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 2008-11-26

00:01:00 Join FOAD_ [0] (
00:03:56 Quit CaptainKewl (" ajax IRC Client")
00:04:06 Join zues [0] (n=7dee2c2d@gateway/web/cgi-irc/
00:04:26zuescan anyone help me i have a ipod 60gig
00:04:35 Quit tyfoo ("Carpe diem")
00:04:36 Quit kugel ("ChatZilla 0.9.84 [Firefox 3.0.4/2008111318]")
00:04:43zuesand it wont boo via usb anymore and i have rockbox installed i have tried resseting it etc
00:05:36bluebrotherwhat's "boot via usb"?
00:05:48zueswell i mean
00:05:54zuesit wont conect to my compuer anymore
00:05:59zuesit was working fine the other day
00:06:09bluebrotheris Rockbox running?
00:06:30bluebrotherand when you plug the usb cable what happens?
00:07:08bluebrotherwell, have you checked the cable / computer? Tried a different usb port?
00:07:28zuesyep tried all that
00:07:59funmanwhat if you plug the usb cable while the Apple OS is running?
00:08:02bluebrotherthen reset the Ipod, put it into emergency disk mode and connect it again.
00:08:11 Quit Seed ("cu, Andre")
00:08:25zuesapple os
00:08:28zuesim usin windows
00:08:42funmanthe apple *ipod* os of course, as opposed to rockbox
00:09:13zuesoh how do i boot into that
00:09:21zuesi tried all the reset tricks that i know of
00:09:38bluebrotherdid you actually manage to reset the Ipod?
00:09:53bluebrotherthere is only one way to hard reset the Ipod ...
00:10:00linuxstbzues: Turn the hold switch on immediately after resetting.
00:10:06zuesum how do i boot back into the orignal software
00:10:08bluebrotherand that is noticed / linked in the FAQ ...
00:10:13*bluebrother points to the manual
00:10:21zuesok hang on
00:10:37zuesso hold menu and center until the aple logo apears then flick hold to on?
00:11:39*bluebrother wonders why people are asking stuff that's written down in the FAQ ...
00:11:41zuesthen wat do i do next
00:11:51zueshey i couldnt find it sorri
00:12:22zueshow long does this take to do lolz its just got a apple logo there
00:12:58zuesdo i leave the switch on?
00:13:40linuxstbIt doesn't matter. But it sounds like your ipod's Apple firmware has a problem. It should start in about 20 seconds.
00:13:52 Quit ender` (" Man created gods. The opposite remains to be proved. -- Serge Gainsbourg")
00:14:01zuesnow do i conect it to pc
00:14:10zuesit has
00:14:12zuesnow wat
00:14:17*bluebrother wonders what lolz-ing is
00:14:19linuxstbit has what?
00:14:46zuesits in the orginal firmwere now wat is next
00:14:47BigBambibluebrother: I further wonder why your DAP not working is funny...
00:15:09bluebrotherme too.
00:15:42linuxstbzues: Yes, connect it to your PC.
00:16:23zuesnothing happens
00:16:44zuesjust turns off and yes the battery is fully charged
00:16:57zuesme bad
00:17:10zuesit is still sitting in the setup screen wat do i do
00:17:18gevaertszues: my guess is that you have a bad cable then
00:17:36zuesreplaced it
00:17:43zueswould useing the power addapter harm it?
00:18:01 Join LambdaCalculus37 [0] (i=41585845@gateway/web/ajax/
00:18:07*bluebrother wonders what the "setup screen" is
00:18:28gevaertsIf it doesn't connect properly while in the Apple firmware, you either have a bad cable, a bad USB port on the PC side, or a broken ipod
00:18:28 Quit FOAD (Read error: 110 (Connection timed out))
00:18:28 Nick FOAD_ is now known as FOAD (
00:18:48zuesi think ipod is broken lolz
00:18:58linuxstb"lolz" ?
00:19:00zueswould insurance cover it
00:19:06 Quit Nico_P (Remote closed the connection)
00:19:12zuesjust my way of saying lol
00:19:18linuxstb"lol" ?
00:19:31bluebrotherasl? asr?
00:19:32*LambdaCalculus37 gets out his Rosetta Stone
00:19:37bluebrotheror is it some kind of "nop"?
00:19:53LambdaCalculus37zues: What is this strange, savage language you speak?
00:20:04bluebrotherbut the assembler wouldn't like nopz ...
00:20:05n1sbluebrother: it's definitely a nop :)
00:20:07zuesjust new zealand slang
00:20:41bluebrotherNZ slang means to add a z to everythingz?
00:21:37 Quit zues ("CGI:IRC (EOF)")
00:21:38 Quit bertrik ("Leaving")
00:22:01*bluebrother is surprised not to read cyaz
00:23:18 Nick linuxstb is now known as linuxstbz (n=linuxstb@rockbox/developer/linuxstb)
00:33:32 Quit aarcane (Read error: 60 (Operation timed out))
00:33:51funmansaratoga: did you look already at the new aac decoder in ffmpeg ? (as opposed to faad). I'm sure it's a floating point implementation, but perhaps it can lead to better performance if converted to fixed point
00:34:01 Quit bluebrother ("sleep, now for real")
00:34:45 Nick linuxstbz is now known as linuxstb (n=linuxstb@rockbox/developer/linuxstb)
00:38:43 Quit Zagor ("Client exiting")
00:41:03 Nick shotofadds is now known as shotofaddz (n=rob@rockbox/developer/shotofadds)
00:41:08 Quit shotofaddz ("cyaz#")
00:46:11amiconnlinuxstb: I wanted to ask whether there is a reason for decoder.c:decode_chunk() and entropy.c:entropy_decode() to have return values in their current implementation
00:47:04amiconnentropy_decode() currently always returns 0, and decode_chunk() returns whatever the last call to entropy_decode() returned (i.e. also 0), and in some cases directly returns 0
00:49:04 Join aarcane [0] (
00:49:40amiconnI think that entropy_decode() doesn't need a return value.
00:53:11amiconndecode_chunk() doesn't need one either in the single core version. For dual core, I am thinking about using the return value to signal whether post-processing (filtering, predicting, decorrelation) is necessary. decode_chunk would only do entropy decoding, and there are cases where no post processing is necessary (those two places: /* We are pure silence, so we're done. */)
00:59:43 Quit aarcane (Client Quit)
00:59:57 Quit faemir (Remote closed the connection)
01:02:18linuxstbamiconn: I think I just started by assuming there would be an error returned, but ended up not detecting any.
01:04:44 Quit n1s ()
01:06:45 Join jeffdameth1 [0] (
01:10:05amiconnjhMikeS: I think there's a "p" missing in the function name.
01:10:37 Join saratoga [0] (n=9803c6dd@gateway/web/cgi-irc/
01:11:06saratogafunman: yes I did, its been discussed a few times
01:11:25saratogain general I am not a fan of trying to convert it to fixed point
01:11:43saratogahowever, when they get SBR support, I think we should probably take that and convert it to fixed if at all possible
01:12:08saratogabut continue to use faad elsewhere, since we have already removed the slowest parts of the LC codec
01:12:28saratogaand replaced them with Vorbis or WMA code
01:13:46 Quit jhulst (Remote closed the connection)
01:15:05funmanFS #9578 : Plugins for Sansa Clip
01:15:21saratogaalso, LC is pretty fast now, actually faster then MP3 on many ARM targets
01:15:35saratogathough HE is still probably 2-3 times slower then it shoudl be
01:15:51funmanDoesn't FAAD support SBR ?
01:16:08LambdaCalculus37funman: Let me try it out.
01:16:33funmanLambdaCalculus37: you have a development environment?
01:17:23 Join JdGordon [0] (n=jonno@rockbox/developer/JdGordon)
01:17:52LambdaCalculus37funman: I got it back. :)
01:18:13LambdaCalculus37By the way, your patch wasn't built against SVN and crapped out with patch -p0 < clip-plugins.diff
01:18:16saratogafunman: yes FAAD supports HE
01:18:25funmansaratoga: I read you have been 'devirginized' of patching your Fuze ? :)
01:18:28saratogaand so do we, though its too slow to be useable on most targets due to terrible implementation
01:18:48saratogayes I've installed the bootloader but done little else with it
01:20:33 Join Hillshum [0] (
01:20:47 Quit jeffdameth (Read error: 113 (No route to host))
01:21:02 Join jhulst [0] (n=jhulst@unaffiliated/jhulst)
01:21:23 Join kharo1 [0] (
01:22:33 Quit kharo (Read error: 110 (Connection timed out))
01:23:14 Quit frinkazoid ("Konversation terminated!")
01:25:19HillshumI've run, and added that one line to my .bashrc, but i get this
01:25:40 Quit funman ("leaving")
01:26:59saratogaHillshum: you may need to log out and back in
01:27:22HillshumI've rebooted rebooted
01:27:48saratogathen you probably didn't edit the script correctly
01:28:25saratogayou can check your path via echo $PATH
01:28:51LambdaCalculus37funman: Can you redo your patch at FS #9578 against SVN?
01:33:23 Quit MethoS- (Remote closed the connection)
01:33:31saratogaHillshum: do any of those folders contain arm-elf-gcc?
01:34:08Hillshum /bin/sh ?
01:35:30saratogai mean of the folders in your PATH, do any of them contain arm-elf-gcc? It would appear that they do not
01:35:44saratogahence the error message that its not in your PATH
01:39:43Hillshum arm-elf-gcc is in /usr/local/arm-elf/bin
01:43:17saratogaHillshum: then add that to your path
01:43:48saratogaprobably in place of "/usr/local/arm-elf-gcc/bin"
01:44:33 Join DerDome [0] (
01:46:26 Quit LambdaCalculus37 (" beer!")
01:47:12 Quit saratoga ("CGI:IRC (EOF)")
01:51:05 Quit Hillshum ("ChatZilla 0.9.83 [Firefox 3.0.3/2008092417]")
01:51:44***Saving seen data "./dancer.seen"
01:53:12 Quit lasser_ (Read error: 110 (Connection timed out))
01:54:34 Part toffe82
01:55:35 Join fdinel [0] (
01:56:35 Quit culture (Connection timed out)
02:01:31 Quit Thundercloud (Remote closed the connection)
02:05:49 Nick fxb is now known as fxb__ (
02:10:42 Quit jhulst (Read error: 110 (Connection timed out))
02:15:56 Join _emp [0] (
02:16:04 Part _emp ("Leaving")
02:19:59Unhelpfulamiconn: i think it was you who had expressed a wish to be able to use, for example, a smooth scaler with greylib for pictureflow?
02:21:53Unhelpfulhow about letting the image reader be passed a scaler to use? that way the smooth scaler needn't be in core on greyscale targets, but you can have a smooth greyscale scaler in pluginlib that can be connected to the chunked reader in core.
02:30:33 Quit kharo1 (Read error: 60 (Operation timed out))
02:31:03 Join kharo [0] (
02:33:56 Join perrikwp [0] (i=4aa794a0@gateway/web/ajax/
02:38:45 Part pixelma
02:39:06 Join pixelma2 [0] (n=marianne@rockbox/staff/pixelma)
02:44:19 Join jhulst [0] (n=jhulst@unaffiliated/jhulst)
02:49:20 Join aarcane [0] (
02:51:12 Join BHSPitMonkey [0] (n=stephen@unaffiliated/bhspitmonkey)
02:56:42 Join ajonat [0] (n=ajonat@
03:00:17 Join martian67 [0] (i=lol3izer@about/linux/regular/martian67)
03:11:43 Quit tarbo (Connection timed out)
03:12:52 Join CaptainKewl [0] (
03:34:05 Join beta__ [0] (
03:35:27 Quit DerDome (Nick collision from services.)
03:35:30 Join DerDome1 [0] (
03:35:40 Nick DerDome1 is now known as DerDome (
03:39:24 Quit CaptainKewl (Remote closed the connection)
03:39:44 Join CaptainKewl [0] (
03:42:14 Nick beta__ is now known as Beta2K (
03:47:45 Quit beta (Read error: 110 (Connection timed out))
03:51:48***Saving seen data "./dancer.seen"
03:58:43 Join johnnyboy [0] (n=johnnybo@
03:59:02 Quit DataGhost (Read error: 104 (Connection reset by peer))
03:59:16 Part johnnyboy
03:59:16 Join DataGhost [0] (i=dataghos@unaffiliated/dataghost)
04:00:38 Quit DataGhost (Nick collision from services.)
04:00:46 Join DataGhost [0] (i=dataghos@unaffiliated/dataghost)
04:05:24 Quit Acky ("Erection reset by queer")
04:13:24 Join blkhawk- [0] (
04:13:52 Join nuonguy [0] (
04:23:02 Join |AhIoRoS|_ [0] (n=ahioros@
04:24:18 Join beta [0] (
04:26:23 Quit Beta2K (Nick collision from services.)
04:26:29 Nick beta is now known as beta2k (
04:29:49 Quit fdinel ("Miranda IM! Smaller, Faster, Easier.")
04:29:52 Quit blkhawk (Read error: 113 (No route to host))
04:30:21 Nick blkhawk- is now known as blkhawk (
04:41:08 Join tarbo [0] (n=me@unaffiliated/tarbo)
04:41:24 Quit DataGhost (Read error: 104 (Connection reset by peer))
04:41:33 Join DataGhost [0] (i=dataghos@unaffiliated/dataghost)
04:44:47 Quit Bensawsome ("The awsome is gone :(")
04:45:23 Join Bensawsome [0] (n=Bensawso@unaffiliated/bensawsome)
04:46:49 Quit tarbo (Read error: 104 (Connection reset by peer))
04:47:04 Join tarbo [0] (n=me@unaffiliated/tarbo)
04:51:37 Join kkurbjun [0] (
04:52:53 Quit tarbo (Read error: 104 (Connection reset by peer))
04:53:06 Join tarbo [0] (n=me@unaffiliated/tarbo)
04:56:47 Join miepchen^schlaf_ [0] (
04:58:41 Quit tarbo (Read error: 104 (Connection reset by peer))
04:58:53 Join tarbo [0] (n=me@unaffiliated/tarbo)
05:04:39 Quit tarbo (Read error: 104 (Connection reset by peer))
05:04:52 Join tarbo [0] (n=me@unaffiliated/tarbo)
05:10:25 Quit tarbo (Read error: 54 (Connection reset by peer))
05:10:41 Join tarbo [0] (n=me@unaffiliated/tarbo)
05:12:11 Quit miepchen^schlaf (Read error: 110 (Connection timed out))
05:16:20 Quit tarbo (Read error: 104 (Connection reset by peer))
05:16:33 Join tarbo [0] (n=me@unaffiliated/tarbo)
05:22:06 Quit tarbo (Read error: 54 (Connection reset by peer))
05:22:19 Join tarbo [0] (n=me@unaffiliated/tarbo)
05:22:49 Join reacocard [0] (n=reacocar@WL-135.CINE.HMC.Edu)
05:24:57 Quit m0f0x ()
05:27:58 Quit tarbo (Read error: 54 (Connection reset by peer))
05:28:14 Join tarbo [0] (n=me@unaffiliated/tarbo)
05:33:45 Quit tarbo (Read error: 54 (Connection reset by peer))
05:33:58 Join tarbo [0] (n=me@unaffiliated/tarbo)
05:34:00 Quit Horschti ("Snak 5.3.3 Unregistered copy. Evaluation period is over. Program will now quit. Thanks for using Snak.")
05:39:38 Quit tarbo (Connection reset by peer)
05:39:51 Join tarbo [0] (n=me@unaffiliated/tarbo)
05:41:46 Quit daurnimator (Read error: 113 (No route to host))
05:45:22 Quit tarbo (Read error: 104 (Connection reset by peer))
05:45:39 Join tarbo [0] (n=me@unaffiliated/tarbo)
05:49:29 Quit JdGordon (Remote closed the connection)
05:51:16 Quit tarbo (Read error: 104 (Connection reset by peer))
05:51:18 Join JdGordon [0] (
05:51:28 Join tarbo [0] (n=me@unaffiliated/tarbo)
05:51:50***Saving seen data "./dancer.seen"
05:57:05 Quit tarbo (Read error: 104 (Connection reset by peer))
05:57:17 Join tarbo [0] (n=me@unaffiliated/tarbo)
06:02:49 Quit tarbo (Connection reset by peer)
06:03:03 Join tarbo [0] (n=me@unaffiliated/tarbo)
06:08:41 Quit tarbo (Read error: 104 (Connection reset by peer))
06:08:54 Join tarbo [0] (n=me@unaffiliated/tarbo)
06:14:27 Join Genbu [0] (
06:14:29 Quit tarbo (Read error: 104 (Connection reset by peer))
06:14:45 Join tarbo [0] (n=me@unaffiliated/tarbo)
06:20:20 Quit tarbo (Read error: 104 (Connection reset by peer))
06:20:32 Join tarbo [0] (n=me@unaffiliated/tarbo)
06:25:19 Quit andrew__1 (Read error: 104 (Connection reset by peer))
06:25:31 Join andrew__ [0] (
06:26:12 Quit tarbo (Read error: 104 (Connection reset by peer))
06:26:26 Join tarbo [0] (n=me@unaffiliated/tarbo)
06:31:59 Quit tarbo (Read error: 104 (Connection reset by peer))
06:32:14 Join tarbo [0] (n=me@unaffiliated/tarbo)
06:33:40 Join planetbeing [0] (
06:37:49 Quit tarbo (Read error: 104 (Connection reset by peer))
06:38:03 Join tarbo [0] (n=me@unaffiliated/tarbo)
06:51:51 Quit ajonat ()
07:07:12 Quit tarbo (Connection timed out)
07:13:17 Join tarbo [0] (n=me@unaffiliated/tarbo)
07:16:08 Quit aarcane ("Leaving")
07:18:27 Join nintendork32 [0] (n=goodbadu@unaffiliated/nintendork32)
07:20:20 Quit Genbu ("Leaving")
07:20:21 Join einhirn [0] (
07:24:17 Quit HBK (Read error: 104 (Connection reset by peer))
07:24:25 Join HBK [0] (
07:24:31 Part nintendork32 ("Waoh nelly not so harsh!")
07:33:07 Join Acksaw [0] (
07:41:30 Join pondlife [50] (n=Steve@rockbox/developer/pondlife)
07:42:08 Quit einhirn (Read error: 131 (Connection reset by peer))
07:44:17 Join Darksair [0] (n=user@
07:44:23 Quit skipper (Read error: 110 (Connection timed out))
07:51:55***Saving seen data "./dancer.seen"
08:09:31 Quit tarbo (Connection timed out)
08:17:23 Join Rob2223 [0] (
08:19:16 Join lasser [0] (
08:24:04 Quit miepchen^schlaf_ ()
08:33:35 Quit jfc (Read error: 54 (Connection reset by peer))
08:33:44 Join jfc [0] (
08:35:01 Quit japc (
08:35:18 Quit Rob2222 (Read error: 110 (Connection timed out))
08:35:34NJoinjapc [0] (
08:37:01pondlifePlease can I ask someone with a clue to document the SVN file properties? A wiki search for "SVN properties" finds no useful result. I'd like to know the SVN command-line in full, please...
08:42:22 Quit BigBambi (Read error: 113 (No route to host))
08:49:00 Join bertrik [0] (
08:49:45 Join ender` [0] (
08:54:04 Join Bagderr [241] (n=daniel@rockbox/developer/bagder)
08:54:30 Nick Bagderr is now known as B4gder (n=daniel@rockbox/developer/bagder)
08:56:34 Quit AndyIL ()
08:57:22 Join midkay_ [0] (
08:59:28 Quit bertrik ("Leaving")
08:59:53 Quit GodEater (" ajax IRC Client")
09:00:00 Join GodEater [0] (i=c2cbc962@gateway/web/ajax/
09:00:00*amiconn points pondlife to , especially
09:02:07 Quit XavierGr ()
09:02:11pondlifeI know we set eol-style and keywords, but to what values?
09:03:49 Quit jhulst (Read error: 110 (Connection timed out))
09:04:23 Join AndyI [0] (i=AndyI@
09:05:07pondlife"svn propget eol-style apps/main.c" gives me no clues.
09:06:07*pondlife wakes up..
09:06:18pondlifeAh - svn:keywords and svn:eol-style
09:06:30pondlifeOK - got it
09:07:44 Quit AndyI (Client Quit)
09:08:52*amiconn thinks that svn:eol-style is basically useless for most source files.
09:09:04amiconnAfaiu, 'native' is the default anyway.
09:11:48 Quit pixelma2 ("-")
09:11:50 Join pixelma [50] (i=pixelma@rockbox/staff/pixelma)
09:12:04 Quit midkay (Read error: 110 (Connection timed out))
09:12:17 Join tarbo [0] (n=me@unaffiliated/tarbo)
09:12:35 Join kugel [0] (n=chatzill@unaffiliated/kugel)
09:14:08 Join AndyI [0] (i=AndyI@
09:15:55pondlifekugel: About to commit your patch
09:16:08pondlifeJust doing some final tests
09:16:09kugelpondlife: cool
09:16:28pondlifeHope it's not going to kill binsize ;)
09:16:51kugelmy messurements showd <1K, most of it is the setting
09:17:07kugelthe settings rather
09:18:52pondlifekugel: Which targets gain a backlight fade? H300, X5, D2, E200,...
09:20:28pondlifeShould credit also go to Ivan Zupan?
09:22:32 Quit JdGordon (Read error: 104 (Connection reset by peer))
09:25:37 Join nplus [0] (n=nplus@141.25.Globcom.Net)
09:25:54 Join JdGordon [0] (n=jonno@rockbox/developer/JdGordon)
09:27:27 Quit kugel (Remote closed the connection)
09:32:51 Join omgun [0] (n=79b45a2d@gateway/web/cgi-irc/
09:35:41pondlifeOops, a little red
09:35:44JdGordonhmm... wont that commit break lang/voice files?
09:36:24pondlifeAh, did it insert?
09:36:40JdGordonwell, the features.txt changed, so its possible it did
09:37:03JdGordonnot really a big deal... just some wrong text/voicing in old builds
09:37:07JdGordonor new builds
09:37:35pondlifeVoice/lang files need to match the build anyway, don't they?
09:37:52JdGordonnot always
09:38:01pondlifeNo, but in genera
09:38:20JdGordonadded strings wont be spoken/xlated, but if feautres change in the middle then yeah things get funny
09:39:32pondlifeI suppose it's inevitable, if you gain an existing feature?
09:41:30 Join Thundercloud [0] (
09:41:52 Join kugel [0] (n=chatzill@unaffiliated/kugel)
09:42:41JdGordonh100 doesnt have backlight brightness control?
09:42:52pondlifeI don't know
09:43:08JdGordonhehe... since when is the e100 port working? :D
09:43:31pondlifeWhy do some targets use a byte for backlight_brightness..?
09:43:51kugelpondlife: eh, it's e200. and c200 also got fading
09:43:53B4gderthat e100 is gonna get some people's hopes up ;-)
09:44:49pondlifekugel: Typo on the 100/200 front
09:45:02pondlifekugel: Any ideas on how best to fix the build?
09:45:14kugelpondlife: yep, wait a minute
09:45:29pondlifeI guess the backlight_brightness could be a byte everywhere?
09:46:18pondlifeOr an int everywhere - like the setting.
09:47:54kugelI don't understand this line in the gigabeat f/x backlight driver "brightness = MAX(1, MIN(12, brightness));"
09:48:21kugelbacklight value boundaries are handled in backlight.c anyway
09:48:25JdGordonshouldnt that just be MAX(12, brightness)?
09:48:48 Quit tvelocity (Connection timed out)
09:48:50 Quit MegafEee ("KVIrc 4.0.0 Insomnia")
09:49:08pondlifekugel: I need to vanish shortly, about 5 mins or so :/
09:49:18 Join tvelocity [0] (
09:49:50JdGordonhmm... I guss that line is to make sure the backlight isnt disabled completly?
09:50:39kugelyea, but backlight.c handles brightness values out of range. There can't be an invalid brightness value (unless _backlight_set_brightness is called directly)
09:51:35pondlifekugel: Is there a missing #ifdef? I didn't think the red targets would be affected at all...
09:51:59***Saving seen data "./dancer.seen"
09:52:41 Join n1s [0] (n=nils@rockbox/developer/n1s)
09:53:00kugelpondlife: no, backlight_brightness is supposed to be handled in backlight.c now. So remove it for those targets
09:55:17 Join Nibbler [0] (
09:55:24kugelbacklight_brightness is common to all targets which have adjustable brightness, so it should be reimplemented for each target
09:58:27pondlifeOK - here goes...
10:00:51kugelpondlife: I prepare a patch
10:01:05pondlifeI've done it - 4 lines removed, I reckon
10:01:25pondlifeIs that what yours does?
10:01:31kugelnot exactly
10:02:32kugel is what I did
10:03:33kugelyour fix should do it as well, but it's not so proper I think
10:03:34pondlifeHeh is soooo sloooow...
10:03:50pondlifeMy fix worked for Gigabeat, not for Fuze :/
10:05:11 Quit omgun ("CGI:IRC (Ping timeout)")
10:05:23kugelmy neither :(
10:05:34pondlifeI have to go now
10:05:57pondlifeIf it's ok with the room, I won't revert - will fix when I get back
10:06:49 Part pondlife
10:09:48 Join beast_ [0] (
10:10:13rasherShould Jack Suter and Janis Meybohm be put in CREDITS for FS #6800 / r19221?
10:12:22n1sB4gder: have you seen FS #9579 ? seems like a simple fix
10:12:45 Quit beast (Read error: 110 (Connection timed out))
10:12:49B4gderlooks fine to me too
10:14:20kugelI don't get this error
10:15:55kugele200 backlight driver links fine (i.e. no undefined reference to backlight_brightness), fuze/e200v2 doesn't. It's almost the same driver though
10:20:12n1skugel the error is that backlight.h now says that there's a symbol called backlight_brightness that's an int while backlight-e200v2-fuze.c has its own backlight_brightness that's a short, so they conflict
10:21:48kugeln1s: I removed the declaration in backlight-e200v2-fuze.c
10:22:28kugeln1s: I now get undefined reference to backlight_brightness, even though backlight.h is included
10:23:05n1sdoes e200v2 and fuze define HAVE_BACKLIGHT_BRIGHTNESS ?
10:24:51kugelbacklight.c is compiled too
10:25:51JdGordonamiconn: (you might not be the right person to ask but....) apps/talk.c:523 is doign a check for #if (CONFIG_STORAGE & STORAGE_MMC)... shouldnt that be HWCODEC or something?
10:26:13 Quit ze (Read error: 110 (Connection timed out))
10:27:30n1skugel: the declaration insinde backlight.c seems to be inside #if defined(HAVE_BACKLIGHT) && defined(BACKLIGHT_FULL_INIT) too, which is only true for !BOOTLOADER
10:27:36 Quit kugel (Read error: 104 (Connection reset by peer))
10:27:39 Quit Nibbler (Read error: 113 (No route to host))
10:28:34 Join kugel [0] (n=chatzill@unaffiliated/kugel)
10:29:18 Quit Thundercloud (Remote closed the connection)
10:29:37kugeln1s: it works for e200v1 bootloader
10:30:59 Quit BHSPitMonkey ("Ex-Chat")
10:31:00n1syes, it works for a lot of bootloaders but the e200v2 bootloader is probably building too much of the backlight stuff
10:31:02 Join Nibbler [0] (
10:31:23kugelwhere's this full init defined?
10:31:39n1si have to grep for that too...
10:32:20kugelI did already
10:32:35kugelit's defined in backlight.c, and it's insidfe !BOOTLOADER
10:32:52kugelso, its neither defined for e200v1 bootloader
10:33:45n1sdoes the e200v1 bootloader even build backlight-c200_e200.c ?
10:33:58kugellet me look
10:39:21kugelhm, no error if I move backlight_brightness out of BACKLIGHT_FULL_INIT
10:39:52n1sof course, but should the bootloaders even mess with brightness?
10:41:30kugeln1s: well, they should at least be able to set the default brightness
10:41:37kugelwith _backlight_on
10:43:06 Join omgun [0] (n=79b45a2d@gateway/web/cgi-irc/
10:43:48n1smeh, anything touching the bootloaders is a mess, lots of custom stuff and many things are special for each target...
10:44:35kugeln1s: moving it out of FULL_INIT seems appropriate. Before the fading, backlight_brightness was known to the bootloader too
10:44:50n1skugel go for that then :)
10:45:26kugelbut the fading shouldn't be
10:47:21kugeland the fading was in the bootloader, that's why we didn't get the error in the e200v1 bootloader
10:48:07 Join LinusN [0] (n=linus@rockbox/developer/LinusN)
10:48:17n1syes, it's built with USE_BACKLIGHT_SW_FADING defined so the backlight-c200_e200.c never saw backlight_brightness
10:52:05 Quit nanok (Read error: 148 (No route to host))
10:57:59JdGordonis the sim not finding the .rockbox forlder for anyone else?
10:58:17JdGordonoh, the root was finally moved?
10:58:22*JdGordon missed the commit :/
10:59:50kugelJdGordon: yea, I was quite surprised
11:00:10kugelyou missed one of the most discussed commits :/
11:00:35kugel <−− fixes reds
11:00:40kugelso if someone wants commit
11:01:05JdGordonoh, I saw the disucssion, just not the commit
11:01:15B4gderwell, the name lasted many years at least!
11:01:30*B4gder thinks of other places to sneak in oldish names instead
11:02:14 Quit kharo ("Leaving.")
11:02:17kugelmake the config file archos.cfg!
11:02:27 Quit omgun ("CGI:IRC (Ping timeout)")
11:02:47kugelor, rename cabbiev2 to archos
11:03:07kugelwhich would be misleading so nicely :)
11:03:40 Join Wonko_ [0] (
11:06:37 Join ze [0] (n=ze@
11:09:10 Join omgun [0] (n=79b45a2d@gateway/web/cgi-irc/
11:10:30JdGordonwould it be possible to have make voice work in a regular build dir? (without doign (v)oice in the advanced bit of confgiure)
11:11:23B4gderthe voice building needs some further info that (v)oice asks for
11:11:33B4gderand checks for
11:12:17JdGordonright, could that get change to a dependancy for make voice instead?
11:14:13 Join Nico_P [50] (n=nicolas@rockbox/developer/NicoP)
11:14:14 Join pondlife [50] (n=Steve@rockbox/developer/pondlife)
11:14:50*pondlife returns and reads the logs...
11:15:11 Quit Nibbler (Read error: 113 (No route to host))
11:17:10kugelpondlife:, it does that properly
11:17:12JdGordonis there any way to sleep() untill a requested voiceing is finished?
11:17:24pondlifekugel: Yes, just got to that bit - will commit
11:17:34kugelgtg, I'm back in 20min
11:17:41 Quit kugel (Remote closed the connection)
11:17:57 Quit tvelocity (Read error: 110 (Connection timed out))
11:18:31 Join tvelocity [0] (
11:21:34*pondlife hopes that didn't cause any new reds...
11:23:23 Join Nibbler [0] (
11:24:03 Join petur [50] (n=petur@rockbox/developer/petur)
11:29:38 Join fyre^OS [0] (
11:31:09 Nick Darksair is now known as Awaysair (n=user@
11:32:31rasherJdGordon: that would mean moving the configuration bits into the makefile. Sounds nasty
11:33:22 Join kugel [0] (n=chatzill@unaffiliated/kugel)
11:33:36kugelehh, yellow :S
11:33:39JdGordoncant make voice have a dependancy on a file voice.make (or something) and if that isnt there then a script is run to generate it?
11:33:55pondlifeyellow is better than red ;)
11:34:08kugelI didn't see that warning :/ because compiling on ubtunu 8.10 creates massive warnings anyway
11:34:43kugelI propose (void)brightness;
11:34:57rasherJdGordon: What would this accomplish?
11:35:31JdGordonnot having to go create a seperate folder to do make voice
11:40:22n1skugel: maybe you should look into why you are getting massive amounts of warnings, as i use latest ubuntu yoo and i don't get extra warnings
11:41:13kugeln1s: massive warnings only when building tools
11:41:44B4gderwhat warnings are they?
11:41:52n1sah, yes i get a few warnings for some tools too
11:42:09n1skugel: why are you rebuilding the tools though?
11:42:53kugeln1s: that's needed when I switch from building an arm build to m68k build
11:43:42kugelnot? Well, I got some erros when I built gigabeat after h300 w/o rebuilding tools
11:43:52kugelmake veryclean helped
11:44:19n1shmm, make veryclean doesn't seem to work here
11:45:26kugeln1s: are you sure? make veryclean doesn't output more than make clean
11:45:57n1skugel: yes, all the binaries in /tools are still there
11:46:02rasherJdGordon: should of course be possible
11:46:05 Quit fyrestorm (Read error: 110 (Connection timed out))
11:47:09n1sB4gder: here's the output of a 'make' in the tools dir mostly a bunch of "ignores return value of fwrite"
11:47:11B4gderbut a bit against how we've "always done it"
11:47:36B4gderI mean, we could allow "make bootloader" too without a recondigure and all others too
11:48:16B4gderwow, they added that on fwrite!
11:48:34rasherB4gder: I'm not sure what benefits the current one gives us
11:48:42JdGordonwell, for voices it would be nice to just do make voice in any build dir without neding to reconfigure... If I'm in a e200-sim but I havnt made a voice for it yet its annoying to create a new dir to do the build
11:48:43rasherB4gder: the current method, that is
11:49:23B4gdernah, they should be pretty much the same
11:49:41B4gderchanging the ways will of course kill a lot of automatic scripts
11:49:58JdGordonwe dont have to break it in configure to do it...
11:50:07kugelis tools built with standard gcc or with the target *-elf-gcc?
11:50:16rasherI don't think we should have configuration code in two places
11:50:26kugelI mean when you did configure and haven't built tools before manually
11:50:28rasherThat's bound to break sooner rather than later
11:50:28JdGordonI only suggest it because no code is compiled for the voice so there is no need for a seperate build dir like bootloader needs
11:50:30B4gderkugel: tools should be native tools only built with native gcc
11:50:50 Quit Zambezi (Remote closed the connection)
11:51:02rasherJdGordon: Go for it!
11:51:27rasherBasically just rip out the configuration function of tools/configure and stick it in a seperate script that outputs something make can include
11:51:28*JdGordon has absolutly no idea how to do it :p
11:51:44pondlifekugel: Would you be ok if I just (void) the parameters to fix the yellow?
11:52:04***Saving seen data "./dancer.seen"
11:52:12pondlifeOr is this warning more serious?
11:52:30B4gdern1s / kugel: the proper tools fix would of course be to actually check the fwrite() return codes to detect failures properly
11:52:31 Quit kugel (Nick collision from services.)
11:52:34 Join kugel_ [0] (n=chatzill@
11:52:41 Nick kugel_ is now known as kugel (n=chatzill@
11:53:11pondlifekugel: Ir does seem odd that _backlight_set_brightness() does nothing with the brightness!
11:53:42*gevaerts tries his Zagor-summoning skills
11:53:48pondlife(Looking at backlight-meg-fx.c)
11:54:52 Nick Awaysair is now known as Darksair (n=user@
11:57:37 Quit petur ("off to lunch and then to customer")
11:57:50kugelpondlife: yes, as it didn't before
11:58:35pondlifeHmm, so the caller actually sets backlight_brightness before calling _backlight_set_brightness() ??
11:59:26kugelpondlife: backlight_set_brightness (without underscore, in backlight.c) sets backlight_brightness, and calls _backlight_set_brightness for target specific stuff
11:59:40pondlifeOK, thanks
11:59:59kugelthe actual brightness change for gigabeat f/x happens in the control_service funtion, just like before
12:00:56pondlifeOK, will add comments too
12:02:17kugeldoes anyone have a gigabeat f/x here?
12:02:32kugelI wonder if the change from char to int has an impact
12:03:38kugelit should, given that backlight_brightness is only the array element of log_brightness[]
12:04:34n1sB4gder: yes, of course but who is motivated to? :)
12:05:15pondlifekugel: Hmm, hope backlight_brightness never == 0.. !
12:05:24pondlifesc606regAval=log_brightness[backlight_brightness - 1]
12:06:00 Quit nuonguy ("This computer has gone to sleep")
12:06:12kugelpondlife: as I said before, backlight_set_brightness() handles brightness value mins and maxes
12:06:36pondlifeGood - I saw maxes were handled, but didn't know about mins
12:06:41 Quit DerDome (Read error: 110 (Connection timed out))
12:07:07kugelpondlife: line 943 in backlight.c
12:07:13B4gdern1s: kugel? B)
12:07:22 Join skipper [0] (n=skipper@
12:07:50kugelB4gder: did I miss something? :/
12:07:58B4gder"n1s / kugel: the proper tools fix would of course be to actually check the fwrite() return codes to detect failures properly"
12:08:10pondlifeMight it give a slightly smaller binary if log_brightness was [13] and the -1s were removed though? Will check...
12:08:36kugelpondlife: no
12:08:43kugelpondlife: you don't need to check
12:08:56pondlifeI've committed a (void) fix
12:08:56kugelMIN_BRIGHTNESS_SETTING is always >= 0
12:09:17pondlife> 0, you mean
12:09:28kugelwhat could be done is using a dummy for [0] and the first usable value would be [1]
12:09:46pondlifeThat's what I meant
12:09:47kugelpondlife: No, >= 1 :P
12:09:48n1shmm, make veryclean only removes some tools, the ones actually used when building, is this correct or should it remove them all?
12:10:14kugelpondlife: oh sorry, I got it wrong then
12:11:37n1spondlife: c++ style comment! :P
12:11:56kugelhmm, [0] is 0, I wonder if that makes even sense since 0 is backlight off
12:12:02pondlifeThat's my day job
12:12:35kugel(and backlight_set_brightness should be able to turn it off)
12:12:46kugelshould *not*
12:13:12n1sMIN_BRIGHTNESS_SETTING should be higher than "off"
12:13:32pondlifeAdding a dummy log_brightness[0] saves 8 bytes, apparently
12:14:22kugelso if someone sets the lowest brightness level on a gigabeat f/x, it should effectively turn off the backlight. Should be checked
12:15:35n1skugel: the lowes settable brightness should not be "off" it should still be visible
12:15:49kugeln1s: yes, that's what I mean
12:16:18kugelbut apparently that's not the case on f/x
12:18:23 Quit AndyI ()
12:18:37n1sthen that should be fixed :)
12:19:19kugelgoing by the code it's seems to off and should be fixed. But I haven't a f/x to actually check if it's really the case
12:20:09kugelpondlife: you possibly don't even need to add the dummy, [0] shouldn't be used by brightness changing (IIUC)
12:20:46pondlifeSo the max value is 11?
12:21:57pondlifei.e. removal of the -1 from the two accesses alone doesn't sound correct to me.
12:22:02 Join Zambezi [0] (
12:22:49kugelhm yea, you're right
12:23:37kugel12 is MAX_BRIGHTNESS, so log_ needs to be [13]
12:23:52 Join moos [0] (
12:24:39kugelbut then there's a value missing. The table has only 11 values given that the 0 shouldn't be used
12:25:08kugelor MAX_BRIGHTNESS is actually 1 too high
12:25:16 Quit n1s (
12:25:16 Quit JdGordon (
12:25:16 Quit GodEater (
12:25:16 Quit lasser (
12:25:16 Quit krazykit (
12:25:16 Quit tim__b (
12:25:16 Quit Bjoern-Erik (
12:25:16 Quit goffa_ (
12:25:23NJoinJdGordon [0] (n=jonno@rockbox/developer/JdGordon)
12:25:23NJoinGodEater [0] (i=c2cbc962@gateway/web/ajax/
12:25:23NJoinlasser [0] (
12:25:23NJoinkrazykit [0] (
12:25:23NJoinBjoern-Erik [0] (
12:25:23NJoingoffa_ [0] (n=goffa@
12:25:43kugelI don't know, we should ask a f/x user if the lowest brightness setting is really "no backlight"
12:26:36NJointim__b [0] (
12:26:55 Join Self-Perfection [0] (
12:27:27NJoinn1s [0] (n=nils@rockbox/developer/n1s)
12:31:59 Join AndyI [0] (i=AndyI@
12:32:23kugelpondlife: thanks for committing :) I hope it'll make users happy
12:32:35pondlifeI like it.
12:34:03 Join ap0 [0] (
12:34:49Lloreankugel: The lowest setting still has the backlight on, on the Gigabeat F
12:35:31 Quit ap0 (Client Quit)
12:35:33amiconnJdGordon: No, the #if is exactly like it should be.
12:35:52kugelLlorean: oh ok, didn't look like that going by the code
12:36:17LloreanI've just tested, the light's definitely on at "1"
12:36:32JdGordonamiconn: care to explain?
12:37:36amiconnThe MMC access is slow, due to how it's hooked to the cpu. Loading the whole voice file at once would take ~5 seconds, so the individual clips are loaded on demand (which is in turn impossible on a hdd target)
12:38:15JdGordonis the internal storage MMC also?
12:42:18kugelpixelma: ping
12:44:01JdGordonrasher: I've got a script to dump the needed export lines to a text file.. any ideas how to mangle the root.make to run the script if its needed?
12:44:26kugelpixelma: are the manuals already build after the backlight commit?
12:44:44JdGordonmanuals are build once per day iirc, so prob not
12:44:51 Quit Llorean ("Leaving.")
12:45:24amiconnJdGordon: 'make voice' in a normal build dir doesn't make much sense, if any
12:45:34kugelpixelma: I remember you said (partly) features.txt decides what goes into the manual. I added some targets to the backlight_fade feature, but they have in fact different options in the setting than the existing backlight_fade targets
12:45:37pixelmakugel: the manuals for download are only built once a day (like the daily builds, at around 6AM our time, IIRC)
12:46:11JdGordonamiconn: why doesnt it make sense? imo it makes more sense than needing to make a new dir for it
12:46:52kugelI assume the manual will describe wrongly the settings possible for pwm fading targets ?
12:47:37 Quit Nibbler (Read error: 113 (No route to host))
12:47:41amiconnYou need to specify the engine, language, and other parameters. It's also not uncommon to have more than one voice configuration per target
12:47:46*amiconn has up to 4
12:48:25JdGordonso if your not using more than 1 engine its not a problem, and its really handy for quick voice builds
12:48:58*linuxstb wonders why voice is amongst the "advanced" options
12:49:54pixelmakugel: if the backlight_fading feature is already used as an option, then it will probably be used in the entire section about it (if there's no other destinction). It could even be that some manuals are broken when something is missing, should be tried...
12:50:50*linuxstb agrees with amiconn that it doesn't make sense to make voice in a normal build dir
12:51:58 Join _Wonko [0] (
12:51:58*JdGordon thinks your both nuts :D
12:52:40JdGordonI already have over a dozen build dirs, how does it make any sense forcing me to make more if I want to build a test voice file?
12:53:20 Join funman [0] (
12:53:43*pixelma has sim build, normal build, voice and manual build dirs
12:54:01n1slinuxstb: because of special casing for sims in a lot of places (often unnecessarily) some sims need different voice files than the targets so you have to make a sim-voice, which is only possible in the advanced menu
12:54:24linuxstbJdGordon: You should be able to create a "voice" subdirectory in each of your build directories, and do it there.
12:54:43linuxstbn1s: Hmm, fixing that should go onto Mr Someone's todo list...
12:55:00JdGordonyou still need to go through configure and find the target...
12:55:01*linuxstb tests and sees that a voice subdir doesn't work...
12:55:02n1sI tried fixing it and they were the same at one point but people keep adding stuff in #ifndef SIMULATOR everywhere :/
12:55:26linuxstbn1s: You should have shouted at them...
12:55:38JdGordonanyway, this isnt going to happen unless I get help for MAkefile so dont worry....
12:55:39linuxstbI've also done a little work removing #ifdef SIMULATOR
12:56:55 Quit japc (Read error: 110 (Connection timed out))
12:57:30 Join Llorean [0] (
12:57:58pixelmakugel: there is one sentence mentioning the time settings which is there for all backlight_fade targets currently. This would need to be opted, probably no breakage
12:58:32kugelthe targets I added just have on and off for both settings
12:59:05pixelmayes, I understood that
12:59:30kugeland how would that work (I'm not too familiar to latex)
13:00:26 Quit Wonko_ (Read error: 110 (Connection timed out))
13:00:26 Nick _Wonko is now known as Wonko_ (
13:00:26amiconnn1s: Sims should never need different voice files than the corresponding target
13:00:38n1samiconn: i agree
13:00:48 Join pondlife1 [0] (
13:02:06kugelcan someone hint me which packages I need to build manuals?
13:02:28pixelmakugel: how are the settings excluded for those targets in the lang files? Is there a different features used?
13:02:41kugelpixelma: they use the same string
13:03:16kugelboth only add the string for the title of that setting. The options are different due to the *_SETTING macros
13:04:11pondlife1kugel: I thought the reason you had 2 settings was to match the existing PWM targets? What's the difference?
13:04:41kugelthe 1 use TABLE_SETTING, the other OFFON setting
13:05:06kugelthey use the same string, and the function to actually set the value is also the same
13:05:27pondlife1So, from the user's point of view, they look the same?
13:05:40kugelas long as they don't enter the setting yes :)
13:05:49 Join robin0800 [0] (
13:05:54pixelmathe PWM fading targets have additional time settings (how long the fade takes9
13:05:59pondlife1Ah, ok.
13:06:03 Quit Self-Perfection (Remote closed the connection)
13:06:07kugelif they enter, they'll find numeral (and off) for pwm, and only off+on for sw
13:06:23pondlife1I'd prefer it if sw could also use the numerals/off...
13:06:31pondlife1Any chance of that?
13:06:44pondlife1Might be easier than fixing the manuals ;)
13:07:48kugelit could be done. But my tests show that there's only a very small interval to chose from. On my e200 it gets un-smooth if the interval is only slightly raised
13:08:22kugeland on h300 (I heard) the backlight controller is too slow to run a shorter intervall
13:09:02 Join Nibbler [0] (
13:09:12kugelit doesn't really make sense to offer a adjustable intervall between the steps.
13:09:14pondlife1Maybe limit the numbers, but still offer it in that form?
13:10:00kugelthe current intervall for e200 is 25 ticks. at 20 ticks it's already looking bad
13:10:23pixelmakugel: I can think of two ways to fix it - how many targets now have backlight fading and how many of them PWM (or the brightness control one, depending which is the smaller group)
13:10:36kugelthere's very little values to chose from, and they all will be about the same in terms of how long the fading took overall
13:10:55funmanhow can the simulator be properly turned off ? keeping the power button pressed to do that depends on HAVE_SW_POWEROFF and it seems undefined in the sim build
13:11:14kugelclick the X on the window :)
13:11:30funmani have no window decorations
13:11:43kugelthen hit ctrl+c in the terminal
13:12:07funmanThat's what I use, and why I asked for 'properly' because it's not very clean :/
13:12:20kugelthere's no proper way
13:13:01kugeldepends on what you mean by "clean", the sim saves settings instantly, so it doesn't need a shutdown procedure
13:14:00funmanok, i just wanted to be the closest possible to the simulated target ;)
13:14:24n1sfunman: implement it!
13:14:28kugelfunman: patches are welcome, you know it! :)
13:14:35funmanI'll try to simulate the real Clip screen
13:14:46n1swould be useful for testing idle shutdown etc, in fact
13:15:31funmanI'm still lost in the rockbox source tree, .make and SOURCES files scattered everywhere; sdl files in uisimulator/ and in target tree, etc etc
13:15:31 Quit kugel (Read error: 104 (Connection reset by peer))
13:15:32*pixelma tries to squeeze the monitor in an attempt to simulate the Archos' screens
13:15:43 Join kugel [0] (n=chatzill@unaffiliated/kugel)
13:15:43 Part pondlife1
13:17:27 Quit pondlife (Read error: 110 (Connection timed out))
13:17:36kugelpixelma: yep, e200 manual describes the options for pwm targets (as expected)
13:18:59kugelpixelma: can the text be made dependent on #defines in config-<target>.h?
13:19:24n1skugel: that's how the whole features.txt thing works
13:20:34pixelman1s: there is no separate feature for PWM and the other backlight fading though (and it's not needed for the lang files)
13:20:49pixelmajust one backlight_fade
13:21:46 Quit jeffdameth1 (Read error: 113 (No route to host))
13:21:48n1spixelma: sure, but if it makes the manual stuff easier we could split it in two: backlight_fade_(pwm|sw) and have the lang string s depend on either
13:21:58kugelpixelma: I guess \opt{HAVE_BACKLIGHT_PWM_FADING} and \opt{USE_BACKLIGHT_SW_FADING} instead of \opt{backlight_fade} will do it?
13:22:18n1skugel: no, it doesn't work like that
13:22:45kugeln1s: oh, I read "\opt{HAVE_BACKLIGHT}" and thought it would work
13:23:39 Join jeffdameth [0] (
13:23:41n1syou will need to define those options somewhere for that to work, HAVE_BACKLIGHT is a special case but features from features.txt is automatically parsed into latex options so we should use them when we can
13:24:21n1sin fact that special case could be dropped with a small change to features.txt
13:24:59pixelman1s: then there are 3 ways (the other 2 I thought of were more "by hand" though) - either directly opting the time settings sentence for the targets which have it, or invent a new option in the platform files, independently of features.tex
13:25:35pixelman1s: don't drop the HAVE_BACKLIGHT thing, it was left there on purpose
13:25:53pixelmaand no "backlight" feature in the lang files
13:26:00n1spixelma: i know, but there might be a nicer way
13:26:24n1siirc it's for the ondio backlight mod?
13:27:26n1sbut i will not touch it so don't worry, anyway i think two 'features' is the easiest way
13:28:13pixelmayes, so one can use downloadable voice files
13:29:18n1shmm, no my ide wouldn't work without special casing in other parts of the manual... unless ther's a way to #undef options in a platform file?
13:29:43pixelmadifference to the RTC mod is that. in case of the backlight it is really only one target that's affected by it (and can have backlight) and there is more than one target that has no RTC and only one can be modded
13:30:47pixelman1s: don't know if that is possible but sure would be a nice and easy solution
13:31:06pixelmaby the sounds of it
13:31:39kugeln1s: so lets make it like this: rename backligh_fade to backlight_fade_sw and _pwm, append a * in every lang file, and opt them seperate in the manual
13:32:03n1sor even easier, have the lang strings depend on a backlight feature or ondio* (with a comment to avoid "cleanups)
13:32:30n1skugel: ye, that's what i had in mind
13:35:30pixelman1s: that could work too, sounds even easier
13:36:41n1sthat would mean a "modded" ondio manual would have the correct settings too :)
13:37:04 Quit robin0800 (Remote closed the connection)
13:39:47pixelmawhile we are at it, while looking at the "Action" definitions in the platform files, I started thinking that it would be nice (if possible) to factor out the XYZ_PAD into own files (only leaving a "link" there in the platform files) because currently all target manuals contain them even in cases where the same keymap is used, e.g. in different Ipod platform files with IPOD_4G_PAD etc.)
13:40:15pixelmas/target manuals/target platform files
13:41:12pixelman1s: do you think that's possible?
13:41:39n1spixelma: yes, that would probably be an improvement, I'd love to try to get some kind of auto parsing of the keymap working at some point but that's probably far off :)
13:41:57n1sshould just be a simple \include afaiu
13:42:33pixelmaas I said Sunday or so, I can't image how that could work but someone could prove me wrong... :)
13:42:56n1sheh, will hope to try :)
13:43:30n1sbut yes, seperate keymap files is a Good Thing (tm)
13:43:53 Join robin0800 [0] (
13:44:12pixelmaok, will try if no-one beats me to it
13:44:26 Nick fxb__ is now known as fxb (
13:45:08n1sjust put a \input{keymap-ipod.tex} in each ipod*.tex and move the keymap to that file, should be all that's needed
13:45:17n1sas an example for the ipods
13:45:47pixelmayes, now that you mentioned the \include I saw the light ;)
13:46:39n1sbtw, it's \input, i was wrong in my first comment :)
13:47:14pixelmaah, thanks
13:48:07 Quit robin0800 (Read error: 54 (Connection reset by peer))
13:49:23kugelok, done. Want me to upload on the tracker or is pastebin enough?
13:52:05***Saving seen data "./dancer.seen"
13:54:42 Quit tvelocity (Read error: 110 (Connection timed out))
13:55:18 Join tvelocity [0] (
13:55:24 Join CaptainKwel [0] (
13:58:28kugeln1s, pixelma:
14:08:25 Quit CaptainKewl (Read error: 110 (Connection timed out))
14:13:37n1skugel, looks good, i'll commit now
14:14:31kugeln1s: hurry, until the manuals are rebuilt with wrong infos!! :P
14:14:53n1slike anyone reads those things anyway :P
14:16:44 Quit moos (Read error: 110 (Connection timed out))
14:17:40 Join robin0800_ [0] (
14:17:44n1sok, done
14:18:41*kugel notices no mention of ams sansas in major changes
14:19:07n1sit's a wiki...
14:19:58n1sit hasn't seens a lot of updates at all lately
14:20:55kugelyea, true. and sad
14:22:29 Quit tvelocity (Read error: 110 (Connection timed out))
14:23:03 Join tvelocity [0] (
14:28:07linuxstbIsn't "major changes" just for things in official builds?
14:28:40 Quit omgun ("CGI:IRC (Ping timeout)")
14:28:47B4gderI think we should update it as soon as the changes are done
14:28:56B4gderas otherwise they will just be forgotten
14:29:13B4gderwe can add markers for official releases
14:29:55linuxstbBut I don't think it should list progress on unreleased ports.
14:30:00JdGordonmaking it actually linked on the front page might make it forgotten less...
14:30:26amiconnn1s: Voice files *are* versioned, it's just not used properly
14:30:51B4gderlinuxstb: ah true, unreleased ports may not be major enough to get mentioned there
14:31:10*B4gder is slow
14:31:17JdGordonmilestones like working bootloader could go there though...
14:32:00 Quit tvelocity (Remote closed the connection)
14:32:08funmanLittle progress on clip sim :
14:32:15n1samiconn: if you see the linked FS task i suggested we should start bumping the version when needed but LinusN objected and said it was a _format_ version so i don't know if we should add a second "cliporder" version? (the same applies to lng files but the problem isn't very big with those)
14:32:44funmanI have some problems with plugins and greylib still
14:33:40JdGordonn1s: how is the problem not as big with lng files? the same thing will happen if the order of english.lang changes... i.e mis-translated strings
14:34:34n1sJdGordon: but installign a build without lng files is much more uncommon than using an outdated voicefile
14:35:04n1sbut yeah, the same solution should be used for both for completness/correctness
14:39:15 Quit robin0800_ (Read error: 104 (Connection reset by peer))
14:39:40 Join __Wonko [0] (
14:41:18 Join CaptainSquid [0] (
14:42:51 Part B4gder
14:44:37 Quit Nibbler ("Ex-Chat")
14:46:15 Join LambdaCalculus37 [0] (
14:49:30 Join robin0800 [0] (
14:52:13 Quit amiconn (Nick collision from services.)
14:52:20 Join amiconn [50] (n=jens@rockbox/developer/amiconn)
14:56:12 Quit JdGordon (Remote closed the connection)
14:56:43pixelmaI thought the order of the phrases in the lang files is not important, at least it's not in the translated lang files compared to english.lang - maybe I'm misunderstanding the problem though
14:57:20 Quit Wonko_ (Read error: 110 (Connection timed out))
14:57:20 Nick __Wonko is now known as Wonko_ (
14:59:22funmanFS #9521 : real screen in the Clip Sim, I welcome comments since I find my diff a bit heavy; perhaps a simpler solution exists
15:01:10 Quit kugel (Remote closed the connection)
15:02:30funmanperhaps I can skip the gradient setting on the 'normal' lcd_surface
15:02:35LambdaCalculus37funman: Regarding FS #9578, I'll try again with -p1.
15:03:28funmanLambdaCalculus37: I generate the patches with git, and git diff uses "diff -u a/xxxx b/xxxx"; so you have to skip that first-level a/ and b/
15:04:36 Quit EspeonEefi ("さよなら")
15:08:02LambdaCalculus37funman: patch -p1 works, but at firmware/target/arm/as3525/sansa-clip/timer-target.h, patch is thinking that a reversed or previous patch was applied. I haven't applied anything to that file.
15:08:16LambdaCalculus37The same for timer-target.h for the Fuze, e200v2, and m200v4.
15:09:03LambdaCalculus37It's asking if it should assume -R.
15:10:09funmando what you want, I just removed all as3525/*/timer-target.h and moved one in ../
15:10:39funmanI tested the diff with patch −−dry-run and it showed no warning/questions or anything, it's strange ..
15:11:22 Quit robin0800 ("Leaving")
15:12:35funmanI just retested, and it applies fine; without questions.
15:13:18LambdaCalculus37If I let patch assume -R, it attempts to create the as3525/*/timer-target.h and fails. Do you want to see the .rej files from the failed attempts?
15:13:30funmanperhaps the use of /dev/null confuse your patch, I use patch 2.5.4
15:14:03funmanNo, just delete all these files and leave one in as3525/
15:14:36LambdaCalculus37I'm using patch 2.5.8 here.
15:15:23 Join Schmogel [0] (
15:16:16funmanif you have git, you can use git-apply to apply the patch
15:20:00 Join robin0800 [0] (
15:20:07LambdaCalculus37I reverted that patch, removed the timer-target.h files, and patched again. The patch wanted to now delete the missing files; I told it not to assume -R and don't apply for those files.
15:20:14funmanbertrik: it seems the Clip battery reports as 95% filled when full, is it the same for you ?
15:20:28LambdaCalculus37funman: The Clip battery reports the same for me.
15:20:29funmanLambdaCalculus37: should be good
15:20:43LambdaCalculus37funman: The rest of the patch applied smoothly.
15:20:58funmanLambdaCalculus37: I noticed it would report as 100% when charging though
15:21:17LambdaCalculus37funman: I noticed that too.
15:21:50*LambdaCalculus37 builds a Clip build
15:23:47 Quit robin0800 (Client Quit)
15:24:09 Join robin0800 [0] (
15:24:24 Join bmbl [0] (n=Miranda@unaffiliated/bmbl)
15:25:20LambdaCalculus37Huh? How do I know have apps/bootbox.make?
15:25:54 Quit robin0800 (Client Quit)
15:26:14 Join robin0800 [0] (
15:26:31funmanisn't bootbox only for archos ?
15:26:52funmanLambdaCalculus37: perhaps you need to remove make.dep to regenerate deps (and rerun configure to enable plugins)
15:27:20LambdaCalculus37funman: Here's the error:
15:29:07 Join sin613 [0] (
15:29:37funmanLambdaCalculus37: < I have this text lines 44 - 47 , do you have the same ?
15:30:46n1spixelma: ?
15:31:43n1samiconn: does rockbox do its own ram setup on pp based players?
15:31:57*LambdaCalculus37 checks his root.make file
15:33:13LambdaCalculus37funman: <−− after the double line is my root.make file.
15:33:44funmanso, the exact same .. are you using GNU make ? (not bsd make?)
15:34:28LambdaCalculus37GNU make 3.79.1 (that's what Xcode installed).
15:35:32funmanI believe make < 3.80 is buggy
15:35:49 Join kugel [0] (n=chatzill@unaffiliated/kugel)
15:35:52LambdaCalculus37funman: And Fink claims to have installed 3.81, but I don't see a difference.
15:36:35funmanMy patch didn't touch that file, if you have problems with unmodified SVN, blame Zagor :)
15:36:49*LambdaCalculus37 blames Zagor :P
15:37:44LambdaCalculus37funman: I know your patch didn't touch that file; I was watching it as it patched. :)
15:39:20*LambdaCalculus37 wonders how different it would be if he were using some more obscure Unix to do this :)
15:39:42funmanimo macos *is* 'some obscure Unix'
15:40:36funmanLambdaCalculus37: is your xcode install up to date ? (i.e. do you have the latest iphone sdk?)
15:40:52LambdaCalculus37funman: We don't have enough OS X using devs/staff to cover any oddities, though. I think besides me, JdGordon and preglow both have Macs.
15:41:49LambdaCalculus37funman: It's version 2.5; IIRC that's outdated. Let me see if 3.0 still supports PowerPC.
15:43:19LambdaCalculus37Yes, I should still be able to use XCode 3.0 on my Mac. I'll see if I can update it later.
15:45:51 Join evilnick [0] (i=0c140464@gateway/web/ajax/
15:48:35 Quit CaptainKwel (Read error: 110 (Connection timed out))
15:52:05 Quit sin613 ("Leaving.")
15:52:08***Saving seen data "./dancer.seen"
15:54:42 Part LinusN
16:05:55 Quit LambdaCalculus37 ("Ka-chunka")
16:12:59 Part robin0800 ("Leaving")
16:14:44 Join blahrus_ [0] (n=blahrus@
16:15:13 Join tyfoo [0] (
16:25:41 Quit J-23 (Remote closed the connection)
16:26:03 Join J-23 [0] (
16:27:37 Quit J-23 (Remote closed the connection)
16:27:50 Quit TMM (Read error: 104 (Connection reset by peer))
16:28:23 Join J-23 [0] (
16:28:55 Join TMM [0] (
16:29:07 Quit blahrus (Read error: 110 (Connection timed out))
16:38:44 Quit kugel (Remote closed the connection)
16:41:41 Join jfc^3 [0] (
16:42:21 Join kugel [0] (n=chatzill@unaffiliated/kugel)
16:46:22 Quit CaptainSquid ("Miranda IM!")
16:46:24 Quit kugel (Remote closed the connection)
16:46:27 Join toffe82 [0] (
16:47:31 Quit Schmogel (Read error: 110 (Connection timed out))
16:48:44 Quit Darksair (Remote closed the connection)
16:49:08 Join robin0800 [0] (
16:49:52funmanhum why is HAVE_WAKEUP_OBJECTS conditional on a model in config.h ? shouldn't that go in config-model.h ?
16:50:23funmanI want to use it in DMA driver for sansa ams
16:51:13 Join Darksair [0] (n=user@
16:52:46jhMikeSfunman: no reason really other than only including it in things that use it (like the USB stack and some other misc. drivers).
16:53:29funmanjhMikeS: wouldn't it be better placed in config-gigabeat.h ?
16:56:03jhMikeSIt might but then writing non-busywaiting drivers should be encouraged and it should get more use. The selection is sort of twisted at the moment.
16:57:47jhMikeSI see JZ4732 also selects it now.
16:57:56funmanI'll add the AS3525 case next to JZ4732 and USBSTACK
16:58:02 Quit jfc (Read error: 110 (Connection timed out))
17:01:21 Quit evilnick (" ajax IRC Client")
17:04:50 Join XavierGr [0] (n=xavier@rockbox/staff/XavierGr)
17:05:36 Quit skipper (Remote closed the connection)
17:10:48jhMikeSfunman: careful though, the scheduler does not deal with FIQ
17:11:24funmanthe as3525 doesn't use FIQ yet, I fail to see the difference with IRQ. Is it that FIQ have higher priorities ?
17:13:43 Join {phoenix} [0] (
17:14:57jhMikeSfunman: it is higher priority and has more banked registers. it's also masked separately of course and the kernel just doesn't deal with it so FIQs calling there would corrupt things.
17:15:27 Join spiorf [0] (
17:18:14jhMikeSFIQ is so broken on imx31 without special messy workaround code (errata) it can't be used there anyway. :\
17:19:00 Join faemir [0] (
17:21:37 Quit J-23 (Remote closed the connection)
17:22:04 Join J-23_ [0] (
17:22:26 Nick J-23_ is now known as J-23 (
17:24:41rasherJdGordon: (fort he logs) You're probably better off going over that with Zagor. Remember voice generation needs to work on Cygwin as well.
17:24:44 Quit robin0800 (Read error: 110 (Connection timed out))
17:25:19 Quit blahrus_ ("Ex-Chat")
17:25:53 Join kachna|lappy [0] (
17:34:25 Join jhulst [0] (n=jhulst@unaffiliated/jhulst)
17:34:43 Quit sbhsu ("leaving")
17:34:55funmanwhat is the difference between CPU_FREQ in config-model.h and CPUFREQ_MAX (or _DEFAULT ? or _NORMAL?) in system-target.h ?
17:34:56 Join sbhsu [0] (
17:36:30funmanI understand that the frequencies in system-target.h are used in cpu_boost*()
17:37:48funmanI'll ask grep
17:38:32 Quit kachna (Read error: 110 (Connection timed out))
17:39:56funmanfor the sansa ams it's only used in system.c : long cpu_frequency = CPU_FREQ;
17:40:45funmanand I don't see set_cpu_frequency() or cpu_*boost*() changing it
17:41:19 Quit {phoenix} ("Konversation terminated!")
17:41:30funmanoh target' set_cpu_frequency() does
17:47:54 Quit funman ("leaving")
17:52:10***Saving seen data "./dancer.seen"
17:52:50 Join pondlife1 [0] (
17:53:43 Quit spiorf (Remote closed the connection)
17:54:02*amiconn pings jhMikeS and repeats his question from y'day
17:55:54 Join skipper [0] (
17:57:59 Join MethoS [0] (
17:59:12 Join CaptainKewl [0] (i=2669ecc2@gateway/web/ajax/
18:02:35 Join Photoguy [0] (n=Adams@
18:06:30 Part pondlife1
18:06:35PhotoguySo the calender plugin is now implimented and compatible?
18:10:13*Llorean loves questions without any context whatsoever.
18:12:12PhotoguyNo matter how ask a question here, someone balks!
18:12:40PhotoguyLast time I tried to kind of lead up to it, and someone was like "Spit it out!"
18:14:17LloreanBut your question doesn't make any sense. How can anyone know what player you're asking about if you don't mention it?
18:14:34LloreanAnd if it is a supported player, it takes you a few seconds to check if it's in the plugin list, so the question seems really weird
18:14:57LloreanSo I can only assume you're asking something a little more obscure than what can be confirmed in the plugin listing, and without more information, I don't know what you're asking.
18:15:43PhotoguyWell, I meant the official build for all normal ports.
18:16:23 Quit AndyI ()
18:17:02LloreanWell, what exactly are you asking? The changelog says it's now working on all RTC equipped targets.
18:17:44PhotoguyAh, ok.
18:18:03PhotoguyThanks for, I just didn't understand...
18:18:06 Join domonoky [0] (n=Domonoky@rockbox/developer/domonoky)
18:19:17 Join kugel [0] (n=chatzill@unaffiliated/kugel)
18:19:31PhotoguyI don't understand why the rules are so...precise.
18:19:54*Photoguy just wondering
18:20:06LloreanPhotoguy: Well, for one thing, it means we don't have to spend ten minutes trying to figure out what you're asking if you speak clearly.
18:20:24PhotoguyYeah, makes senes.
18:20:27LloreanFor another, we don't have to spend *any* time if you read the changelog for yourself first... It helps everyone if you do the prerequisite searching.
18:20:54PhotoguyI think the person "in charge" is a very presice, neat person.
18:21:15 Quit GodEater_ (Read error: 60 (Operation timed out))
18:21:48LloreanThere's not one person in charge.
18:22:05PhotoguyHence the " "
18:22:20PhotoguyI should of said people
18:22:43 Join AndyI [0] (i=AndyI@
18:22:52LloreanConsidering the quotes were around "in charge", that indicates that you suggest there could be an alternate meaning for "in charge", NOT for "person"
18:23:24pixelmathere is a nice link in the topic of this channel, which leads you to a page with info about the rules and the reasoning behind them ;)
18:24:00 Join HBK- [0] (
18:24:34PhotoguyI'm not knocking the rules, I'm just amazed at the order of this community, compared to others.
18:24:44 Quit HBK ()
18:24:45 Nick HBK- is now known as HBK (
18:24:59 Join {phoenix} [0] (
18:25:06LloreanIn all honesty, though, being precise about what you type is going to help a lot if you ever need support. People will frequently get frustrated and give up on you if you continue to type things other than what you mean, and assume people will realize your sentence was incorrect and guess what you meant.
18:25:18*jhMikeS pongs amiconn and should probably look at the logs for the question
18:27:20 Join GodEater_ [0] (n=ge@rockbox/staff/GodEater)
18:27:47kugeljhMikeS: You noticed backlight fade is committed? You might add you gigabeat s patch now
18:28:03*jhMikeS can't seem to locate amiconn's question
18:28:36jhMikeSkugel: yes. I guess that _should_ be quick work.
18:29:53*kugel thinks the whole backlight stuff could need some clean up
18:30:59toffe82jhMikeS: I think he wanted to know how to change the cpu frequency to have full speed if I remember
18:31:33 Join Nibbler [0] (
18:31:53kugelah right, he wanted to know how to make the beast run with full power
18:32:25kugelso that he can show how the beast outperfoms every single target on the ape wiki page :p
18:32:27jhMikeStoffe82: thanks. I'll answer him.
18:32:27PhotoguyWaste of battery!
18:34:05jhMikeSamiconn: You have to change the MCU divider to be /1 instead of /2. The keypad reading delays are too short for that so the keys will be misread (increase those delays).
18:34:50PhotoguyHey, how does the Cowon D2 work, as far as keymapping, since there aren't many.
18:35:01LloreanPhotoguy: It has a touchscreen and uses it.
18:35:31LloreanjhMikeS: If Ape performance scales linearly, c5000 may be realtime now.
18:35:44PhotoguySo, everything is controlled via touchscreen?
18:36:04LloreanPhotoguy: The touchscreen covers Up, Down, Left, Right and 'Select'
18:36:12PhotoguyAh, cool.
18:37:13jhMikeSLlorean: Memory bus access will still be the same. I did try an mpegplayer benchmark for ED and it was around 80-90fps iirc.
18:37:34PhotoguyCan you slelct individual menus, like a normal touchscreen?
18:38:01LloreanPhotoguy: Rockbox doesn't have a graphical menu. It just uses the touchscreen as directional buttons in the menus.
18:41:49jhMikeSamiconn: Set PDR0[2:0] (MCU_PODF) to 000
18:44:25 Quit kugel ("ChatZilla 0.9.84 [Firefox 3.0.4/2008102920]")
18:47:07*amiconn can't find where this is set
18:47:27amiconnI would have expected it in system-imx31.c, but even grep doesn't help...
18:48:15jhMikeSnothing changes that register right now from bootup
18:48:31amiconnOh? How do we set the desired clock then?
18:49:23*amiconn also can't find the keypad reading delays
18:49:28pixelmaPhotoguy, Llorean: in some plugins you can already point at specific items (also in a few plugin menus, e.g. pegbox) at least that worked with the mouse in the sim
18:49:40 Quit Nibbler (Read error: 113 (No route to host))
18:50:14 Join kugel [0] (n=chatzill@unaffiliated/kugel)
18:52:47PhotoguyThat;s nice.
18:52:54jhMikeSamiconn: the clocking is all as the loader sets it
18:53:21jhMikeSamiconn: the delays are in button-imx31.c in KPP_HANDLER
18:53:22*amiconn found the delay meanwhile
18:53:40 Join herrwaldo [0] (
18:54:17amiconnjhMikeS: Hmm, relying on the loader isn't nice, especially not on an OF loader. If we know how, we should set up everything the way we want it
18:56:28 Join BigBambi [0] (n=Alex@rockbox/staff/BigBambi)
18:57:47jhMikeSamiconn: yes, that would be better. I'm not sure why it sets up the core for 528 vs 532.
19:01:02jhMikeSit gives a nice, round 66MHz ipg_clk after /8 though
19:01:19 Quit bmbl ("Woah!")
19:01:37 Join bmbl [0] (n=Miranda@unaffiliated/bmbl)
19:02:20 Quit jhulst (Remote closed the connection)
19:06:24 Quit GodEater_ (Remote closed the connection)
19:06:41amiconnjhMikeS: Buttons seem to work without adjusting the delay
19:07:36jhMikeSamiconn: hmmm, on mine they misread quite often at full speed
19:09:52 Quit Photoguy ("On to something better...")
19:11:14BigBambijhMikeS: Coming from the other side, is the beast "easy" to slow down?
19:11:14amiconnAPE scales almost linearly with CPU clock.
19:12:08 Join LambdaCalculus37 [0] (
19:13:06 Join nuonguy [0] (
19:13:07jhMikeSBigBambi: You could set the MCU divider the other way but there could be restrictions on clock ratios (which should be outlined in the datasheet).
19:13:31BigBambiOK, cheers. I'm thinking about battery life :)
19:13:40BigBambiI'll have a shufti
19:15:11amiconnTheoretical speedup factor would be 2.000, and I get 1.980 (-c1000) ... 1.985 (-c5000)
19:15:25amiconnThat means -c5000 runs 104.2% realtime
19:15:34 Quit Darksair ("To Arch or Gentoo? That is the question...")
19:16:43LloreanThat's 104.2% realtime in a pure decode test.
19:16:57 Join Horscht [0] (n=Horscht@xbmc/user/horscht)
19:17:04BigBambiI know, but it is still impressive
19:17:08amiconnIt doesn't play without skips though, at least not with my peakmetered wps
19:17:14LloreanOn lower speed targets we've needed as much as 120% realtime to get realtime playback with the usual stuff going on. Since we're talking percentages though, that 4% represents a lot of CPU overhead
19:17:25Lloreanamiconn: Try default settings?
19:17:51 Join Self-Perfection [0] (
19:18:13amiconnAbout 1 short (<0.5 sec) skip per minute
19:18:32amiconnBuffering takes nearly forever
19:18:36jhMikeSare there further prospects for optimizing?
19:22:30 Join Thundercloud [0] (
19:24:26amiconn2 skips even with iCatcher, and buffering ran up to 3:40 (for a 4:05 song without a following track)
19:24:35amiconnI guess disk DMA would help
19:25:28amiconnjhMikeS: Btw, after a while my Up button stopped working. All others still work
19:26:17 Join bertrik [0] (
19:32:19 Join tvelocity [0] (
19:34:22amiconnNo skips when the track is fully buffered (pause until buffered, then unpause), even with peakmeters.
19:40:22Unhelpfulcan anybody say about how often i ought to yield during scaling? every N ticks would probably be easiest, i can either check if it's time to yield each line, or more often if need be.
19:41:36jhMikeSyou should try to yield every tick minimum so threads get properly scheduled
19:46:10Unhelpfulhrm... any thoughts on the best way to do that? every n input pixels would be easy to implement, but n would have to vary quite a bit between targets...
19:48:45jhMikeSif (last_tick != current_tick) { last_tick = current_tick; yield(); }
19:49:08 Join saratoga [0] (n=9803c6dd@gateway/web/cgi-irc/
19:50:31 Join funman [0] (
19:50:55jhMikeSactually, reverse the tick save and yield order or else it may overdo it
19:51:31 Join miepchen^schlaf [0] (
19:52:13***Saving seen data "./dancer.seen"
19:56:19 Join aarcane [0] (
19:57:51 Join Zagor [242] (
20:01:16Unhelpfuleven checking that per-pixel would probably be ok, considering everything else that goes on per-pixel. yielding is about the only thing left, aside from hardware testing on hardware i don't have
20:01:21bertrikfunman, my clip RTC is still running fine without getting reset
20:01:43bertrikmaybe I'm just lucky or maybe something got fixed
20:01:45funmanbertrik: mine too since I have put the good setting
20:02:23funmanperhaps it will run fine for 71 years and then reset? (it would start at Jan, 1st 2080)
20:02:29J-23hm, how is >1GB memory built on v2 Sansas?
20:02:53funmanJ-23: the same undocumented iNAND chips I believe
20:03:07 Quit Self-Perfection (Remote closed the connection)
20:04:06funmanperhaps if I find a cheap 4GB clip that would help
20:04:13*bertrik suspects a century bug somewhere in the rockbox code
20:04:52funmanthe best price I can find is €50 , without shipping : not so good
20:05:02bertrikI collected some CSD data from various ams sansas a few weeks ago. I should put them online, they may help.
20:05:17funmandomonoky: did you continue looking at i2sout?
20:05:19kugelfunman: have you looked into the bank switching in the sd-pp driver?
20:05:38funmankugel: quickly, I just stopped when I noticed it used a 'vendor specific command'
20:06:19funmanperhaps the command is the same for sansa ams, perhaps not
20:06:20domonokyfunman: a little bit, but i got lost when i tried to find out why the pcm_xx functions arent called...
20:07:38kugeljhMikeS: how would I block the lcd driver while reading buttons?
20:08:05bertrikcheapest I can find is E43,18 for a 4 GB silver clip, but I think they add E9,- for shipping
20:08:36kugelon my fuze I can't get the button reading to work in the main binary since (I at least strongly assume) the lcd is messing with the gpios while I'm trying to read the buttons from the same port
20:09:00funmankugel: what if you add a delay before returning from the function ?
20:09:06domonokykugel: is the lcd workin in interrupts or in normal code on fuze ?
20:09:08 Join HBK- [0] (
20:09:15 Quit HBK- (Client Quit)
20:09:26kugelthere's nothing with irq in the lcd driver
20:09:33 Quit Jabone (Read error: 145 (Connection timed out))
20:09:54domonokyif both lcd and buttons dont use interrupts, you can not get interrupted with cooperative multitasking..
20:10:23domonokyso lcd can not run at the same time, as you juggle with the button pins...
20:10:43Unhelpfulkugel: you were one of the people interested in preserving *everything* that resize-on-load did already. any suggestions for how to test the scaled-for-remote case? only thing i can think of is if there's a way to get album art on the remote?
20:10:45kugelso why does the button code work in the bootloader?
20:10:57domonokytiming might be more a problem. insert a delay after you switch gpio_afselect/gpiodir..
20:11:09kugelI think I tried that
20:11:13kugelI might try again though
20:11:49kugelI control the lcd with the buttons, e.g. inverting colors with the up button. That's the symptoms I have
20:13:13bertrikwhat I did for an embedded project at work, is reading only one row at a time in an interrupt, then select the next row. This way, there is plenty of time for the column signals to settle.
20:13:47domonokykugel: and show us your changes.. maybe you missed something..
20:13:48bertrikdoesn't need to be an interrupt of course, can also be done polled
20:15:45 Quit AndyI ()
20:15:52 Join Jabone [0] (
20:15:54*domonoky also inserted delays into the m200v4 button-driver.. we were to fast for the button circuit
20:15:59 Join AndyI [0] (i=AndyI@
20:16:02kugelbertrik: I tried reading only one button, the result was that the function returned that button without me pressing it *sometimes*. Pressing the button was as useless as when reading all buttons
20:16:17kugeldomonoky: Ok, I'll have a look into that
20:16:21 Quit LambdaCalculus37 ("Ka-chunka")
20:17:45kugelUnhelpful: I was the one that has the opinion that resizing should be implemented in a generic way, so that other parts can use it as well. That was about the only feature I wanted that the other patch does
20:17:59 Quit J-23 (Remote closed the connection)
20:18:16kugelUnhelpful: well, but you could use the icons which are shown in front of files to test
20:18:42kugelI don't know if showing album art on the rwps is possible
20:19:49 Quit AndyI (Read error: 104 (Connection reset by peer))
20:19:52 Join AndyIL [0] (n=pasha_in@
20:20:47kugeldomonoky: missing delay could be very well the reason, assuming the afsel aren't already changed when reading. The GPIO pins are set to 1 directly after, which could explain the heavy lcd weirdness
20:21:38domonokykugel: so test it with some big delays after changing afselect..
20:21:47kugelwill do
20:22:01lasserfunman: still thinking about buying a 4gb clip? I found some for E35,99 + E6 shipping. They're shipped from US.
20:22:22*kugel would stop thinking domonoky thinks he's stupid if he stopped adding several dots after his sentences
20:22:41funmanlasser: the fact is I already have 2, so I'd prefer to sell/give one before buying another
20:23:06*domonoky always adds ....even to not dump people *hehe*
20:23:52bertrikwhat is the function of the LCD pins shared with the button GPIOs?
20:24:56jhMikeSkugel: was there any reason other than SW fading that backlight_brightness should be externalized?
20:26:12kugeljhMikeS: the backlight_brightness variable is common to all targets which feature backlight brightness.
20:26:26Unhelpfulkugel: the scaler's generic, but right now, only album art and sliding_puzzle ever ask for resize
20:26:54kugelUnhelpful: you could load the icons depending on the font size, and select a huge font
20:26:59 Join tessarakt [0] (
20:27:30kugeljhMikeS: I don't really see a reason why each target should implement setting that variable on it's own
20:28:33funmanBy the way if you can have a (even quick) look at clip plugins (FS #9578) and real clip screen (m200v4 also?) for the sim (FS #9521) I'll appreciate it
20:29:08*bertrik wonders why lcd_write_command* and lcd_write_data is defined in the lcd.h public header file
20:29:25jhMikeSkugel: true enough
20:29:31kugelexternalizing that was pretty necessary for my patch, and given that it doesn't hurt other targets, I rather avoided to #ifdef it all the time
20:29:56 Join toffe82_ [0] (
20:31:40 Join fml [0] (n=4fd3c355@gateway/web/cgi-irc/
20:32:09 Join stoffel_ [0] (
20:32:28fmlI like the new display fade in/out feature on the sansa, but why is it not possible to set the fade out time?
20:33:07fmlI'd set it to a bit longer than it's now. Fade in should be short IMHO, so it's ok.
20:33:40jhMikeSkugel: well, that was an easy addition in any case but the actual set_backlight_fade_* functions have to be in the target or else its too complex
20:34:19kugeljhMikeS: you mean in backlight-thread-fading.c?
20:34:57*kugel assumes you mean backlight_set_up_fade_up and _down in backlight.c
20:36:06jhMikeSkugel: yes. no point in trying to share those since I need to use hardware bits anyway.
20:36:35 Join massiveH [0] (
20:36:54kugeljhMikeS: or do you mean the functions which set backlight_set_fading_type ?
20:37:12 Quit skipper (Read error: 145 (Connection timed out))
20:37:47kugelwell, why do you need backlight_set_up_fade_up/_down at all? I thought your patch does it in hardware w/o using the thread
20:38:27*gevaerts reads about ape on a full speed beast
20:38:56gevaertsWould half a second skip per minute translate to just over 1% CPU?
20:39:32fmlkugel: hello. Are you the Mr. Fading Guy? Could you please tell me why the new settings are boolean and not int (i.e. durations)?
20:41:04jhMikeSkugel: no, the ones in backlight.c to set the options. also, I think I need to maintain a protected backlight_brightness or else improper values could be written.
20:41:13kugelfml: because you can't configure the duration. For 2 reasons: the resulting duration is dependent on the brightness level you set, and because the the fading gets un-smooth or too fast to be noticed respectively very quickly
20:41:18 Quit Thundercloud (Remote closed the connection)
20:41:38 Join jhulst [0] (n=jhulst@unaffiliated/jhulst)
20:41:54kugeljhMikeS: backlight_set_brightness if the desired value is within the boundaries
20:42:49kugeljhMikeS: nothing prevents you to check the brightness and set backlight_brightness again in your _backlight_set_brightness function though
20:42:57jhMikeSit's a multithreading issue since set_backlight_brightness is called from wherever while on/off is on the backlight thread
20:44:29 Quit toffe82 (Read error: 110 (Connection timed out))
20:47:53kugeljhMikeS: I don't quite understand. How could improper values be written?
20:48:35 Join Strife89 [0] (n=michael@
20:49:02fmlkugel: but it should be possible to build in some slowness depending on the duration. I.e. each level of fading would stay for a little longer than now (one tick?)
20:49:22jhMikeSkugel: racing between two threads for access to the spi
20:49:23kugelfml: which target?
20:49:45jhMikeSnot that it's a problem for the spi itself
20:50:06kugelfml: I tested different intervals on my e200, and it turned out that making it slower isn't look nice at all
20:50:16fmlkugel: sansa e200, but I think this is target independent
20:51:11fmlkugel: are the stages too visible then? Or what?
20:52:05 Join bluebrother [0] (n=dom@rockbox/staff/bluebrother)
20:52:19kugelthis, and the backlight flickers a bit
20:52:20 Quit fml ("CGI:IRC (EOF)")
20:52:48 Join fml [0] (n=4fd3c355@gateway/web/cgi-irc/
20:52:49 Join AndyI [0] (i=AndyI@
20:52:57kugelfml: the interval has to be chosen well, since we effectively change the brightness in steps. This is, by definition, not as smooth as pwm fading is
20:53:49kugelso, to make it smooth as possible, but not too short on the other hand, there's only very little intervals to chose from. And those wouldn't effectivly affect the fade time noticeably
20:53:58fmlkugel: yes, but are those steps visible that much? If yes then I understand. But I haven't seen it myself.
20:54:32funmanis there a doc on logf()?
20:54:32 Join sarixe [0] (
20:54:32kugelfml: the steps are just slighlty visible. The flickering is the worse part
20:54:55fmlkugel: what's the duration of one step now?
20:55:11kugelthat's really distracting, as if you would watch on a crt monitor with 25h
20:55:20fmlkugel: you mean, when you change the bg brightness the display flickers?
20:55:21kugelfml: HZ/25 for e200
20:56:13kugelfml: yes, you don't notice that when you change only one level, but if you change more levels in a row. I've chosen a value which is a very good compromise imho
20:56:43kugelHZ/20 (bigger interval) is already noticeably worse, and the fading time isn't much higher
20:57:01 Quit CaptainKewl (" ajax IRC Client")
20:57:22fmlkugel: ah, ok then. How many levels are there? E.g. at the max brightness?
20:57:58kugelfml: take a look at the brightness settings
20:58:03kugelthe max value is 12
20:58:05pixelmafml: you could look at your brightness setting, it's dependent on the target
20:58:19kugelfor the e200 that is
20:59:00fmlkugel: so, for e200's max brightness, we have 12 / 25 ~ 0.5 sec, right?
20:59:24kugelfml: no, HZ represents ~1s, not the brightness level
20:59:57fmlkugel: there are 12 levels, each HZ/25
21:00:17kugelbut only if you maxed the brightness
21:00:39kugelif you have it at e.g. 6, you'll end up with only 6 steps
21:01:33fmlkugel: yes, I talked about the max brightness. But I have it set to 3! =:-O
21:01:54kugelfml: and it's still a nice effect, isn't it?
21:02:18fmlkugel: yes it is
21:03:09 Quit AndyIL (Connection timed out)
21:05:50Strife89Blast, forgot to unapply the backlight patch.
21:06:18Strife89Has the backlight fading patch been committed?
21:08:38fmlkugel: but what if we'd set one step to 1 HZ? Then we'd have 3 sec for my level. Would it flicker?
21:09:18kugelfml: do you really think changing the backlight every full second can be considered as smooth?
21:10:04 Quit stoffel_ ("leaving")
21:10:07fmlkugel: no, but it's nearer to the fade out duration I'd like to have
21:10:50kugelthat's no fading
21:11:51kugelfml: please understand. It's not designed to last long. It's designed to be smooth
21:13:25fmlkugel: ok. I just saw that on H120 it's longer and smooth. So HW fading is much finer, right? Then I'm quiet.
21:14:07kugelfml: yes, *very* much finer
21:15:32funmanwill a simple codec like wav fit in a 512kB codec buffer ?
21:16:10funmanI'm thinking about the fact that pcm_* functions are not called may be due to codec not fitting in the 512kB buffer on my clip (instead of the usual 1MB)
21:18:05domonokywav codec is probably the best to try on our low-mem targets..
21:18:09 Quit bluebrother ("leaving")
21:19:06linuxstbOnly codecs that use malloc will have problems - mp3 doesn't for example, so should be fine.
21:19:43linuxstb(assuming the codecs actually compile and link OK)
21:20:09funmanwell using wav doesn't call pcm_play_data() either
21:20:58domonokydoes anyone have a idea, to easily see, if the codec really starts ?
21:21:27amiconnfunman: Many codecs will fit into an even smaller buffer than that
21:21:34 Join HBK- [0] (
21:21:43amiconnIirc the iFP port used a 128KB codec buffer
21:22:14 Quit HBK- (Client Quit)
21:22:17funmanyou make me feel good :) now to find what's going wrong
21:22:36amiconnJust build all codecs for a normal target of the same architecture (i.e. armv4 if you want to know that for the ams targets), and then check the .map files
21:22:53linuxstbThere's even a script now to tell you the codec binsize/ram usage
21:23:23amiconnDon't forget the IRAM - it'd probably best to do that for a target where we don't use any (i.e. gigabeat f)
21:23:57funmanwell at the moment I'm focused on PCM, not on the codecs
21:24:15funmanIf I can leave to the mountains with rockbox playing mp3s I'll be happy :)
21:24:17 Quit MethoS (Read error: 104 (Connection reset by peer))
21:24:22linuxstbMaybe trying a plugin that uses sound will be easier.
21:24:28domonokywe first have to get sound, before we can optimise the codecs :-)
21:26:00funmanlinuxstb: ah I didn't think about that, good idea
21:26:22 Quit Dieterbe_ (Client Quit)
21:26:33domonokyfunman: metronome might a good test :-)
21:26:41 Join Dieterbe [0] (
21:27:03funman/ 0
21:27:28funmanat 3004*666*0
21:27:41funmanbut the instruction I see is "mov r4, r0" ?
21:28:17 Join MethoS [0] (
21:28:19 Join Reptile211 [0] (
21:28:31funmanhm just after a bx __divsi3 though
21:28:49*funman blame the person who wrote set_cpu_frequency()
21:30:58funmanstill, calling set_cpu_frequency(0) sounds strange
21:37:41domonokythe only place which could call set_cpu_frequency(0) is panic.c. Most targets do either nothing on this, or switch to default freq..
21:37:56funmanthe iriver bootloader as well
21:38:43funmanperhaps that means one of my panics happened ..
21:39:14 Join Seed [0] (
21:39:21domonokyyes, looks like a panic happend...
21:39:31funman int divider = frequency ? (CPUFREQ_MAX / frequency) : 16 /* minimal */ ;
21:39:47funmanshould solve the problem
21:39:58 Quit aarcane ("Leaving")
21:40:20funmanI thought of using a switch/case but I doubt set_cpu_frequency is speed critical, and that can wait until we are sure which frequencies we want to use
21:41:10funmanindeed metronome plugin calls pcm_ functions, but not the codecs
21:41:18 Join Thundercloud [0] (
21:41:40domonokywe have some other problems with the codecs, i think.
21:42:45 Join buk_ [0] (
21:43:21amiconnset_cpu_frequency() is meant to set one of a few special frequencies. Passing any other value is meant to set the default
21:43:22 Join culture [0] (
21:43:56amiconnThe latter is what panic() relies upon
21:44:24 Join MegafEee [0] (n=Linux@unaffiliated/megaf)
21:44:41 Quit jeffdameth (Read error: 104 (Connection reset by peer))
21:46:39funmanamiconn: right, while the sansa ams port is in the works I prefer to leave a function which will use the right setting whatever the 3 frequencies are modified to.
21:47:00funmanat the cost of 2 divisions, not something critical
21:48:03funmanmy approach is: 1/ run 2/ run fine - I hope you won't see a problem in this
21:48:20 Join zackyramone [0] (n=chatzill@
21:48:42funmandomonoky: are you using metronome to write i2sout support? I can find other work to do :)
21:48:52zackyramoneis there a way to use rockbox on creative zen mozaic?
21:49:20domonokyfunman: at moment i try to get the plugins running on m200v2, so i can take a try on i2sout :-)
21:49:20funmanzackyramone: if it isn't listed on the website, No.
21:49:44amiconnfunman: No, but it would probably make sense to switch to some sane (i.e. low) default when called with 0
21:50:09zackyramonefunman: the website said something about mozaic being compatible with zen vision m or something like that...what about that/
21:50:37amiconnOn most targets, adjusting the CPU clock requires adjusting some other delays; one or two divisions wouldn't be enough
21:50:50amiconns/most/a number of/
21:51:35funmanamiconn: sure, it's already committed in my tree to the minimal frequency, just waiting to be committed to svn :)
21:52:15***Saving seen data "./dancer.seen"
21:52:21funmanamiconn: I choose to keep a common PLL setting, AS3525 leaves 4 bits for a divider of this PLL
21:53:25funmanzackyramone: if the mozaic is compatible with the vision m, the vision m port is still in development, and so, not ready yet to be used. Just check the webpage regularly; it will be listed here as soon as it's ready to use
21:54:03zackyramonefunman: thanks :)
21:54:10 Quit zackyramone ("ChatZilla 0.9.84 [Firefox 3.0.4/2008102920]")
21:54:52pixelmameh, building the manual now takes longer because it looks like it needs to make the complete dependency file to parse the features late :\
21:55:00funmanin fact there is 4 bits for an integer divider (1->16) and 2 for a fractional divider (1, 5/8, 6/8, 7/8)
21:55:00 Quit fml ("CGI:IRC (EOF)")
21:56:16amiconnfunman: Both the coldfire and the PP chips use plls as well to set up the cpu clock. But there's more to it: DRAM speed, LCD waitstates/ delays, timer prescaler adjustment (on coldfire) ...
21:57:36funmanamiconn: on the as3525 the peripherals use a different clock, which can be not linked to the PLL (and isn't currently)
21:58:41 Part Reptile211
22:01:37kugeldomonoky: delay didn't help
22:02:01kugelI haven't tried putting it between the both afsel sets
22:02:24funmanamiconn: in fact I was thinking about this, and was hoping to have delays in various drivers not be dependant of the current cpu freq
22:03:20funmanand timer frequency as well, but luckily this one doesn't depend on the peripheral clock since it uses directly the 24MHz crystal (which is used as a source for PLLs and various clocks)
22:08:02 Quit bmbl ("Woah!")
22:08:23 Join bmbl [0] (n=Miranda@unaffiliated/bmbl)
22:12:57pixelmais the dep file always (re)made even when nothing's changed?
22:13:28 Quit sarixe ("Ex-Chat")
22:13:33Zagorpixelma: it is never remade unless you remove it or explicitly ask for it
22:15:32pixelmawonder why it now takes so long until the actual manual build starts even on second run, but maybe it just feels longer than before :(
22:19:23 Join bluebrother [0] (n=dom@rockbox/staff/bluebrother)
22:21:00 Quit tyfoo ("Carpe diem")
22:23:23 Join tyfoo [0] (
22:24:37pixelmaI'm quite sure now that it takes longer than before the changes in the build system... /me cries
22:32:30 Part kkurbjun
22:33:34 Quit {phoenix} (Remote closed the connection)
22:38:35*funman comforts pixelma : it must be Zagor or GNU make's fault
22:39:15Zagorpixelma: it does take longer. it now reads all dependencies and all makefiles before starting to compile the first file.
22:39:29Zagorthere is a very simple solution to make it MUCH faster though: dump cygwin
22:39:44 Quit MegafEee (Read error: 110 (Connection timed out))
22:41:27pixelmawell, I guess even in Linux it will take longer than before, no?
22:41:49Zagorperhaps now it's 1.5 seconds and before it was 1.1
22:43:04*gevaerts finds something he thinks may be a bug
22:43:20gevaertsRun "make clean;make dep" and watch the output
22:43:38Zagoryeah that is a bug
22:43:48 Join `x_X-` [0] (
22:43:55funmancan I get a preview of the output without running it ?
22:44:20`x_X-`I need help :<
22:44:27gevaertsAnyway make dep takes 41 seconds here, so that's probably 20 seconds for a single one
22:44:39funman`x_X-`: we all need help. what do you need?
22:44:57`x_X-`Well my main problem is that I don't have a "GBSYSTEM" folder.
22:45:19Zagorgevaerts: yeah that's for _making_ the deps. pixelma talks about reading them.
22:45:35gevaertsah, ok.
22:45:40*gevaerts misread that bit
22:45:59domonoky`x_X-`: there is a wiki page with a fake GBSYSTEM folder which is enough if you only need rockbox..
22:46:07funman`x_X-`: ok, and how is that a problem?
22:46:22*`x_X-` goes to find
22:46:34gevaertsfunman: the gigabeat needs them to boot
22:46:47funman`x_X-`: please use plain english (for non english native people who read the logs and the channel)
22:47:18`x_X-`Sorry, I thought "orly" was a common term
22:47:42 Quit massiveH ("Leaving")
22:48:08funman`x_X-`: I personally know it, but I guess a lot of people who don't read 4chan don't :P
22:48:35`x_X-`Lol. I can't stand 4chan. But anyway...I'm showing hidden Folder/Files and it still doesn't appear to be there.
22:48:43 Quit Strife89 ("Yard work.")
22:48:49`x_X-`When I try to uncompress the "dummy" one, it says it's duplicate?
22:49:02linuxstbShow system files as well.
22:49:11amiconnZagor: Why does making the manual need to read all the source dependencies?
22:49:13`x_X-`Oh? There's an option for that?
22:49:21amiconn(or voice, for that matter)
22:49:27funman`x_X-`: in the same menu for view hidden files, yes (another checkbox)
22:49:58`x_X-`:o Like Magic, it appears.
22:50:28Zagoramiconn: ah, it doesn't. that's a bug. I missed that it was a manual build.
22:51:43pixelmait needs the features.txt - not sure how much that matters
23:06:14*amiconn is annoyed by the "speed" of 'make voice'
23:09:21*n1s finds it funny that 'make help' needs the dependencies built too :)
23:10:03kugelZagor: another question: Why doesn't make build the manual when I configure a "manual build"?
23:10:18n1salso i don't know if this is worth bothering with but if i ^C while building dependencies and tries make again it thinks they are good and of course fails
23:10:22kugelI need make manual. I'd think with make I get the target which I configured for
23:10:27amiconn4:27 until make.dep even starts to appear (0 bytes). Another 5:30 until make.dep is complete.
23:10:41n1skugel: that is older and intended for simplicity
23:10:43amiconnHmpf, and then it fails!
23:11:06pixelmakugel: that's been there before since the automatic parsing of features.txt into a features.tex with usable options was introduced
23:11:26kugeln1s: i find it misleading. Do I even need to configure for manual or is make manual working for every build type?
23:11:52n1skugel, in fact 'make manual' works perfectly in a regular build dir :)
23:12:05n1smaybe we should drop that configure target...
23:12:29pixelman1s: by the way - separating the keymaps works
23:12:35Zagoramiconn: the slowness is entirely your own choice.
23:12:42amiconnNo, it's not
23:12:54n1spixelma: ah, nice :)
23:12:54Zagoryes it is
23:12:56amiconnHow else would I make voice files using SAPI voices?
23:13:23amiconnAnd why does 'make voice' need the dependencies?
23:13:23amiconnAnd btw, it's broken
23:13:40kugeldoes voice need the regular build environment?
23:13:48funmananyone who messes with build systems is doomed
23:13:51kugelor can basically just mingw used w/o crosscompilers installed
23:14:12amiconnZagor: 'make voice'
23:14:26amiconnLast 3 lines of 'make voice':
23:15:09Zagorfunman: nah, you just need tough skin on your nose :)
23:15:58kugelthe build system is just "sophisticated" :p
23:16:08pixelmakugel: you can build voices in Linux too, just needs a TTS engine. The downloadable voice files are built under Linux with "festival" IIRC
23:17:07pixelmathe SAPI voices I know of though are nicer/cleaner than the downloadable ones and some espeak ones I tried
23:17:12Zagoractually make is very simple. all it does is compare the date on two files, and run a command if file b is more recent than file a. I don't really understand why "everybody" thinks make is black magic.
23:17:38kugelpixelma: I mean of cygwin/interix/vm is need at all? Do you need the target specific compilers for voice builds?
23:18:13Unhelpfullatest scaler is done. i could still use some testing from people who own hardware for any of the greyscale/mono targets, and it wouldn't hurt to have binsizes for coldfire targets, for which i lack a toolchain.
23:18:27funmanZagor: if you build thousands of files, it is black magic
23:18:27amiconnZagor: I think it is because that almost nobody (including me) really understands in what way all the rules depend on each other
23:18:50Unhelpfulat this point, aside from figuring out some gotchas related to mono targets, where i kind of doubt a scaler is going to do much good, i think it's about "done"...
23:18:52 Join shotofadds [0] (n=rob@rockbox/developer/shotofadds)
23:19:07funmanthat's where autotools, cmake, and others are here; to lower the complexity of writing rules
23:19:10amiconnUnhelpful: Mono displays do not need scaling at all in the core
23:19:53Zagorfunman: no, those tools are there to aid multiplatform builds
23:20:05Unhelpfulamiconn: i'm inclined to agree. this is designed to be general-purpose, but i can special-case mono-only targets to not use any scaler.
23:20:21funmanZagor: not automake.
23:20:36amiconnI see the scaler as an option that's ifdefed out for mono displays
23:20:50funmanmultiplatform builds are only a part of these solutions.
23:21:17shotofaddsgevaerts: when you were playing with md5sum yesterday did you hack together a version that just displays the md5 on screen? I'm about do the same myself and wondered if you've already done it...
23:21:32gevaertsshotofadds: I tried, but it didn't work at all
23:21:39gevaertsSo I reverted the file
23:21:58amiconnZagor: I think the problem is that 'make voice' is looking for apps/features (root.make line 247), but the 'features' file is located directly in the build dir, and there is no apps/
23:22:08 Join avis [0] (n=ident@pdpc/supporter/student/avis)
23:22:24shotofaddsok, I'll try it later. "can't be that hard" :g
23:22:35amiconnWhy it does print ...genlang-features is beyond me though
23:22:53Zagorgenlang-features is another file
23:22:57 Quit HellDragon (Client Quit)
23:23:33*kugel wonders what will happen if Zagor happens to be unavailable to fix build system quirks
23:23:43kugelseeing you're the only one really getting it
23:23:51Bagderhe's not
23:23:58n1skugel: we'll just live with them like we always did :)
23:24:00Zagorbagder and linusn groks it too, at least
23:24:27gevaertskugel: make isn't that hard if you look into it properly
23:25:00funmanZagor: rockbox can be built on different platforms; did you consider using such programs (autotools, cmake, ..) before revisiting the Makefile?
23:25:08amiconnn1s: Right now I cannot live with it... I cannot make fresh voice files for my targets...
23:25:17 Join HellDragon [0] (
23:25:21 Quit avis (Client Quit)
23:25:40funmanI guess most of the checks autoconf does are in the C library and 3rd party libraries (useless to rockbox), but perhaps some of the logic could be useful?
23:25:50Zagorfunman: rockbox can only be built on systems with gcc and with a very narrow set of include files. autotools would have given us nothing more than extra complexity.
23:26:10 Quit domonoky (Read error: 104 (Connection reset by peer))
23:26:10Bagderand we use some of the logic already
23:26:35n1sBagder: would the automated manual builds break if we drop the "Manual" target from configure (it's not needed, a manual can be built from a target or sim build configure) (or should i just leave it?)
23:26:41funmanZagor: that is autoconf; what about automake ? (Makefile generation from 'simpler' Makefiles)
23:26:57Bagdern1s: I'll check...
23:27:20Zagorfunman: I don't see the point.
23:27:24Bagderfunman: that's not without debate
23:27:25linuxstbn1s: What if someone doesn't have a target gcc? I guess they could build a manual in a sim build directory?
23:27:26Unhelpfulamiconn: branches are cheap. i'll put together a version that directly outputs on the mono-only targets. there's already a direct-output fallback for color if scaling is not requested, because of the amount of manipulation in the scalers.
23:27:34gevaertsfunman: have you looked at the various *.make files? They aren't that complex
23:27:36funmanZagor: as the files are simpler, less maintenance
23:27:47Bagderfunman: that's not true
23:27:49funmanBagder: I'm all for debates!
23:27:59Zagorare they? where does all my information go?
23:28:00n1slinuxstb: yes sim dir works fine (and you need a native gcc to build the manual anyway)
23:28:06Bagdersimpler files are easy to edit, sure
23:28:13Bagderthen they generate huge complex files
23:28:23Bagderand when they malfunction, you're in it
23:28:35funmangevaerts: I had a look already, but I'm still a bit lost to know which .make file I have to look for a particular rule
23:28:38Unhelpfulseirously, fixing auto* output can be nasty.
23:28:58*linuxstb agrees - Make is simple enough to not need a layer on top of it
23:29:07funmanBagder: I know where to look in (very) complex generated Makefiles so I may be biased
23:29:18amiconngevaerts: The problem with following Makefiles (for me at least) is that the order in which stuff appears in the file(s) is totally unrelated to the order in which things are actually done
23:29:21pixelman1s: what do I need to do to get the file ID info in the new tex files - set the svn props and put a % $Id at the top?
23:29:21funmanZagor: I guess it's compressed?
23:29:44Bagderit really isn't
23:29:49Zagorthe current makefiles are some of the most simple you can imagine. the average is about 25 lines _with_ rockbox header. only root.make and codecs.make are long.
23:30:11amiconnpixelma: % $Id$
23:30:17Bagderthe auto* tools are good, but for more complex scenarios than rockbox imho
23:30:21amiconn(the trailing $ is important)
23:30:29pixelmaah, thanks
23:30:42Bagdern1s: yes it will break, but will be easy to fix
23:30:54Bagderthe manual builds I mean
23:31:26n1sBagder: ok, i'll wait to do it though, too tired for reliable commits now :)
23:31:28linuxstbn1s: Hmm, isn't features possibly different for the sim though?
23:31:49amiconnlinuxstb: If it is, I'd consider that a bug
23:31:58gevaertsZagor: on linux I get an empty BUILDDIR/apps/features and a non-empty BUILDDIR/features, so voices build, but I wouldn't be surprised if they were wrong
23:32:10linuxstbamiconn: I agree - but isn't that the same as the voice issue we were discussing earlier?
23:32:19 Join Reliom [0] (
23:32:25Zagorgevaerts: yeah I'm looking into that
23:32:56amiconnlinuxstb: Probably. I didn't try that recently. Last time I tried, my (target) voice files worked fine in the respective sims
23:33:19amiconn(swcodec only, of course, as hwcodec sims have no audio output)
23:33:24 Quit `x_X-` ("It's Business Time ^_^")
23:33:49 Join tvelocity[a] [0] (n=tony@
23:34:32Reliomanyone know how to force the ipod 5g backlight to turn off when plugged in? the option under settings doesnt seem to work
23:35:28n1slinuxstb: you could just configure a target build and ignore the compiler warnings though
23:35:34pixelmaReliom: if I remember correctly, you can't
23:36:20Reliom:\ really annoying dont want to damage it from being plugged in and i dont use it much so the battery dies from being unplugged
23:36:47linuxstbn1s: So features is processed with the native gcc? (I'm guessing features processing the only use for gcc in a manual build)
23:36:48Reliomits dead by the time i go to use it in other words
23:37:59amiconnjhMikeS: Is that a 3rd type of fading?
23:38:46gevaertsReliom: seems to work here
23:39:15bluebrotherBagder, n1s: I had that intention a while back too (removing the manual option) ...
23:39:16n1slinuxstb: not sure really it seems to use $(CC) so i guess i'll just have to try it before i make this change
23:39:31 Quit Reliom (Read error: 104 (Connection reset by peer))
23:39:54bluebrotherbut why does it break the manual build? make manual-pdf works fine in my build dir
23:39:56 Join Reliom [0] (
23:40:32pixelmadid someone say manual builds are broken?
23:40:40n1sbluebrother: the automated builds probably configure "Manual" builds so if the option goes away that will not work
23:41:41n1sbluebrother: do you have any comments on FS #9550, i tired googling a bit but it seems noone has this issue...
23:42:04n1sor my google fu isn't up to scratch
23:42:08bluebrotherthat's possible.
23:42:36 Quit tyfoo (Read error: 60 (Operation timed out))
23:42:38bluebrotherI searched about that a bit a while back but haven't found a real solution. It's definitely related to the use of textsc
23:42:45pixelman1s: should I seperate all keymap things into own files or only ones which are used in more than one platform file? Would probably be the cleaner solution though
23:42:52n1sI think we should drop the sc style if we can't find a fix because searching is important
23:42:53gevaertsZagor, amiconn: I think should fix the voice build
23:43:22n1spixelma: i think consistency is nice
23:43:29bluebrotherwell, I found the soul package to greatly help with this (and also with the html issue)
23:43:53gevaertsReliom: (in case you missed it) backlight off while connected seems to work here, although my ipod video has a slightly oldish build installed (a few weeks)
23:44:03bluebrotherunfortunately there's something wrong with our latex sources that make soul break with the toc generation.
23:44:06pixelmaand is it worth an own "keymap" directory or should it stay in the platform one?
23:44:21linuxstbn1s: The manual Makefile uses $(HOSTCC), so I think it's fine.
23:44:22bluebrotherusing a minimal latex example works.
23:44:27Zagorgevaerts: yes, pretty much. but actually it is the use of $@ in apps.make that causes the bug. $@ should not be used in rules with multiple targets
23:44:38 Quit perrikwp (" ajax IRC Client")
23:44:52gevaertsZagor: yes, I wondered about that too
23:44:53Reliomi had and old build alot older then that and just updated it both have the same problem
23:44:59Reliommaybe i need to reflash it
23:45:15 Join perrikwp [0] (i=4aa794a0@gateway/web/ajax/
23:45:28bluebrotherso figuring out what's wrong with out latex sources is my current goal
23:45:29n1slinuxstb: ah, i looked at apps.make for some reason :/ should really go to bed
23:46:00n1sbluebrother: ah, that would be nice, yes :)
23:46:03linuxstbn1s: So did I ;) But I also ran "make V=1", and that showed /usr/bin/gcc being used, so I was confused...
23:47:14bluebrothersoul also has the advantage that we could use textmarker-like hilighting if we find applicable use ;-)
23:48:20*gevaerts compiles a new build to try on his ipod
23:48:38saratogadoes test_codec work on the Clip?
23:48:52 Quit tvelocity (No route to host)
23:49:28funmanif 'work' == build and run until the call to read(); yes.
23:50:00saratogawhy does the call to read fail?
23:50:10linuxstbfunman: What happens? test_codec loads the entire file to RAM, so you will need very small test files...
23:50:22funmanI didn't look into this, I just figured a lcd_puts() before read() would print, and after would not
23:50:38funmanlinuxstb: oh, since all my files are ~3 or 4 MB that may explain
23:51:15saratogahow much RAM is there on the clip?
23:51:23saratogaerr free ram
23:51:24linuxstb2MB plus 320KB
23:51:30saratogaabout 1 MB I guess?
23:51:47gevaertsReliom: are you sure the settings are set correctly?
23:51:49funmanI have 512k for codecs and 512k for plugins in my build
23:52:08linuxstbYou should be able to greatly reduce the plugin buffer size.
23:52:19***Saving seen data "./dancer.seen"
23:52:31amiconnThe IRAM size looks like the IRAM would make a perfect codec ram
23:52:35Reliomi set it to turn off after so many seconds on each and set all of them to just be off permanently and still turns on and stays on when i plug in
23:52:39funmanI bet plugins don't malloc(), and would fail to link if I choose a too small buffer ?
23:52:56linuxstbfunman: Correct.
23:53:02funmans/plugins/plugins not build on monochrome targets/
23:53:24amiconnSome plugins use plugin_get_buffer() though and will fail if that buffer is too small
23:53:27gevaertsReliom: what is the normal backlight set to? Maybe it doesn't detect the connection properly?
23:53:36funmanwhen I get comments on FS #9578 I'll consider reducing the plugin buffer size to keep 1MB codec buffer :)
23:53:45Reliomwas set to 5 seconds i changed it to off
23:53:45amiconnMost plugins should work fine with just 64KB or even 48KB of plugin ram on arm
23:53:56Reliomit turned off when i changed the setting but back on when plugged in
23:54:07saratogathe codec buffer could be made smaller, I don't think anything even needs 1 MB, except for seeking in very long AAC files
23:54:19gevaertsThat's weird...
23:54:22 Join tvelocity[away] [0] (
23:54:29 Quit tvelocity[a] (Read error: 60 (Operation timed out))
23:54:32linuxstbsaratoga: And some vorbis files IIUC
23:54:41amiconnThere are very few exceptions, but we can use the same technique as we do on archos to make them run (requires stopping music)
23:54:48funmananyway, I believe test_codec needs a pcm driver first; and I prefer working on a correct one than on a dummy one
23:54:49gevaertsReliom: Maybe try checking the filesystem for errors. That's been known to fix weird problems
23:54:49saratogado we actually know how much RAM vorbis needs?
23:55:03 Quit bluebrother ("leaving")
23:55:04saratogaas I recall floor0 is thought to require more, but i think we don't even decode that correctly
23:55:06n1ssaratoga: no
23:55:15gevaertsfunman: are you sure? It doesn't actually play anthing
23:55:19Unhelpfulamiconn: modified per your suggestion, and uploaded. mono-display targets don't use the scaler at all, mono remotes on greyscale targets still use the scaler and share loop code with it - i can branch later and change that to see if it's a binsize savings or not.
23:55:35pixelmaReliom: if you plug it in does your Ipod stay in Rockkbox or does it reboot to the OF?
23:55:43n1ssaratoga: as in no, we don't know how much ram it needs and yes we do decode floor0 files correctly
23:56:04Reliomit goes to the "Do Not Disconnect" screen
23:56:08Reliomfor rockbox
23:56:44Reliomif i eject it it says safe to disconnect and still stay lit
23:56:45n1ssaratoga: there was an old bug report with a floor0 file that crashed before the buffer merging (needed to allocate more than half a meg) that now plays fine with the merged buffer
23:57:01linuxstbUnhelpful: How would you describe the quality of your image scaler?
23:57:30gevaertslinuxstb: "unprecedented"
23:57:42amiconnThe 'Do Not Disconnect' screen is not rockbox'. It's the OF's diskmode, which rockbox reboots into when it detects an usb connection
23:58:16Unhelpfullinuxstb: very good on color targets, less impressive on greyscale ones, where it's only nearest-neighbor scaling.
23:58:16gevaertsReliom: ah, ok. I was assuming that you were using it in no-disk-connection mode (i.e. just powered over usb)
23:58:18pixelmaReliom: the "Do not connect" is not Rockbox but Apples, so it looks like it doesn't detect that you only plug a charger. Try holding "select" while plugging
23:58:25funmangevaerts: no I'm not sure; I must see why test_codec (presumably) causes a data abort first
23:58:37gevaertspixelma: I think it's menu
23:58:40Reliomhmm ok

Previous day | Next day