDC]Dragon checks
11:18:06amiconnYou would then have to include /opt/sh1/bin into your search path if it's not already there
11:18:33amiconn82.4 MB uncompressed
11:19:04amiconn...containing sh binutils 2.16, gcc 3.3.6 and gdb 6.1.1
11:19:06[IDC]DragonI'm different, have the compiler in /rockbox/bin
11:20:29darkskiezd'ya reckon i can get a working mainboard from somewhere, like if someones dropped it an forked the hdd/screen
11:20:34[IDC]DragonI thought I just need the 3 sh-elf-*.exe files
11:20:36darkskiezi have, i hope, i working hdd and screen
11:20:42amiconnI used to have /rockbox as well, but moved to using /opt for everything extra
11:20:57amiconnI have the m68k toolchain under /opt/m68k
11:21:15amiconn[IDC]Dragon: No, you'll need all the libraries etc as well
11:21:19[IDC]Dragonme too, I just found
11:21:55amiconn54.9 MB uncompressed for m68k (binutils 2.16, gcc 3.4.4, no gdb)
11:22:27[IDC]Dragonmy m68k dir has 28 MB
11:23:38 Join Wett [0] (
11:24:18amiconnYou probably have an older revision of gcc 3.4, and it was built to support mcf52xx only (old gcc build patch)
11:24:28[IDC]DragonI was wrong, sh-elf stuff runs from /opt/sh1, too
11:24:43amiconnMy version supports all m68k targets except 68040 (new gcc build patch)
11:24:47[IDC]Dragonwhich has 37 MB
11:24:56amiconnsh-elf-gcc -v ?
11:25:12[IDC]Dragonfor me not needing it, I have more than enough ;-)
11:26:07amiconnsh-elf-as −−version ?
11:26:56amiconnold ;)
11:27:35amiconnMy whole opt folder is 44.8 MB zipped
11:27:39[IDC]Dragonyes, I'd like to have the "official" compiler when releasing bootbox
11:28:38 Join Aramil [0] (
11:28:43[IDC]Dragonlet's see what Eric has for us
11:30:35[IDC]Dragon2.16 binutils, but stil gcc 3.3.5
11:31:12amiconnI'm putting my opt folder zip onto my web server, see pm for link :)
11:31:33[IDC]Dragonplease take out the 68k
11:32:10*[IDC]Dragon is updating cygwin now
11:32:25[IDC]Dragonso I'll be fine with binutils, too
11:32:40*kurzhaarrocker thinks it is incredible that compilers & co can be serveral dozen MB big
11:34:17[IDC]Dragonsh-elf-as −−version is 2.16 now
11:36:59 Quit tvelocity (Connection timed out)
11:46:43 Join webguest35 [0] (
11:47:21 Quit webguest35 (Client Quit)
11:49:19 Join Zagor [0] (i=foobar@pdpc/supporter/sustaining/Zagor)
11:51:25[IDC]Dragonhi Zagor, long not "seen"
11:52:01Zagoryeah. life somehow seems to keep me busy... :-)
11:52:07[IDC]Dragonme too
11:59:06preglowi don't see why the emac needs to be initialised from within mpa.c
11:59:21preglowit should be initialised locally wherever it's needed
12:02:24***Saving seen data "./dancer.seen"
12:03:03 Part Aramil ("Leaving")
12:04:37*preglow summons Slasher
12:06:35Slasherpreglow: Hmm, i am not sure but there will be a glitch after boot if the mac is not initialized
12:06:50 Join Mark__ [0] (
12:06:51preglowthen i've screwed up some place
12:07:04preglowi know why, of course
12:07:10preglowi'll move it to libmad init code
12:07:10Slasherhmm :)
12:07:11preglowwhere it belongs
12:07:24Slashergood, it's definately a better place :)
12:08:16preglowwe might need a swap-back routine for the codecs
12:08:34preglowfor instance, if you play wavpack, and use mp3s for voice, the results might be rubbish
12:09:18preglowbad example, flac would be a better example
12:09:38preglowsince sets the emac in integer mode
12:09:52preglowsince flac sets, yes
12:10:12Slasheroh, that might be a problem
12:10:21preglowproblem is that mad_f_mul is used everywhere in libmad code
12:10:26preglowand it expects the emac to be set in frac mode
12:10:38preglowi can't set the emac to frac mode in every mad_f_mul, that would slow things down
12:11:01Slasherhmm.. maybe the codec swap routine should check the mac mode and reset it back what it was when switching the codec back
12:11:09preglowthat's a good idea
12:11:12preglowvery nice
12:12:24preglowdoesn't need to check either, just saving it and reverting it would do nicely
12:12:38preglowafaik, at least
12:14:54preglowthink i'll put the emac init in mad_frame_decode, that _should_ do the trick
12:16:26preglowand there's _got_ to be a more structured way of doing the codecs than a while (1) and gotos ;)
12:16:31Slasheryes, that should be enough because the codec swap can happen only during pcmbuf_insert -callback
12:16:44Slasherhehe :D
12:17:02preglowbut anywho, i need to put up some furniture and go vote
12:18:07preglowall these practicalities are stealing my rockbox time :/
12:28:56 Join Moos [0] (
12:29:14MoosGood evening
12:29:22 Join webguest33 [0] (
12:29:53 Join DangerousDan [0] (
12:30:03 Join webguest63 [0] (
12:31:56 Quit webguest63 (Client Quit)
12:33:20amiconnpreglow: Some time ago I experimented with the MAC for faster 64x64 bit multiplication.
12:34:32amiconnI remember some codecs had mac init problems, causing periodic bursts of static when mandelbrot.rock with the macified 64 bit multiplication was running while music was playing
12:35:00amiconnIirc one of these was mpa.codec, and vorbis.codec didn't have that problem
12:35:46amiconnIt's a non-issue now, because I scrapped the macified routine as using standard mulu.l and mulu.w was faster anyway
12:36:13amiconn(not due to the instructions being faster, but due to using less registers, hence less stack access)
12:36:52 Nick Mark__ is now known as Mark__`afk (
12:44:29 Quit webguest33 ("CGI:IRC (Ping timeout)")
12:44:51 Join hicks [0] (
12:48:39 Quit DangerousDan (Read error: 104 (Connection reset by peer))
13:10:43 Join DangerousDan [0] (
13:18:46 Quit Nibbler (Read error: 113 (No route to host))
13:19:27 Join Nibbler [0] (
13:23:45 Join Sucka [0] (
13:26:33 Quit paugh ("speaking of reboots..")
13:47:22pillwould it be possible for rockbox to open .nfo files using terminal 9 font?
13:50:20HClno idea
13:52:26 Join edx__ [0] (
14:02:26***Saving seen data "./dancer.seen"
14:08:54 Quit edx (Read error: 110 (Connection timed out))
14:09:09 Nick edx__ is now known as edx (
14:12:06 Quit Sucka ("a bird in the bush is worth two in your house")
14:14:26 Nick Mark__`afk is now known as Mark__ (
14:14:55preglowamiconn: you raise a good point
14:15:22preglowi'm starting to believe macsr should perhaps be saved in the thread context
14:27:15amiconnHmm, we could consider macsr belonging to the volatile environment, as gcc doesn't save this in functions (because it doesn't know about it)
14:27:50amiconnHowever, not all threads use the mac, so the additional overhead might not be desirable
14:38:24 Quit [IDC]Dragon (Read error: 110 (Connection timed out))
14:42:27 Quit kurzhaarrocker (Remote closed the connection)
14:42:56 Join drfeelgood [0] (
14:43:23drfeelgoodgood morning!
14:44:14drfeelgoodgot a small problem w/ my H120 and Rockbox
14:44:43drfeelgoodI just upgraded to the 09/01 bleeding edge build as of 8am EST
14:45:04drfeelgoodi also tried to upgrade to the latest bootloader
14:45:20drfeelgoodi patched the firmware and all like it says on the site
14:45:39drfeelgoodwhen i disconnected my H120, it recognized the new v. of rockbox and asked to reboot, which i did
14:45:56drfeelgoodi then turned the player off and tried to reboot into the iRiver fw
14:46:15drfeelgood(this is before I could flash the latest patched firmware)
14:46:28drfeelgoodNow, it hangs on the "read file System" display
14:46:41LinusNis the battery charged?
14:46:53 Join muesli- [0] (
14:47:04drfeelgoodmy reading according to rockboot v4 loading screen is 4.01v
14:47:16LinusNok, start the player with the usb cord inserted
14:47:25LinusNit will enter usb mode
14:47:32LinusNthen you should be able to perform a scandisk
14:48:23LinusNthe iriver firmware has issues with the file system
14:49:02amiconnLinusN: What's your opinion about saving macsr in the thread context?
14:49:22LinusNamiconn: not sure
14:49:44LinusNbut if it simplifies for the codecs, i'm all for it
14:50:28drfeelgoodLinusN: do I want to have scandisk automatically fix errors or attempt recovery of bad sectors?
14:50:40LinusNnot automatically
14:50:57drfeelgoodso if i don't choose that, it'll prompt me at the end?
14:51:30drfeelgoodok, i just ran scandisk.
14:51:47drfeelgoodIt ran through and at the end popped up a box saying "disk scan complete"
14:51:51drfeelgoodor something to that nature
14:52:09LinusNit found no faults, but still the iriver firmware hangs
14:52:15LinusNthat sucks
14:52:21drfeelgoodthat was my theory
14:52:52drfeelgoodwhat doubly sucks is that I rely on the iRiver fw for remote support still
14:53:08 Join Sucka [0] (
14:53:17drfeelgoodi wasn't thrilled with the patch for rockbox, so i'm just gonna wait it out until it's built in by you guys
14:53:41LinusNdrfeelgood: there is a tool called "directory snoop"
14:53:54 Join webguest78 [0] (
14:54:08LinusNthat might give us some clue why this happens
14:54:29drfeelgoodwhere do i get it?
14:54:47webguest78Hey Linus (if you're here) could the songrating WPS tag work like the new playmode does? So I can put different text (or image!) depending on the rating?
14:56:14LinusNwebguest78: that could be arranged
14:58:17webguest78maybe it should be a different tag, so people that want the old version don't have to do %?rr<0|1|2|3|4|5|6|7|8|9|10>
14:59:40drfeelgoodalright linus, i got directory snoop - anything specific i need to do?
14:59:42LinusNit should work both ways
15:00:02LinusNdrfeelgood: i haven't run directory snoop in years, so i don't really remember
15:00:13 Join muesli__ [0] (
15:00:14webguest78Ah, didn't know how it worked so I thought maybe it wouldnt work
15:00:24amiconnwebguest78: Isn't the plain rating tag %rr ?
15:00:40amiconnSo it would automatically be different from %?rr
15:01:07LinusN%rr will show the rating
15:01:15WettSorry to bother you, but i'd like to understand how works the codecs architecture... I read mpa.c but there a too much things I don't know : how is feed the input buffer, etc.
15:01:22amiconnIn fact all conditional and multi-conditional tags could work that way.
15:01:22LinusN%?rr<|||||||> will work too
15:01:27webguest78So it's only in the case of %?rr that it acts differently?
15:02:02amiconnWithout ? they would return a number of 0..n (0..1 for booleans), and with ? they would use the <..|..|..> form
15:02:17amiconnThat should allow for some nice code reuse...
15:02:28LinusNthe volume would be a pain though :-)
15:03:22amiconnWell, you aren't forced to use the %?tag<..|..> form
15:03:40LinusNof course not
15:07:02muesli__could we have a condition to use a pic for the codec?
15:09:30webguest78if it works like amiconn suggests you could
15:09:42 Quit OnkelJonas (Read error: 104 (Connection reset by peer))
15:09:59LinusNit already works like that
15:11:33webguest78so it could be made to work like that with not so much effort?
15:11:38LinusNsoon done
15:14:33 Join Aison [0] (
15:16:52drfeelgoodLinus, i know this sounds a little drastic, but if I backed up the entire drive and reformatted it, think that would alleviate the error?
15:18:59drfeelgoodwell, thanks for your help.
15:19:22drfeelgoodjust dragging and dropping all the files off my drive onto my C: drive is enough of a backup, right?
15:20:08drfeelgoodwell, i'll be off to reformat my drive.
15:20:18drfeelgoodThanks for the troubleshooting help!
15:20:21LinusNok, %?rr<|||> and %?fc<||||> is now possible
15:21:12webguest78talk about quick action to user-requests ^_^
15:21:26B4gderlet's hope we add new codec names to the end then ;-)
15:22:42WettJust in case of anyone didn't see my question a little above, i'd like but don't find the way to play sound in a plugin :)
15:22:57 Quit muesli- (Read error: 110 (Connection timed out))
15:23:13MoosSalut Wett :)
15:23:27Mooshow are you with your video plugin?
15:23:37Wettsame as yesterday :(
15:23:45Moosstill playback problems?
15:23:55*B4gder doesn't know how to play sound in plugins
15:23:55WettI'm not used with this buffered and stream stuff
15:24:31WettI'd like to understand what mpa.c does exactly, reading the code is not enough...
15:25:47LinusNthe plugin interface can only currently only play pcm audio on the iriver
15:26:22Wettyes, I know that, so I wanted to deal with any codec directly
15:26:38Wettto be exact, with libmad
15:26:53LinusNthat is unfortunately not possible in a plugin
15:27:30LinusNwhy do we have AFMT_APE and AFMT_REAL?
15:27:48LinusNand AFMT_AAC?
15:28:05B4gderfor codecs?
15:28:14LinusNseems silly to have defines for formats we don't support
15:28:26B4gderI thought APE was a tag format
15:28:27Wettoh, I see... So I can't do anything until this is upgraded...
15:28:36LinusNWett: or make it happen
15:29:04LinusNi'll remove the unsupported formats
15:29:35B4gderLinusN: add a comment about adding new ones to the end of the list for WPS enum reasons
15:29:35webguest78cool, my rating is now shown as roman numerals
15:29:45webguest78(not exactly what I was aiming for, but for testing..)
15:30:05 Nick Mark__ is now known as Mark__`afk (
15:30:41Wettok, so i'll have to do it myself. But the problem is the same than before : I don't understand everything in this codec arch. :D
15:33:06 Join linuxstb [0] (
15:33:46linuxstbWett: I don't think you need to know about the codec architecture - your plugin won't use it.
15:34:20linuxstbYou should look at mpa.c as an example of how to use libmad.
15:34:39linuxstb(unless Rockbox is changed to give plugins access to the codecs)
15:35:20Wettthat's what I did, but there are a lot of things I don't get ! Don't know where are the buffers, how they are feeds, etc
15:35:40linuxstbHave you looked at the mpa2wav plugin?
15:36:05linuxstbThat simply works by reading the whole MP3 file into a memory buffer, and then decoding it using libmad, and writing to a WAV file on disk.
15:36:19Wettyep :( This is the first time I get into audio stuff, seems it's may be a little too early
15:36:41LinusNlinuxstb: that plugin is gone
15:36:48linuxstbI think mpa2wav is simpler than mpa.c - mpa.c has been developed quite a lot, and contains a lot of irrelevant code
15:36:57linuxstbLinusN: I know - but it's in CVS still (Attic)
15:37:19Wettyes, dragon gave me the link yesterday
15:38:22WettCan I ask precise question about libmad functions here ?
15:38:55linuxstbI don't have much time, but shoot.
15:40:57Wettthe main function seems to be mad_frame_decode() which take a frame and a stream struct. where is stored the input and output buffers ?
15:42:06 Join ashridah [0] (
15:42:24LinusNgotta go
15:42:26 Part LinusN
15:42:59Wettcan I just give the codec some frames, without any mp3 headers or others informations about frequency, etc. ?
15:44:03linuxstbThe output data is stored in Frame, which is then converted into PCM samples by the mad_synth_frame function. This creates a Synth struct containing the actual PCM samples.
15:44:15linuxstbFrame also contains the samplerate etc of the decoded frame.
15:45:09linuxstbI'm not sure I can remember how the input buffer works (I'm looking at the source now), but yes, I think you just pass the data blindly and don't have to worry about initialising anything.
15:45:55linuxstbYou seem to use the mad_stream_buffer function to give libmad data. You call this before you call mad_frame_decode.
15:46:40linuxstbSo you start by calling mad_stream_init, mad_frame_init and mad_synth_init.
15:46:48Wettgreat ! Can I pass data not frame aligned ?
15:47:09WettI mean, using mad_stream_buffer
15:47:29linuxstbYou then call mad_stream_buffer to give some data, followed by mad_decode_frame and then mad_synth_frame
15:47:57Wettok I see better what they're for
15:51:20Wettabout input buffer filling, if I get stream.next_frame!=0, that means next time codec will need a new frame, right ?
15:52:10Wettwell, you're busy I don't bother you anymore :) thanks a lot !
15:52:33B4gderrockbox talk isn't bothering us
15:52:40linuxstbSorry - I am busy. I'll try to come back tonight when I'm not supposed to be working at the same time :)
15:52:59Wettlol ok ^^
15:53:24linuxstbI may be able to find a simple libmad example program that I based mpa.c on. I haven't used libmad for a few months, so I've forgotten the details.
15:53:37 Quit linuxstb ("CGI:IRC")
15:53:58webguest78hm, why does the preload tag have coordinates? wouldn't it be more flexible to put the coordinates when you're displaying it?
15:54:07webguest78the image preload tag, that is.
15:54:19B4gderjust to make the conditional use of it easier
15:54:42webguest78But what if you want to show the same image twice, in different places
15:54:45B4gderotherwise it would be a whole lot of nested |
15:54:52B4gderwebguest78: then you preload both
15:55:01webguest78I see
15:55:22B4gdera future fix will make sure that the second one isn't loaded but uses the first buffer again
15:55:53 Quit muesli__ (Read error: 110 (Connection timed out))
15:59:42amiconnB4gder: I think giving the coordinates within the display tag instead of the load tag is the better way
16:00:29amiconnIt's simpler to reuse images in multiple positions, and it could even be extended to clip regions
16:00:36 Quit drfeelgood ()
16:01:25amiconnThis way we wouldn't need multiple .bmp files for multiple wps graphics. All graphics belonging to one wps design could be part of one single .bmp
16:02:27***Saving seen data "./dancer.seen"
16:04:21amiconnLinus' last change introduces a problem on archos... :(
16:12:42webguest78is 2-bit greyscale planned for wps bitmaps?
16:17:13webguest78Are comments in WPS only allowed at the top?
16:19:33 Nick Aison is now known as NewOrleans (
16:19:41 Quit NewOrleans ("has quit USA (Excess Flood)")
16:21:25 Quit edx (Read error: 104 (Connection reset by peer))
16:21:41 Join Aison [0] (
16:27:36 Join edx [0] (
16:32:17 Quit webguest78 ("CGI:IRC (EOF)")
16:45:58 Join drfeelgood [0] (
16:47:05drfeelgoodhello everyone! question, what do I use to format the player if windows won't allow me to format the player?
16:47:08 Join kurzhaarrocker [0] (
16:47:30kurzhaarrockergrrrr. How am I supposed to update a patch when sourceforge doesn't work!
16:47:30drfeelgoodi've tried using a program LinusN recommended, SwissKnife, but that hangs whenever i try to access the drive
16:47:35ashridahwihch playm odel?
16:47:47ashridahdrfeelgood: which player are we talking about?
16:48:01dwihnodrfeelgood: there is also some application demo from paragon software
16:48:27ashridahdrfeelgood: iriver's original firmware has a format feature built in
16:48:41drfeelgoodsry, i can't access that
16:48:41ashridahjust go to the general menu
16:48:58drfeelgoodto make a long story short, i couldn't access the iriver fw
16:48:59ashridahsounds like the player's borked then to me
16:49:05drfeelgoodso i was going to reformat
16:49:13 Part kurzhaarrocker
16:49:22drfeelgoodcopied everything off, started a reformat, program hung part way through
16:49:24ashridahi'm not sure how formatting'd help tbh
16:49:50ashridahiriver's firmware tends to work to a point even if the hd's been trashed.
16:49:51drfeelgoodlinusN and i decided it was a file system problem, as per the Iriver FAQ for non-booting orig. fw
16:50:35drfeelgoodthank god, i forgot about that
16:50:39drfeelgoodlet me try something
16:50:42ashridahinteresting. as i say, most of the hd issues i've seen with iriver's firmware have still resulted in an iriver firmware that booted (but wigged out when trying to browse the tree)
16:50:48drfeelgoodhow long does the iRiver f/w take to reformat the drive
16:51:05ashridahdrfeelgood: about 20-30 seconds
16:51:13ashridahit's not long,
16:51:15drfeelgoodwell, we'll see
16:51:23ashridahit repartitions too
16:51:42drfeelgoodi just reformatted, forgot i could ignore rockbox, who wigged over my HD and boot into iRiver
16:51:55*ashridah nods
16:52:08ashridahthat's what i meant. get into iriver's firmware, and use its built in format tool
16:52:26drfeelgoodwindows sees the drive again
16:52:50drfeelgoodnow, when i was trying to copy all the files off BEFORE the reformat, one of my config files wouldn't copy
16:53:06drfeelgoodit said "Can't read from source disk/file" or something along those lines
16:53:17drfeelgoodmight that have been messing up the iriver f/w?
16:53:42drfeelgoodwhen windows wouldn't delete the file, i just went into dos and trashed it since it wasn't my primary cfg file
16:53:50ashridahhrm, as i say, i've found the iriver firmware to be reasonably robust
16:54:05ashridahi'd have been inclined to run chkdsk against the unit
16:54:16drfeelgoodi ran windows built in scandisk
16:54:25drfeelgoodwhen you open the drive properties under tools
16:54:28drfeelgoodno problems
16:54:35drfeelgoodnone from norton's disk doctor either
16:54:50drfeelgoodthat's what made Linus and I decide on reformatting it
16:55:02drfeelgoodwell, i'm gonna copy my files back over
16:55:07drfeelgoodthanks for the trauma support
16:55:32drfeelgoodhow stable is rockboot v5?
16:57:08 Quit Febs (Read error: 110 (Connection timed out))
16:58:08 Quit drfeelgood ()
17:00:41preglowamiconn: a simple 32 bit save extra per thread doesn't sound like any overhead worth mentioning to me
17:01:14preglowamiconn: the alternative is having a 'codec_restore' function that has to be called every time a codec needs to generate data
17:01:50preglowamiconn: plus, this isn't just a codec problem, it's a problem for ordinary plugins using emac as well, the only easy way to fix this is to ad macsr to the thread context, or require everyone to reinit the emac after every yield()
17:03:29 Join Febs [0] (
17:03:59preglowhmm, there might be a problem with the dsp part of the thread clobbering macsr as well
17:04:48amiconndsp runs in audio thread, while codecs run in codec thread, correct?
17:05:13preglowi have no idea whatsoever
17:05:26preglowif the only thing running in the codec thread is the actual codec, then the context save will do fine
17:07:52amiconnSo there are 2 questions: (1) In which thread does the dsp code run (2) is saving MACSR enough? What about the accumulators and MASK
17:08:37amiconnStill, each part needs to init the emac at the start
17:08:43preglowwe will never depend on mask the way we depend on macsr
17:08:59preglowand we've got a zero accumulator policy
17:09:04preglowyou leave them at zero when you're done with them
17:09:19preglowi decided this because of the movclr instruction
17:09:32 Quit ashridah ("sleep")
17:09:59preglowamiconn: at the start, yes, but i don't need to worry about macsr in every single mad_f_mul anymore
17:10:00amiconnOn SH1, the multiplication accumulator (used for both normal multiply insn and mac) is declared scratch, so we can do this on coldfire as well
17:10:28preglowamiconn: i'd much rather like them to be required to be zero, then you don't need to clear them before you use them
17:10:41preglowamiconn: it's just a matter of always using movclr.l instead of move.l
17:11:26amiconnI think it doesn't matter
17:11:51amiconnBoth methods have the same drawback that you can't rely on the accumulator content after yield()
17:11:56preglowwell, it clearly does, my way saves code space and is slightlyfaster
17:12:09amiconnMaybe another thread changed it (to zero or whatever)
17:12:19preglowamiconn: for example, mad_f_mul can't be coded the way it is if i can't rely on the accumulators being zero
17:12:41preglowamiconn: i'll have to add a move.l #0, %acc0 to every single mad_f_mul in all of libmad
17:12:57preglowso it does matter
17:13:12amiconnI mean if we say accumulators are scratch, this policy can be added and wouldn't influence thread switching
17:13:26amiconnSo be it the zero policy...
17:13:31preglowahh, yes
17:13:35preglowbut it already is that way ;)
17:13:51amiconnI hope interrupt code will never use the emac...
17:14:09preglowcan't see a reason for it to do so, still, it wont be a problem
17:14:17amiconnIt would, for sure
17:14:29preglowwhy? it just has to save the entire emac state
17:14:57amiconnHmm, but __attribute__((interrupt_handler))
17:15:11amiconndoesn't do this, as gcc doesn't know about emac...
17:15:29preglowahh, yes, i completely forgot it's in c
17:15:33preglowbrb, wash clothes
17:15:53amiconngcc should really learn about emac...
17:16:47preglowoh, i agree
17:16:52preglowi tried looking at it to add it
17:16:55preglowbut it's way over my head
17:17:45amiconnNeither do I have any idea about adding such things to gcc...
17:18:05amiconnI would like to add my ed 64bit multiplication...
17:20:56 Join yngwi [0] (
17:21:30 Join TCK [0] (
17:24:06preglowi would like to add a complete intrinsics set
17:24:14preglowthat schedules instructions optimally
17:24:20preglowshould be a simple job for gcc, but i dont know how
17:24:43preglownot much to schedule, just stuff as many mac.l as possible in a row, and wait 2 cycles before you fetch after a mac
17:24:53preglowbut brb, gotta wash kitchen
17:28:23B4gderwash kitten? ;-)
17:28:47 Quit Febs (Read error: 110 (Connection timed out))
17:33:27 Join LinusN [0] (
17:43:12amiconnLinusN: Your latest change causes a problem on archos...
17:44:22amiconnThe AFMT_* values no longer match the strings in id3.c on archos, because the string for mpeg layer 1 is #ifdef'ed, but you removed the #ifdefs for the AFMT_* values
17:44:30LinusNah, yes
17:45:52amiconnThe endmarker could also be removed, and the char * array shortened by one as a result
17:47:42 Quit DangerousDan (Read error: 104 (Connection reset by peer))
17:47:53LinusNgotta go
17:47:55 Part LinusN
17:48:15amiconnThe endmarker is only used to determine the number of entries, and that can be done with sizeof(codec_labels)/sizeof(char*)
17:55:38preglowamiconn: but yeah, you agree adding macsr to the thread context would simplify things quite a bit?
17:59:50 Join amiconn_ [0] (
18:02:30***Saving seen data "./dancer.seen"
18:08:13 Quit Sucka ("a bird in the bush is worth two in your house")
18:08:41 Join Sucka [0] (
18:12:48 Quit yngwi ("Chatzilla 0.9.68a [Firefox 1.0.6/20050717]")
18:17:37 Quit amiconn (Read error: 110 (Connection timed out))
18:17:37 Nick amiconn_ is now known as amiconn (
18:36:14 Join webguest66 [0] (
18:36:44 Quit webguest66 (Client Quit)
18:37:40 Join dpassen1 [0] (n=dpassen1@resnet-233-61.resnet.UMBC.EDU)
18:40:35*amiconn is really disappointed about today's changes :(
18:44:27preglowwhat about them?
18:44:27 Join DangerousDan [0] (
18:50:00 Quit darkskiez ()
19:03:18 Nick Mark__`afk is now known as Mark__ (
19:03:18Mark__ _( Y )_
19:03:18Mark__ ()~*~()
19:03:18Mark__ (_)-(_)
19:05:47preglowyes, that's quite something you've got there
19:29:29 Join _Wett [0] (
19:29:36 Join bagawk [0] (i=1000@unaffiliated/bagawk)
19:33:31 Nick Mark__ is now known as Mark__`afk (
19:37:03 Nick Mark__`afk is now known as Mark__ (
19:37:03Mark__ _( Y )_
19:37:03Mark__ ()~*~()
19:37:03Mark__ (_)-(_)
19:38:00preglowanyone for banning people who use away nicks?
19:38:16bagawkB4gder, I assume you are 'bagder'?
19:38:37Mark__i really dont know why the bunny script keeps bleeding over to this channel
19:38:46Mark__i'll /part for now to stop it
19:38:48preglowyou _have_ a bunny script? :P
19:38:53Mark__ _ _
19:38:53Mark__ ( \_/ )
19:38:53Mark__ (=^_^=)
19:38:53DBUGEnqueued KICK Mark__
19:38:53Mark__ (-)_(-)
19:39:03 Quit Wett (Read error: 110 (Connection timed out))
19:39:07preglownow, there's a sensible thing to have
19:39:08bagawkNot that is pointless...
19:39:09Mark__its also a teddy and penguin script
19:39:13Mark__ (o_
19:39:20preglowagreed, please stop spamming
19:39:35CtcpIgnored 1 channel CTCP requests in 0 seconds at the last flood
19:39:35*Mark__ slaps Mark__ around the head with box of matches.
19:58:19pikeho bagawk
19:59:17bagawkpike, Hello
19:59:36bagawkpike, I had no idea you were the same pike
19:59:44 Join LinusN [0] (
19:59:53pikeso many of us?
20:01:00bagawkHey LinusN
20:01:19bagawkLinusN, I was thinking of adding a parts avail page to the wiki
20:01:21 Join arkascha [0] (
20:01:39bagawkLinusN, I have a number of parts around people could use (screens, buttons etc)
20:01:56bagawkLinusN, Is that okay by you?
20:02:33LinusNamiconn: u there?
20:02:34***Saving seen data "./dancer.seen"
20:02:52preglowwhat kind of parts?
20:03:51bagawkpreglow, archos parts
20:05:10arkaschagot a clear newby question about a firmware build under GNU/linux. Firmware has only 8kb and produces a roro checksum error. Any easy hint?
20:05:38preglowso it should
20:05:45arkaschasorry, iriver
20:05:52preglowgcc and ld version?
20:06:05arkaschagcc 3.4.4 and a moment...
20:06:21arkaschald 2.15.94
20:07:55preglowno link errors?
20:08:02arkaschano error, no warning
20:08:23 Quit dpassen1 ()
20:09:07preglowwhat does 'file' say to rockbox.iriver?
20:09:19preglowforget it
20:09:24preglowit's raw, of course
20:09:26arkascharockbox.iriver: data
20:10:14arkaschasorry for the newby stuff, just tried rockbox first time yesterday, it rocks ! thanx !
20:10:15LinusNdid you follow the instructions here:
20:10:34arkaschagotta check, a second
20:10:57arkaschayep, that's it
20:11:07LinusNespecially rhis:
20:11:31preglowyou _could_ try upgrading to binutils 2.16
20:11:36arkaschayep, read that one
20:12:43arkaschaI installed binutils 2.16 especially for rockbox, since I usually have 2.15
20:13:06LinusNm68k-elf-as −−version
20:13:35 Nick Mark__ is now known as Mark__`afk (
20:13:35DBUGEnqueued KICK Mark__`afk
20:13:44arkaschasorry, I am a newby, I did specify m68k as target
20:13:53arkascha, that what you mean
20:14:09LinusNno, type what i wrote
20:14:35LinusNbtw, the target should be "m68k-elf"
20:14:42arkaschawait a second, even more beginner stuff: you don't say the rockbox.elf file is the right firmware?
20:15:08arkaschaok, thought so. I did specify m68k-elf as target sorry
20:15:33LinusNno errors whatsoever?
20:15:56arkaschabash: m68k-elf-as: command not found - of course
20:16:00arkaschano compile errors...
20:16:14LinusNcommand not found?
20:16:20LinusNthen it can't work
20:16:20preglowthat's wrong
20:16:26preglowyou should have the entire binutils toolchain
20:16:32preglowi wonder why the hell it built
20:16:46arkaschasorry LinuxN: GNU assembler 2.16 was the correct result
20:16:58arkaschaThis assembler was configured for a target of `m68k-elf'
20:18:03LinusNsounds correct
20:18:35LinusNand m68k-elf-gcc −−version?
20:18:54arkascham68k-elf-gcc (GCC) 3.4.4
20:19:16LinusNsource from cvs?
20:19:44arkaschadownload from link latest version, not cvs checkout
20:20:02 Nick Mark__`afk is now known as Mark__ (
20:20:02DBUGEnqueued KICK Mark__
20:20:08LinusNok, so you have a rockbox-20050901/ directory?
20:21:05arkaschano, called rockbox-2.4, but I don't remember if I renamed it, gotta check that
20:21:25LinusNthat won't work
20:21:27arkaschasorry, said I am a bloody beginner
20:21:41LinusNyou must use a daily build
20:22:03preglowhow'd you actually manage to compile 2.4 for m68k?
20:22:13arkaschaconfigure / make ... .-)
20:22:52preglowyes, seems we solved it then, hehe
20:23:06arkaschaI'll try the daily build and be back, *thanx* for the help !
20:23:52preglowyou should be fine with a daily
20:24:03bagawkLinusN, I added a parts page, but it does not show up as a link on the main wiki page?
20:24:20LinusNbagawk: should it?
20:24:31bagawkLinusN, See yes
20:24:37bagawkLinusN, Look at the main wiki page
20:25:02LinusNthey don't end up there automatically, you know
20:25:17LinusNand please rename it to a proper WikiWord
20:25:30bagawkLinusN, ahh yes
20:25:58LinusNSpareParts maybe?
20:26:08amiconnLinusN: Now I am here
20:26:18LinusNamiconn: dosappointed?
20:27:30amiconnLinusN: Yes
20:27:51amiconnOne reason already killed by me
20:27:59 Join Lear [0] (
20:28:20LinusNthe codec type?
20:29:32preglowsomething tells me the code size is another thing
20:29:46 Quit cYmen ("wusch")
20:30:38LinusNyes of course
20:30:51LinusNfortunately, the builds are still green
20:31:08preglowso we've gotten to a point where we really can't add anything to the archos builds without doing massive size opts first?
20:31:12amiconnThe reintroduction of the line selection pointer eats more than 1KB
20:31:23amiconn..for no reason, imho
20:31:33LinusNin your humble opinion, yes
20:31:35preglowi dont think your reason was the reason for it being reintroduced, heh
20:31:36amiconnThere are some optimisation possibilities
20:32:27amiconnI hope to be able to get some in before the release
20:32:54LinusNtime to read a fairy tale to my son
20:33:03LinusNcu later
20:33:05amiconnOne of them will hopefully fix the long line problem in the quickscreens
20:33:10LinusNamiconn: nice
20:33:12 Part LinusN
20:34:35 Nick _Wett is now known as Wett (
20:38:14amiconnpreglow: Regarding binary size: Now we have ~550 *bytes* to play with on the fmrecorder (worst-case target)
20:38:32preglowwell, it was bound to happen sooner or later
20:39:18amiconnYes, but it can be made happen far later with some clever optimisation
20:39:31preglowyes, but it will happen
20:39:34preglowand i wonder how we'll deal with that
20:39:41amiconnThe reintroduction of the pointer arrow was a plain rollback of not-very-clever code
20:40:31amiconnWe will need quite some extra space if we want to use the pcm codec
20:40:41amiconnI do want that!
20:41:17preglowyes, of course
20:41:21preglowit's a major feature
20:41:25preglowhow much are we talking?
20:46:12amiconnThe codec itself is ~3K words == 12 KByte
20:47:01amiconnIt can be compressed to ~7.5 KByte as the MAS only uses 20 bit words, at the cost of some extra code
20:47:31amiconnOf course some extra code is needed for downloading the codec to the MAS, set it up etc.
20:47:56amiconnThe playback engine must also be extended to handle MPEG <-> PCM switching
20:49:14amiconn...and the recording engine needs the same extension
20:50:31preglowtime to get optimising then ;)
20:52:01amiconnI agree with several others that the playback engines should be unified
20:52:35amiconn...but for this, the swcodec playback engine first has to get more stable, and some major annoyances must be solved
20:53:17amiconnThe good thing if we do this is that we could handle codecs on archos in a similar way as on iriver
20:55:14amiconnThis would save quite some space, no inclusion of the pcm codec in the core...
20:56:02preglowbut mpeg.c will need quite a rewrite for this
20:56:05 Nick Mark__ is now known as Mark__`afk (
20:56:05DBUGEnqueued KICK Mark__`afk
20:57:15amiconnThe goal is to remove mpeg.c, imho
20:57:56amiconnSome hw access functions will probably stay in the firmware layer, the rest should be application layer as on iriver
20:58:05amiconni.e. using playback.c
20:59:05preglowwell, what are the major annoyances with the playback engine?
20:59:57amiconnThe biggest one is the removal of nonexistent files from the playlist
21:00:13amiconnIt should silently skip them and leave the playlist alone
21:00:24amiconnmpeg.c does this
21:00:52preglowthat should be a quick fix? if i don't remember incorrectly, it was only recently added
21:01:02preglowand i agree that is the way it should be handled
21:01:18amiconnIt was added because the code didn't handle missing files well another way of handling these must be introduced
21:02:55amiconnThe archos playback code uses a "skip index" in its internal tracklist that indicates how many steps in the playlist to go forward in order to get from the current track to the next one
21:03:14amiconnThis index is usually 1, unless files are missing
21:03:23Slasheramiconn: that will be fixed after the feature freeze
21:03:27Slasher(the playlist thing)
21:04:03Slasheri don't want to introduce the skip index to playback.c, it's ugly solution..
21:04:46amiconnWhy do you think it's ugly?
21:04:59Slasherthere is many ways you could do it wrong
21:05:08Slasherin fact it would be quite hard to get it working right
21:06:13Slasheri think a better way would be that playlist engine can mark the incorrect entries temporarily and skip them on playlist_peek
21:06:52amiconnIt is working quite well on archos. We used to have quite some strange problems with nonexistent files before it was introduced
21:11:19preglowand btw
21:11:33preglowhow is it possible that a simple selectable cursor uses 1kb of code?
21:11:58amiconnDid you have a look at the implementation?
21:12:14amiconnIt's a fact that it increases binary size by >1KB
21:15:38amiconnSame is true for iriver btw, only that it's almost neglectible there
21:16:02amiconnThere is no such hard limit on iriver as is on archos
21:18:19amiconnThe thing is that the archos firmware is designed to load update firmwares from disk. The first method for loading rockbox was therefore relatively simple - it is built in a way that the archos firmware in the rom loads it as such an update firmware
21:19:07amiconnThere is a hard limit in this process - archos rom won't load a firmware >200 KB
21:19:38 Join webguest61 [0] (
21:19:43amiconnHowever, due to the mentioned design, not all archoses are flashable, as they don't need to be.
21:21:28amiconnArchos saved some pennies by not using 39VF020 flash roms in all units. Some have 37VF020 s which aren't in-circuit flashable.
21:22:06amiconnThat means we can't get around the 200KB limit by flashing as that wouldn't work for all units
21:22:53amiconnAll of the above isn't true for the Ondios, but that doesn't help us here. The worst cases are fmrecorder and recorderv2 regarding code size
21:23:28 Quit bagawk ("Leaving")
21:24:22webguest61Hi; a quick question as I'm a li'll bit confused: will it be possible to play some videos on iRiver iHP 1**?
21:25:23 Nick Mark__`afk is now known as Mark__ (
21:25:23DBUGEnqueued KICK Mark__
21:30:51preglowyes, know about the limit
21:31:09preglowwebguest61: might be
21:31:22preglowwebguest61: just need someone to come and implement it
21:31:53 Join ender` [0] (
21:32:15webguest61preglow: I see; good to know it can be done :)
21:32:21preglow'course it can be done
21:32:35preglowall depends on the video format you're speaking of, of course
21:33:07preglowproblem with h1x0 is that it hasn't got a display with a lot of colours
21:33:12preglowso we need to do some grayscale magic
21:33:18preglowwhich also takes some cpu
21:33:23preglowamiconn: you got any numbers on that, btw?
21:33:30amiconnpreglow: Do you think it would be good to add macsr to the context and change the iram distribution now?
21:33:33webguest61yeah, I was just thinking of any, even basic formats
21:33:44amiconnThen you could do some codec magic if you want...
21:33:49 Join TCK- [0] (
21:33:54preglowamiconn: well, of course i do, the macsr is almost a bugfix, and iram, well, i can work on codecs then :)
21:34:14amiconnI'll see what I can do ;)
21:34:47amiconnwebguest61: I'm almost sure the .rvf format as used on the archos will work on H1x0 as well one day
21:34:51preglowi'm not quite confident enough in myself to go around chaning the linker files to do it myself yet
21:36:11preglowhmm, btw, can i assume that only one instance of a codec will be used at a time?
21:36:19preglowso i can make libflac data static, for instance
21:36:21preglowit mallocs a lot
21:36:57preglowcodec swapping wont be touched, since it actually loads two instances of the actual plugin unless i've been mistaken
21:37:05amiconnI think you can assume that, as for the second instance used in talk.c the whole codec gets swapped
21:37:12preglowyes, indeed
21:37:28preglowi can't think why we'd need two actual data structure based instances
21:37:39preglowi actually already assume this in wavpack and a52
21:38:09preglowif i can toss the mallocs out the window, i'm quite confident i can get flac up to snuff
21:38:36preglowand that without actually touching too much code, something that would have a real impact on whats left of my sanity
21:41:29amiconnI should really look into asm optimising memcpy() ... sigh...
21:41:43preglowthat would certainly be good
21:42:02amiconnI think this will help a number of codecs to gain some speed
21:42:11preglowoh yes, and very probably the voice ui as well
21:42:27amiconnPerhaps I should switch to memmove() before doing the asm
21:42:34amiconnI want to do that anyway
21:44:07 Join Bazz [0] (
21:45:08Bazzi'm trying to build the rockbox simulator for iriver h120, configure goes fine, but when i make i get: <stdin>:1:11: config.h: No such file or directory, and then failuers in uibasic.c
21:45:10Bazzany ideas
21:45:20preglowamiconn: whatever floats your boat, would be great anyway
21:50:03 Quit TCK (Read error: 110 (Connection timed out))
21:55:59 Nick Mark__ is now known as Mark__`afk (
21:55:59DBUGEnqueued KICK Mark__`afk
21:59:32 Join Inverno [0] (n=desgaste@
22:00:07Invernohey any devs here?
22:00:09 Quit arkascha (Read error: 110 (Connection timed out))
22:00:11amiconnpreglow: MACSR now part of thread context, about to commit...
22:00:55Invernoi tried a build last night and rockbox seemed to be using the hardrive much more and literally eating chunks of battery life out of my iriver h-140
22:02:20preglowamiconn: excellent
22:02:33preglowInverno: when?
22:02:37***Saving seen data "./dancer.seen"
22:02:42 Join cYmen [0] (
22:02:43preglowInverno: when browsing? playing?
22:03:04amiconnThe good thing is that it only adds one instruction and one longword access to each of context store and context load
22:03:13InvernoIm not very sure now but it seemed like all the time
22:03:16Slasheramiconn: Hmm, you changed that codec index to start from 0.. there was a reason why it started from 1, i hope that doesn't break the playback code. But i might investigate tomorrow, now sleep ->
22:03:25Invernowait ill fire it up
22:03:41amiconnSlasher: What might fail?
22:03:57Slasheri am not sure, but now i really need to go. cu tomorrow
22:03:58amiconnOnly the code for unknown codec is zero now
22:05:34Invernodoesn't seem to be acting up now but it was when I was editing a playlist
22:05:43Invernowith playback at the same time
22:06:27Invernoit whent from two third full to half battery in some 20 minutes ;| (low volume using headsets and mp3/ogg playback no flacs)
22:06:54amiconnpreglow: Context change committed :)
22:07:53LearHm.. Normal playback seems fine (very quick look on a fresh build).
22:08:30LearBut I made a full rebuild, still no cursor option... Odd...
22:09:35amiconnHaha, funny effect:
22:09:58amiconnI am playing some flacs for a test, have voice menu enabled, and entered the menu and left it again
22:10:21LearLooks like linus forgot to roll back settings_menu.c... :)
22:10:32amiconnSeveral seconds later it said "general settings" (in german), with some further seconds spacing...
22:11:12preglowwell, flac doesn't leave much time to libmad
22:11:16preglowcpu time, that is
22:13:13amiconnYes, especially with flacs encoded with -8
22:13:28preglowoh well
22:13:35preglowi _should_ be able to work wonders for flac
22:13:37preglowwe'll see
22:13:51amiconnI'm just fiddling with the linker scripts....
22:14:21Learbetter than wavpack even?
22:14:41preglowLear: no, not really
22:14:45preglowbut who knows
22:15:33Learjust wondering, since that uses ~45% boost, iirc...
22:15:53preglowi've got files that doesn't boost at all
22:16:28LearI haven't tried much, granted, just a track or two...
22:16:50amiconnMaybe it's a matter of lossless vs. lossy mode
22:17:04amiconnI've encoded a whole album for testing as lossless
22:17:09amiconnDoesn't boost at all
22:17:42Learno, I did a lossless compress, don't remember the settings though...
22:17:55amiconnI used maximum compression
22:18:25preglowamiconn: it is
22:18:35preglowamiconn: lossy files require much more
22:19:35Learok, it was more like 30-35% (for a ~900 kbps file)...
22:20:25preglowLear: but i've seen varying cpu usage with lossless files as well, yes
22:26:04 Join ender1 [0] (
22:33:05 Quit ender` (Read error: 110 (Connection timed out))
22:39:26 Quit Bazz ("Client exiting")
22:40:44amiconnSomething is wrong with the wavpack codec
22:40:59amiconnThe voice ui doesn't work when playing wavpack
22:41:08amiconnIt works with all others I can test
22:42:23preglowlemme think
22:42:26preglowwhat's the glitch?
22:43:01amiconnIt simply doesn't talk, even when paused
22:43:21amiconnAfter stopping playback the voice ui speaks the last queued clip
22:43:55 Quit Inverno ("Client exiting")
22:48:14Learhm.. where does the codec swap happen anyway?
22:48:31Learduring pcmbuf_insert?
22:50:40LearYep, but only if the DSP is runnig. Which isn't the case for wavpack (except for resampling). Should be easy to fix.
22:50:41amiconnHmm, mow much stack is displayed as used directly after boot with cvs rockbox?
22:51:03Learwhich stack? which platform?
22:51:22amiconnErgh, h1x0, main stack
22:51:31Learone minute...
22:51:55Lear(that's before your context fix, btw).
22:52:12preglowi firmly believe the dsp should always be enabled
22:52:13amiconnGah, voice UI causes rockbox to hang after returning from rockboy
22:52:16preglowthat's a nonsense option
22:52:35preglowthe dsp layer knows far better than the codec itself whether dsp operations are needed
22:53:39Learbut that's not really the issue in the wavpack case...
22:54:26Learif dsp processing is off, codec_pcmbuf_insert_split_callback isn't called (the way it is done now), where the swap is done...
22:54:51Learbut currently the voice requires dsp to be always on, no matter what.
22:55:05 Join Domonoky [0] (
22:55:28preglowso i say let's remove the dsp enable flag
22:55:30preglowit's nonsense
22:55:43amiconnLear: thx
22:55:47Domonokyis it possible to compile a simulator build with Logf ?
22:55:56LearI've done that...
22:56:01amiconnIt's possible to reduce main stack on h1x0 to the same size as on archos
22:56:01BagderDomonoky: yes
22:56:02LearWas a little while ago though.
22:56:17amiconnUsage is 37% after returning from rockboy (the biggest stack eater)
22:56:18Domonokyhow do i configure this ?
22:56:22 Join zezayer_ [0] (n=zezayer@
22:56:34BagderDomonoky: like you do it for target, select devel build
22:56:43Bagderthen select sim and logf
22:56:55Learand dsp_process isn't quite made with that in mind, so I suspect it wouldn't be as efficient as it could be then (when always on, that is).
22:57:04 Nick zezayer_ is now known as zezayer (n=zezayer@
22:57:32Domonokyah.. thx
22:57:46preglowLear: so, as it is now, if a codec doesn't enable dsp, not even replaygain will be done?
22:57:59CtcpIgnored 5 channel CTCP requests in 1 hour and 42 minutes at the last flood
22:57:59*amiconn hands preglow an extra 16KB of iram for codecs :)
22:58:10*preglow thanks amiconn and wraps sleeves up
22:58:28preglownow for a better way to measure performance increases
22:58:39Learnope, I don't think so... But for all formats that supports replaygain should enable it...
22:58:56preglowi think it's stupid
22:59:02preglowit should be rewritten to always be enabled
22:59:08amiconnWhy not simply always enable it?
22:59:13Learand wavpack does that too.
22:59:16preglowif i enable eq, i don't want a friggin codec to override it for me
22:59:18 Join matsl [0] (
22:59:35Learwhat eq? :)
22:59:40amiconnEven formats without replaygain support need it when playing at samplerates != 48kHz
22:59:51amiconn...and stereo width would need it, and...
22:59:55Learand so they enable it...
22:59:59 Quit Aison ()
23:00:48Domonokyhm.. i get an error, when compiling with german as standart language.
23:00:49Learbut you're right, dsp.c could contain the on/off logic.
23:01:01Domonokya missing quote..
23:01:11 Quit cYmen ("zZz")
23:01:12Domonokyseem easy to fix :-)
23:02:15LearStill, don't quite understand why the voice requires the dsp to be on; it doesn't seem to use it anyway...
23:02:31preglowwelcome to evolutionary software design
23:02:36amiconnLear: Now only 484 bytes left for fmrecorder after re-adding the pointer menu option...
23:02:48preglowwell, it had to be done
23:02:58LearOnly uses the dsp-enabled pcmbuf_insert callback to do the swap...
23:03:04amiconnLear: The sample rate is well below 44.1kHz for voice
23:03:08Learamiconn, doesn't surprise me. :)
23:03:13 Nick ender1 is now known as ender` (
23:03:45Learwell, so the voice codec requires the dsp, doesn't mean the music codec should be forced to have it...
23:04:00amiconnI think the dsp does the mixing (?)
23:04:24Learnope, done in pcmbuf.c (as is the case for crossfading).
23:04:42amiconnAre there two dsp instances?
23:05:01Learsort of, two states (data structs)...
23:05:13LearI guess that
23:05:18Lear's the same.
23:05:39LearIt needs to keep config, resample state, dither state...
23:06:08preglowi wonder how easy it'd be to add a realtime measurement to the audio thread debug screen
23:08:19Learboost rate is almost that. You don't want 90%+ boost, because buffering gets really slow then...
23:08:33preglowyeah, but it takes a while to reach a number i can use
23:08:45Learoh well, time to hit the sack. bye.
23:09:01 Quit Lear ("Chatzilla [Firefox 1.0+/undefined]")
23:10:36 Nick Mark__`afk is now known as Mark__ (
23:10:36DBUGEnqueued KICK Mark__
23:10:36Mark__ _( Y )_
23:10:36Mark__ ()~*~()
23:10:36Mark__ (_)-(_)
23:10:45preglowMark__: please, shut it off
23:10:52Mode"#RockBox +o Bagder " by ChanServ (ChanServ@services.)
23:10:56preglowit doesn't add to the conversation
23:15:16solexDid anybody already try to use the new wps image features?
23:15:19amiconnDomonoky: Fixed, thanks
23:15:44Bagderwe don't _use_ features, we just add them ;-P
23:15:59solexthe wiki says "The image tag must be on its own line"
23:16:08CtcpIgnored 1 channel CTCP requests in 0 seconds at the last flood
23:16:08*Bagder never records, never edit playlists and uses default of most stuff etc
23:16:11*amiconn never used images in the wps yet
23:16:30solexbut the example below ("using images") suggests something else:
23:16:37amiconnBagder: There is one thing I never used the default for - the line selector ;/
23:16:40BagderI don't think that "own line" is true
23:17:42solexforget it, my girlfriend says I need to sleep now :)
23:17:43 Join webguest05 [0] (
23:18:12solexit's true!
23:18:33preglowshe's fooling you!
23:18:34solexit was a tough ikea day.
23:18:39preglowoh god
23:18:41preglowyou need to sleep!
23:18:47preglowi had one of those two days ago
23:18:49preglowslept like a child
23:18:53Bagderthose days require sleep
23:19:29solexthe ikea part wasn't that hard, but tidying up the room to build the table..
23:19:47*solex ligths his last cigarette for the day
23:20:48*preglow joins in
23:21:20solexcool, remote smoking
23:21:37preglowno, it's pretty local
23:21:40preglowsmells like it, at least
23:22:06*preglow strokes all his new iram
23:22:21 Nick solex is now known as solex_away (
23:22:32preglowif it feels good, it can't be bad!
23:23:42amiconnpreglow: Btw, when adapting the iram sizes, I overlooked the 2 #defines in playback.c first. Gave very funny effects with the voice UI enabled
23:24:01amiconn(codec iram was swapped incompletey)
23:27:30 Join wacky_ [0] (
23:27:40wacky_hey guys, anyone knows about the iAudio port development effort ?
23:27:46 Join Paul_The_Nerd [0] (
23:29:58preglowyou'd have to ask austriancoder
23:30:18Bagderhe's been very quiet recently
23:34:01wacky_I just bought an X5L, and was very pleased to see that Rockbox was being ported there too :)
23:34:22wacky_since my iRiver H120,s LCD was broken!!
23:34:45BagderI'm sure you can get good money for the rest of it
23:34:45 Quit TCK- (Read error: 110 (Connection timed out))
23:35:38wacky_well.. I'd rather give good money to have it repaired or replaced.. but I just can't !
23:39:33wacky_who would buy that ?
23:39:53Bagderpeople who fry their players with bad chargers
23:40:07Bagdersince they'll have a good lcd left
23:41:51wacky_hmm.. but all I want is the Rockbox
23:41:53wacky_running somewhere!
23:41:56 Nick Mark__ is now known as Mark__`afk (
23:41:56DBUGEnqueued KICK Mark__`afk
23:42:05wacky_your software is my desire !!
23:42:20Bagderthen join up with ac and the iaudio dudes and make it happen
23:42:25Bagderhe seems to need some help
23:42:41wacky_Imagine I could have a remote capturing software running on a tiny machine like an iAudio or an iRiver! wow..
23:42:53wacky_well I'm of no use for hardware stuff
23:43:01wacky_it seems the iAudio port needs a bootloader also
23:43:13wacky_and it seems that the only one who has ever worked with the bootloader stuff, is Linus
23:48:44 Join webguest83 [0] (
23:49:01webguest83i hope someone's here...
23:49:10webguest83is there a way to change the boot logo easily?
23:49:25Bagderit depends
23:49:29Bagderon what you call easy
23:49:44webguest83ill consider "compiling" an easy task
23:49:47Bagderyou need to replace a piece of source code
23:49:52webguest83but "programming" = not easy
23:50:03webguest83apps/icons.c is it?
23:50:09Bagdersounds right
23:50:30webguest83those hex numbers seem to be it...
23:50:37 Quit zezayer ("Chatzilla [SUSE 1.0.6-4.1/20050715]")
23:50:41Bagderbmp2rb is the tool
23:50:41webguest83and im stuck there :)
23:51:13webguest83size limit?
23:51:25wacky_why is it a so widespread habit to include pictures as hex stuff in source code ?
23:51:32wacky_wouldn't file loading be easier ?
23:51:41Bagderbefore ata is inited?
23:51:49dwihnoA great idea!
23:51:55 Quit DangerousDan (Read error: 104 (Connection reset by peer))
23:52:00webguest83ok how bout this
23:52:14webguest83is it safe to completely remove that picture loading lines?
23:52:16Bagderwebguest83: I don't think there's a limit other than screen size, but you may need to edit the code if you change size
23:52:19wacky_oh :) right
23:52:21webguest83so it boots up faster?
23:52:30webguest83well.. hmm
23:52:42BagderI don't think you'll see any difference by removing the pic
23:52:44Zagorwebguest05: sure. i bet you'll gain at least 30 ms :-)
23:52:45Bagderbut sure, try
23:53:18webguest83lol 30ms..
23:53:35webguest83oh, and when i said "size" i mean the dimensions
23:54:07Bagderme too
23:54:41webguest83oh ok :p
23:55:17 Join DangerousDan [0] (
23:55:34 Quit DangerousDan (Client Quit)
23:56:01 Nick Mark__`afk is now known as Mark__ (
23:56:01DBUGEnqueued KICK Mark__
23:56:07 Nick Mark__ is now known as Mark__`away (
23:56:07DBUGEnqueued KICK Mark__`away
23:58:42webguest83ah, the dimensions are in the code itself..
23:58:48webguest83ok i got everything, ty :)
23:58:55 Quit webguest83 ("CGI:IRC")

Previous day | Next day