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

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

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

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

#rockbox log for 2008-12-13

00:00:05 Join saratoga [0] (n=9803c6dd@gateway/web/cgi-irc/
00:00:58gevaertsBagder: ping
00:01:08bertrikkugel, did you also try it with SDA #defines as 5?
00:01:32kugelbertrik: the lcd driver sets pin 5
00:01:39kugelunlikely that this one will owrk
00:01:55bertrikI think there's a bit of a bug in funman's patch, it passes an int pointer where a byte pointer is expected, don't know if this has any actual impact
00:02:14kugelno, it doesn't
00:02:30kugelI tried changing that
00:02:35bluebrotherhelpo: please define "weird instructions"
00:02:37advcomp2019helpo, the wiki has the step for the e200R
00:02:58bertrikactually, at least you get a value other than 0 or 255, which indicates that there is an actual device somewhere responding to out request
00:03:05 Join Thundercloud [0] (
00:03:08kugelhelpo: we just know unknown instruction, so I guess we cannot help
00:03:25 Quit MethoS- (Remote closed the connection)
00:03:32*kugel was kidding
00:03:46*bluebrother didn't understand this joke
00:05:24kugelbertrik: possibly. funman said I shall try different delays, the -4 indicates a failed read
00:05:27*amiconn had yet another hard freeze experience on his H180 today :\\
00:05:44bertrikkugel, ah of course
00:07:06jhMikeSamiconn: On SH, using 1u for a shifted bit is more efficient? I should check the kernel.
00:08:00bluebrotherhelpo: please explain your problem
00:09:02 Quit helpo (" ajax IRC Client")
00:09:32*gevaerts seems to not have been running a current build after all...
00:09:44jhMikeShrm, nothing shifty there without priority
00:10:23 Join helpo [0] (i=4b3a331d@gateway/web/ajax/
00:10:38helpoalright i figured out the problem
00:10:44helpomy device is not supported
00:10:49helpo'Note: Rockbox doesn’t function on the newer v2 models. They can be identified by b checking the Sandisk firmware version number under Settings!Info. The v1 firmware is named 01.xx.xx, while the v2 firmware begins with 03.'
00:11:25helpoi have version 2 and i was fallowig all the directions correctly but rockbox was not detecting my media player
00:11:35kugelbertrik: what delay would you think is sufficient?
00:11:48gevaertshelpo: that would indeed explain things
00:12:06kugelbertrik: I also tried this one { int old = TIMER1_VALUE; while(TIMER1_VALUE + 4 > old) ; } as posted by funman a little later, same results though
00:12:09jhMikeSgevaerts: I don't know if what I committed makes any difference for mrobe100. It didn't matter for GBF but it's more correct anyway.
00:12:09bertrikkugel, try something very slow, i2c can't really be too slow, so perhaps increase it by a factor 100
00:12:17helpoawwww, the sad part is that rockbox wont work... ;[
00:12:54gevaertsjhMikeS: testing now
00:13:09rasherhelpo: Work is underway to support the v2 models, but it's still far from ready
00:13:29helpoany talk about dates?
00:13:45helpoversion 3 came out so i though it would be supported....
00:14:39saratogaversion 3 predates the first V2 code I believe, so no suprise it didn't add support
00:15:11 Quit EspeonEefi ("さよなら")
00:18:34gevaertsjhMikeS: I can't reproduce the problem at all anymore. This seems to be a bit more random than I thought at first
00:19:59jhMikeSno check between r19409 and r19410 to see it that really mattered?
00:21:20gevaertsI tried r19410, r19396 (latest daily), and 3.0, and it now always works
00:21:57jhMikeSmaybe mrobes need some time to warm up? :p
00:22:21*gevaerts thought the same :)
00:23:42 Quit Thundercloud (Read error: 54 (Connection reset by peer))
00:23:46*gevaerts finds something
00:24:30 Join Thundercloud [0] (
00:25:01 Join MTughan [0] (n=MTughan@
00:25:06gevaertsjhMikeS: in 3.0, if the headphones are plugged in while booting, it works. If they are plugged in later, it fails. Now trying other versions
00:25:25bluebrotherBagder: can we get content-type: text/plain for diff and patch files on the tracker?
00:25:52MTughanJust installed Rockbox on my iPod yesterday, seems to work. But I just replaced some files, and the old ones still show up. Does an iPod cache file names, does Rockbox, or is something else going on?
00:26:28 Quit gregzx ("ChatZilla 0.9.84 [Firefox 3.0.4/2008102920]")
00:27:08bluebrotherMTughan: seems you haven't set the database to auto-update
00:27:26MTughanbluebrother: I just forced it to update, and the old files show up in the file view as well.
00:27:47MTughanAlthough Auto Update wasn't on, thanks for the tip.
00:28:11gevaertsMTughan: also enable dircache, for even better auto-update (I think)
00:28:44MTughangevaerts: All right, I'll look it up.
00:28:44pixelmaI believe update and auto-update don't detect deleted files if the database is not set to "Load to RAM"
00:28:55bluebrotherdidn't auto-update require dircache?
00:29:02pixelmaor however that option is called
00:29:31pixelmabluebrother: no, I think it's "just" slow without
00:30:01MTughanOh, I forgot to copy over the files in one directory, happened to be the one I was looking in... Others updated fine.
00:30:14MTughanWell, thanks for the help and database config flags at least. :P
00:33:01MTughanAnd dircache is enabled... One more thing. Is there any "soft-off" mode in Rockbox, so the player doesn't have to load from scratch every time?
00:33:42MTughanWell, Vorbis and FLAC playback override that... Thanks for the help.
00:33:46Unhelpfuli know that the beast HW at least supports some sort of suspend, since the OF uses it. :/
00:33:47 Part MTughan
00:34:00Unhelpfulsome of the ipods don't have a "real" off, do they?
00:34:19saratogabefore the 3G they do not
00:34:24pixelmathe 1st and 2nd generation don't
00:34:42*gevaerts now doesn't understand this bug at all anymore
00:36:42 Quit jhulst (Remote closed the connection)
00:36:53gevaertsI could reproduce this earphone-while-booting in 3.0, r19396, and r19410 for a while, but now it suddenly always works again
00:37:12***Saving seen data "./dancer.seen"
00:40:16jhMikeSgevaerts: perhaps the initial state isn't read at boot?
00:40:36jhMikeSit has hp detection?
00:46:44amiconnjhMikeS: It's a little more complex than that. SH1 has no shift-by-n instructions, only fixed shifts. So libgcc provides subroutines for shift-by-n
00:47:24amiconnThere are 3 of those: arithmetic right shift, logical right shift, and arithmetical left shift (which is identical to logical left shift of course)
00:47:24gevaertsjhMikeS: I added FS #9628, which has all the information I know
00:49:14amiconnThere are two points to consider. (1) Logical right shift by n is a bit more efficient than arithmetical, due to the available instructions. (2) If it is possible to get rid of any of these 3 completely, a bit of binsize will be saved, because the corresponding libgcc routine will no longer be linked
00:49:24jhMikeSgevaerts: doesn't sound like something I can address without a device.
00:50:05jhMikeSamiconn: It always uses a function or inlines simple instanes (like division).
00:50:09amiconnThat said, with r19409 all uses of arithmetical right shift by n are already removed, so __ashrsi3 is no longer included in the rockbox core.
00:50:10jhMikeS*on ARM
00:50:15gevaertsjhMikeS: I guess your initial state theory is not improbable. I'll see if I can get at it tomorrow
00:50:53Unhelpfulamiconn: to a certain degree, we can try to make all logical shifts left, or right, by changing which operand is shifted...
00:50:54 Join jhulst [0] (n=jhulst@unaffiliated/jhulst)
00:51:20Unhelpfulthere are going to be expressions that doesn't work for, and it'd be a pain to go through all the shifts in RB, i'd think... :/
00:51:46amiconnBut of course you could still save a tiny amount of binsize and execution time on SH by using logical right shifts instead of arithmetical ones where possible
00:52:12amiconnI mean fixed shifts - the variable ones are already optimised this way.
00:52:43amiconnE.g. >> 4 translates to 2 instructions for unsigned, but 4 instructions for signed
00:53:30Unhelpfulamiconn: you've explained an oddity that i encountered trying to compact the bmp reader a bit - it seemed that the 15/16 bit cases could be merged by using calculated shifts - but you say makes it clear that it would be cheaper on SH to branch the code and use fixed.
00:53:58 Quit helpo (" ajax IRC Client")
00:57:56Unhelpful*what* you say.
00:58:12amiconnUnhelpful: Depends - if the fixed shift would need too many instructions when inlined (I don't know the exact limit gcc uses), that fixed shift will also be turned in a subroutine call (though without a parameter)
00:59:33rasherI agrees with Zagor about the database app (PCTOOL) - basically create a version of the sim with a tiny main() routine and accepting some commandline arguments
00:59:41rasherWould be easier to maintain
01:00:10amiconnIf you see calls to e.g. __ashiftrt_r4_NN (NN being a number from 1 to 31), this is what I mean
01:00:28bluebrotherisn't the main problem that nobody is actively maintaining the database app?
01:00:57bluebrotherwe could consider adding database building to rbutil. In that case a small sim wouldn't be too helpful
01:01:08Unhelpfulamiconn: urgh, that could lead to terrible space-efficiency, if all of those functions are being copied into the object at link :/
01:01:32amiconnThose functions are actually interleaved code
01:02:14Unhelpfulsomething along the lines of a series of single shifts with a label at each instruction, basically?
01:02:16rasherbluebrother: Making it a, say 25-line "patch" to the sim would make it so much easier to maintain
01:02:31rasherbluebrother: I don't see how the current situation makes things any easier for rbutil
01:02:53amiconnThe whole bunch of __ashiftrt_r4_0 to __ashiftrt_r4_31 totals 41 instructions (82 bytes)
01:04:01 Quit saratoga ("CGI:IRC (EOF)")
01:04:08amiconnYes, with some special cases short-circuiting >>31, >>24 (and up) and >>16 (and up)
01:05:16 Quit bertrik ("Leaving")
01:10:50 Quit culture (Read error: 110 (Connection timed out))
01:13:40 Quit bluebrother ("leaving")
01:16:14 Join midgey [0] (n=tjross@
01:21:24 Quit Bensawsome (Nick collision from services.)
01:22:38 Quit einhirn (Read error: 110 (Connection timed out))
01:22:42 Join Guest39166 [0] (
01:25:17 Nick Guest39166 is now known as Bensawsome (
01:27:17midgeyrasher: around?
01:27:30 Join perrikwp [0] (i=4aa794a0@gateway/web/ajax/
01:28:15rashermidgey: for a short while
01:28:35midgeyi've been looking at some of the lang updates on the tracker
01:28:42midgeyi'd like to get them in before 3.1
01:28:50amiconnwtf? Gcc uses __ashrsi3 for a variable shift (in the grelib) where clearly all involved variables are unsigned!?
01:29:49rashermidgey: What else besides FS #9553?
01:29:54amiconnDoesn't matter much, but it's really strange behaviour...
01:30:42midgeyFS #9239 and FS #9587 (if we can get the patch to apply)
01:32:14rasherAh yes, I guess we might as well add the turkish one. Maybe someone should check if there's anything in the old one that's not in the new one, since it's translated from scratch
01:32:39rasherI'd be happy to have FS #9587, but I've no idea what's happened to that patch
01:33:01midgeyi looked at fixing it manually, but it's way too huge
01:34:02Unhelpfulamiconn: tbh, i really didn't understand the purpose of that transformation it performed in ata.c... it turned a left shift into a right one, and if the code had been taken as written, there wouldn't be a need for worrying about sign :/
01:34:24midgeyi'm looking at the turkish one right now, it looks like some of the voice strings are gone
01:35:02amiconnTransforming it into a right shift saves a register
01:35:36midgeyerr, scratch what i just said, i formed the patch backwards
01:36:56amiconnAnyway, gcc sometimes produces weird and inefficient code. The greylib code I was looking into (grey_mono_bitmap_part inner loop and surroundings) is just one example...
01:37:56 Join mc2739 [0] (
01:37:59amiconnRather more often than just sometimes...
01:38:27 Join domonoky1 [0] (
01:39:36Unhelpfulamiconn: saves a register at the cost of requiring a static library symbol that was unneeded before, but gcc doesn't do inter-object-file optimization yet :/
01:40:18amiconnWell, one static library symbol is needed either way, just not the same. The left shift would need __ashlsi3
01:41:56Unhelpfulright, and gcc can't make a smart decision about it, unless it learns to revisit that decision at link time.
01:43:59amiconngcc isn't involved at all in linking
01:49:14 Join slact [0] (
01:49:25 Quit jhulst (Remote closed the connection)
01:50:30Unhelpfulstrictly binutils at that stage, right... :|
01:52:24 Quit kugel ("ChatZilla 0.9.84 [Firefox 3.0.4/2008111319]")
01:55:16 Join fdinel [0] (
01:55:22 Quit domonoky (Read error: 110 (Connection timed out))
01:56:43 Quit ender` (" Washing your car to make it rain doesn't work.")
01:57:54midgeyrasher: it looks like the turkish patch does lose a few strings
01:58:15midgeyit also has a lot of untranslated english
01:58:18rashermidgey: Maybe it's worth it to construct a script that merges the two.. I'll look into it
01:58:41rasherAnd remove strings that are identical to english as well
01:58:49midgeywell, the ids are there, they've just been replaced with english
01:59:29midgeyhe also likes to use Title Case a lot
01:59:32rasherShould be possible to get a reasonable result with an automatic merge if it compares to the english phrase as well to see which one has it right
01:59:44rasherI think that's less of an issue
02:00:05midgeyi wasn't sure if we had a consensus on that
02:00:33rasherHe probably shouldn't use it, but it's not a deal-breaker
02:01:12rasherJust tried the portuguese patch against all revisions of portuguese to see if my site had an old revision for some mad reason - still doesn't apply
02:01:30midgeyi tried a few old revisions
02:01:47pixelmaare you sure how cases are handled in Turkish? Just saying because there are some languages that do differently (well at least German does ;) )
02:02:00rasherpixelma: I did say probably
02:02:10midgeyit's inconsistent at least
02:02:22midgeybut i have no idea how turkish works :)
02:02:41 Quit moos ("Rockbox rules the DAP world")
02:03:38pixelmaI bet you'll find words in German which are verbs (then spelled lower case) and also nouns (upper case), but don't know a thing about Turkish...
02:03:56 Quit n1s ()
02:04:03midgeyi suppose it's best to trust the translator
02:05:12pixelmamaybe ask directly about this?
02:05:30rasherHrm, I think the portuguese can be mostly saved, actually
02:05:51 Quit Rob2222 ()
02:06:04midgeyyou figured out why it's not applying?
02:06:16 Quit Seed ("cu, Andre")
02:06:29rasherNot really, no
02:06:48rasherBut most of the failing bits is all additions
02:07:07slacthey, is anyone interested in a black-background version of the EscapeBoD 2 theme for ipod 5G ?
02:08:39slact , ,
02:08:40 Quit domonoky1 (Read error: 104 (Connection reset by peer))
02:09:37slactit's just luminance-inverted but with non-inverted button gradients
02:09:51slact(and minor tweaks-to-taste)
02:12:56Unhelpfuldoes that need multifont, or is the small text all sysfont?
02:13:42rasherLooks like sysfont
02:13:50slact(yteah, it's a sysfont)
02:14:04slact(sorry, the "neither" was to adifferent conversation)
02:18:43 Join Rob2222 [0] (
02:21:09 Quit Chronon ("ChatZilla 0.9.84 [Firefox 3.0.4/2008102920]")
02:22:56 Quit Nico_P (Remote closed the connection)
02:23:40 Quit midgey ()
02:31:35 Join XavierGr [0] (n=xavier@rockbox/staff/XavierGr)
02:37:13***Saving seen data "./dancer.seen"
02:41:26 Quit XavierGr (Nick collision from services.)
02:41:37 Join XavierGr [0] (n=xavier@rockbox/staff/XavierGr)
02:47:48*rasher fixed the portuguese patch
02:48:31rasherStill no flaming idea what went wrong
02:49:09 Quit fdinel ("Miranda IM! Smaller, Faster, Easier.")
02:49:13 Quit lasser (Read error: 145 (Connection timed out))
02:50:36 Quit herrwaldo ("Konversation terminated!")
02:53:17jhMikeSgevaerts: what's the cleanup function for usb storage? I need to undo audio_get_buffer.
02:54:24 Join mib_lx2l6kdh [0] (i=d8ef5c7f@gateway/web/ajax/
02:54:30 Nick mib_lx2l6kdh is now known as MarcGuay (i=d8ef5c7f@gateway/web/ajax/
02:55:09MarcGuayjhMikeS: LIttle problem with the keyclick/recording fix. It doesn't reactivate after you leave the recording screen.
02:55:25MarcGuayOnly by restarting playback does it come back into effect.
02:55:33jhMikeSMarcGuay: I was just fixing that right now.
02:57:25MarcGuayIs it considered normal that if you scroll quickly it doesn't click for every entry you pass? (with Keyclick Repeats set to Off, at least)
02:58:05jhMikeSif repeats are off, yes it's normal
02:58:30 Quit mofux (Read error: 104 (Connection reset by peer))
03:01:39 Join kharo1 [0] (
03:02:46 Quit kharo (Read error: 110 (Connection timed out))
03:03:57 Join midgey [0] (n=tjross@
03:05:42 Part toffe82
03:08:59*jhMikeS wonders if wheel messages shouldn't just ignore that setting
03:09:01rashermidgey: Got it. No idea what happened though.
03:09:11midgeyi noticed
03:09:33midgeyi was trying mess with the patch by hand. i'm not sure what was wrong
03:09:45rasherIf you check the diff, you see that it's vaguely different from the one he posted in some "from"-lines
03:10:13rasherLike there were edits to the original portugues.lang, which seems odd though
03:10:14rasherOh well
03:11:05 Join MTughan [0] (n=MTughan@
03:11:10midgeyodd, i tried copy and pasting from the original lang file into the diff and rebuilding it manually. didin't seem to help
03:11:23 Join Xillius [0] (
03:11:28Xilliushello guys
03:11:41MTughanRockbox can play FLAC files, right? Because it doesn't seem to be.
03:12:00Xilliusanyone knows about an xm/mod plugin support for rockbox sansa ?
03:12:29 Quit jeffdameth (Read error: 113 (No route to host))
03:13:59 Join jeffdameth [0] (
03:14:16UnhelpfulMTughan: can you be a bit more specific than that? it supports flac on all targets with software decoding, as far as i know.
03:14:59MTughanI'm running an iPod 5th gen (aka Video), with the latest stable of Rockbox. The FLAC files were transcoded from AIFF by ffmpeg.
03:15:27MTughanRockbox just seems to skip by all the files in the folder without actually playing any of them.
03:15:36scorcheXillius: mod should play fine...
03:16:26Unhelpfuland there are no messages displayed at all about what happens? in my experience, when it fails to decode something, it says why...
03:16:41MTughanIt just stays on the frontend.
03:17:25MTughanDisplays the file name for the title, but the artist and album fields read "(root)".
03:17:56Xilliusby default it should play it ?
03:17:57pixelmaMTughan: do you know how your .flac files are tagged (which tag format)?
03:18:13MTughanpixelma: Not sure. I used VLC to save the metadata.
03:18:17scorcheXillius: mod support was added back in May
03:18:47Xilliusyeah havent updated since then...
03:18:50Xilliusthanks for the info
03:19:23Xilliusany idea about xm/it ?
03:19:43scorcheXillius: next time please update before you come in here to ask...
03:21:45MTughanpixelma: Does the container format matter to Rockbox? I used OGG.
03:22:26Unhelpfuldefinitely, rockbox generally only supports whatever the "native" container is for a codec
03:22:50scorcheXillius: i am not sure about xm or it...i know there is a patch in the tracker though
03:23:03MTughanUnhelpful: As I know, OGG is the one used with FLAC.
03:23:24UnhelpfulMTughan: no, flac has it's own native format.
03:24:08Unhelpfulthe ogg container thing is something that got added when it was incorporated into the xiph codec family, and there's really no benefit from it except for slightly larger files that play on fewer devices.
03:24:09MTughanThanks scorche.
03:28:24 Join Llorean [0] (
03:28:25slacti'm trying to register on the rockbox TWiki, and it's giving me a bad error when I enter the validation code
03:28:46MTughanHmm, using the FLAC container still doesn't seem to be working.
03:28:51slactsomething about it not having write permissions
03:29:57Unhelpfulhow did you convert them?
03:30:17MTughanI used ffmpeg. Want the output options?
03:31:13Unhelpfulif you named the output file .flac, and didn't go out of your way to force a container format, it should be right. why not use the reference encoder, though?
03:31:19 Quit Thundercloud (Read error: 54 (Connection reset by peer))
03:31:33MTughanWhere's the reference encoder? I just used ffmpeg out of convenience.
03:31:59MTughanI guess off the FLAC page, eh...
03:32:07Unhelpfulthe one that comes from generally the binary is named "flac", or "flac.exe" on windows
03:32:23MTughanI'm running on OS X, so a BSD underbed.
03:33:08Unhelpfuli'm not sure what the "right" way is to get flac installed on OS X
03:33:30MTughanThere seems to be command line tools available. I'm downloading them now, will explore.
03:33:47Unhelpfulgood luck
03:34:21Unhelpfulwe should probably move this to #rockbox-community, it's become a conversion issue, now that we've established that the problem was probably the ogg container.
03:34:50MTughanMaybe. They play fine in VLC.
03:35:34Unhelpfulyes, ogg flac will play fine in vlc, to my knowledge, but it won't in rockbox
03:36:01MTughanYeah, I could see that. I'll see how far I get with the reference encoder.
03:41:26MTughanNice, that seems to work. Thanks Unhelpful.
03:41:35MTughanEven keeps the metadata from the original...
03:42:02Unhelpfulglad it worked. enjoy your rockbox :D
03:42:14MTughanI imagine I will... any ideas on TWiki erroring out on registration?
03:46:24 Quit kharo1 (Read error: 60 (Operation timed out))
03:49:06 Part MTughan
03:49:44 Join kharo [0] (
04:01:49 Quit Xillius ()
04:03:53 Join jhulst [0] (n=jhulst@unaffiliated/jhulst)
04:04:50*jhMikeS hates it when he thinks of a better, cleaner solution to something _after_ committing a bunch of changes. argh! :\
04:09:38 Quit Rob2222 ()
04:10:01 Join blkhawk- [0] (
04:10:58 Quit blkhawk (Read error: 60 (Operation timed out))
04:10:59 Nick blkhawk- is now known as blkhawk (
04:14:10 Quit MarcGuay (" ajax IRC Client")
04:19:08slactTWiki detected an internal error - please check your TWiki logs and webserver logs for more information.
04:19:10slactRCS: failed to create file path: Permission denied
04:20:57 Join CaptainKewl [0] (
04:36:28 Join ameyer [0] (
04:37:16***Saving seen data "./dancer.seen"
04:51:47 Join Darksair [0] (n=user@
04:55:38slact..maybe i just neeed write access or something?...
04:55:41slactwho do i ask for that?
04:57:46Unhelpfulyou need to be registered to *get* write access...
04:58:27slactmaybe i did register...
04:58:51slacti can't tell, it just gives me a blank plage when i try to attach stuff
04:59:39Unhelpfuloh, now, for that, you do need write access
04:59:56Unhelpfulyou really don't know if you're registered or not? :/
05:00:43slactwell, it gave me that error when i entered the emailed validation code
05:01:12slactbut it didn't say "GTFO" when i logged in. just blank pages when i try to attach, it seems...
05:01:31Unhelpfulif you're able to log in, i would say you're registered.
05:03:44slactmight you be able to point me to the login page?
05:04:26slactooh, wait
05:04:34slactit's giving a normal Acess Denied error now, i think...
05:05:20Unhelpfultry to edit a page?
05:05:56slactthat looks stateless, so i think the link'll work
05:07:04Unhelpfuldid you register with first+last name, as suggested?
05:07:08 Quit miepchen^schlaf (Read error: 110 (Connection timed out))
05:07:21slactyes, got a validation email and everything
05:07:39Unhelpful(obviously with your names)
05:07:50slactthen it doesn't exist
05:08:30 Join blkhawk- [0] (
05:09:01 Quit Darksair ("Use the Force, Luke!")
05:10:05Unhelpfulhrm. maybe try the email provided, if you seem to be unable to complete your registration?
05:10:30 Quit blkhawk- (Client Quit)
05:11:12 Join blkhawk- [0] (
05:12:05slacti'm not sure what you mean
05:12:59Unhelpfulif that page doesn't exist, you're not a registered user. there's an email at the bottom of the registration page for issues.
05:13:23slactah, that
05:13:32slactguess that's what i'll do then...
05:14:10Unhelpfulhe's on irc as well, so you could ask Bagder... probably should take it to the community channel, though, this is getting less and less about rockbox itself by the minute
05:14:28 Quit blkhawk- (Client Quit)
05:16:49jhulstslact: What's your wiki name?
05:17:23slactLeoP . Against the rules for the moment, i know...
05:18:06slact(i don't much care if the name gets deleted, i'm just trying to post a theme)
05:18:19jhulstslact: Per the registration page, you have to ask for write permissions in this channel
05:18:32jhulstslact:, Step 2
05:18:49jhulstRegister with the correct name and I'll give you write access
05:18:53slactyes, but TWiki gave mea a mean error when i entered the activation code provided in the email
05:19:09slactso i wasn't sure if it was time for Step @ yet.
05:19:37Unhelpfuland you're quite unlikely to get them if you're not going to follow the registration rules. we need to know who actually contributes things to the wiki, especially stuff like themes, primarily because of copyright issues.
05:19:52jhulstslact: Try registering with the correct name and I'll give you access, we'll see if that works
05:21:04slactvery well.
05:23:05slact( that is, my wiki username should be LeoPalmer )
05:25:29LloreanYou need to go through registration with that name.
05:26:02slacti did.
05:26:48LloreanHave you received a confirmation email for your attempt to register with the name "LeoPalmer" (NOT your LeoP attempt) yet?
05:27:07scorcheLlorean: yes...see back-log..
05:27:32Lloreanscorche: He responded to jhulst's "What's your wiki name?" question with "LeoP" though
05:28:21 Join Darksair [0] (n=user@
05:29:38 Join sajes_ [0] (n=sajes@
05:30:42Lloreanscorche: I just wanted to be sure the registration with the new name was actually complete.
05:31:53slactwell, that's what's in question
05:32:10Lloreanslact: So did you receive a confirmation email for the LeoPalmer name?
05:32:27slactand entered the validation code. and it gave me a mean error.
05:33:10slactTWiki detected an internal error - please check your TWiki logs and webserver logs for more information. RCS: failed to create file path: Permission denied
05:33:54slacttrying to enter the code a second time, however, sends me to a normal "code invalid" page. So clearly it did /something/, just not sure what.
05:33:59*scorche nods and sits around for certain people to wake up
05:40:04 Join LambdaCalculus37 [0] (n=rmenes@rockbox/staff/LambdaCalculus37)
05:40:08 Quit amiconn (Read error: 60 (Operation timed out))
05:40:47 Join amiconn [50] (n=jens@rockbox/developer/amiconn)
05:46:08 Quit sajes (Read error: 110 (Connection timed out))
05:47:38 Join gregorovius [0] (
05:55:57 Join pixelma_ [0] (n=pixelma@rockbox/staff/pixelma)
05:55:59 Quit pixelma (Read error: 110 (Connection timed out))
06:00:22 Quit LambdaCalculus37 ("Ka-chunka")
06:02:02 Quit n17ikh|Lappy ()
06:05:34 Quit mc2739 ("ChatZilla 0.9.84 [Firefox 3.0.4/2008102920]")
06:07:57 Quit XavierGr (Read error: 60 (Operation timed out))
06:12:45slactwell, anyway,"> is the theme i wanted to upload. If someone else cares to do it, be my guest.
06:28:56 Quit aarcane ("Leaving")
06:31:56 Quit GodEater (D-lined)
06:33:32 Quit perrikwp (D-lined)
06:37:18***Saving seen data "./dancer.seen"
06:53:05 Quit Darksair (Connection timed out)
06:56:19 Quit Aurix_Lexico (Read error: 110 (Connection timed out))
07:00:19 Join perrikwp [0] (i=4aa794a0@gateway/web/ajax/
07:04:07Unhelpfulamiconn: so, you said coldfire has a div instruction, and arm does not. i see that sh1 does? what about a different brightness() for arm, with three multiplies and a shift instead of two and a div?
07:04:28 Join crackmonk [0] (
07:14:29 Quit esthar ("KVIrc 3.4.0 Virgo")
07:14:43 Quit crackmonk (Remote closed the connection)
07:14:47 Join n17ikh|Lappy [0] (
07:42:49J-23hm, what's "Eraser" build?
07:44:21 Join Rob2222 [0] (
08:04:19 Join Duchess [0] (
08:07:24 Join Seed [0] (
08:08:01DuchessHi I'm having an issue using ipodpatcher, if someone feels up to helping me out.
08:12:24LloreanDuchess: Did RBUtil fail to work for you?
08:15:09DuchessWell, heh, thing is I'm actually trying to use it as a general tool not really related to rockbox, but I figured since I found the download off the rockbox website, I would ask you guys about my issue.
08:15:10 Join Rob2223 [0] (
08:16:32LloreanDuchess: Well, it's much less likely anyone will be able to help with questions about using it for things it's not meant to.
08:17:03DuchessNo no, I'm not trying to use it inappropriately, ipodpatcher *is* a general-purpose tool
08:17:36DuchessI'm trying to use it to backup firmware.
08:17:50LloreanFrom one of the list of supported iPods?
08:17:59LloreanSo what's the problem?
08:18:49DuchessIs it okay if I copy-paste the output?
08:18:58Lloreanuse a site like
08:19:09LloreanThe channel guidelines are pretty clear what to do about multi-line pastes.
08:19:47 Quit Rob2222 (Read error: 60 (Operation timed out))
08:20:22LloreanDid you try "sudo ./ipodpatcher" instead then?
08:20:34DuchessI tried with no luck
08:20:56LloreanIt gives you the exact same message about permissions even when run with su?
08:22:08DuchessNo it gets stranger if I do it that way, for some reason it won't even recognize the command
08:22:50LloreanAre you doing it from the location the executable is actually located in, or at least typing the correct path to it?
08:23:29DuchessI type the right path, yeah.
08:23:45LloreanIf it doesn't recognize the command, you're typing something wrong.
08:23:51LloreanSince that's your shell reporting the error, and not iPodpatcher.
08:24:06 Quit midgey ()
08:24:07DuchessI might be, though I'm just stumped why it's not working at the outset anyway.
08:24:16LloreanBecause you don't have the required permissions.
08:24:18LloreanIt can't work without them.
08:24:55DuchessHuh. Weird.
08:25:04LloreanWhat's weird about it?
08:25:13LloreanThere are certain restrictions OSes put on what applications can do.
08:25:33Unhelpfulnormal users do not have acces to raw disk devices, normally. that would be insecure and dangerous
08:25:46 Join write_erase [0] (
08:26:06DuchessTrue, I think the main issue is that I'm doing this at almost 2:30am, haha.
08:26:19write_eraseHi... do rockbox support remote control serial protocol ?
08:26:40ameyerstrangely, I *think* my debian setup gives me raw disk access *shudder*
08:27:01Lloreanwrite_erase: What player do you have?
08:27:33write_eraseipod 3rd gen ... which has serial IIRC
08:28:05LloreanThere's no support for it at all in Rockbox.
08:28:52write_erase There is manual here: -> iPod 3rd gen
08:29:02write_eraseWhy not supported ?
08:29:06LloreanNo support for *serial* on it.
08:29:13DuchessAlright well, thanks for the help guys, I'll tinker with this for a while and if it doesn't work I'll give it up.
08:29:18*Duchess waves
08:29:20 Part Duchess
08:29:38write_eraseyou mean the rockbox firmware don't support serial remote control ?
08:29:47LloreanNot on the 3G iPod.
08:29:47 Quit denes_ (Read error: 110 (Connection timed out))
08:30:02 Join denes_ [0] (
08:30:11write_erasenot 3G ipod but ipod Gen3
08:30:25Llorean3G means "3rd Gen"
08:30:43write_erasedoes the remote works on other devices ?
08:31:03LloreanSome devices with remotes have support, yes.
08:31:15LloreanPlus there's some experimental work being done for newer iPods.
08:31:46 Quit gregorovius ()
08:33:04write_eraseWhere can I get the feature chart ? so I can see what' supported ?
08:33:24J-23What targets support lcd_set_{back,fore}ground?
08:34:31Lloreanwrite_erase: What feature chart?
08:34:47LloreanThere's just the features in the manual
08:35:07LloreanJ-23: I believe any that are more than 1bpp, but I'm not certain.
08:35:23J-23should it work on simulator?
08:35:40 Quit jhulst (Remote closed the connection)
08:35:48*J-23 thinks what is he doing wrong
08:35:50*amiconn points J-23 to the helpful wiki
08:37:11Unhelpfulwrite_erase: you could try here:
08:37:20***Saving seen data "./dancer.seen"
08:50:53J-23I set colors with lcd_set_{fore,back}ground, but I still all colors are as in theme
08:56:35 Join lucent [0] (
08:56:57lucentI own a Sansa Clip and Sansa Fuze 8GB now :)
08:57:17lucentis either hardware of interest to capable developers?
08:57:19Unhelpfulamiconn: it's been bumped off the bottom by all the bugfixing, but the gain was on the order of 4.5KB on color targets, and i think closer to 1.5-2KB on greyscale
08:57:23lucentI don't need both
08:57:34amiconnUnhelpful: My point is that we have smooth scaling, we should offer it on all targets that can afford it
08:57:47amiconnThe greyscale targets certainly belong to this group
08:58:19Unhelpfulyou think so? it wouldn't be hard to code that, a quick clone-and-hack job.
08:58:20amiconnAnd while the greyscale LCDs are merely 2bpp, you get a higher effective depth through dithering in the wps, but that's not my only point
08:59:00amiconnIf the bmp reader + scaler would allow a pluggable output stage, coverflow could use the greylib on greyscale targets, for an effective depth of ~7bpp
09:00:26Unhelpfulamiconn: doesn't yet, but that's a matter of adding a function pointer w/ the same signature as resize_on_load to bmp_read_*
09:00:31amiconnThe greylib currently has no bmp loader, a fact that I wanted to change for a long time already :/
09:01:00amiconnHmm, I'd prefer the data post-resizing
09:01:21amiconnI.e. load, smoothly resize, and then pass the data to the output stage
09:01:42Unhelpfulwhat i'm saying is you could have an 8-bit greyscale resizer in pluginlib
09:01:43amiconnThink of the greylib as just another pixel format
09:02:24amiconnActually the simplest one you can think of. Just 8bpp in natural order.
09:02:34Unhelpfulright now the inside loop of the scalers is direct-output in a target format... although, they do both generate whole lines at a time, we could add an output_line function...
09:02:48amiconnAH, hmm
09:03:05Unhelpfulsorry, actually, nearest doesn't, but you're talking about doing away with nearest, pretty much.
09:03:22amiconnNot sure whether it's worth having this in the core then, but imho it would then make sense to share the source file between core and pluginlib
09:05:03Unhelpfulwell, if we support smooth scaling on greyscale targets, there's really only one scaler, then. we can pass an output_line function in that copies lines to the target bitmap
09:05:30Unhelpfulthe one in core can be 2bpp or 16bpp, depending on target, and an 8bpp greyscale can be in pluginlib
09:05:38lucentweird, I see a patch supporting radio mode on Sansa Clip - I thought that rockbox wasn't able to run on the Clip though due to limited resources?
09:05:54 Quit gevaerts (Nick collision from services.)
09:06:05 Join gevaerts [0] (n=fg@rockbox/developer/gevaerts)
09:06:09Lloreanlucent: Rockbox is actually coming along nicely on the clip
09:06:30lucentwha? :) I would like to try, if it is "reasonably" safe to revert
09:06:40LloreanYou'll need to compile the tools and stuff for yourself.
09:06:43lucentmy Sansa Clip 2GB is now my 2nd player
09:06:51lucentyeah I could figure that part out
09:06:57lucentI'm surprised though
09:07:13lucenthow does the interface work on that tiny little display :P
09:07:26LloreanIt's still just a list of text, like it is on anything else.
09:07:31 Nick pixelma_ is now known as pixelma (n=pixelma@rockbox/staff/pixelma)
09:07:41lucentthanks for that info
09:07:57lucenthas it gotten far enough to play back audio?
09:08:21Unhelpfulwe could use exactly the *same* scaler if we always scale in color, and then convert to greyscale only at output... i wish i'd kept my multiply counts for the color scaler, i could try to figure which was more efficient :/
09:08:24Lloreanlucent: A little bit.
09:08:30*lucent smiles :)
09:08:30LloreanIt usually crashes after a short period of playing
09:08:53lucentif another user testing the code is handy, I will try
09:09:01amiconnUnhelpful: I'd think that scaling in greyscale would be faster
09:09:18Lloreanlucent: We need more people actually working on it. Investigating the code and figuring out where the crashes are happening, and why
09:10:22lucentoh, I'm a good bug tester but not much of a coder then
09:10:25Unhelpfulamiconn: well, remember, brightness costs 3 multiplies, though... and the smooth scaler doesn't actually have any per-input-pixel multiplies, right now.
09:11:13amiconnThat'd be an interesting comparison
09:11:29lucenttrying to find out how to check out the rockbox code that I would need to build to get a clip firmware
09:11:42lucentwhere on the rockbox website is that information mentioned?
09:12:16Unhelpfuli need to do some work here, but i'll be back in a bit, and recount those multiplies first :)
09:12:30Lloreanlucent: If you just follow the link to the wiki, the main page should contain links to an awful lot of useful pages.
09:13:08lucentLlorean: wiki, got it. thanks
09:13:47 Join n1s [0] (n=nils@rockbox/developer/n1s)
09:14:32 Join blkhawk- [0] (
09:17:12lucent is what I am looking for :)
09:21:24lucentLlorean: is Clip an ARM core?
09:26:06 Quit blkhawk- ("Caught sigterm, terminating...")
09:26:08jhMikeSn1s: I have a nice FM menu item on my beast but I can only listen to God on it (cosmic background radiation). :d
09:26:41n1sjhMikeS: hmmm
09:27:42n1si think i get worse reception on the beast then the h300 but it's hard to tell because fm is kind of weak here
09:27:44jhMikeShmmm, weird. I take that back, it's tuning now. :\
09:27:55n1soh :)
09:29:10jhMikeSStill, strange volume leaps when switching regions.
09:29:49n1sjhMikeS: that is the softmute, when it decides it is not tuned to a valid station it drops the volume by 16dB so it's actually a feature
09:30:34 Join GodEater [0] (i=c2cbc962@gateway/web/ajax/
09:30:49n1sbut it is kind of weird because it is supposed to set a bit to tell if you are on a valid station but that seems to never get set...
09:30:50jhMikeSBut it is tuned. It goes silent or quiet, then when I go back to the screen, it pops back up.
09:33:24jhMikeSn1s: Nice job on the tuner though. Don't let my probing for strange behavior say otherwise.
09:34:01n1sstrange, i don't think i've had that but it seems very position dependent so moving just a few cm can mean going fron clear signal to just noise
09:34:18n1sjhMikeS: thanks, and feel free to probe :)
09:35:17Unhelpfulamiconn: ok, the color downscaler is 9*dw*sh + 12*dw*dh multiplies
09:35:55n1si hope i can get proper seeking/scanning working soon but therere's so much to do before christmas :/
09:36:10Unhelpfulcalling brightness per output pixel would make that 9*dw*sh +15*dw*dh multiplies
09:37:12jhMikeSn1s: tuning is a bit sticky. you set the proper GPIO settings on the tuner?
09:37:23Unhelpfulconverting the scaler to work in 8bpp greyscale, and calling brightness on input pixels, means 3*sw*sh + 3*dw*sh + 5*dw*dh
09:37:40Unhelpfulerm, sorry, last term is 4*dw*dh.
09:38:27n1sjhMikeS: not sure really, from what i understand the gpios are only for interrupts from the tuner, which we currently do not use
09:38:28 Quit gromit`` (Read error: 110 (Connection timed out))
09:38:31jhMikeSif you didn't then that would explain why the MCU GPIO pins don't register correctly
09:38:55n1sor maybe i don't understand wht you mean :)
09:39:49Unhelpfulfor input dimensions 2x the output dimensions, that's 2/3 as many multiplies for color->grey->scale, instead of color->scale->grey
09:40:09jhMikeSn1s: You have to set the GPIO lines on the tuner in its registers to be functional was all I was saying. You're probably familiar with that alread.
09:41:15n1sjhMikeS: yeah, ok, as i said i only think they are used for interrupts, do you think setting them up will affect tuning?
09:41:38Unhelpfulat 4x, color->grey->scale is about 25% more multiplies... but i hope we're not going to make greyscale targets do that, anyway.
09:42:20jhMikeSn1s: so that loop isn't needed and it could get rid of the sticky feeling from the UI
09:43:12 Quit Seed ("cu, Andre")
09:43:27n1sjhMikeS: aha, yes it can indeed, bertriks patch in FS9611 does this and it works fine, or at least the same
09:44:15 Part tkooda ("thanks!")
09:44:22jhMikeShmmm. and we can avoid i2c traffic to check the mono/stereo bit.
09:45:03*jhMikeS listens to The Doors on 94.7
09:45:32LloreanSome stations wouldn't come in at all on the Clip radio patch, for me.
09:49:38jhMikeSn1s: we definitely need the volume ranges combined to match pcm. -54db is pretty loud still.
09:50:15ameyerFor the record, I just read some of the backlog and just wants to point out with the "read audio data from flash" (FS9322?) patch, the clip is reasonably stable, although there still appear to be some issues with the PCM driver and the sd driver
09:50:32n1sjhMikeS: yes, i just didn't know which volume to use :/
09:50:50n1swill look at it later, gtg now
09:51:53jhMikeSn1s: The volume attenuates the ADC after min is hit for the HP amp to get a 96dB range. It could actually be much larger than that.
09:53:47Unhelpfulhrm, all of the in-the-loop code would be cut to nearly 1/3 the size, too, i'd imagine... which would bring the smooth scalers on greyscale in line with the cost of the NN scaler now
09:54:50*jhMikeS wonders if a 190dB volume range on the beast would be excessive?
09:56:59 Join rvvs89 [0] (n=ivo@pdpc/supporter/base/rvvs89)
09:58:05ameyerjhMikeS: pretty much
09:59:27LloreanjhMikeS: Is most of that downward?
10:00:13ameyerif full scale is 0 dB, -186dB would be full scale * 2^-31
10:00:40jhMikeSLlorean: -184 to +6 if DAC+HP are combined
10:00:52LloreanjhMikeS: What's the current range?
10:00:57ameyer-190dB would be full scale*10^(-19)
10:01:07jhMikeSLlorean: -90 to +6
10:01:36LloreanI guess it more or less boils down to "which is most similar to what we do on other players?'
10:01:57ameyerand would be percieved as approximately full scale * 2^-19 IIRC
10:01:58jhMikeSLlorean: The beast is really a loud player though
10:02:03LloreanThis is true.
10:02:15LloreanI know the lower limit on the F is just a little higher than I'd personally like it to be, even with my replaygained music.
10:02:34LloreanI don't really see the harm in offering users ridiculous lows, since nothing forces them to go that far down anyway.
10:03:03ameyermaybe a bit of binsize or ram usage, although probably not
10:03:07jhMikeSThe line can go in effect -104.25 to +6 combined with HP so it's a good range
10:03:16Lloreanameyer: The beast is a beast in terms of RAM too.
10:03:51ameyerand I'd imagine it'd be miniscule anyway
10:04:02ameyereven on something like the clip with 2 MB
10:08:11jhMikeSLlorean: Really, I've wondered about limiting the maximum on the beast. It's just ridiculously loud at the upper end and my headphones aren't even terribly efficient.
10:11:12LloreanjhMikeS: People on other players ask us occasionally if we can make it louder.
10:11:26LloreanI say we give them the freedom to deafen themselves, if they feel they really must.
10:12:52UnhelpfuljhMikeS: don't forget car adaptors and the like - one of ours is quite quiet, and needs either the beast up to -10, or the stereo up well above where it's used for other channels.
10:12:56jhMikeSI blasted myself by mistake once though just because I forgot I had it cranked. :p
10:13:19LloreanjhMikeS: fixed.cfg =P
10:14:24 Join gregzx [0] (
10:17:34 Join lasser [0] (
10:21:08 Join bertrik [0] (
10:24:27*pixelma wonders whether fixed.cfg is described in the manual or at least in the wiki
10:24:52LloreanIt wasn't last time I looked, but that was some time ago.
10:25:12LloreanIt's one of those things that went from "first idea" to "final form" in like, 30 minutes of discussion in here.
10:25:13pixelmait is
10:25:25LloreanWith a few really bad ideas in between.
10:27:06pixelmain "Advanced topics - Specifcations for .cfg Files."
10:28:44pixelmaand weird... when I c-p parts of the manual. all lower case "f"s are lost
10:32:28LloreanWasn't there an issue like that previously?
10:33:10 Join bmbl [0] (n=Miranda@unaffiliated/bmbl)
10:37:19 Quit rvvs89 ("leaving")
10:37:21***Saving seen data "./dancer.seen"
10:38:20Unhelpfulamiconn: as far as brightness itself goes, i'm inclined to think we keep the existing version, with the div, on SH, where it's smaller code and we don't dither or have >1bpp output. we could probably go even more low-quality, really.
10:39:49amiconnDid you compare sizes on SH?
10:40:17Unhelpfulyes, >>8 is more code, i'm guessing the extra multiply is what puts it over.
10:42:21Unhelpfulthe >>4 version is smaller on arm than >>8, we should probably use that on PP targets if the main display is color, and the >>8 version on greyscale main-display targets
10:43:42Unhelpfuli don't know how that split works out on coldfire... we should probably stick to whichever is decent and smallest there, as well, *if* the main screen is color
10:44:49 Join esthar [0] (
10:45:42pixelmaLlorean: that was with the textsc function (small caps) which was used for settings (leading to a broken search in some pdf viewers and bloated html), bluebrother made the manual build with some other function and font and said this fixes it. For me all lower case "f"s are lost, no matter if they are in simple text, headline or else. When pasting into Notepad++ with enabled view of control characters I get an "FF" instead
10:45:54 Join MethoS- [0] (
10:47:06 Quit MethoS- (Remote closed the connection)
10:48:21 Join stoffel_ [0] (
10:50:36 Join seraph [0] (
11:02:29 Join AndyI [0] (i=AndyI@
11:13:08 Quit AndyIL (Connection timed out)
11:20:27 Join hobbs [0] (n=nandrew@p3m/member/hobbs)
11:24:05 Join JdGordon [0] (n=jonno@rockbox/developer/JdGordon)
11:25:26Lloreanbertrik: I tried FS 9611 briefly, and at least one frequency didn't seem to come in at all. Strangely ones both above and below it did.
11:25:36LloreanAnd it's one I can pick up quite well in the OF.
11:28:17 Quit gregzx ("ChatZilla 0.9.84 [Firefox 3.0.4/2008102920]")
11:28:55bertrikLlorean, weird. I don't have an explanation for that.
11:29:01 Quit reacocard (Read error: 104 (Connection reset by peer))
11:29:25bertrikI have the feeling I'm still missing something in the initialisation.
11:29:52 Join gregzx [0] (
11:30:07*amiconn has the impression that the tea5767 is a far better tuner than the sii one
11:33:43 Join reacocard [0] (
11:37:16bertrikI can't compare because the only other target with a tuner I have is a c200v1, which has a lvXXXX tuner
11:37:40bertrikat least the si470x is easy to use, the one in the c200 is a nightmare
11:38:25pixelmamy last week's comparison between OndioFM (TEA) and c200 (LV) was that the Ondio had a better reception
11:39:14pixelmaand the auto search function gave "cleaner" results
11:40:03pixelma(from a user's perspective)
11:40:11 Join fredddy [0] (
11:41:38 Join ender` [0] (
11:42:26 Join culture [0] (
11:48:55Unhelpfulamiconn: at a guess, i'd say greyscale scaler is maybe 1-2wk worth of spare-time work to get right, based on how the color scaler went. maybe less, because i'm just adapting, maybe more, because of holidays. but i'm starting on it in a branch, and we'll see where it goes :)
11:51:10 Nick fxb__ is now known as fxb (
11:51:59 Join Schmogel [0] (
11:54:53 Quit stoffel_ ("leaving")
11:58:02 Join moos [0] (
12:01:05 Join miepchen^schlaf [0] (
12:01:36 Join mofux [0] (
12:02:26 Quit _lifeless (Remote closed the connection)
12:02:37 Join __lifeless [0] (n=lifeless@
12:07:40bertrikLlorean, what frequency was that. On my clip, I don't hear anything around 96 MHz, not even noise. The received signal strength indicator shows a fairly strong signal. I can imagine that the radio is picking up a clip clock frequency or a harmonic thereof
12:10:24 Quit Schmogel (Read error: 104 (Connection reset by peer))
12:11:32 Join robin0800 [0] (
12:15:46 Join Schmogel [0] (
12:17:07Lloreanbertrik: 106.9
12:18:24 Join Thundercloud [0] (
12:26:00 Quit perrikwp (" ajax IRC Client")
12:26:19 Quit advcomp2019 ("Ex-Chat")
12:27:59 Join domonoky [0] (n=Domonoky@rockbox/developer/domonoky)
12:30:02 Quit tessarakt ("Client exiting")
12:37:24***Saving seen data "./dancer.seen"
12:44:14 Part jon-kha ("[IRSSI] so real, you'll wet yourself!")
12:44:30 Join jon-kha [0] (
12:44:44 Quit fredddy (Remote closed the connection)
12:46:03 Join fredddy [0] (
12:46:12 Quit Schmogel (Read error: 54 (Connection reset by peer))
12:50:37pixelmahmm, looks like upscaling AA still doesn't work on my M5 (it does in the sim) with r19410. Weird.
12:53:05*pixelma switches WPSs and tries what happens on downscale
12:54:54pixelmathat works
12:57:47robin0800bertrik: any chance you could update your sansa c200/e200 power patch −−- looks like it missed 3.1 freeze!
12:58:05pixelmablah, sorry my fault... forgot to rename the cover.50x50.bmp on target
12:59:07*pixelma goes to corner
13:09:24 Quit seraph ("Ex-Chat")
13:17:27 Quit miepchen^schlaf ()
13:20:36 Join advcomp2019 [0] (n=advcomp2@unaffiliated/advcomp2019)
13:21:20 Join AndyIL [0] (i=AndyI@
13:25:18 Join casainho [0] (
13:25:42casainhohello :-)
13:26:22casainhocan someone help me on build GNU Debugger? is there any script on RB sources to build it? like the ""?
13:26:50n1scasainho: GNU debugger as in "gdb" ?
13:28:06casainhoLlorean: thank you!!!!! :-)
13:28:37casainhoI will build it right now, since I have now my JTAG hardware and OpenOCD working :-) :-)
13:32:27 Quit AndyI (Read error: 110 (Connection timed out))
13:32:48 Quit jhMikeS (Nick collision from services.)
13:32:54 Join jhMikeS [50] (n=jethead7@rockbox/developer/jhMikeS)
13:36:26bertrikrasher, you shouted something like 'wfms' yesterday, was that a joke or am I completely missing something?
13:37:24 Quit Thundercloud (Remote closed the connection)
13:38:23moosbertrik: I think everyone wants it, there was first attempt in the past to make it working (JdGordon?)
13:39:19JdGordonyeah, I had a very very early attempt
13:39:47bertrikit reuses the wps for radio or something like that?
13:40:24JdGordoncant even remember how I did it (or if it even worked) other than I chated and filled in a fake id3 struct with the info
13:40:26JdGordonwhich wasnt very good...
13:40:44JdGordona wfms would be a very nice addition thouhg
13:41:26moosindeed, and I think everyone is ok with this
13:43:19bertriksounds like a nice job for Mr. Someone
13:43:28JdGordonwe need to decide if we want to reuse tags or create new ones, and do we do it as a seperate file or as a wps mode
13:43:33JdGordonsame as a rec screen wpos
13:44:04JdGordon+ how to handle statusbars in general (they want to be wps-ifityed also)
13:44:05moosand status bar
13:44:07LloreanJdGordon: I'd say just use the same .wps file.
13:44:13mooshehe to slow :)
13:44:24LloreanJust as a %fm conditional, so you can change a few things around for the FM screen.
13:44:58 Quit casainho ("ChatZilla 0.9.84 [Firefox 3.0.4/2008111318]")
13:45:13JdGordonthe downside with that is wps files for screens which on some targets have fm and some dont have alot of extra cruft
13:45:37LloreanWho cares? The memory's wasted already, since we pre-reserve memory for the WPS.
13:45:54JdGordonnot if it ends up needing us to increase that buffer size
13:46:16JdGordonalhough with some cleverness we can ignore bits which are only used for fm
13:46:16LloreanAllow a few of the ID3 tags to show station name and frequency, and the progress bar to graphically represent the "tuner" with 0% being the lowest available frequency and 100% being the highest. =P
13:46:38LloreanTags aren't going to make us increase the buffer size. New graphics are.
13:46:43 Join faemir [0] (
13:47:01*JdGordon does like the idea of having the entire interface specified in a single wps file with the position of the menu viewport there also
13:47:05LloreanAllowing it to reuse WPS graphics will probably save memory overall.
13:47:07n1shaving both in the same file means that we don't need to spin up the disk to go to fm which would be nice imho, unless we use seperate buffers for fm but that would be a waste
13:47:20 Join write__erase [0] (
13:47:40LloreanPlus, FM is in a sense "the music player" still, so having it encourage a unified look for the two is nice.
13:48:29write__eraseHi... rockbox rocks... But I realy nead the remote control feature. Any reason it's not implemented on ipod 3G ? I'd do the job if there are no particular reason
13:48:57JdGordonthere is a patch which enables this
13:48:57Lloreanwrite__erase: The reason is that nobody's done the work. It's not a simple job.
13:49:05LloreanJdGordon: Not for 3G
13:49:38write__eraseLlorean, but rockbox implements serial remote, ru
13:49:40write__eraseright ?
13:49:53Lloreanwrite_erase: For some players, sure.
13:49:54write__eraseThe problem is it's not enabled for 3G
13:49:58LloreanIt's not just a case of enabling it.
13:50:10LloreanYou also need to write all the drivers for it. It's not the same for every player.
13:50:12write__eraseyeah ... low level uart control :-)
13:50:29*bertrik agrees with Llorean so far w.r.t. the wfms
13:50:49write__eraseThere should be the code in linuxipod kernel
13:51:06 Join Aurix_Lexico [0] (
13:51:08LloreanWell, we'd love to see a patch for it.
13:52:30write__eraseNo indicator for battery charge on ipod3g... funny ...
13:52:31JdGordonbertrik: well, if we just reuse tags then it shouldnt actually be too hard to implement.. the problem is the rest of the radio code needs work to play nice with it
13:52:39JdGordonrecorder/radio.c is very messy imho
13:53:23Lloreanwrite__erase: Well, the 3G isn't common. We'd love someone new working on it.
13:54:24 Quit lucent (Read error: 110 (Connection timed out))
13:54:31write__eraseno svn ?
13:55:16Lloreanwrite__erase: Could you explain the question? "no svn?" doesn't have a context for me to know what you're asking.
13:55:48write__erasesorry I'm looking for the sources, through CVS, SVN or any source control protocol
13:55:56JdGordonbertrik: hmm.. it might not even be all that much work
13:55:57LloreanIt's in SVN.
13:56:22Lloreanwrite__erase: In the wiki index there should be a section "For developers" with lots of information.
13:59:00bertrikJdGordon, I'm not so familiar with radio.c yet, I didn't realise that radio required this much code. I think the interface towards the tuners looks nice, clean and simple.
13:59:50JdGordonyou do? it is?! or are you only taking about the actual hardware interface?
14:00:20JdGordonI started working on a patch to move more of that code out of apps into firmware but its not quiet finished
14:01:39bertrikI mean the interface in tuner.h
14:02:09JdGordonah yes, that part is nice
14:03:47bertrikwow, there is support for multiple tuners ...
14:04:04JdGordonnot at the same time though...
14:04:27 Join Raynes [0] (
14:05:24 Join _jhMikeS_ [0] (
14:05:24 Quit jhMikeS (Nick collision from services.)
14:09:40JdGordonn1s: where is your beast radio patch?
14:09:48JdGordonand the ams radio patch?
14:10:12*JdGordon searches gmail and finds them
14:10:20n1sJdGordon: beast radio went in a couple of days ago
14:10:24JdGordonfinds that the beast 1 was commited already
14:10:38JdGordonwhich is nice, didnt break my radio patch
14:11:41n1sJdGordon: I think that to get seeking working nicely i will have to extend the tuner interface a little, will that be nicely doable after your changes?
14:12:50JdGordoni guess so... right now im trying to remove the tuning logic from apps... so ideally it would just say "tune/seek to freq and let me know if its tuned"...
14:13:57JdGordon is the interface I have so far
14:14:19JdGordonwhich is mostly cut/paste straight out of radio.c
14:14:54JdGordonthat fm region sturct also will be removed from the .h as its only needed for tuning (which apps/ doesnt need)
14:14:54 Join herrwaldo [0] (
14:16:23n1swhat is the difference between radio_set_frequency and radio_scan_frequency ?
14:17:10JdGordonvoid radio_set_frequency(int freq)
14:17:13JdGordon curr_freq = freq;
14:17:15JdGordon tuner_set(RADIO_FREQUENCY, curr_freq);
14:17:17JdGordonbool radio_scan_frequency(int freq)
14:17:23JdGordon curr_freq = freq;
14:17:25JdGordon return tuner_set(RADIO_SCAN_FREQUENCY, curr_freq);
14:17:30JdGordonalthough looking at it, its not very nice :p
14:17:46JdGordonset should force the freq wheras scan would get the nearest station to it
14:17:50 Nick sajes_ is now known as sajes (n=sajes@
14:18:56n1sso scan starts at curr_freq and just goes to the closest station?
14:20:09n1shmm, then maybe this can work without changing the interface, how does radio.c find out which freq it scanned to?
14:20:43JdGordoncall radio_get_frequency or radio_get_station
14:22:18n1sdoes it do this currently or do you intend to make it do this?
14:24:23JdGordonthats the intention
14:25:30n1ssounds good, the si4700 seeking is basically: set a bit for seeking up or down, 3 different signal measures by which it validates a station and set it off
14:26:04n1sand when it finishes you can check if it decided it failed and at which freq it ended
14:27:07 Join XavierGr [0] (n=xavier@rockbox/staff/XavierGr)
14:27:17JdGordonhow do you know when its finished?
14:28:57n1sultimately i want to use an interrupt, but you can also check a "finished" bit
14:29:07 Join stoffel_ [0] (
14:29:50 Part Raynes ("Reality is that which, when you stop believing in it, doesn't go away")
14:32:52 Join bluebrother [0] (n=dom@rockbox/developer/bluebrother)
14:34:36 Join mcuelenaere [0] (i=mcuelena@rockbox/developer/mcuelenaere)
14:35:33JdGordonn1s: do you tihnk we could do that in a tick task for the other tuners?
14:37:28***Saving seen data "./dancer.seen"
14:38:47n1sJdGordon: i'm not familiar with how the other tuners work but i imagine it would be possible.
14:39:14 Join kugel [0] (n=chatzill@unaffiliated/kugel)
14:39:21JdGordonwould it be a good idea thouhg?
14:39:38*JdGordon wants to make the code for the radio screen as simple as possible
14:39:50 Quit XavierGr (Nick collision from services.)
14:39:59 Join XavierGr [0] (n=xavier@rockbox/staff/XavierGr)
14:39:59 Join Jaykay [0] (
14:41:11 Quit esthar (Read error: 60 (Operation timed out))
14:43:27n1sfor me, tell the tuner to scan, get interrupt/ check for "finished" status, read new freq, update screen seem like a good plan
14:43:41 Quit stoffel_ ("leaving")
14:46:02 Join esthar [0] (
14:47:00n1si don't know if a tick task is needed, a check; sleep loop would do fine i would think
14:47:19mcuelenaeregevaerts: thanks for the jz4740_usbtool.c fixes
14:47:29Jaykaywhat is the best way when i notice that something in the wikipages is wrong/out of date?
14:47:46JdGordonregister and fix the problem?
14:47:47n1sJaykay: fixing it
14:48:19JdGordonn1s: well, a tick task would mean apps could get the tuner scanning and then continue doing nothing for all tuners
14:48:25Jaykayok so i have to register myslef in the wiki (ß)
14:49:17n1sJdGordon: yes, maybe that is better
14:49:30n1sJaykay: you need to register to edit, yes
14:50:33*JdGordon would like a partial playlist viewer in the wps (and preset viewer in the fm screen)
14:50:57Jaykaythe registration page sasy i should ask for writing permissions, so now i ask for write permissions :)
14:51:00n1sJdGordon: would certainly be nice on portrait screens
14:51:31n1sJaykay: what's your wiki name?
14:52:20JaykayJohannesLinke if i did everything in the right way
14:53:43rasherbertrik: a (poor) joke
14:55:27n1sJaykay: ok, should have write permissions now, promise not to spam :)
14:55:41Jaykayok i promise :)
14:55:42 Quit robin0800 (Read error: 104 (Connection reset by peer))
14:56:16 Join Darksair [0] (n=user@
14:57:00 Quit Darksair (Client Quit)
14:57:25 Quit Aurix_Lexico (Remote closed the connection)
15:00:38 Quit kachna ("Konversation terminated!")
15:06:25bluebrotherhmm. How much additional room should be left on the device after installing? 1MiB? 0.5MiB?
15:06:41LloreanDepends an awful lot, I should think.
15:06:56LloreanYou won't need much at all for the .cfg file and the nvram.
15:07:10LloreanCompared to that though, if you want to use database or dircache it can get pretty big, right?
15:07:20bluebrotherwell, I'm adding a space check prior to installing for rbutil
15:07:28n1sbluebrother: we have a couple of nasty bugs triggered by the disk getting full when trying certain things...
15:07:52n1syeah, the database can take several megs
15:08:08bluebrotherand my idea was to compare free space with the extracted size. That is a bit off due to cluster sizes on the player.
15:08:40Lloreanbluebrother: Maybe warn if <15mb just in general.
15:08:54bluebrotherbut if I consider that we should have 1MiB left after installation it should be safe. But should we have a bigger safety?
15:09:16LloreanBetter to play it really safe, at least until Rockbox itself is able to deal better with running out of space, right?
15:09:22bluebrotherLlorean: could be done, but I think that would be a separate thing. For now I want to warn / abort during installation.
15:09:23n1sbluebrother: maybe let them install if it fits but if you have, say < 10MB free say "Very little disk space free, certain functions in rockbox may not work as intended" or something
15:10:13LloreanI think just a warning "Your player is nearly full. While Rockbox can still be installed, some features will create files, and if you run out of disk space may experience unpredictable failures."
15:12:47soapBagder: can I have the registration email for TWiki user BenZavala? I wish to contact him about a recently uploaded theme. He says "PM me" but leaves no publicly visible email address nor mention of forum / IRC nick.
15:14:03 Join fdinel [0] (
15:14:18 Quit jeffdameth (Read error: 110 (Connection timed out))
15:14:22Lloreansoap: All the album images used in RockFlow?
15:15:44 Join jeffdameth1 [0] (
15:17:01soap? No. Not if you're asking the question I think you are asking.
15:17:13 Join mcuelenaere_ [0] (n=mcuelena@rockbox/developer/mcuelenaere)
15:17:23soapJust a slight issue with the structural content of his theme. Nothing major at all.
15:17:42 Quit mcuelenaere_ (Read error: 104 (Connection reset by peer))
15:17:46 Join mcuelenaere__ [0] (
15:17:55soapUnlike the most recent 5G upload which broke half the rules.
15:18:12soap(at least he left a contact address)
15:20:34Jaykaywhy is there ? there isnt such a page for any of the other players....
15:20:45Jaykayand the first paragraph is out of date.
15:20:49n1sdoes/will the new theme site require some kind of account ore a valid contact email?
15:21:09soapJaykay: that was from early in the e200 dev cycle.
15:21:31Jaykayso this page should be deleted or what?
15:21:55JdGordonn1s: maybe, probably.. we'll find out in 4-6 years time :D
15:22:18soapIt should be labeled as a historical artifact - but I don't see a reason to delete.
15:22:33n1sJdGordon: :P
15:23:04domonokyn1s: if you help getting the new theme page up and running, you could decide if it will do that :-)
15:23:11Jaykaysoap: its useless^^
15:23:12soapMarcGuay had gone through and tagged many of the outdated pages - if you want do to so with this one you should look at how he did those other old ones.
15:23:38 Join kachna [0] (
15:23:51Jaykaycan you give me a link to one of those tagged pages?
15:24:01soapyou can search as easily as I can.
15:24:13soapand I do not know one off hand.
15:24:16n1sdomonoky: i have written exactly 0 lines of php in my life and is perfectly happy with the default theme/wps so i'm not sure i'm the right guy for that :)
15:24:27Jaykaybut i dont know what i should search for
15:24:37soapI just remember his massive cleanup efforts. I would also post a link on that page to SansaRuntime
15:24:44 Quit sajes (Read error: 104 (Connection reset by peer))
15:25:35 Join petur [50] (n=petur@rockbox/developer/petur)
15:26:21bluebrotherhmm. My current implementation has one flaw: it requires the space of the extracted files to be actually free, regardless if there will be files overwritten (like during an update).
15:26:50bluebrothershould I simply ignore this? The biggest archive would be doom afaics, and that has ~15MiB.
15:27:48n1scouldn't you just check the return code from the failing copy?
15:28:08bluebrothersure, but my current attempt is to check the available space before installing anything
15:28:33bluebrotherif I'd just look if space runs out during extraction I might have a partly extracted build on the player
15:29:59bluebrotheror even two mixed builds, if an update fails due to this. Which I imagine could happen if someone updates from a quite old build and the new one just needs that few bytes more he is missing
15:29:59n1syeah, since it's such small sizes anyway, i think it can be ignored but it doesn't feel entirely clean
15:30:55bluebrotheryep, same here ... but checking the space the extracted files already take on the device, if present at all, would be a bit of additional work :)
15:32:57 Join rvvs89 [0] (n=ivo@pdpc/supporter/base/rvvs89)
15:35:12 Quit mcuelenaere (Read error: 110 (Connection timed out))
15:39:59 Nick JdGordon is now known as JdGordon|zzz (n=jonno@rockbox/developer/JdGordon)
15:40:28 Quit Horscht (Read error: 110 (Connection timed out))
15:41:40 Nick mcuelenaere__ is now known as mcuelenaere (
15:43:56 Join Horscht [0] (
15:46:39 Quit soap (Nick collision from services.)
15:46:50 Join soap [0] (
15:47:28 Join Aurix_Lexico [0] (
15:47:32 Quit soap (Nick collision from services.)
15:47:42 Join soap [50] (n=soap@rockbox/staff/soap)
15:48:37gevaertsmcuelenaere: you're welcome. I was trying out some code analyser I found on the net, and it pointed them out (as well as dozens of false positives elsewhere)
15:50:10mcuelenaereahh cool, how's the program called?
15:50:10gevaerts_jhMikeS_: You need usb_storage_disconnect() (which wasn't there when you asked)
15:50:21 Join dfkt [0] (i=dfkt@unaffiliated/dfkt)
15:50:28gevaertsmcuelenaere: this was with cppcheck
15:55:06 Join {phoenix} [0] (
15:57:37 Join einhirn [0] (
15:59:51 Join einhirn_ [0] (
16:03:06 Quit faemir ("Leaving")
16:09:46 Join mcuelenaere__ [0] (
16:10:22 Join mcuelenaere_ [0] (
16:10:26 Quit mcuelenaere (Read error: 104 (Connection reset by peer))
16:10:31 Nick mcuelenaere_ is now known as mcuelenaere (n=mcuelena@rockbox/developer/mcuelenaere)
16:14:02 Join perrikwp [0] (i=4aa794a0@gateway/web/ajax/
16:18:32 Quit einhirn (Read error: 110 (Connection timed out))
16:28:09 Quit mcuelenaere__ (Read error: 110 (Connection timed out))
16:32:35J-23is it possible to use PCRE in Rockbox plugins?
16:33:16 Quit Jaykay ("ChatZilla 0.9.84 [Firefox 3.0.4/2008102920]")
16:34:43 Join LambdaCalculus37 [0] (n=rmenes@rockbox/staff/LambdaCalculus37)
16:36:30linuxstbJ-23: You need to find a C library that implements them, and compile that as part of your plugin.
16:37:29***Saving seen data "./dancer.seen"
16:37:32J-23so I can even use pcre.h from my distro?
16:37:52n1sJ-23: if your distro runs on your player
16:38:30linuxstbJ-23: Google tells me that PCRE is a regular expression library in C. So you need to port that library to Rockbox.
16:39:06linuxstbBut looking at it, I'm sure there are simpler regexp libraries if you really need one in Rockbox.
16:39:54J-23Doesn't FS #4755 use PCRE?
16:40:35linuxstbDoesn't look like it.
16:40:46linuxstb(based on the files modified/added by the latest version of the patch)
16:41:00J-23ah, somebody just posted pcre.h for MacOS
16:42:04linuxstbMaybe the PC-based conversion tool uses it.
16:42:45linuxstbBut that has nothing to do with using it in a Rockbox plugin.
16:42:51linuxstbWhy would you want regular expression support?
16:43:43J-23I ask just to know
16:44:38 Quit einhirn_ (Read error: 104 (Connection reset by peer))
16:51:42 Quit gregzx ("ChatZilla 0.9.84 [Firefox 3.0.4/2008102920]")
16:56:08 Join Schmogel [0] (
17:06:10 Join miepchen^schlaf [0] (
17:07:24 Join faemir [0] (
17:16:30 Quit LambdaCalculus37 ("Ka-chunka")
17:17:21 Quit n1s ()
17:33:03 Join lucent [0] (
17:33:13 Quit XavierGr (Read error: 104 (Connection reset by peer))
17:33:15lucentthanks @ ameyer
17:33:19soapregex playlist generator plugin!
17:34:54lucentanyone else running rockbox on Sansa Clip? could suggest to me which patches to look into?
17:37:17domonokylucent: all ams-sansa players are still under heavy development, so nothing for normal user, but if you want to help out fell free :-)
17:38:11lucentdomonoky: I had a Sansa Clip 2GB Red and recently acquired a Sansa Fuze 8GB Silver, so my Clip can become a sacrificial lamb ;P
17:40:07bluebrothera red clip? Are those the v2 clips?
17:43:12lucentI'm not sure, how would I determine this?
17:44:19lucentthe 2GB Clip that I own is running the OF V01.01.30A at the moment
17:44:28 Join Strife89 [0] (n=michael@
17:44:29bluebrotherdon't know. Maybe they have a "v2" marking on the back?
17:45:00lucentoh, the silkscreen wore off on my device a long time ago :P
17:45:19bluebrotherhmm, firmware version at least was the only indicator one could count on. Assuming the v2 also have a firmware v3 would make that hardware v1
17:46:51lucentYes, looks like it's a "revision 1 clip" according to sandisk sansa clip forums post
17:47:43domonokylucent: then try out what we already have in SVN.. and try to improve it. :-)
17:48:27domonokystatus for the different ams-sansas: :-)
17:48:57 Quit {phoenix} (Remote closed the connection)
17:49:10lucentdomonoky: do you suggest adding any additional patches for my "V1 Clip" like FS 9332 ?
17:49:44gevaertslucent: first build and install a standard build
17:49:51lucentokay will do
17:49:57lucentI need to set up my toolchain today
17:52:17 Quit __lifeless (Remote closed the connection)
17:54:35lucentI wonder what happens, I have all my music on the Clip in "MTP" mode... I have a copy of this music elsewhere but I am curious to know what happens to the MTP music when loading rockbox?
17:55:52 Quit bmbl ("Woah!")
17:56:57bertriklucent, I have a clip too but I don't know, probably nothing. I avoid MTP where possible.
17:57:59lucentbertrik: with the OF using MTP made "database refresh" faster
17:58:05lucentso that's my motivation for using it ;)
17:59:30bertrikah interesting. I was looking for a way to speed it up or skip it altogether
17:59:32lucentsome interesting things are in motion for Ubuntu users too, this might be offtopic for rockbox, but the VFS layer in Gnome is getting updated and should support browsing MTP devices as files and folders
18:00:24lucentso Rhythmbox's MTP support plugin would need an overhaul to directly access files on MTP players based on that new access scheme
18:00:51gevaertsThat is indeed off-topic here :)
18:00:52lucentthat's what I was set to work on until I heard rhythmbox is sort of running on the Clip =)
18:01:27 Join toffe82 [0] (n=chatzill@
18:01:35bertrikrockbox you mean?
18:03:08kugelpetur: what about ?
18:03:27lucentno, it was offtopic though, I meant Rhythmbox (Gnome destkop environment's default music player and software application for loading music onto MTP capable players)
18:05:11 Join Jaykay [0] (
18:05:39 Quit slact ("ChatZilla 0.9.84 [Firefox 3.0.4/2008102920]")
18:07:39 Quit fdinel (Read error: 110 (Connection timed out))
18:11:32 Join Strife1989 [0] (n=michael@
18:11:57 Quit Strife89 (Nick collision from services.)
18:12:05 Nick Strife1989 is now known as Strife89 (n=michael@
18:12:12lucentHmm... do I need to apply rockbox-multilibs-arm-elf-gcc-4.0.3.diff patch to build rockbox for the clip?
18:13:04gevaertsI think applies all patches itself
18:13:38lucentoh, I didn't want to install to /usr or /usr/local, I was preferring to keep everything in a non-root owned directory and change my environment variables beforehand
18:13:47lucentI'll go read that script now though
18:14:06gevaertsChange the script then. The destination directory is easy to change
18:14:07krazykitlucent, you can change a variable right at the top of to point it wherever you like :)
18:14:38lucentah! that is great
18:14:38peturkugel: just closed ;)
18:14:56 Quit tvelocity (Remote closed the connection)
18:15:10kugelone bug less \o/
18:19:55 Join karashata [0] (n=karashat@
18:22:36Jaykayi suggest closing of because hass more functions and is more up to date.....
18:24:45 Quit Strife89 ("Bye, guys!")
18:25:16Jaykaybtw how do i write those links that only the fs#xxxx is shown, as bbcodes in some forums?
18:28:02gevaertsJaykay: while FS #2999 does indeed seem (very) outdated, I'm not sure if both support the same targets
18:28:14 Join EspeonEefi [0] (i=eefi@STRATTON-SIX-FOURTY-SIX.MIT.EDU)
18:28:46Jaykayfs2999 supports only uslt tags, sncviewer much more (including uslt)
18:29:28gevaertsYes, but do both of them work on all players?
18:29:45toffe82where do I enable test hdd, don't remeber if it is the good name :)
18:30:21gevaertstoffe82: add test_disk.c to apps/plugins/SOURCES
18:30:37kugelgevaerts: going by SOURCES, the latter seems to compile on any target
18:30:41 Join mofux_ [0] (
18:30:47toffe82gevaerts: ok
18:32:19gevaertskugel: even the Player?
18:32:35 Quit mofux (Read error: 60 (Operation timed out))
18:33:07kugelor maybe not. I haven't applied, but it's next to vbrfix and viewer
18:33:13 Join gregorovius [0] (
18:37:32***Saving seen data "./dancer.seen"
18:38:10 Join casainho [0] (n=chatzill@
18:38:26casainhoBagder: are you there?
18:41:09 Join Thundercloud [0] (
18:41:13 Join BHSPitLappy [0] (n=BHSPitLa@unaffiliated/bhspitmonkey)
18:41:13 Quit karashata ("G'bye everyone!")
18:43:07casainhobluebrother: hello :-)
18:43:13 Join Kitti [0] (n=himka_co@
18:43:17 Part Kitti
18:43:26casainholinuxstb: hello :-)
18:47:49 Join Seed [0] (
18:47:58casainhobluebrother: can I know if you use debug?
18:48:11casainhoand what IDE you are using?
18:48:26bluebrotherif I use debug? What do you mean?
18:48:29casainhofor debug, I wanted to ask about JTAG
18:48:31bluebrotherand my IDE is vim :)
18:49:16bluebrotherwell, what's the exact question about JTAG?
18:49:59casainhobecause until now I just saw tutorials for ARM7 of people using JTAG debugging, GBD and Eclipse
18:50:09kugelkeyclick's nice now after the rework
18:50:23casainhoI got it working today :-) - here a screenshot:
18:50:45casainhobluebrother: so, do you use JTAG debugging?
18:51:11bluebrotheryes, but not for Rockbox. Though that doesn't make much of a difference
18:52:22lucentbertrik: how to go back to the OF on a Sansa Clip after Rockbox is loaded?
18:52:25casainhoso, do you use commands to OpenOCD?
18:52:32Jaykaygevaerts: gevaerts: now whats about fs#2999?
18:52:41bertriklucent, hold the home button while booting
18:52:44casainhobluebrother: or you are not using OpenOCD?
18:52:49bluebrotherwell, when I used gdb I used it directly.
18:52:51lucentbertrik: ah. Very nice
18:53:08bertriklucent, the "hold" button works too IIRC
18:53:24bluebrotherno, haven't used OpenOCD yet. Back the time I had an emulator for ARM I had a BDI2000 which connects to gdb directly
18:53:37J-23what makes e200v2 buttons work in bootloader, but not in Rockbox?
18:53:49bluebrotherplanning to get a JTAGkey for ARM though.
18:54:07casainhobluebrother: ah, okok −− I don't see to much information on RB wiki about this subject... maybe people at RB do not use JTAG debug because of missing connection to players...
18:54:16 Join karashata [0] (n=karashat@
18:54:17kugelJ-23: lcd and shared physical connections (that's what I assume)
18:54:35casainhobluebrother: I got one "cheap", from Olimex −− AE
18:54:52casainhobluebrother: I got one "cheap", from Olimex −− ARM-USM-TINY for about 55 euros
18:55:09 Quit karashata (Client Quit)
18:55:17casainhoI will now try to understand better to use debug :-)
18:55:19casainhobye bye
18:58:01bluebrotherthe JTAGkey tiny is ~30EUR
18:58:12bluebrotherwhich is the reason why I want to get that ;-)
18:58:15bertrik+ 30 EUR shipping I think
18:58:27 Join karashata [0] (n=karashat@
18:59:02bluebrotheryep, unless you have connections you can use ;-)
19:00:57 Quit moos ("have a good night")
19:01:31lucentbertrik: okay, and how to make the clip use I've made?
19:01:41lucentthat part is not so clear for me
19:03:06 Join Chesteta [0] (
19:03:53 Quit karashata ("G'bye everyone!")
19:03:56Chestetahi; im "the guy" who wanted to help figuring out the keymap of the e200v2... if anyone is willing to help me in the endevor id appreciate it :)
19:04:09lucentyay more ams-sansa people
19:04:39Chestetahaha ive been looking at the development thread twice a day for a month waiting for things to work
19:04:51Chestetanow i can actually help (hopefully)
19:05:03 Join Nico_P [50] (n=nicolas@rockbox/developer/NicoP)
19:05:26lucentI'm going to try rockbox out on my Sansa Clip 2GB v1 today (hopefully)
19:05:30Chestetalucent, sarcasm or you have one too?
19:05:51lucentChesteta: I own a 2gb clip v1 and a Fuze 8gb
19:06:15Chestetalucent: nice
19:06:18 Join karashata [0] (n=karashat@
19:06:37bertriklucent, did you already install the bootloader?
19:06:48lucentbertrik: ah, no I haven't seen documentation yet on the bootloader
19:06:56lucentbertrik: where do I look for this?
19:07:02Chesteta(sorry, answered lucents question)
19:07:39Chestetalucent: I have installed the bootloader if you need help
19:08:00lucentyes, where do I find out how to install the bootloader on a Clip?
19:08:36casainhocan someone please tell me where in Makefile of bootloader I can setup the Optimization of GCC? because I am doing debug with GNU Debugger and I am getting some errors and I think the problem is on otpikization
19:08:38 Quit freqmod_qu (Remote closed the connection)
19:08:52kugelChesteta, lucent: you build the bootloader file using the ordinary rockbox build system
19:09:03bertrikNobody wrote the documentation yet AFAIK. Putting rockbox on a Clip is not recommended for general use yet, so maybe we shouldn't even document it.
19:09:29kugelthen you pass the bootloader file, and (any) OF file to rbutil/mkamsboot/mkamsboot (which you need to "make" once before)
19:09:32domonokylucent: build a bootloader. merge it with a OF binary with mkamsboot (in rbutil dir) and copy the resulting binary with the right name to the player
19:09:48 Quit Nico_P (Remote closed the connection)
19:10:08 Quit mcuelenaere ()
19:10:32Chestetawow my irc just lagged quite a bit, got a full page of txt at once
19:12:07lucentstill not sure how to build the bootloader
19:12:18Chestetaheres how i did it: 1) download device firmware (say for example you wanted to update to the latest sandisk firmware) 2) build mkamsboot (navigate to that directory and run the build command)
19:12:19lucent../tools/configure −−target=bootloader or some such?
19:12:48domonokyjust run configure, it will guide you
19:13:25lucentdomonoky: I've built configure's "55) Sansa Clip" target... am I all set?
19:13:32Chesteta3) create a bootloader for your device (mkdir bootbuild-in root directory (instead of just build)) then run ../tools/configure and select your device and then select "B" instead of "N" (b for bootloader)
19:14:29Chesteta4) run a build command in the bootbuild folder. you should have a file in there with a .sansa extension; this is the rockbox bootloader
19:14:40 Join freqmod_qu [0] (i=quassel@2001:700:300:1430:213:d3ff:fee9:5ed0)
19:14:45lucentah okay I have that file from your step # 4
19:14:51lucentI'm starting to see how this works
19:16:20Chesteta5) copy the .sansa file you build in the bootbuild directory and the bootloader you downloaded in step 1 (the sandisk "stock" firmware) into the mkamsboot directory
19:16:48lucentcool. Now I'm following :)
19:16:58lucentthanks a heap Chesteta !
19:17:10Chestetathen you will run the command "mkamsboot [sansafirmwarename.bin] [rockboxfirmwarename.sansa] [outputfilename.bin]
19:17:15 Quit mofux_ (Read error: 110 (Connection timed out))
19:17:17Chestetano problem lucent :)
19:17:24 Quit Chesteta ()
19:17:28 Join mofux [0] (
19:17:36 Join Chesteta [0] (
19:17:36casainhobluebrother: are you there?
19:19:05Chestetalucent: it took me about 4 hours to figure that out :) its isn't ever stated in steps on the site... and i hadn't used the svn for quite some time
19:22:20lucent[ERR] No room to insert bootloader, aborting
19:22:21lucentSegmentation fault
19:22:25lucentniiiice... :P
19:22:32Chestetaoh; forgot: lastly, step 6) you will connect to your sansa (in MSC mode i believe) and put the newly generated file (i'd rename it to be the name that sandisk initally named it) in the root directory... once copied unplug your device and it will install the new firmware.
19:22:37gevaertslucent: careful, the rockbox.sansa from the normal build is not the bootloader
19:22:49lucentgood call
19:22:52lucentmy mistake!
19:22:59 Quit herrwaldo ("Konversation terminated!")
19:23:04bertrikand don't just unplug, unmount/safe remove first
19:23:12Chestetayea; lucent: be sure you use the bootloader build, should be 64k
19:23:45 Quit Thundercloud (Remote closed the connection)
19:24:21bertrikBTW, a failed bootloader installation can brick your device, there is no known recovery mechanism
19:24:25lucentmuch better now that I follow directions correctly
19:25:17bluebrothercasainho: yep ... just doing other things the same time
19:25:43Chestetalucent: bertrik is correct: with the e200v2 there is a way to open it up and recover the device (some contacts to trip on the circuitboard) but there isnt for the clip
19:25:51 Join herrwaldo [0] (
19:26:57bertrikwasn't there an issue with the sd card driver for ams sansas > 1 GB?
19:27:11lucentcool, bootloader appears to work
19:27:21lucent"*PANIC* SD:DATA TIMEOUT,"
19:28:14lucentbertrik: confirmed?
19:28:22Chestetadamn I got that a while ago; stopped happening lately though :)
19:28:42Chestetai have and 8gb too so idk
19:29:12Chestetalucent: you did build rockbox (normal build) and put .rockbox in the root directory right?
19:29:37lucentChesteta: yep
19:30:55domonokyimportant: as these ams-sansa targets are in a early state, we dont want bug-reports. fix them yourself :-)
19:31:03Jaykaygevaerts: whats about fs 2999?
19:31:11lucentdomonoky: noted, thanks
19:31:32gevaertsJaykay: I'm inclined to close it, but I'd like to wait for a few more opinions
19:31:38 Quit herrwaldo ("Konversation terminated!")
19:32:19Chestetais anyone willing to help me work on the keymaping for the e200v2?
19:32:19Jaykayok.... but when will you get them?
19:32:43gevaertsJaykay: when people get here :)
19:32:59ChestetaI have the device and some freetime but dont really khow where to start :)
19:33:16Jaykaywow, so i guess this task will keep open for the next few years^^
19:33:28kugelbertrik: I found a delay which doesn't make i2c_read_data() return -4 (returns 0 instead, as it should), but still no button
19:33:54 Join mofux_ [0] (
19:34:03Jaykayi think sometimes you should simply close a task even if you are not sure.... if anybody needs the patch than he will ask for reopening.....
19:34:52gevaertsJaykay: I mostly agree, but this is a part of rockbox I'm not very familiar with. If I haven't heard anything by tomorrow night, I'll close it though
19:35:14Jaykayah that sounds good :)
19:35:19domonokyChesteta: shouldnt the e200v2 just use the e200 keymap ? (so no real keamap work needed i would think)
19:35:22bertrikkugel, what did it read in that case?
19:35:53 Join midgey [0] (n=tjross@
19:36:13 Quit Jaykay (Read error: 104 (Connection reset by peer))
19:36:26 Join herrwaldo [0] (
19:37:02kugelbertrik: 0
19:37:37kugelbut I initialised btn with 0, so it might not have changed at all
19:39:01bertrikI think the data line was probably stuck at 0. The i2c read routine read a 0 and thought it was an ACK, then read more data which is not suprisingly also 0.
19:40:15kugelwell, funman made it be initialised with 0
19:41:15Chestetadomonoky: oh alright; in the developer thread it was mentioned about testing the keymap (by atomikpunk, reply #498). I was going to try and help with that but perhaps it has already been taken care of
19:44:00kugelbertrik: the hardwaremapping page reads that SDA is GPIOB1
19:44:02 Join Jaykay [0] (
19:44:12kugeland A0/A5 are SCL
19:44:47kugelbertrik: funman either confused those or used "wrong" variable names
19:49:03lucentChesteta: I think I'm stuck unless there's a patch for the > 1GB SD driver :)
19:49:47bertrikkugel, I wrote the entry about fuze i2c I/O in the hardwaremapping page... I may have made a mistake. Or maybe this code was linked in the firmware but is not actually used. Or we're overlooking something else.
19:49:48domonokylucent: you could shrink the partition to below 1Gb :-)
19:49:55lucentha, maybe!
19:50:20lucentdomonoky: there's no partition listed though
19:50:32lucentdomonoky: do you infer that I create one?
19:50:32Chestetalucent: I dont know of any; I know how to use the svn and use patches but as far as making them (actual programing) i dont, you just need to press the left arrow key while powering on to get to the sansa firmware :)
19:50:35 Quit mofux (Read error: 113 (No route to host))
19:51:11Chesteta(or atleast on the e200v2 it works that way...)
19:51:37gevaertslucent: I don't know if creating a partition table on the clip is a good idea, but just shrinking the filesystem should work
19:51:42Aurix_Lexico./util/MTP/sendfirm won't compile for me, it says there is a mismatch in the amount of arguments with a function from libmtp
19:52:00kugelbertrik: when I change it, i2c_read_data() returns -4 again...
19:52:44gevaertsAurix_Lexico: what version of libmtp do you have? It compiles fine here
19:53:33gevaertsI have 0.2.6
19:53:56Chestetalucent (cc domonoky): i tried using the windows disk management console and also paragon partition manager and the sansa e200 would not let me create a different sized partition... this is when I was getting the PANIC** (sd message)
19:53:57bertrikkugel, does the delay have to have a very specific value to return 0? i2c delays can be made very long and i2c should still work.
19:54:50casainhobluebrother: I would like to know how I can define -O0 (optimization of GCC) on bootloader makefile... because I am doing debug and I need it...
19:54:50kugelI added a rather long delay, really
19:55:11gevaertsAurix_Lexico: I can confirm that it fails with 0.3.*. Can you install 0.2.something?
19:56:01Aurix_LexicoI'll try
19:56:05bluebrothercasainho: you need rather -g to create debug symbols. Disabling optimizations helps during debugging too, but is not strictly necessary (at work we debug with compiler optimizations enabled. Can be confusing at times)
19:56:05 Quit miepchen^schlaf ()
19:56:46casainhoah, okok - so I see on makefile an option to do debug that uses the -g option
19:57:07lucentChesteta: I do have many files on my Sansa Clip 2GB, and then I added the .rockbox structure
19:57:11lucentmaybe it's just out of range
19:57:12casainhobluebrother: in what kind of company do you work? what devices you do?
19:57:40Jaykayi request committing of and
19:58:09Chestetalucent: oh; I erased everything (backed it up first) and then put rockbox on there... that is probably why its working now :)
19:58:24Jaykaythe first one clears the button queue after zooming in mandelbrot, the second makes aplles in snake appear immediatly
19:59:07gevaertsJaykay: not during the freeze
19:59:36gevaertsWhile those may be nice improvements, I don't think they qualify as bugfixes
19:59:37bluebrothercasainho: in the bootloader Makefile the CC option -g is used if DEBUG is defined.
19:59:48bluebrotherso I guess a make DEBUG=1 whould do it
20:00:02casainhookok - thanks
20:00:12bluebrotherthough I haven't tried :)
20:00:22kugelgevaerts: I think the second one is a bugfix. Or do you think it's intended to delay the next apple?
20:00:24Jaykaygevaerts: why not bugfixes?
20:00:50gevaertskugel: it could well be
20:01:07 Join {phoenix} [0] (
20:01:29gevaertsJaykay: both of them are at least debatable in my opinion
20:02:04Jaykayok then you can simply commit them since they are not really new features ;)
20:02:09Jaykayok i stop....
20:02:23kugelJaykay: btw: don't request committing. You may suggest it, but only if you accept to do the work that might need to be done before committing
20:03:16gevaertsThey are changes in behaviour for things that are (in my opinion) not clearly wrong. If someone else wants to commit them, fine, but I won't during the freeze
20:03:36Jaykayi thought the "i suggest..." is annoying because im writing it every time...... and i tried both patches and they work fine (the second after the sync, which i did)
20:04:13lucentChesteta: confirmed, now that I have done a "format" and reloaded the .rockbox directory structure, the rockbox firmware loads and runs correctly
20:04:31lucentChesteta: so there is a limitation in the bootloader that it must load from < 1GB , that's my guess
20:04:32Jaykaygevaerts: so ill suggest comitting after 3.1.... hehe
20:04:35Chestetalucent: nice! do your buttons work :) mine dont :(
20:05:15lucentChesteta: yes, buttons are working, it does have a panic and SD timeout though
20:05:23lucentI tried playing some music and got the panic
20:05:26 Join saratoga [0] (n=41becb3b@gateway/web/cgi-irc/
20:05:44gevaertsJaykay: sure :) We won't go wild with untested commits then either, but for those two, why not?
20:06:13Chestetalucent: i believe that you are correct about the bootloader limitation; as an experiment try putting over 1gb of files (songs) on your device and then add a few more (to be sure they are put in the +1gb section of the chip) see if you can play them
20:06:13bertriklucent, one very useful patch for ams sansa players at this moment if the buffering_flash patch, I'll look it up
20:06:17lucentChesteta: though "Play" is mapped to the center button
20:06:26Jaykayanother thing: is it allowed to post patches in the tracker who relates to other patches which arent in svn?
20:06:40Jaykay( for rockboy)
20:06:48bluebrotherJaykay: yes, as long as those are in the tracker.
20:06:53lucentbertrik: I applied the buffering flash patch, thanks!
20:07:17bluebrotheror at least, I don't see a reason why it shouldn't be allowed
20:07:22gevaertsJaykay: for those I'd post the patches to that specific task
20:07:32Aurix_Lexicodowngrading libmtp worked, I now have the rockbox bootloader on my zen vision:m and not much to do with it
20:07:46Chestetalucent: whats the fs number of the patch?
20:08:07Jaykaygevaerts: i didnt plan to post a patch..... i know as much as nothing about programming in c^^
20:08:27lucentChesteta: should be FS9332
20:08:29gevaertsJaykay: ah, ok. I slightly misunderstood you then :)
20:08:40Chestetathanks :)
20:09:13Jaykayso f2 4857 is out of date because rockboy isnt in the tracker?
20:09:49Chestetalucent: so can you play music?! :)
20:10:12saratogaJaykay: how is it out of date?
20:10:48bertrikoggs and wav play nicely, mp3s are known to skip/stutter a lot on the clip
20:11:20Jaykaymaybe out of date are the wrong words (as you may have noticed i cant write english...) but rockboy is not in svn (?) and not in the tracker so fs 4857 is not needed
20:11:23lucentChesteta: I don't have any music on there right now, let me try...
20:11:35lucentI wasn't really interested in playing music with it yet
20:11:36Chesteta... now if only my buttons worked... lol
20:11:52BagderJaykay: rockboy _is_ in svn
20:13:35Jaykayok i was confused because its not shown in the plugin section....
20:14:22lucentAre "##MUSIC#" and "##PORT#" hidden directories?
20:15:23bertriklucent, yes those are weird, rockbox sees them, but both windows and linux do not
20:15:26Jaykaybut there isnt anybody working on rockboy...?
20:15:43kugelsaratoga: which target have you referred to when you said you like the backlight fade (again, I'm glad you like it :) )
20:15:48BagderJaykay: nope
20:16:03lucentbertrik: I've heard rumors of ways to make those visible from Windows platform using flash drive recovery tools
20:16:25lucentbertrik: it's not normally iterated with the mass storage driver in Linux kernel
20:16:55kugelbertrik: I looked at the generic i2c, and i2c_read_data == -4 basically means that the sda pin is 0
20:17:04 Quit Chesteta (Read error: 113 (No route to host))
20:17:16 Join Chesteta [0] (
20:17:22saratogakugel: sansa e200
20:17:28Chestetasorry... wifi issues
20:19:50saratoga Jaykay: you can see whos working on files by looking at that SVN link i posted above
20:19:58lucentChesteta: confirmed, music plays back
20:20:16lucentChesteta: I'm still getting frequent panics with the SD issue mentioned before, and the database
20:21:20*bertrik hasn't seen any sd panics on his 1GB clip in the past week
20:21:59Chestetalucent: is that after putting 1+ gb of data on the device?
20:22:21lucentno, did a "format" using utility in OF
20:22:32bertrikkugel, no, -4 is returned when the i2c slave does not ACK the device address, which means it didn't send a 0 back after receiving the address
20:22:36lucentthen added .rockbox/* and a single mp3 in the MUSIC dir
20:23:08Chestetaok... I was also wondering if we could partition our devices into multiple small drives
20:23:30Chestetawould be a pain to navigate :)
20:23:32kugelbertrik: getack returns 0 "if (iface->sda())";
20:23:45gevaertsChesteta: wouldn't help anyway. The problem is at a lower level
20:23:48saratogai think the problem is that the SD driver can only read one bank of the SD card
20:24:00saratogapretty sure the rockbox fat driver is workign fine
20:24:09ChestetaI tried as i stated before but with no luck... there may be other partition tools that work better
20:24:10kugeland sda() is "return SDA_PIN"
20:24:41lucentsaratoga: Clip has.... an embedded SD?
20:24:49lucentI'm confused about that part
20:24:57kugelbertrik: (and i2c_outb returns getack, and if getack is 0, i2c_read_data returns -4)
20:24:59Chestetasaratoga: so if we format in fat (instead of fat32) it may work better?
20:25:21bertrikyes, there seems to be some kind of NAND flash <-> SD controller built-in
20:25:29kugelsaratoga: yep, it looks like bank switching needs to be implemented, I surely will have a look as soon as buttons work
20:25:47lucentkugel: which player do you have access to?
20:25:55kugelfuze 4gb
20:26:13lucentit's scary to try a bootloader without knowing if the buttons will work
20:26:53kugelthe buttons work in the bootloader. Besides a faulty bootloader can't break the player, since the dual boot isn't handled by the bootloader
20:27:22Chestetalucent: you got it working once before :)
20:27:47lucentI've got an 8gb fuze and would be willing to test any work you put in on the SD driver
20:28:11bertrikkugel, so sda() returned something else than the expected 0 and caused getack to return 0, which is interpreted as failure. So sda() was high, not low as was expected.
20:29:00kugelbertrik: ah right, I missread
20:29:26saratogaChesteta: no the fat driver is working fine, hence it doesn't matter
20:29:40saratogawe've been using it for years on targets with 10s of GB
20:29:46lucentChesteta: fat driver probably covers all "versions" of FAT
20:29:49kugelso our sda_pin is high. but they key is that the sda_pin doesn't change upon button press I guess
20:29:57Chestetasaratoga: oh right on, understood ;)
20:30:10Jaykayive got a question: why are there sometimes revisons whose diffs are only showing -no changes-? (as the last revision)
20:30:36 Join admin1lbo [0] (
20:30:55bertrikwhite space changes or svn property changes I think
20:31:06kugelbertrik: why does it sda_hi() after checking sda() ?
20:31:21Jaykaywhat is svn property?
20:31:45rasherJaykay: r19423 only changed whitespace
20:31:55Jaykayok thanks
20:31:59admin1lboquick question how do you install rockbox on an sansa e200 with the utliity in fedora 10
20:32:41bertrikkugel, I don't know. I think the generic driver is a bit weird.
20:33:11kugelI'd think it should be done before, and the sda() checks if it worked
20:33:24kugelbut I haven't written it and I don't understand it at all
20:33:39bertrikkugel, are you familiar with i2c at all?
20:33:44kugelnot at all
20:34:18 Quit faemir ("Leaving")
20:34:32kugelbut returning 0 if it's high and then setting it high is surely weird
20:37:36***Saving seen data "./dancer.seen"
20:37:57bertriksda is a bidirectional pin in the protocol, so master and slave can both set it. In this case, first the master checks the sda state, produces a clock edge, then sets it high.
20:39:20bertrikthe generic i2c driver is a bit weird because it explicitly set sda as output and then outputs a high level, this could produce a bus level conflict if the slave tries to drive it low at the same time.
20:41:04bertrikin the i2c protocol that I'm familiar with, the SDA and SCL lines are never set to a "hard" 1 and can only output a hard 0. Setting the pins to high-impedance mode (inputs) allows a set of pull-up resistors to pull the lines high.
20:41:58kugelbertrik: "could produce conflicts" doesn't sound good
20:42:20kugelmaybe you can rewrite it a bit so it meets your knowledge and I see if it works better?
20:43:37 Join faemir [0] (
20:44:09 Join n1s [0] (n=nils@rockbox/developer/n1s)
20:45:36 Quit mofux_ (Read error: 110 (Connection timed out))
20:45:58 Join mofux [0] (
20:49:06 Quit saratoga ("CGI:IRC (EOF)")
20:50:48kugelbertrik: weird that I get -4 no matter of the delay if I change scl/sda accordingly do teh hardwaremappings page, isn't it?
20:51:15kugelbut neither result in a button detection so...
20:53:35Jaykaytwo questions regarding the lang files: is there any sense in writing sometimes sound_settings and sometimes sound settings (without _)
20:53:40Jaykayand what is superbass?
20:53:58admin1lbohelp need a way to install rockbox on my sansa e200
20:54:45Jaykayadmin1lbo: download rbutil
20:54:55admin1lbotried it and it doesn't do diddly
20:55:06lucentadmin1lbo: bootloader piggybacks onto existing firmware
20:55:10Chestetaadmin1lbo: do you have an e200 or an e200v2?
20:55:24admin1lboand bootloader wont install or run
20:55:30gevaertsadmin1lbo: so what exactly goes wrong?
20:55:38admin1lbotried both the utility and doing it manually and nothing
20:55:39admin1lboit wont boot
20:55:51gevaerts"nothing" is a it vague
20:55:54admin1lborelease 3.0
20:56:02admin1lboit wont boot
20:56:28*domonoky thinks rbutil shoudnt do "diddly" whatever that is :-)
20:56:37gevaertsSo does it show anything on the screen?
20:57:22gevaertsSo why didn't you say that?
20:57:35admin1lboI did I quote' it wont boot'
20:57:43Chestetabut it does boot
20:57:48Chestetait boots into the original firmware
20:57:58admin1lbok you got me there
20:57:59gevaertsAnd now you say it _does_ boot into the OF. Please make up your mind before asking for help
20:58:02lucentChesteta makes an excellent point
20:58:02Chesteta"it doesnt boot" could mean the device does not power up
20:58:19lucentI have to run and go to work
20:58:26Chestetapeace lucent
20:58:26lucentthanks for all the advice and help today, folks
20:58:49 Quit admin1lbo (Remote closed the connection)
20:58:58lucentRockbox is really neat, I am anxious to make it my daily firmware for music playback :)
21:00:56Jaykayis anybody so nice and answers my questions? ;)
21:01:26 Part denes_
21:02:33*bluebrother wonders what diddly means
21:02:47rashermidgey: Not too happy with the way Turkish turned out, but I think that's all there was in it
21:03:02midgeybetter than nothing
21:03:14rasherJaykay: Do you mean in the desc field? That doesn't matter - it's simply a comment
21:03:18lucentbluebrother: usually it's vulgar and refers to sex, but in the silly or obscure context of "nothing meaningful"
21:03:48lucentand with that I'm out... ;)
21:03:49rasherJaykay: Superbass is a feature of the archos sound decoding wotsit that adds "Super bass" to the sound
21:04:50Jaykayrasher: when i compile rockbox for i.e. the sansa e200 i guess the "translation" for superbass is still in the lang file?
21:05:48rasherJaykay: Nope - only the targets which have the "masf" feature get that in their lang file
21:06:07Jaykayso same thing with mdb?
21:06:41rasherJaykay: Yup - the text to the left of the colon denotes which targets include that phrase
21:06:44 Join maddler [0] (n=maddler@
21:07:10Jaykayah i see
21:08:27n1spetur: doesn't the recording screen still fall back to sysfont when the text doesn't fit on the display?
21:11:01 Join miepchen^schlaf [0] (
21:11:04midgeyrasher: did I miss something in the Serbian update?
21:12:19rashermidgey: you could've deprecated LANG_RESTARTING_PLAYBACK
21:12:39Jaykayim sorry if im annoying.... why is fm-radio translated with ukw-radio in german?
21:12:43Jaykayits the same
21:12:44midgeybah, i'll fix that
21:16:37pixelmaJaykay: why not? I believe UKW is (or at least was) the more common abbrevation, just saying - I didn't translate that myself so can't tell you the reason but I don't see any need to change it
21:16:41rasherJaykay: you'll have to ask one of the germans about that. Jörg Hohensohn did the translation, but he's not around here much.
21:17:59Jaykaypixelma: i think fm is the more common abbrevation^^
21:18:10Jaykayand im german:)
21:18:21Jaykaybut ok there is no need in changing this
21:22:16 Quit herrwaldo ("Konversation terminated!")
21:24:36pixelmaJaykay: but don't let that stop you to suggest improvements, I know some things could possibly be translated better
21:25:07 Join thegeek [0] (
21:25:43Jaykayok so my suggestion for improvement is to change ukw-radio in fm-radio^^
21:26:10Jaykaybut since its the same its equal....
21:26:29Jaykayi mean there is no sense in changing this
21:26:32 Join herrwaldo [0] (
21:27:15 Quit gregorovius ()
21:28:00pixelmaI mean in general :)
21:29:40Jaykayanother question(...) why are there so much "prhases" in the lang files which are deprecated? why dont you delete them?
21:31:10 Join einhirn [0] (
21:31:10*bluebrother has heard UKW much more often than FM in german
21:31:11domonokydeleting them make it incomaptible to previous versions, especially for older voice files.
21:31:53 Quit casainho ("ChatZilla 0.9.84 [Firefox 3.0.4/2008111318]")
21:32:56*Jaykay still thinks that fm is more often used, but he also thinks its not worth discussing about it
21:33:05Jaykaydomonoky: thanks^^
21:33:28bluebrotherJaykay: you could ask wikipedia about that. I bet those guys know best ;-)
21:34:58Jaykayi already did, but they dont say how often wich version is used ^^
21:35:05 Join nk [0] (n=44dd351c@gateway/web/cgi-irc/
21:37:29 Join Zagor [242] (n=bjst@rockbox/developer/Zagor)
21:38:37Jaykayso my cd player says fm (yes he can play radio;))
21:39:16domonokymay sansa just says "radio" :-)
21:40:46Jaykaynot bad^^
21:41:08Jaykaywhich language?
21:41:59rasherNot a bad idea - the FM is more or less redundant
21:43:02Jaykaybut actally there isnt any phrase whish just says "radio"....
21:43:07Jaykaybut doesnt matter
21:43:42Jaykayi agree with rasher, in fact fm and ukw is useless^^
21:44:44 Quit mofux (Connection timed out)
21:45:02 Join mofux [0] (
21:45:19pixelmathe DAB equipped targets (some D2s and the Logik DAX) also only have DAB exclusively right?
21:45:54pixelmajust making sure, if so I agree that the "UKW" could be dropped
21:46:01 Quit Chesteta (Read error: 113 (No route to host))
21:47:04*kugel votes for "Rundfunk" :)
21:47:23*Jaykay agrees with kugel
21:48:08Jaykayso kugel what is more used: fm ore ukw? :)
21:49:26kugelmy radio says FM, but I'm used to both. I don't feel strong about either
21:55:50dfktfrequency modulation or ultrakurzwelle? tough choice on a global scale...
21:56:49rasherI'll bet FM works in a lot of languages
21:57:05pixelmawould work in German too
21:57:20pixelmaor does ;)
21:57:22 Join ZincAlloy [0] (n=d9eefff6@gateway/web/cgi-irc/
21:58:57Zagor5 second backlight timeout is a lousy default
21:59:18rasherIt seemed a bit fast for me
22:01:07Zagorit's way too short. 30 is reasonable."backlight" on many targets pretty much means "display". when the backlight goes out, you see nothing. hence it's a power mgmgt mode, which should only kick in after user idles a while
22:01:43kugelalso, new users seem to be challenged with that short timeout
22:02:00rasherI whole-heartedly agree
22:02:21Jaykayi would think 15 is better
22:02:22 Quit mofux (Connection timed out)
22:02:23Jaykaynot 30
22:02:30rasherMaybe the default should be dependent on READABLE_WITHOUT_BACKLIGHT (don't we have a define like that?)
22:02:32kugelI can understand. They try to read the main menu and have hardly time to read
22:02:44 Join mofux [0] (
22:03:33kugelrasher: like I introduced in this patch? FS #8523
22:04:33rasherkugel: Ah, that might be what I was thinking of.
22:04:53Zagoryeah I don't think we have anything like that in svn
22:05:05 Quit nk ("CGI:IRC (EOF)")
22:05:15kugelNope, that's why I introduced it :)
22:06:39rasherkugel: have you tried a battery bench with and without this patch, for example with caption backlight on to simulate backlight usage during playback?
22:07:23kugelrasher: No, but there was a report in another task that "showing wps" vs "showing main menu" makes up a 15% difference in runtime
22:08:00kugelwith backlight off all the time, that is
22:08:18Jaykaynot bad....
22:08:39Jaykayi guess aeverythings updated in the wps even if the bakclight is off
22:08:45Jaykayincluding the statusbars?
22:08:49kugelJaykay: hence the title of that task
22:09:17kugelthe wps is updated since there're targets where the screen is readable without backlight
22:09:19bluebrother"Rundfunk" does not include TV?
22:09:40 Quit n17ikh|Lappy ()
22:09:50Jaykayi think this was a joke, rundfunk is really old-fashioned.....
22:09:54kugelbluebrother: hmm, right. I guess ZincAlloy's "Hörfunk" matches even better
22:10:22bluebrotherurgh, that sounds even more 50s
22:10:33kugelyes :)
22:10:48bluebrother"MP3-Abspieler" :)
22:11:00peturn1s: yes, but only in cases where you have a very very large font selected (insanely) compared to your screen size. So I wouldn't care about it using sysfont at that moment, it is just better than showing a splash
22:11:34Jaykaymp3 is still to new for abspieler^^
22:11:40rasherpetur: what does it use the sysfont for? Displaying text, or numbers?
22:12:21Jaykaykugel: i thought the patch only disables chaning of alternating textes?
22:12:22bluebrotheris a READABLE_WITHOUT_BACKLIGHT really necessary? Are there any targets with color display around that are actually readable without backlight and non-color targets that are unreadable?
22:12:58kugelJaykay: No, I think the title tells what it does
22:12:58bluebrothersomeone should make a "old german" translation ;-)
22:13:03rasherbluebrother: I think someone reported the Nano to be readable, but I'm not sure
22:13:08Jaykaykugel: ok
22:13:12rasherCould be another target I'm thinking of
22:13:22bluebrotherrasher: we could ignore the nano :P
22:13:22midgeyisn't the video readable too?
22:13:26Zagorbluebrother: don't some ipods have transflexive screens?
22:13:29kugelwhy should a patch aim for disabling alternating sublines?
22:13:43pixelmayes, the Ipods with colour screens are quite readable without backlight
22:14:01Jaykaykugel: i think you mentioned the reason already, battery runtime?
22:14:53bluebrotherZagor: I don't know, hence the question ... but I only experienced color -> unreadable, non-color -> readable
22:14:56pixelmaVideo, Nano, 4G Photo/Colour (although I didn't find the latter as good as the two former but could only compare from memory)
22:15:34kugelZagor: Hmm, getting back to that patch, I'm under the impression that this should be wanted in svn?
22:15:51Jaykaybut withoud the fast-changing lines....
22:15:54Zagorkugel: the patch or the define?
22:15:58kugelJaykay: of course..
22:16:05kugelZagor: both
22:16:12Zagorkugel: which patch is that?
22:16:34kugelit needs work
22:16:40kugelbut from the idea, I mean
22:16:48pixelmaIMO it would be a real pitty to turn the LCD off on the Video/Nano without backlight
22:16:48rasherIt seems like something that should be done
22:17:05Zagorkugel: yes, I like the idea
22:17:43kugelbut I think the check shouldn't be for backlight, but rather for lcd_enabled
22:17:46Zagor...if we can show that wps updating actually costs anything
22:18:34bluebrotherdid someone make measurements about the power saving when disabling the lcd?
22:18:39Jaykayis there any sense in disabling the lcd completely after maybe a minute without backlight?
22:18:41Jaykayon players with those not readable lcds....
22:19:02Jaykaybluebrother: i dont think so, if its appreciated i would do this....
22:19:12bluebrotherI'm wondering if it really makes much of a difference.
22:19:20ZagorJaykay: on unreadable lcd:s it would make more sense to disable instead of just turning off the backlight
22:19:40Zagoragain, if we find it makes a power difference
22:20:36Jaykayzagor: but the lcd is not disabled right?
22:20:56Jaykaythis should save even more power
22:21:03kugelJaykay, Zagor: Targets where the display is unreadable without backlight do already turn of "lcd_enable(false)" in their backlight_off function
22:21:04pixelmadisabling the LCD is already done on some of those unreadable ones (on the Sansas and X5 for sure)
22:21:06Zagorkugel: thanks, I missed that comment. 15% is a lot.
22:21:16bluebrotherthat sounds a lot. Having actual current numbers would be still interesting though.
22:21:33kugelZagor: and I heard a very similar report some time ago
22:22:07pixelmabertrik benchmarked runtime on a c200 (IIRC) before committing the LCD disable
22:22:10kugelsome time as in a few month, way after that comment
22:22:41kugelpixelma: the disabling doesn't stop the wps updating though
22:24:05bluebrotherwhich leaves the question how much power the wps needs, and if / to what extend that depends on the theme used
22:24:41Jaykayanother thing: is there any way to disable the database rebuild on the e200 without connecting it with a computer?
22:25:15Zagorbluebrother: it probably is very depending on the wps used.
22:25:37bluebrotherZagor: that's my guess too.
22:25:40Jaykayits kinda annoying if you just want to check something out in the OF but changed a file with rockbox
22:25:54kugelprobably thinks like scrolling also affect it
22:25:58pixelmakugel: I know, but thought one question was also about power savings from turning the LCD off
22:26:01ZagorJaykay: OF is annoying per definition :)
22:26:15 Join jhulst [0] (n=jhulst@unaffiliated/jhulst)
22:27:30Jaykayzagor: i agree^^ but i think rockbox should make comparisons with the OF as easy as possible, because the finding of resons why rockbox is better will be easier....
22:27:56ZagorJaykay: well we can't change how the OF works
22:28:54Jaykayrockbox can prevent the OF of e200 from rebuilding the database when the player is connected with a computer....
22:29:41ZagorJaykay: yes, but that is an exception.
22:30:02Zagorwe managed to find some bits to fool the OF into skipping that rebuild
22:30:06Jaykayback to the wps updating: shall i do some battery benchs with(out)the patch and different themes?
22:30:30rasherJaykay: that'd be grand. Try a completely blank wps as well
22:30:44JaykayZagor: so it would be easy to do that i.e. when a second button is hold down....
22:30:56Jaykaythis would make some things easier
22:31:18Jaykayrasher: what do you mean with a completely blank wps, and where do i get this?
22:31:43bluebrotherjust use an empty wps file. It should show simply nothing :)
22:31:48kugelthat test could take some time though :)
22:32:07kugelI think main menu vs cabbiev2 is most interesting so start with that please :)
22:32:14pixelmaa totally empty .wps will show the status bar
22:32:34rasherJaykay: A file with simply "%wd" on the first line
22:32:54Jaykaypixelma: not when the statusbar is disabled right?
22:33:41pixelmaah true, it respects the global setting. Didn't think of it as I never disable the status bar in menus
22:33:45kugelJaykay: you may use a .ape music file, to shorten the whole test
22:34:55Zagorkugel: well that will decrease the significance of wps differences
22:35:30Jaykayi agree with zagor, additionally i dont have music in .ape^^
22:36:51kugelZagor: but using mp3 or flac will not show reliable results either, as they need less cpu than unboosted
22:37:08kugelso the wps updating is basically free
22:37:38***Saving seen data "./dancer.seen"
22:38:21ZagorI'd say the best test is showing the wps, but pause the music .
22:38:28Jaykayhow about mp3's with 320kbits?
22:38:45Jaykayzagor: this would need more than a day..... for a single test
22:38:45Zagorwe want to know how much difference different wpses do, nothing else
22:38:48UnhelpfulZagor: that could take a while to show real drain...
22:39:11*rasher points Unhelpful at FS #9631
22:39:21kugelZagor: the cpu runs at 30MHz at minimum. Everything that doesn't need 30MHz will not show any difference I think
22:39:24bluebrotherrasher: sudoku_generate takes a char* as second argument but you pass it a const char*. Thus you discard the const qualifier. Make sudoku_generate take a const char* and everything is fine
22:39:55kugelZagor: on the e200 that is
22:40:01pixelmabut when music is paused there will be no progress bar or time display updates (maybe also a factor who knows)
22:40:12kugelUnhelpful: I don't know if's related, but I have a particular album art which data aborts in pictureflow, but not in the wps (cabbiev2)
22:40:49pixelmanot to mention peak meter when someone uses it
22:41:18Zagorboosting is done in specific places in the code, is it not?
22:41:22kugelthe test music needs to drive the cpu a bit, else it doesn't add cpu load
22:42:12kugelat least I think so.
22:42:45Bagderunless we "sleep" the chip when we can
22:42:54Bagderwhich we at least used to do
22:43:55Jaykaykugel: cant the cpu switch to 24mhz? i saw it somewhere im entwicklerbereich (debug)
22:43:56ZagorI don't see how wps changes can affect power consumption at all, unless the lcd activity uses more power
22:44:17Zagormore wps cpu use doesn't affect cpu_boost() frequency
22:44:18rasherbluebrother: ah!
22:44:26kugelJaykay: 30 and 80 are the performance levels
22:44:52Zagorunless, as Bagder says, we sleep away idle time
22:44:52Jaykaykugel: when you press select in cpu frequenzy, it switches to 24mhz
22:45:13Zagorin which case I maintain that cutting the codec out of the test will yield much better accuracy
22:45:13bluebrotherZagor: copying data (especially graphics) around for display?
22:45:27kugelZagor: well, when the cpu needs to calculate the updates, it can't sleep
22:45:30Jaykaybut why doesnt the cpu use 24mhz when it has nothing to do?
22:45:47kugelyea, you're probably right then
22:45:57 Quit karashata ("G'bye everyone!")
22:46:09BagderJaykay: because the clock switch isn't free so it's about finding the "best" compromise
22:46:41kugelJaykay: probably because someone has chosen 30MHz when he did the port
22:46:47Bagderand we only have one boost level
22:47:04Bagderkugel: not only that, it has also been tested and proven
22:47:50kugelBagder: 24MHz works well too though. But it's way too slow without some gui boost implementation
22:48:09Bagdersince for example the gui doesn't boost, we need speed enough to do a non-sluggish gui
22:48:13rasherI wouldn't call that working well
22:48:32BagderI agree, it's a weird definition of "workign well"
22:49:06Bagderpersonally, I'd like interaction to cause boost
22:49:12ZagorI think he means it's been tested and works, but is too slow for the gui
22:49:34Bagderright, but I meant tested and works fine for sound + ui and the rest
22:49:41kugel"well" as in the hardware does that well (not counting the software in). Also, the only situation I tried 24MHz was with gui boost
22:49:42bluebrotherurgh. Updating from Qt 4.4.0 to 4.4.3 increases the rbutil binary by about 1MiB :(
22:49:43Zagoryeah, I know :)
22:50:00Jaykay24mhz _works well_ when you just navigate through menus or files....
22:50:01*bluebrother goes checking if he overlooked something to disable when building
22:50:03 Quit mofux (Read error: 110 (Connection timed out))
22:50:09bertrikI'm a bit surprised that 24 MHz is apparently not enough for non-sluggish GUI
22:50:25 Join mofux [0] (
22:50:29Bagderisn't that most nostable on things like ipod video with its huge screen
22:50:59Zagoramiconn: did you look any more at clk div boosting on pp targets?
22:51:04kugelyea, it didn't add sluggishness to the e200 (if I remember correctly)
22:51:13kugelI've tested it shortly without gui boost
22:51:24peturrasher: everything....
22:51:31Bagder... which basically is an argument for different speeds on different targets
22:51:35 Join mofux_ [0] (
22:52:00 Quit mofux (Read error: 60 (Operation timed out))
22:52:31kugelaccording to Buschel, the difference between 24MHz and 30Mhz doesn't yield as much runtime as one might think
22:53:39Unhelpfulkugel: we're not using scale-on-load in PF yet
22:54:01kugelUnhelpful: no, but you changed the load function too. and pf loads bitmaps
22:54:05Jaykayi could do another two battery benchs, only showing the menu. one time time with 24, one time witz 30mhz
22:54:30kugelJaykay: you can't run 24MHz without changing the source
22:55:11kugelin case you have 24MHz, you're obviously using a custom build. Only tests with the official build are valuable
22:55:21Jaykayso why does "cpu frequency" show 24mhz when pressing select?
22:56:04Unhelpfulkugel: i've not found any cases of the unscaled loader failing in a bit... is there a particular AA that you could add to the scaler FS task, perhaps?
22:56:27kugelUnhelpful: Sure!
22:56:37Jaykaykugel: ?
22:57:01kugelJaykay: you're running a custom build. Download and install the official one before you do any tests please
22:57:19Jaykayno im running the official build......
22:57:46Jaykayi did apply a patch for snake and for mandelbrot, they dont have anything to do with the
22:57:52Jaykaycpu frequency
22:58:27 Quit Gareth (Excess Flood)
22:59:02Unhelpfulbe until at least tonight before i can check it... if it breaks in sim, you could try a sim build with ROCKBOX_VERBOSE_BMP_LOADER (i think)
22:59:25kugelJaykay: You're right. I didn't know this is possible
23:00:41kugelJaykay: don't touch cpu frequency if you make test. Let rockbox handle boosts
23:00:56kugeltests where you force a freq will hardly be meaningful
23:01:12midgeyrasher: did you mean to upload FS #9633 as a bug? ;)
23:01:59kugelUnhelpful: Oki doki
23:02:44Jaykaykugel: but if i do two benchs with 24 and 30mhz and only showing the menu, we would know wether 30mhz uses much more power than 24.... if this is the case then the developers could set the standard cpu frequency to 24mhz
23:03:13Jaykayso rockbox is only switching too 30mhz if its really needed
23:03:20bluebrotherdon't we have three frequency steps on H100? IIRC there's 11MHz too (but that's only used in the fm screen)
23:03:42rashermidgey: err nope
23:04:04Unhelpfulactually, a sim at all should tell you if it's any of the things that will cause an unscaled load to fail, all or nearly all of those have regular DEBUGF
23:05:02midgeyZagor: when I log into flyspray, I can see dates on any of the comments
23:05:05Unhelpfulalso, if anybody has a handy reference for C constant size/type suffixes... i need 16-bit, and all i remember is that U==unsigned :/
23:05:15 Part Jaykay
23:05:46Zagormidgey: I assume you mean can't? :)
23:06:16midgeywe'll chalk that up to stress from finals :)
23:06:51hobbsUnhelpful: I don't believe there _is_ a suffix for that
23:06:56hobbsUnhelpful: only U, L, and LL
23:06:59Unhelpfuldo you see "bagder" next to all comments, despite the fact that he's not commented on the task, or committed it, etc?
23:07:04Zagormidgey: what do you see instead? nothing?
23:07:22rashermidgey: did you set "Detailed date format" to something silly in your prefs?
23:07:26Unhelpfulhobbs: cast it in the operation, then?
23:07:31hobbsUnhelpful: I suppose so.
23:07:39kugelUnhelpful: s maybe?
23:07:50kugel123s, that'd be my guess
23:08:07Unhelpfulsyntax highlighting says otherwise :/
23:08:16midgeyrasher: haha looks like it got overwritten with my username during the updgrade
23:08:57midgeywhat's it supposed to be set to?
23:09:19Unhelpfulsurely not either "midgey" or "bagder" ;)
23:10:05Unhelpfulblank works for me
23:10:09*rasher has it set to %Y-%m-%d %T, but that's probably not standard
23:10:24midgeyif i try blank it reverts to my username
23:10:25amiconnZagor: Nope. I know it's doable, and should be relatively easy. But I see no need for it, and it limits the choice of clock frequencies
23:11:03Zagorthe gain would be to lower the cost of boost
23:11:08 Join Gareth [0] (
23:11:29midgeysetting it to the same as rasher worked...
23:12:39amiconnZagor: Yes, but imo there's no real need. We can't do that on all targets anyway, and hence we need always need a mode of operation that limits the boost/unboost rate
23:13:00amiconnOn a lowmem target things are different, of course
23:13:23Unhelpfulhrm, need to look @ the SH asm ref again
23:14:11Zagoramiconn: not necessarily. we could use two different models: very low idle freq with lots of boosting for div-clk targets and a higher idle freq with less frequent boosting for coldfire
23:15:07Unhelpfulamiconn: i went back to the "old" bmp loader for reference when i started on HQ greyscale, and i see a fair number of differences that i didn't introduce, in terms of how things are calculated. the comments seem to suggest that you did a lot of the size optimization, i might need to bother you (more than i have) at some point about getting loader size down :/
23:15:12amiconnMaybe, but then there are 2 facts which contradict each other
23:15:25ZagorI don't know if the gain would outweigh the added complexity though
23:16:23amiconnIf we want to lower the idle freq and boost for ui, it would make sense to go 24MHz. But then we'd also want the extra bonus of disabling the pll when at 24MHz - which then means we cannot boost quickly...
23:17:04kugelUnhelpful: interesting. The cover doesn't show in the sim, but on target
23:17:29kugeland pictureflow shows a wrong cover in the sim
23:17:30Zagoramiconn: those are not contradictory. we simply cannot disable pll in that scenario.
23:17:33amiconnUnhelpful: Imo the non-scaling loader size difference isn't that large. But I might have a look anyway
23:17:35Unhelpfulkugel: now i *know* that's not me ;)
23:17:51amiconnRight now I'm going to compare sizes for the >>8 brightness versus svn
23:18:07*amiconn is curious
23:18:09kugelUnhelpful: You're the bmp man now, no matter of the changes you did :)
23:18:31Zagoreverything is of course a matter of "how much do we gain" versus "how much does it hurt"
23:18:33Unhelpfulah, that's a size lose pretty much everywhere, i have a smaller mono brightness that's probably good enough for mono.
23:18:45kugelUnhelpful: pf segfaults when I click on that wrong image
23:19:13Unhelpfultbh i've seriously considered that we should do (0,1,0)/1 for mono ;)
23:20:27kugelUnhelpful: Why does the image show up on target?
23:20:52Unhelpfulkugel: i don't know, did you clear cache on both? it might be wise to do that manually, also...
23:21:01 Quit mofux_ (Read error: 110 (Connection timed out))
23:21:07kugelbtw: it shows in pictureflow, it data aborts after clicking it
23:21:10amiconnUnhelpful: Saving binsize is always good, but it's not everything. A speed gain may be worth some binsize, as well as an accuracy gain
23:21:24kugelUnhelpful: pictureflow cache you mean? yes I did that
23:21:33 Join mofux [0] (
23:21:36 Quit dfkt ("-= SysReset 2.53=- Ph'nglui mglw'nafh Cthulhu R'lyeh wgah'nagl fhtagn.")
23:21:59Unhelpfulamiconn: on mono-only targets, we never dither, hence i'm a bit less concerned with grey accuracy there. also, they have the lowest memory :/
23:22:23pixelmakugel: wrong picture reminds me of that bug that was said to be fixed yesterday (the one with the clear cache)... do you have album art for all your albums and/or did you change music files/directories recently, updating the database but not pictureflow?
23:22:27Unhelpful(2,5,1)/8 has mean-squared-error of ~22 grey levels
23:22:33amiconnThe mono targets do have a greylib though, and if we want to make the line output pluggable...
23:23:30Unhelpfuloh! i thought that was only on the 2bpp targets. what uses it on mono, sliding_puzzle?
23:23:34kugelpixelma: I updated the database and then cleared the pf cache (using its option)
23:23:48kugelI'm fairly sure that my fix really fixes it
23:26:21amiconnUnhelpful: Not sliding_puzzle, no. Atm there is no plugin that uses the greylib and bmp files, mainly due to the fact that there is no bmp loader for the greylib yet
23:27:15amiconnBut while bmp scaling and dithering etc doesn't make much sense in the core on mono targets, having it available for plugins would be a good thing, imo
23:28:07pixelmakugel: are there some albums without art, only showing the default? I only have a very few and when the bug happened to me I could usually see the art two times - one in the right place and the other a few steps before or after depending on how content had changed, probably on the old "spot" it would have been. Didn't try since yesterday
23:28:10amiconnNot all mono targets are lowmem, btw
23:28:47kugelpixelma: I think that just happened to me, yes
23:29:01Unhelpfuli know about some of the "modern" mono targets. maybe the "nice" load+scale should be available in pluginlib, and the stripped one in core, on lowmem targets?
23:29:09kugeldeleting the pf folder manually results in "?" instead of the same cover twice
23:29:19 Quit jhulst (Remote closed the connection)
23:30:12amiconnI dunno how you measured, but I get an increase by a mere 16 bytes on recorder for the >>8 brightness
23:30:23pixelmakugel: sounds like the bug hasn't been fixed. Maybe it's not only related to the prepared bmps but also on how pf "remembers" the order?
23:30:37amiconnThis is within measuring uncertainty (section alignment is 16 bytes on SH)
23:30:39 Join saratoga [0] (n=9803c6dd@gateway/web/cgi-irc/
23:31:43kugelpixelma: pf deletes the old cache now, but only for covers it's going to create. My test showed that it doesn't delete one of the the cache files
23:32:23amiconnSame amount on arm (mini g2), and +24 bytes on coldfire (h120)
23:32:37Unhelpfuli was using "actual" size from rockbox-info, but perhaps i need to look @ the symbol w/ nm?
23:32:45amiconnBut it's faster (no division) and more accurate
23:33:01amiconnI also used actual size from rockbox-info.txt
23:33:31kugelpixelma: indeed
23:34:19Unhelpfulbizarre. i get +16 for the /256 brightness, and -32 for (2,5,1)/8
23:34:31Unhelpfulon sh, that is
23:34:36pixelmaZincAlloy: wanted to ask you... now there's the possibility to show something else while adjusting the volume, do you think this could be used in cabbiev2 (I'd find it quite useful if the volume icon could change and dB will be shown as the statusbar does, when there's space enough the dB could also show near the icon only while adjusting). It probably needs some viewport overhaul of the theme though and would take some time
23:34:40amiconnYes, +16 for >>8
23:34:46amiconnWhy is that bizarre?
23:35:19Unhelpfuli guess not, it matches yours
23:35:41saratogaI saw the dicussion of CPU frequency
23:35:43*Unhelpful claims tiredness
23:35:48saratogathis is actualyl something I've explored quite a lot
23:36:08saratogabasically, you will use a couple extra MHz doing LCD updates on most color targets
23:36:15saratogaand WPS updates too
23:36:20saratogaso the boost pecentage goes up
23:36:34saratogaon PP this means 1 mA of additional battery load for every 3MHz
23:36:47saratogaso updating the WPS takes about 1 mA of additional power
23:37:19saratogahowever, my MP3 on COP patch worked around this problem by shifting a lot of the decode use for MP3 to the COP
23:37:20kugelpixelma: it did work before I moved the deletion a little down in the function. And that broke it without me noticing
23:37:31saratogameaning theres extra MHz available for WPS updates without boosting
23:37:50saratogaso they're free, provided you have PP and only use MP3/FLAC/MPC
23:38:29saratogahowever this is still not very efficient, and I have some ideas how it could be further improved
23:39:15saratogaI would eventually like to change how boost works, such that the normal frequency depends on how recently the buttons have been touched
23:39:46saratogathis way we could disable the WPS on targets like the Sansa, and reduce the CPU clock accordingly to save more power
23:42:35saratogaamiconn: regarding PLL, I'm assuming the gain Toni saw from it was because he actually reduced the core clock to 24MHz, and that the gain Buschel saw [<1mA] is correct
23:42:42saratogain which case we might as well just leave it on
23:42:44 Join freddy_ [0] (
23:42:56ameyerwhat's the normal power usage at 30 MHz for the Sansa anyway?
23:42:59saratogathough obviously its worth further testing on my part
23:43:10saratoga30mA with current SVN according to Toni
23:43:31Zagorsaratoga: I don't understand. how does boost percentage go up with lcd updates?
23:44:01ameyerso 3 MHz = about 3% battery life hit?
23:44:22saratogaZagor: CPU time is spent in the driver and in the WPS code
23:44:47toffe82on the test_disk_log of test_disk, the value to put on the wiki are the last three ?
23:45:01ameyeron color targets, couldn't you disable the WPS when the LCD was off?
23:45:11ameyerI know, probably easier said than done
23:45:17Zagorsaratoga: do you really mean boost, or do you mean "not sleeping"?
23:45:26saratogaZagor: boost
23:45:30saratogasleeping saves almost nothing on PP
23:45:38 Quit fredddy (Read error: 60 (Operation timed out))
23:46:03kugelrasher. pixelma:
23:46:53Zagorsaratoga: then, again, please explain how frequent lcd_update causes more cpu_boost() calls to be done. I don't see how that can be.
23:47:13saratogaZagor: perhaps this will clarify what I mean:
23:47:52saratogaZagor: time spent in the LCD driver and WPS code requires a higher average CPU clock then would otherwise be needed, this increase in CPU clock increases power draw from the battery
23:48:59Zagorsaratoga: so you mean with less lcd_update we could lower the standard clock speed? you're not talking about the cpu_boost() call?
23:49:02*ameyer wonders what the power consumption of a CFmodded iPod Mini 2G is at 30 MHz
23:49:03saratogathis shows what boosting does to CPU load:
23:49:21kugelsaratoga: after that what Bagder said, even doing stuff at 30MHz isn't free, if it wakes up sleeping cores
23:49:25saratogaZagor: the codec thread will boost the CPU more frequently the less of the CPU's time it gets
23:49:43saratogakugel: sure but sleepign the core barely saves anything
23:49:58kugelthat's what I thought too
23:50:28amiconnUnhelpful: Do you know what *is* weird? Using a >> 16 version is effeictively +/-0 on SH1...
23:50:31saratogasee the first graph I posted, it shows one CPU going to sleep while the other is still on
23:50:51Zagorsaratoga: aha! thank you
23:51:20amiconnTry it. Factors are 19596, 38469 and 7471
23:51:34 Quit petur ("Zzzzz")
23:52:38amiconn+80 bytes on arm though, and +36 on coldfire
23:52:43saratogai think the best solution would be to have the button code set a time stamp the last time a button changed, then have the normal CPU frequency change between 24 and maybe 45MHz depending on how recently a button was pressed
23:52:55kugelsaratoga: so my assumption that using a codec which effectively runs under 30MHz most of the time will yield inaccurate measurements when measuring "show wps" vs "show main menu/blank wps"?
23:52:58saratogathe WPS update speed could also respond
23:53:26kugel+ was true
23:53:26saratogakugel: yes using mp3 for this will not work
23:53:27 Part blkhawk ("Leaving")
23:53:35saratogayou should probably use AAC or just disable MP3 on COP
23:53:36 Join jrsharp [0] (
23:53:44saratogaAAC being the same speed as MP3 without the COP
23:53:45amiconnsaratoga: The idea being discussed was switching to just using the post divider(s) for boostin gon PP
23:54:07saratogaamiconn: I'm not familar with how boosting actually works, how would that change things?
23:54:11amiconnThen boosting becomes essentially free, and can be done for the ui if a button is pressed
23:54:29saratogaah you mean it eliminates the transistion time?
23:55:05amiconnRight now we have to wait for pll relock, which takes ~500µs on PP. Just switching the post divider (or even using the post-divider selector) takes maybe 1µs
23:55:29saratogathat'd be nice
23:55:40amiconnThing is, we can do this on PP and on AS3525. We cannot do this on coldfire
23:56:05amiconnSo we have to maintain 2 boosting strategies if we decide to go that route
23:56:20amiconnIt also means that we are limited in the choice of frequencies
23:56:50ameyerbut what would this new strategy do to power consumption while unboosted
23:57:01saratogathough even 500us isn't bad, thats only 15uA at 30MHz
23:57:31saratogaor rather 15uA of average additional boost current
23:57:45saratogai think my units are getting mixed up here
23:57:54ameyershould be in watts, I think
23:57:58amiconnBut we would be able to boost/unboost more often, i.e. boost for ui operation (less sluggishness even witha low unboosted clock on G5), and having a smaller pcm buffer
23:58:15saratogayeah that would be great
23:58:38ameyerwhat about power consumption while boosting part of the time?

Previous day | Next day