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

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

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

#rockbox log for 2009-07-12

00:00:08amiconngevaerts: Could you perhaps test on Video, mrobe100 and Clip?
00:00:20amiconnkugel: Yes, that's the idea
00:00:51amiconnFor functions with previously saved r4..r12 *and* lr it was easy - just don't save r12 because it's scratch reg anyway
00:01:13amiconnBut functions which didn't use all the regs needed some register rearrangement
00:01:25*kugel finds it confusing to read rl in code that doesn't use it as the lr
00:02:29amiconnConsistency - this way you'll see that something happens to the link register. It's just a preference though
00:03:17kugelwhat speaks against keeping r11 and save this instead of lr?
00:03:35amiconnThe way I did it saves an instruction
00:04:04amiconnIf you save lr, you can pop it into pc together with restoring the other registers
00:04:33amiconnBasically, if you need a non-scratch reg you should *always* use lr first
00:05:22gevaertsamiconn: that patch doesn't apply. It seems to be broken off halfway
00:06:35amiconnOh? The patch has 1544 lines. All of them are shown in pastie...
00:07:43kugelbroken here too
00:07:48amiconngevaerts: Pastie's "raw" or "download" swallows the last 3 lines
00:07:48gevaerts shows it fine, (which is the way to get it raw) does not...
00:08:07amiconnblergh - and doesn't werk atm
00:08:30pixelmadid you really try pastebion? ;)
00:08:47*pixelma wonders what that has to do with a cinema
00:09:35amiconnWeird - now the latter works
00:09:38*amiconn goes pasting
00:09:49kugelappending the lines gives a failed hunk
00:09:54 Join matsl [0] (
00:11:07 Join Ubuntuxer [0] (
00:13:47linuxstbUbuntuxer: Hi. Since you committed Clix, I'm been meaning to ask if you can refer to FS tasks when you commit patches? Also, when you close a task as "accepted", can you mention the SVN revision it was committed as?
00:14:14 Join tessarakt [0] (
00:14:17 Join tessarakt2 [0] (
00:14:19 Quit tessarakt2 (Client Quit)
00:15:15linuxstbUbuntuxer: Also, looking at it now, I think you forgot to set svn:keywords on clix.c
00:15:27linuxstb(see the UsingSVN wiki page)
00:15:34Ubuntuxerok, I'll do it next time, sorry
00:15:47*amiconn is puzzled. He tested on c200 this afternoon and it worked. Now mpegplayer crashes...
00:19:48pixelmaUbuntuxer: also - since you committed chamges to the mazezam controls on some targets, it would be really nice if that couild be changed in the manual as well
00:20:39Ubuntuxerok, I'll have a look...
00:20:39kugelamiconn: stkov here
00:22:33amiconnAnd in mpegplayer I guess?
00:22:42gevaertsAlso on ipod video
00:23:02amiconnSame thing I observe on the beast and c200. Weird
00:23:07*amiconn goes bug hunting
00:23:09 Quit ender` (" If I capture the hero's starship, I will keep it in the landing bay with the ramp down, only a few token guards on duty and")
00:24:23 Join fml [0] (n=4fd3df32@gateway/web/cgi-irc/
00:24:52fmlIn the manual, can all the entries in "features" be used as the switch for \opt and \nopt?
00:26:28fmlFor example, I see \opt{swcodec} in some places. I'll just try it out.
00:29:38kugelfml: yes
00:30:35 Quit flydutch ("/* empty */")
00:32:20 Join funman [0] (n=fun@rockbox/developer/funman)
00:34:33funmansvn is slow for me
00:35:05CIA-71New commit by bertrik (r21792): S5L8700: fix buttons used in debug menu
00:36:43 Join Blue_Dude [0] (
00:37:17fmlAren't commits to the manual observed by the CIA?
00:38:13funmanyes they are
00:38:27*funman kicks CIA-71
00:39:14fmlfunman: he-he, they should be fired. Mine slipped through! :-)
00:39:44 Quit AndyI (Read error: 60 (Operation timed out))
00:40:47linuxstbfunman: I was wondering about your *patcher Makefile changes - doesn't the mkdir fail if the directory already exists?
00:41:15 Join mc2739_ [0] (
00:41:25linuxstbfunman: i.e. "mkdir -p" would be better?
00:41:57 Join AndyI [0] (i=AndyI@
00:42:17 Quit matsl (Read error: 110 (Connection timed out))
00:43:01 Join fdinel [0] (
00:44:50funmanlinuxstb: hum yes i didn't think of a previously existing directory (and i felt a bit akward of using mkdir in a makefile ..)
00:45:48funmanso yes -p should be used (doesn't error if the directory already exists)
00:46:08linuxstbI would also use "cp -p", but there's probably no good reason to (just personal preference)
00:47:02funmani have no problem with that, can you commit a fix?
00:47:02 Quit fml ("CGI:IRC (EOF)")
00:47:43linuxstbFor both mkdir and cp?
00:49:27 Quit mc2739 (Nick collision from services.)
00:49:28 Nick mc2739_ is now known as mc2739 (
00:49:36 Join fml [0] (n=4fd3df32@gateway/web/cgi-irc/
00:50:18fmlHrm... After all the pitchscreen related commits I wonder whether it was worth it. The reason for my doubt is a high red delta. :-/
00:50:25CIA-71New commit by dave (r21793): Small tweaks to r21778/r21779 - use -p for both mkdir and cp commands in dmg creation rule
00:51:26linuxstbfml: What actually changed? r10359 has lots of words to read...
00:51:45*linuxstb should probably just try it...
00:52:59pixelmaunfortunately the commit message didn't have much to read, the FS task though
00:53:23fmllinuxstb: 1) Correct display of the speed 2) Changing pitch in 0.1 semitone steps 3) Correct changing from % to semitones and back 4) Persisting the screen state (pitch changing mode)
00:54:40funmansaratoga: test_fps/battery_bench with a slower pclk would be interesting now that data cache is enabled on Sansa AMS
00:55:28fmlMaybe it's just a not optimal implementation. I didn't pay much attention to the bin size when I was looking at it, to be frank
00:55:32 Join thegeek [0] (
00:56:15***Saving seen data "./dancer.seen"
00:56:25saratogafunman: I thought kugel said the tests of it were done with MMU enabled?
00:56:26amiconngevaerts, kugel: The crash has nothing to do with my patch - svn crashes too
00:57:07 Quit tessarakt ("Client exiting")
00:57:10*amiconn suspects r21781
00:58:44funmansaratoga: oh right, i'll look in the forum thread for the benches
00:59:15*amiconn wonders whether people actually test pontentially affected things before committing :\\
00:59:25 Quit Ubuntuxer ("Leaving.")
01:00:35fmlamiconn: what is the problem?
01:00:40amiconnwrong pitch
01:00:53saratogafunman: I still want to do more testing though with DMM instead of battery bench
01:01:03FlynDicesaratoga: got just a minute to chime in... On the battery benches I did the higher pclk configs performed significantly better, by 3 hrs or so
01:01:04 Join Ridayah_ [0] (
01:01:04amiconnThe scale changed. Several plugins set pitch as well
01:01:22saratogaFlynDice: why is that?
01:01:43saratogadoes higher pclk reduce boosting independently of everything else?
01:01:44 Join _0x90u[Sleep] [0] (
01:01:59funmanFlynDice: do you rememeber where your benches are (on flyspray?), i can't find them on the forum
01:02:13fmlamiconn: oh! I didn't check the plugins. Do the set it to 100% and then restore?
01:02:16amiconnsplitedit, video, wavplay, wavrecord, mpegplayer
01:02:30amiconnThey just set it to 100%. No restore
01:02:41funmansaratoga: a fact is that SDRAM clock frequency is equal to pclk
01:02:42FlynDiceI don't know... I tried what I thought would be the best power saver in 62/31 fastbus only and 248/62 sync bus beat it by 3 hours
01:02:49FlynDiceI'll go find it
01:02:53amiconnBut they use the permille values, i.e. 1000. splitedit also sets 50% and 150%
01:02:58funmanin theory we could have it twice faster but in practice rockbox didn't boot
01:03:03 Nick _0x90u[Sleep] is now known as _0x90u (
01:03:56funmanFlynDice: in svn we use fastbus clocking when unboosted
01:04:22funmansaratoga: why do you prefer to use a DMM over a battery_bench? is it more precise?
01:04:30saratogafunman: it is
01:04:36saratogaplus it doesn't take 18 horus a test
01:04:37FlynDiceYes but I changed it to fastbus only to test and see what I could get, was rather disapointed
01:04:40saratogaso I can do dozens of them a night
01:05:04funmanwhat is 'fastbus only' ?
01:05:16saratogaplus I can do time resolved power measurements, so I can see how much power is used boosted, how much unboosted, how much playing mp3, how much paused, etc
01:05:59saratogabattery benches can be misleading because they average boosted and unboosted power together, so changes in boost ratio get conflated with changes in power consumption per clock
01:07:19saratogai wonder though why changes in memory clock would have a big impact, particularly with mp3 I would expect us to be mostly running off of IRAM with little/no DRAM activity
01:07:47 Quit Ridayah (Read error: 110 (Connection timed out))
01:08:07FlynDicesaratoga: here's my graphs post
01:08:19saratogai'm going to do boosted/unboosted measurements of power consumption using SVN, if someone wants to make other suggestions about what they want tested let me know
01:09:23saratogacan we lock the mem clock to 31MHz and try pclk at 31 and 62MHz ? that would be very interesting to me
01:09:44 Join Thundercloud [0] (
01:09:52funmansaratoga: I think the IRAM clock is pclk
01:10:16FlynDicefunman: for fastbus only I simply made the busmode stay as fastbus in the boosted state
01:10:22funmansaratoga: the SDRAM clock can be equal to pclk or twice faster
01:10:23saratogahmm so both get faster/slower
01:10:41fmlamiconn: I have a patch: Could you have a quick look? I don't really know what the plugins do. Just set the pitch to 100/50/150%?
01:10:46funmanFlynDice: so you used the same frequency for boosted cpu (i.e. no boosting) ?
01:10:56saratogathats really undesirable, ideally we would lock IRAM at full speed even if the CPU goes lower then 60MHz
01:11:11funmanthe cpu frequency must be higher or equal to pclk
01:11:28FlynDicefunman: no I ran unboosted at 31 MHz fastbus and boosted at 62MHz fastbus
01:12:15CIA-71New commit by amiconn (r21794): Fix plugins for the changed pitch scale from r21781
01:12:50amiconngevaerts: Could you try again with latest svn + patch? Patch doesn't change...
01:13:45FlynDiceno more time for now!
01:15:08fmlamiconn: ah, you were faster! But wouldn't it be better to work with the _100 value (like in my patch)?
01:16:03fmlamiconn: the effect is the same, just the readability would be "better" IMO.
01:16:34fmlPrecision is technical term, whereas "_100" is a "semantical" value meaning 100%
01:17:16*amiconn thinks a single multiplication is more readable than using fractions
01:18:42fmlamiconn: but it's hard to figure out what multiplication it should be. We probably have to introduce a macro for converting percents to the right value
01:19:28amiconnBah, typical. Hung build and no Swede around
01:19:35amiconnNew system isn't building either
01:19:45linuxstbgevaerts: Have you looked at FS #4755? Any idea how committable (or not) it is?
01:20:33gevaertsamiconn: ipod video seems to be fine
01:20:55 Quit fml ("CGI:IRC")
01:21:30saratogafunman: somone on ABI mentioned graphical corruption from time to time in MPEGplayer, not sure if you've heard of this before
01:21:42gevaertslinuxstb: no. I happen to be subscribed to it, but I've never actually looked at the code
01:21:44saratogaon AMS of course
01:22:07funmansaratoga: i have not heard of it, but i frequently see it on my fuze ;)
01:22:10gevaertslinuxstb: IIRC Last year roolku told me that it needed cleanup
01:22:48linuxstbI see mcuelenaere posted a "cleaned up" version 5 days ago...
01:22:49saratogaok then
01:22:51funmansaratoga: i am not sure if i mentioned this LCD corruption problem to kugel (I also saw it in plugins, not only mpegplayer)
01:23:43gevaertslinuxstb: ah indeed. I do think that tools/converter is a bit too generic though
01:23:51funmannow i hope we can put some light on what's wrong in the current SD driver
01:24:16linuxstbgevaerts: Yes, and perhaps not for tools/ either. (if tools/ is "build tools")
01:24:19saratogaregarding voltage scaling or some new bug?
01:24:51funmansaratoga: if you have an e200v2 could you have a look at fs#10413 ?
01:25:34soapsaratoga, thoughts on the forum thread move and title edit?
01:25:37*linuxstb looks at the patch and sees "* Copyright by ???" in the first file...
01:25:40funmansome users experiment deadlocks using current build (see
01:25:47gevaertslinuxstb: also true. utils?
01:25:55funmanthat, and filesystem corruption
01:26:02linuxstbgevaerts: Yes, I think so.
01:26:09linuxstbHmm, plus files #2 and #3 have no (C) holder...
01:26:52saratogafunman: sure but I didn't even realize we could flip the lcd on the e200v2
01:27:15*linuxstb decides to post a comment
01:27:25funmanmc2739: you should shout about your work and not quietly post new patches to flyspray ;)
01:27:42saratogasoap: the one about the headphone jack? i dont' mind
01:28:00soapjust saw you were reading the moved thread and was curious.
01:29:22saratogafunman: whats wrong with the driver now when its flipped?
01:29:34gevaertsamiconn: mr100 seems fine as well
01:29:40funmansaratoga: no idea i don't have an e200v2
01:29:51mc2739funman: It isn't that important a patch, not as important as other things that are being worked on
01:30:27saratogamc2739: I'm trying mpegplayer now with the screen flipped and it seems to work without the patch
01:30:32saratogawhat would it actually fix?
01:30:43saratogaoh the menu are upside down
01:30:48mc2739try pausing
01:30:56saratogayes now i see
01:31:14saratogai'll try it out later, i have to get back to work now
01:31:25mc2739the video content does not flip but the controls do
01:32:16kugelfunman: I think we all see the corruption from time to time
01:32:37funmanmc2739: have you seen fs#10371 (recording on AMS) ? Perhaps I've missed something stupid ..
01:33:26funmankugel: i imagine it could be a race condition in the button/lcd synchronisation code, but i have nothing to back this feeling
01:34:04_0x90uDoes anyone know if there is a simple custom bootloader for the STMP3500? So that I could possible load a different .sb if say the play button is pressed during startup.
01:34:12kugelcould be, but I can't see how
01:34:53mc2739kugel: funman: I don't think I have seen any corruption. Is it reproducible?
01:35:07kugelno, it's random
01:35:12funmanmc2739: it's much more frequent in mpegplayer though
01:35:16saratoga_0x90u: I dont' know that anyone has looked at the STMP chips very much
01:35:41kugelfunman: it surely is, it updates very frequently too
01:35:52kugelif it's not on the e200v2, maybe it's some delay missing
01:36:20_0x90usaratoga: Ok, well I'm looking into it. Might be able to shead some light on the subject of the STMP.
01:36:49saratogais there a way to run code on any player based on it?
01:37:12mc2739funman: I have seen FS10371 but have not had time to try it yet
01:37:28funmanmc2739: well i think there is nothing to try, only read :/
01:38:03funmanhow much work would it be to merge e200v2 and fuze lcd drivers ?
01:39:05amiconngevaerts: Clip is okay as well?
01:40:13gevaertsamiconn: no, but that's apparently due to a bad bootloader (plain svn also didn't work, but it does now after installing a new bootloader). I'll know more soon
01:41:11funmanmc2739: is the e200v2 battery the same than fuze battery ? I see that powermgmt-e200v2.c was copied from e200v1
01:42:19saratogaIIRC its similar but not exactly the same
01:42:29gevaertsamiconn: looks fine as well
01:42:36saratogasince the e200v2 one is a user replaceable sandisk battery while the fuze has an OEM one soldered in
01:43:05mc2739funman: I do not think that the battery is the same
01:43:11amiconngevaerts: Thanks for testing!
01:43:49CIA-71New commit by amiconn (r21795): ARM asm LCD and ATA driver functions: Don't save r12 as it is a scratch reg. Saves a bit of stack and execution time.
01:44:50 Quit dfkt ("-= SysReset 2.53=- Ph'nglui mglw'nafh Cthulhu R'lyeh wgah'nagl fhtagn.")
01:47:12amiconnBleh, svn is borked
01:49:29 Join Ox9OU [0] (
01:49:50 Quit rasher ("leaving")
01:50:04pixelmasomeone around with a c200 at hand (not a too patched build)?
01:51:01pixelmacould you try whether bubbles crashes for you as soon as the bubble should be shot?
01:51:22 Quit funman ("free(random());")
01:53:04 Quit _0x90u (Read error: 60 (Operation timed out))
01:53:58gevaertspixelma: I finished the first level, but this is about 20 revisions behind. I'll try a current build as soon as it's built
01:54:40pixelmamine's running r21778 now but with kugel's pla_rework patch
01:54:57pixelmaand bubbles uses PLA
01:55:10gevaerts21776 here
01:56:12bertrikis there a minimum required time that needs to be possible to set with timer_set?
01:56:39bertrikif I implement it in a simple way, the timer overflow after 83 ms
01:58:12gevaertspixelma: latest svn also works
01:59:01pixelmaok, seems to be patch related then. Thanks :)
02:01:08pixelmaand also - teru is right that the repeat actions don't work correctly (observerd in robotfindskitten as he said and aldo aiming left/right in bubbles)
02:01:51kugelHow can my patch make bubble crash?!
02:02:42pixelmametronome seems to have lost the "tap" but now has the "play" instead, svn is the other way round
02:02:43kugelthat's the only patch you have applied?
02:03:39pixelmawell, other than that only the HID disable and I have a few changes in the manual...
02:04:02gevaertsHID disable really shouldn't change anything at all for plugins
02:04:26pixelmaI try an update
02:04:26gevaertspixelma: play seems to more useful than tap anyway :)
02:05:35*gevaerts thinks that metronome should use the volume buttons for volume on the gigabeat f
02:06:49kugelpixelma: tap could be long select
02:07:07gevaertskugel: long?
02:07:07pixelmahow would you tap with "long select"?
02:07:17bertrikvery slowly ... :)
02:07:29kugelno idea, that's what the code says :)
02:07:33pixelmaespecially since short select changes play/tap mode
02:07:44*gevaerts wonders if this is an indication of what sort of music kugel likes :)
02:07:47kugelI admit it doesn't make sense
02:08:57pixelmaoh, then it works (but changes play/tap mode at the same time an indicator that it's not "guarded" by button release events, probably=
02:09:07pixelmaalthough not really usable
02:09:57*gevaerts thinks that metronome is just weird
02:10:07kugelin general?
02:10:13gevaerts"Rec to SYNC" on gigabeat f.
02:10:26*pixelma thinks metronome should use vol up/down for changing the volume on the c200 too, then Play (Up) would be free for "play" and Select
02:10:28gevaertsthat button doesn't exist
02:10:32pixelmafor tap
02:10:38kugelit says that on any target, but Rec only does something on the irivers
02:11:29pixelmagevaerts: I've been told that this is a feature that basically only works on the H100 (H300 too?) and just wasn't ported correctly - or taken out from the other targets
02:11:52gevaertsindeed. The code has proper #ifdefs everywhere except for the text
02:13:36kugelWhat is this strange sorting on new.cgi?
02:13:45kugeland why is the fuze not on the tabe
02:13:58pixelmarrr... I would like to update but nothing happens (and seeing the trouble I had with earlier in the evening) :\
02:14:04kugelah found it
02:14:43gevaertspixelma: I had problems an hour ago as well
02:15:47kugelamiconn: still stokv
02:16:28kugeloh, wait
02:16:35kugelgrml lagging..
02:18:27kugelor just git-svn?
02:18:43pixelmano, read a few lines up ;)
02:19:05kugelyea, but I could access the site just fine
02:19:35kugelok, it's definitely
02:20:50kugelPLA doesn't seem to be a good choice for metronome
02:20:54pixelmaseems like it is impossible for me to update now - I got a nice "data abort" in bubbles though. Happens when the bubble has to be shot (also after the timeout without shooting manually)
02:21:59pixelmaquite a "low" address, 00001140 (0)
02:22:37Ox9OUnoob question.. Is there a decomplier/complier for the STMP?
02:22:47saratogawhat CPU does it have?
02:22:58saratogaif ARM, then we have compilers
02:23:22Ox9OUNo idea. I haven't figured that part out yet.
02:23:36saratogawell then no sense asking about compilers until you know the CPU
02:23:39kugelamiconn: shouldn't you've updated other langs too?
02:24:02pixelmathey weren't in the initial patch
02:24:39Ox9OUI'm trying to find the CPU arch. for the STMP3400
02:24:47kugelsemitone was added? I seem to recal it exists already
02:25:37saratoga0x90U: the rockbox wiki says Motorola DSP56004.
02:26:49Ox9OUOk thanks bro
02:26:52saratogaaccording to our wiki, there is no compiler
02:27:10Ox9OUwhat about ASM?
02:27:33saratogatheres an assembler
02:27:39saratogawhy don't you read the wiki entry?
02:27:44kugelnew pitchscreen is weird
02:28:01Ox9OUOk nvm. I found the DSP56000 Software Development Tools. Sorry
02:28:06pixelmakugel: "#ifdef HAEV_SCROLLWHEEL" in bubbles.c? Could this be problematic?
02:28:17kugelyes it could
02:29:45pixelmaand could you change the comment below, while at it? Not all sansas have a scrollwheel ;)
02:29:49kugel"Tempo" definitely needs renaming now in the german translation
02:30:55pixelmayes, the problem is that it is the most descriptive of what "pitch" does on hwcodec
02:31:15kugelwhat does it there?
02:32:43 Nick fxb is now known as fxb__ (
02:32:54pixelmawell, it changes pitch by playing the data faster or slower (so not independently)
02:33:07kugelup/down indicatins making playback faster/slower. But up and down actually do nothing, and the scrollwheel doesn't make anything faster or slower eaither
02:33:28kugelpixelma: so what it did on swcodec too before timestrech?
02:34:19kugelit doesn't really indicate changing tones
02:35:50 Join Bishop [0] (
02:36:39kugeland it's just wrong on swcodec now
02:37:09BishopWhich OSS replacements exists for zen/muvo/nomad?
02:39:07kugelpixelma: hm, the old Pitch is now Rate, does hwcodec still show Pitch?
02:40:50kugelit's buggy also :(
02:41:06pixelmahaven't checked yet. Fixing the typo in the new bubbles.c didn't help fixing the crash
02:41:15 Quit bertrik (Read error: 113 (No route to host))
02:41:48*kugel can't really believe the patch is causing this
02:42:20linuxstbBishop: AFAIK, none... Although some work has been done to port Rockbox to some Zens.
02:43:37kugeloh, I think I'm seeing the problem
02:44:17kugelpixelma: try swapping the mapping for BUBBLES_START and BUBBLES_SELECT
02:44:36kugelah no wait
02:46:20 Join r0b- [0] (
02:46:48r0b-ok rockbox iw reporting 1 ammount of freespace but windows is showing a different ammount
02:47:02kugelpixelma: try this (incremental patch, use with p1):
02:47:20r0b-rockbox shows 322MB free but windows only shows 215MB free
02:52:03 Quit n00b81 (Read error: 110 (Connection timed out))
02:53:15 Join n00b81 [0] (
02:55:08r0b-what could cause this?
02:55:46gevaertsr0b-: run chkdsk on it
02:56:10r0b-whats the command
02:56:19***Saving seen data "./dancer.seen"
02:56:40r0b-i run windows
02:57:07kugeljust check it for errors in, the properties of that drive
02:58:22pixelmar0b-: before doing that - did you trigger an update in the Rockbox Info screen? Button depending on target
02:59:34 Join GreatBeaver [0] (
02:59:34 Join BHSPitLappy [0] (n=BHSPitLa@unaffiliated/bhspitmonkey)
03:00:03pixelmar0b-: when you went to the Rockbox Info screen and it showed you the stats (322MB), you can also press a button to let it update (Select on many players)
03:00:21r0b-i used the Stats app
03:01:20linuxstbr0b-: Then try the Rockbox info screen...
03:01:42r0b-ok there we go
03:01:46r0b-now its right
03:01:49pixelmathe "Stats app"? Do you mean the stats plugin`I'm wondering because I can't see info on free disk space there
03:02:26r0b-i wasnt thinking
03:02:26linuxstbr0b-: The problem is that there is a bit of info stored on the disk called (I think) "FSINFO", which Rockbox uses and keeps up to date, but Windows doesn't.
03:02:49 Join Biatch [0] (
03:03:46BiatchWhich player hardware is most supported? Is there a (de facto) primary development platform?
03:04:22linuxstbBiatch: Not really, no.
03:04:47r0b-(anything but sansa's)
03:05:04linuxstbBiatch: Generally, the longer the hardware has been supported by Rockbox, the better the support (but that's a generalisation)
03:05:05dzr0b-: my e260 v1 works just fine :P
03:05:27r0b-i like my E250 v1
03:05:46dznewegg's got recert ones for $35ish I think
03:05:49BiatchIf I would to purchase a new player and flashing with rockbox, which should i buy? (price no matter)
03:06:12soap"new" as in fresh off the factory floor with no previous owners?
03:06:14linuxstbYou can't buy a new player and install Rockbox...
03:06:49Biatchnew, yes completely new manufactured.
03:07:04r0b-if you want alot of sttuff i like my Sansa E250v1 refurb
03:07:31BiatchWhy cant i install rockbox on a new player?
03:07:47r0b-new players have encrypted firmware
03:07:51linuxstbSee the list of hardware Rockbox supports on the front page of our website...
03:08:02linuxstbAnd then compare with what you can buy new.
03:08:22BiatchIs there any player currently selling without encrypted firmware?
03:08:25linuxstbr0b-: That's not true at all...
03:08:25r0b-honestly the older players ROCK!
03:08:33soapr0b-, _few_ new players have encrypted firmware.
03:08:54r0b-sorry not thinking
03:09:01 Part wincent ("Kopete 0.12.7 :")
03:09:07 Quit tvelocity (Read error: 110 (Connection timed out))
03:09:15Biatchzune is out i guess. lol.
03:09:22linuxstbBiatch: Simply that porting Rockbox to new devices generally takes longer than the lifetime of that device.
03:09:29 Quit kugel (Remote closed the connection)
03:09:57dzisn't the zune made by microsoft?
03:09:59r0b-honestly the older players if there good they will last longer then a modern player
03:10:31dzso why would you want to buy a microsoft product just to put open source firmware on it? :P
03:11:00Biatchi dont. want to.
03:11:13Biatchbuy it.
03:11:53 Join tvelocity [0] (
03:12:03 Quit Bishop ()
03:12:05r0b-i went with my mp3 player because it can run rockbox
03:12:21r0b-and just for the bragging rights that i can play doom and non of my friends can on there mp3 players
03:15:11r0b-hoenestly being able to play doom while running around is pretty interesting
03:15:36dznow if the controls just weren't quite as awkward
03:15:57 Join mc2739_ [0] (
03:16:11 Quit mc2739 (Nick collision from services.)
03:16:13 Nick mc2739_ is now known as mc2739 (
03:16:21r0b-i got the controls a little easier
03:17:28r0b-but even using the "Rotate Screen" option still is a little odd
03:17:44r0b-i think i know how i can make the game easier on me
03:17:52dzgetting it set up so the sansa's scroll could be used for turning would be nice (would probably need some acceleration)
03:31:02 Quit Thundercloud (Remote closed the connection)
03:34:05 Quit Blue_Dude (Read error: 110 (Connection timed out))
03:35:23 Join Blue_Dude [0] (
03:38:50 Join daurnimator [0] (n=daurnima@unaffiliated/daurnimator)
03:41:26 Quit gevaerts (Nick collision from services.)
03:41:38 Join gevaerts [0] (n=fg@rockbox/developer/gevaerts)
03:47:23 Quit evilnick (
03:47:23 Quit saratoga (
03:56:06r0b-dz thats gangerous thinking
03:59:16r0b-dz you aorund
04:04:59r0b-if there was a way to map to the reverse scrollwheel motion it would work
04:05:57r0b-but as you cant map the game buttons to Button Off
04:05:58r0b-it wont work
04:06:47dzsome of the bindings for the sansa are odd, such as 'record' actually being submenu
04:06:57r0b-yea i know
04:07:15r0b-if i had the crap to compile it properly id change that :P
04:07:29dzI've got a toolchain set up
04:07:44r0b-is it able to run in windows?
04:07:50dzalthough I can't guarantee it'll work properly, since it's with gcc-4.3
04:08:13dzno, I run linux at home
04:08:15r0b-tell me what i need
04:08:23r0b-i only got CYGWIN or a VM
04:08:25 Quit Biatch ()
04:08:41dznot sure for windows
04:09:05r0b-well what tools do i need using Linux?
04:09:17dzbut you can feed me patches (or probably preferably post them on flyspray)
04:09:38dzyou need arm-elf-gcc and friends in binutils
04:09:45dzthere's a shell script to set it up on the wiki somewhere
04:09:49r0b-well dz i havent really learned full C
04:10:00dznor have I
04:11:21dzif I get adventurous tomorrow I might try to fix that backlight flickering bug (tried poking around at it a while ago but couldn't quite wrap my head around the code)
04:12:35r0b-could you try to get scrolling working in the dictionary?
04:12:38r0b-dict plugin
04:21:02 Quit Lss__ (Read error: 104 (Connection reset by peer))
04:31:40 Quit fdinel (Read error: 104 (Connection reset by peer))
04:34:02 Quit tvelocity (Remote closed the connection)
04:45:48UnhelpfulFS #10430 updated with new asmdefs mechanism for handling C struct sizes and offsets in asm... comments welcome, especially from anybody familiar with the build system or asm. i also tweaked tools/ a bit to prefix all headers which are not found during deps with the build dir, since the asmdef.h files will need this as the existing sysfont.h does.
04:55:42 Join kamlurker [0] (
04:55:45Dhraakellianstrafing/sidestepping needs to be easier in doom
04:56:20***Saving seen data "./dancer.seen"
04:56:21Dhraakellian(while still being able to turn and aim)
04:57:34dzI'd be happy with being able to both turn and shoot and turn and move forward at the same time
04:58:02dzwhich is rather difficult on the e200 due to the raised scrollwheel
05:02:28 Quit sbhsu (Read error: 60 (Operation timed out))
05:02:34 Join martian67_ [0] (n=martian6@about/linux/regular/martian67)
05:03:40*Dhraakellian has a fuze, so yarr
05:04:06 Quit martian67 (Read error: 60 (Operation timed out))
05:04:17Dhraakelliansame number of buttons, really, but power is less convenient to hit (well, slide) than record
05:05:07Dhraakellianotherwise, though, the Fuze controls are much nicer, so if they could be properly mapped...
05:11:25kamlurkerdoes Rockbox currently support audio playback during fast-forward?
05:15:03kamlurkerokay. thanks.
05:15:22UnhelpfulDhraakellian: just add mouse/keyboard support for targets with usb host hardware. ;)
05:16:11 Join sbhsu [0] (
05:18:44 Quit r0b- (Read error: 110 (Connection timed out))
05:22:52 Quit Blue_Dude ("ChatZilla 0.9.85 [Firefox 3.0.11/2009060215]")
05:25:43 Quit BHSPitLappy (Read error: 60 (Operation timed out))
05:26:11 Join r0b- [0] (
05:27:19 Quit n00b81 ("Leaving")
05:32:34 Quit efyx_ (Remote closed the connection)
05:39:21 Quit martian67_ (Read error: 60 (Operation timed out))
05:40:01 Quit mc2739 ("ChatZilla 0.9.85 [Firefox 3.0.11/2009060215]")
05:41:14 Join martian67_ [0] (n=martian6@about/linux/regular/martian67)
05:42:08 Join sbhsu_ [0] (
05:50:10 Quit sbhsu (Read error: 110 (Connection timed out))
06:07:19 Join EpicReviews [0] (
06:07:41 Join evilnick_home [0] (
06:08:05EpicReviewsif I didn't like RockBox, could I use the Sansa Firmware Upgrader to revert back to the default firmware?
06:11:36 Quit sbhsu_ (Read error: 60 (Operation timed out))
06:11:56 Join legendarymidget [0] (n=79dc436b@gateway/web/cgi-irc/
06:13:10 Quit legendarymidget (Client Quit)
06:13:19 Join legendarymidget [0] (n=79dc436b@gateway/web/cgi-irc/
06:13:39EpicReviewshi legendarymidget
06:13:47EpicReviewslet's carry out a conversation here
06:13:55soapEpicReviews, RBUtil has an uninstall function.
06:14:03evilnick_homeEpicReviews: You could use rb-util to uninstall Rockbox
06:14:31 Quit legendarymidget (Client Quit)
06:14:48 Part EpicReviews
06:17:50 Quit GreatBeaver ("ChatZilla 0.9.85 [Firefox 3.0.11/2009060215]")
06:18:05 Join faemir [0] (
06:21:37 Join sbhsu [0] (
06:24:30 Quit evilnick_home1 (Read error: 113 (No route to host))
06:30:05 Join m67_l3 [0] (n=martian6@about/linux/regular/martian67)
06:30:08 Quit Ox9OU ()
06:30:59 Quit m67_l3 (SendQ exceeded)
06:31:33 Join m67_l3 [0] (n=martian6@about/linux/regular/martian67)
06:38:01 Quit martian67_ (Read error: 110 (Connection timed out))
06:45:19 Join EpicReviews [0] (
06:45:56EpicReviewsdoes rockbox only support the v1 series of Sansa? will this work on v2?
06:47:57 Join ReKleSS [0] (
06:48:17advcomp2019EpicReviews, only the v1's but the v2 aka ams sansas is being worked on
06:48:43EpicReviewsis there like a daily build that would work?
06:50:16advcomp2019nope.. the only way is build it yourself
06:50:33EpicReviewsok I'm hopeless there lol
06:50:44EpicReviewswell is there like a newsletter I can sign up for?
06:50:59EpicReviewsso I can know when it come s out?
06:52:21advcomp2019you can look at the mailing list or keep an eye on the rockbox site
06:53:24EpicReviewswell I'll just hopefully find it later
06:53:30EpicReviewswhen it comes out... if ever
06:53:56 Quit faemir (Remote closed the connection)
06:55:07EpicReviewsIt's kinda funny, I love Linux, but I really like the interface of Windows Media Player, and it amazes me that no one's made a look-alike media player/manager
06:56:24***Saving seen data "./dancer.seen"
06:56:29 Join BHSPitLappy [0] (n=BHSPitLa@unaffiliated/bhspitmonkey)
06:56:47advcomp2019that is getting off-topic for your info.. you can read the guidelines
06:57:20EpicReviewsok :\ not a fun room I see
06:57:23 Part EpicReviews
06:58:58 Quit kamlurker ("ChatZilla 0.9.85 [Firefox 3.0.11/2009060310]")
06:59:53 Join martian67_ [0] (n=martian6@about/linux/regular/martian67)
07:00:45 Quit martian67_ (SendQ exceeded)
07:01:27 Join martian67_ [0] (n=martian6@about/linux/regular/martian67)
07:04:04 Join sbhsu_ [0] (
07:04:37 Quit sbhsu (Read error: 101 (Network is unreachable))
07:08:50 Quit m67_l3 (Read error: 110 (Connection timed out))
07:15:12 Join martian67 [0] (
07:25:09 Quit martian67_ (Read error: 110 (Connection timed out))
07:28:23 Quit martian67 (Read error: 54 (Connection reset by peer))
07:29:03 Join martian67 [0] (n=martian6@about/linux/regular/martian67)
07:35:03 Quit amiconn (Nick collision from services.)
07:35:07 Join amiconn_ [0] (i=quassel@rockbox/developer/amiconn)
07:35:16 Nick amiconn_ is now known as amiconn (i=quassel@rockbox/developer/amiconn)
07:35:22 Join pixelma_ [0] (i=quassel@rockbox/staff/pixelma)
07:35:22 Quit pixelma (Nick collision from services.)
07:35:40 Nick pixelma_ is now known as pixelma (i=quassel@rockbox/staff/pixelma)
07:46:34 Join Horschti [0] (n=Horscht2@xbmc/user/horscht)
08:05:29 Quit Horscht (Read error: 110 (Connection timed out))
08:09:54 Quit evilnick_home (Read error: 104 (Connection reset by peer))
08:10:59 Join evilnick_home [0] (
08:46:23 Join domonoky [0] (n=Domonoky@rockbox/developer/domonoky)
08:51:16 Join Rob2223 [0] (
08:56:26***Saving seen data "./dancer.seen"
09:09:01 Quit Rob2222 (Read error: 110 (Connection timed out))
09:10:11 Part toffe82
09:21:59 Quit domonoky (Read error: 104 (Connection reset by peer))
09:41:09 Join bertrik [0] (
10:06:02 Join flydutch [0] (
10:07:53bertrikI'm looking into a way of implementing timer-s5l8700.c for hardware that has a timer of only 16 bits
10:07:55 Quit n17ikh ()
10:09:03bertrikthere is a clock selection register (/2, /4, /16 and /512) , a prescaler register and a count register
10:12:55 Join bmbl [0] (n=Miranda@unaffiliated/bmbl)
10:18:10 Quit flydutch ("/* empty */")
10:26:19 Join flydutch [0] (
10:38:04bertrikI think I got it (and not too complex)
10:40:39markunbertrik: other targets have 32 bits?
10:40:49*pixelma wonders about the one change (in the diff) by fml in the manual (the simplify conditions one)
10:41:26bertrikmarkun, yes, the other timer drivers I've seen so far have 32 bits indeed
10:43:50bertrikthe prescaler can go up to 1/1024 (10 bits range), the clock select can go to 1/64 (5 bits) and the timer itself gives me 16 bit range, for a total of 31-bit range
10:43:58bertrikso that's OK I think
10:45:43bertrikI looked a bit into the clock frequencies for boosting. I think a good and simple start would be to use peripheral clock PCLK=50 MHz and use processor clock FCLK = 50 MHz / 200 MHz for normal / boosted
10:46:15bertrikAs far as I can tell, we can toggle between unboosted/boosted CPU by setting a single bit somewhere
10:48:24markunbertrik: don't know if it's possible for you to do some measurements to see how much current the meizu draws boosted and unboosted, on the gigabeat for example it didn't make any difference.
10:49:26markunalso a samsung chip
10:50:02bertrikI think the power controller automatically switches to USB power when plugged in and I can measure USB current.
10:51:15pixelmafml: (I know he reads logs) - why did you change the "masf" opt to an "swcodec" nopt? Usually opts are preferable over nopt and this way it's a bit more accurate... masf will also exclude it from the Player (although not necessary since it's already put inside a pitchscreen target only section). The "masf" also comes from features.txt.
10:51:32 Join petur [50] (n=petur@rockbox/developer/petur)
10:51:47bertrikI have a hard time believing it really doesn't make a difference, or maybe we're not doing something right then
10:51:53pixelmathe only thing is that it is not as descriptive
10:52:44*amiconn summons Bagder or Zagor
10:55:03 Join GodEater_ [0] (n=godeater@rockbox/staff/GodEater)
10:55:51amiconnbertrik, markun: The sh1 timers are also just 16 bit with prescaler (/1.../256 in powers of two).
10:56:28***Saving seen data "./dancer.seen"
10:58:54 Join borris1 [0] (
10:59:14pixelmafml: there also is the possibilty to use an "hwcodec" opt
11:01:13 Join kugel [0] (n=kugel@rockbox/developer/kugel)
11:01:36 Join matsl [0] (
11:02:24 Join martian67_ [0] (n=martian6@about/linux/regular/martian67)
11:04:11CIA-71New commit by Ubuntuxer (r21796): set missing svn keywords
11:04:28 Quit GodEater_ (Remote closed the connection)
11:07:20 Quit BHSPitLappy ("Ex-Chat")
11:14:38 Quit martian67 (Success)
11:14:45 Join m67_l3 [0] (n=martian6@about/linux/regular/martian67)
11:15:31 Quit m67_l3 (SendQ exceeded)
11:15:46 Join Zagor [242] (n=bjst@rockbox/developer/Zagor)
11:15:57Zagoramiconn: you called?
11:16:56CIA-71New commit by kugel (r21797): Remove svn:executable from these files.
11:17:03amiconn"Build should have been done 629m ago...."
11:17:13amiconnAnd several commits are not on the frontpage
11:17:16Zagoramiconn: ah
11:17:50Zagorindeed. odd.
11:18:27Zagorlooks like one of bagder's cronjobs is broken
11:23:10 Join robin0800 [0] (
11:23:49 Quit martian67_ (Read error: 110 (Connection timed out))
11:25:08 Join m67_l3 [0] (n=martian6@about/linux/regular/martian67)
11:27:00robin0800 am Iright in thinking WMP only syncs in MTP mode and this is causing music not to show in rockbox but perhaps the database can read it if player is back in MSC mode?
11:27:42Zagormeh, I broke ping again...
11:28:53amiconnZagor: Perhaps repo locked or some 'svn up' got stuck? was *extremely* slow last night
11:29:17amiconnI had several occasions of hung 'svn up'
11:30:02Zagoryeah but you regularly have problems with our servers when others don't so I rather suspect network issues
11:30:32amiconnOthers reported slowness too - see backlog
11:32:37Zagorwell, it has load 0.23 now so obviously that isn't the problem
11:32:40kugelrobin0800: no, you're wrong. WMP can also sync in MSC mode
11:33:03 Quit at0m (Read error: 104 (Connection reset by peer))
11:33:13 Join at0m [0] (
11:33:52tmztrobin0800: you're in windows now?
11:34:19tmztkugel: in that case would it still be in a hidden (burried) directory?
11:35:37kugel /music ? I don't know if WMP hides it, but the OF on sansa does
11:36:20amiconnZagor: Probably not now, but it seems that once svn gets stuck, it won't continue even if the reason for getting stuck is gone
11:36:41 Join Ubuntuxer [0] (
11:36:41amiconnIt may even need a kill -9 followed by svn cleanup
11:36:46ZagorI just did a manual svn up, and it's not stuck
11:37:07Zagorthe update has simply not ran for some reason
11:37:39tmztkugel: I don't mean hidden attribute, I know about that. I mean is it in the same place as it would be for mtp, a directory with an essentially random name and unrecognizable filenames
11:37:44UbuntuxerI can't set the svn keywords, it doesn't work.
11:38:05Ubuntuxersvn pset svn:keywords "ubuntuxer Id 2009-07-12 17:42:37Z 21720" apps/plugins/clix.c
11:41:06pixelmarobin0800: but files put on the Sansa in MTP mode won't show up in Rockbox's database (so never should have there though)
11:41:31robin0800tmzt, no i'm out of the house and on a laptop with linux
11:41:35pixelmashould have been
11:42:50pixelmaZagor: svn was slow for quite a few people last night (including kugel and me, should be in the logs)
11:43:42robin0800pixelma, the post says automatic sync in WMP is that possible in MSC mode?
11:43:47CIA-71New commit by kugel (r21798): Correct svn:keywords and svn:eol-style on a few more files.
11:44:26kugeltmzt: no.
11:44:37pixelmarobin0800: never really used WMP so can't tell (just for the one try)
11:45:09robin0800pixelma, I'll check tomorrow
11:45:14kugelI used it many month ago, but it worked
11:46:29kugelamiconn: I'm still getting stkov on my fuze
11:46:36 Join martian67 [0] (n=martian6@about/linux/regular/martian67)
11:46:45pixelmaUbuntuxer: you set the keywords "Author Id Date Revision" (as an example), the data will be filled out with Ubuntuxer etc,
11:46:55pixelmaby svn for you
11:47:11tmztkugel: stkov?
11:47:20kugelI just filled out with "Author Date Id Revision" as the apps/ files have this
11:47:33kugelI don't know if it matters
11:48:17pixelmawell the actual keywords and the order was made up by me now, off the top of my head
11:48:53kugelUbuntuxer: "svn propset svn:keywords "Author Date Id Revision" apps/plugins/clix.c"
11:49:06linuxstbpixelma: You quoted the same as the UsingSVN wiki page... But I don't think it matters.
11:49:07Ubuntuxerah, ok thanks
11:49:25 Join QncOPFGl [0] (
11:49:34pixelmaUbuntuxer: you can look at how the other files do it - with proplist (or plist prbably). I use it with -v for a bit more verbosity
11:49:59linuxstbUbuntuxer: I would use svn pset svn:keywords "Author Id Date Revision" apps/plugins/clix.c - that's what the UsingSVN wiki page says to use.
11:51:14CIA-71New commit by Ubuntuxer (r21799): set missing svn keywords to clix
11:51:41Ubuntuxerthank you for your help linuxstb!
11:51:43 Join fml [0] (n=4fd3c6a6@gateway/web/cgi-irc/
11:51:45 Quit matsl (Remote closed the connection)
11:51:53kugellinuxstb: I just changed 126files in firmware/ to "Author Id Date Revision" because all apps/* files use it
11:52:12kugelto "Author Date Id Revision" I mean
11:52:22linuxstbkugel: Bah...
11:52:55kugelwe should rather change UsingSVN
11:52:56 Quit m67_l3 (SendQ exceeded)
11:53:16linuxstbkugel: What about all the other files though?
11:53:30pixelmais the order really important?
11:53:39fmlpixelma: hello! Yes, that was my reasoning when changing the condition in the manual. The whole pitch screen related block is protected by "pitchscreen". And then, I think, all the devices can be diffrentiated by the "swcodec" feature which gives us two disjoint sets. Looking at "masf" could potentially give weird results.
11:53:40kugelI don't, I don't think it's very important to change those
11:53:48UbuntuxerCould it be that the plugin maze doesn't have a manual?
11:54:26pixelmafml: nopt is a custom hack though and sometimes tends to add unwanted newlines or spaces
11:54:49fmlA note for all: since the size delta of the pitch scree was so big I intend to look into it and, if I find something, rework parts of it to reduce the size. But should I revert the whole patch now?
11:55:26kugelfml: the new pitchscreen is strange
11:55:50fmlpixelma: ideally, I'd have only one table with some \opts / \nopts inside. But that would be too much hacking and hard to read. Hence I left two tables.
11:56:04kugelit resets values if you change the "pages", but only if you aren't doing it quickly, but then it also has strange values if you go on the same page again
11:56:42kugelthe new icons are superfluous
11:56:59fmlkugel: in what regard?
11:57:20pixelmafml: the button tables need a rework in general to become more readable. Reminds me that I wanted to put a proposal on the wiki page explaining the style (I think it was LatexGuidelines are so)
11:57:26kugeland I bet many languages translated "pitch" with "speed", they're all totally wrong now
11:57:48kugelthough, that's not you fault of course
11:57:57pixelmathat can only be done bit by bit though I believe
11:58:21kugelfml: the new icons? well, theres already a << icon, now there are two << icons next to each other
12:00:00pixelmafml: and by the way - when reading the diff, I first thought that by using \nopt{swcodec} this part would show up in the Player manual now because the surrounding pitchscreen opt is a bit away from it (just saying that different people read things differently)
12:02:49 Quit Zagor ("Leaving")
12:03:21fmlkugel: I agree. The new icons should indicate that time stretch mode is active. But it's already obvious since there are two values displayed. Or are there daps that allow timestretch but have very small screens?
12:03:54pixelmathe Clip has a tiny screen
12:04:02kugelhow does that matter?
12:04:05pixelmaand the m200
12:04:15kugelthere are 2 icons on the left and 2 icons on the right
12:04:29kugel1 per side is totally enough
12:04:55*pixelma bites the bullet and takes the 5 minutes to install a new build on the Ondio to see if the new pitchscreen actually works
12:04:55fmlkugel: I know. See above for the reason. But if the two values are shown on all platforms we can eliminate the second icon pair.
12:05:28kugelI don't understand that
12:05:45kugelyou change pitch with up/down and speed with left/right, don't you?
12:05:47pixelmaI wonder if it has been tested on hwcodec at all before (except compiling)?
12:06:41fmlpixelma: not by me, no.
12:07:20kugelWhy do you need 2 icons on the left the left button is only doing 1 thing?
12:07:55fmlkugel: to indicate the mode
12:08:13fmlkugel: (rate vs. pitch/speed)
12:08:17kugelit doesn't indicate the mode at all to me
12:09:06fmlBut the two displayed values already do it. The question is whether all platforms show two values.
12:09:26kugelThe icons are still a bad indicator
12:09:57kugelA page counter at the top left makes more sense (ie. showing 1/4, 2/4 etc)
12:14:12 Join n17ikh [0] (
12:14:31pixelmahrmm... the Ondio daily manuals is broken though
12:15:26pixelmaand some of the others - Ipod Video for example
12:16:53pixelmathe ones that don't have a shortcut to the pitchscreen from the WPS and access it from the WPS context menu...
12:17:28pixelmathey break in the WPS button table since \ActionWpsPitchScreen isn't defined
12:18:10pixelmafml: ^
12:18:22fmlpixelma: yes I'm looking into it
12:18:48pixelmathe opt list was there for a reason
12:19:26bertrikI think I did everything right, but the s5l8700 still count 2x faster than expected
12:20:04 Join _zic [0] (
12:20:29fmlpixelma: yes, now I understand.
12:21:32pixelmais there a hanging build in the table for others too? I mean the one with the "old" system
12:22:09kugelthere's something borked, my commit also doesn't show on the front page
12:22:27pixelma"Build should have been done 695m 7s ago, at 00:46:53"
12:22:39 Nick Horschti is now known as Horscht (n=Horscht2@xbmc/user/horscht)
12:25:07 Part borris1
12:26:14CIA-71New commit by alle (r21800): Fix the manuals (the problem was introduced in r21791): some players do not have a button to go from WPS to the pitch screen
12:26:50pixelmafml: the new pitchscreen seem to work on my Ondio, just that the center viewport is a bit big and cuts off the ones above and below (even looks like overlapping but doesn't cause update problems, "just" that you can't see the options in these two)
12:27:13pixelmacan't see them completely, I mean
12:27:38fmlpixelma: does it have the timestretch feature?
12:28:13fmlpixelma: is it properly centered vertically? I.e. is it exactly in the middle of the screen?
12:28:30fmlpixelma: I mean the visible text
12:28:36kugelamiconn: nevermind, working now
12:29:33pixelmathe Ondio is hwcodec, but the Clip's screen is almost equally small (128x64 for the latter 112x64) the former), I use a 9-point font though which is unlikely someone does on the Clip
12:30:18kugelthe default one is the same height as sysfont (8px I think)
12:30:23pixelmafml: looks centered correctly but "Rate: value in percent" is now two lines
12:30:51pixelmathe "Rate" isn't needed on the hwcodec screens
12:31:16fmlpixelma: hrm... Does RB automatically wrap texts in viewports?
12:31:42fmlpixelma: I agree, if there is no timestretch feature then the word is not needed
12:31:51kugelpixelma: rate is the old pitch I think
12:31:57 Quit Ubuntuxer ("Leaving.")
12:32:41pixelmaprobably, but even if it wouldn't, then your problem would just be a different one - that you can't see everything completely
12:33:27pixelmakugel: yes, and targets that can only do the old style don't need to state this "mode"
12:33:52kugelbut then the same mode has two names accross tagets
12:33:52pixelmaas fml said, sorry
12:33:57fmlpixelma: but it was displayed in one line before, right? The patch added a space after the colon. hance I assume that it gets wrapped
12:34:28kugelfml: there's no automatic wrapping
12:35:11fmlkugel: then I have to look into the code more carefully to see what's done there
12:35:57 Join evilnick_home1 [0] (
12:36:13pixelmahmm, the old one had the same problem
12:36:47fmlpixelma: you mean two lines? Or partially non visible texts?
12:37:18pixelmaI can't tell the difference between the old and the new one though now on hwcodec, just that it keeps the mode?
12:38:46fmlpixelma: yes, that should be the case. Just the %/semitone mode should be preserved.
12:39:05 Join mcuelenaere [0] (
12:39:55pixelmafml: yes, the old one also has two lines and the center viewport overlaps the viewports above and below a bit (where you can read "semitone up" etc.)
12:40:10pixelmatwo lines in the center viewport
12:40:49pixelmafml: can this cause really that much binsize difference?
12:41:22fmlpixelma: ok, so the patch didn't introduce something new in this regard
12:42:16fmlpixelma: internal computations of semitone<->rate have been changed too. This might have been done not in the optimal way. I'll look into it.
12:44:04CIA-71New commit by bertrik (r21801): S5L8700: implement timer driver
12:45:42kugelbertrik: I wonder if you could have a look at purging tuner.c from target specific code
12:46:17pixelmakugel: bubbles crashes on my Ondio with your patch too (even with the changes you proposed later in the night), the patch also changes the "shoot bubble" button from "Up" to "Mode" which is the nice exception (and example) for how controls can differ between plugins: "Up" to shoot the bubble makes sense in this game, but not as a general "Select" button in other plugins
12:46:33bertrikkugel, I think I'll be working on the s5l8700/meizus in the coming time and not so much on other things
12:47:30kugelno samsas too?
12:47:39kugelpixelma: so is that good or bad?
12:48:22kugelhave you managed to install a svn version now to see if it crashes? It doesn't crash on my fuze
12:49:03pixelmakugel: the "Up" and "Mode" (for select) caused a lot of trouble when introducing PLA, since bubbles was the first one and later people thought "Up" was the "do something" button on the Ondio. Well, I prefer "Up" for shooting the bubbles, in other plugins "Mode" might be better suited for "action!"
12:49:44pixelmaI can compile a version without the patch too now
12:50:21 Quit Sajber^ (Read error: 104 (Connection reset by peer))
12:52:03 Quit evilnick_home (Read error: 113 (No route to host))
12:52:04 Join DarkDefender [0] (
12:56:30***Saving seen data "./dancer.seen"
12:58:00 Quit robin0800 (Read error: 110 (Connection timed out))
12:59:10 Quit petur ("linsucks")
12:59:17 Join robin0800 [0] (
13:00:07 Nick fxb__ is now known as fxb (
13:06:06 Join Thundercloud [0] (
13:14:21pixelmakugel: yes, the same revision without the patch doesn't crash in bubbles...
13:14:34kugelwhich revision?
13:15:22 Quit robin0800 ("Leaving")
13:18:02pixelmaand you should really look into the button repeat problem with the patch
13:18:31kugelthe robotfindskitten one?
13:19:13pixelmait looks like a general problem, e.g. aiming left/right in bubbles doesn't work correctly either
13:19:18 Quit fml ("CGI:IRC (EOF)")
13:21:29pixelmaor adjusting bpm in metronome
13:24:32 Join faemir [0] (
13:28:56*bertrik has only one screenful of undefined linker references left for a normal meizu m3 build
13:33:04 Join dash32 [0] (
13:39:40 Join robin0800 [0] (
13:56:41CIA-71New commit by kugel (r21802): Optimize chopper a bit by making a often used macro an inline function(which means its parameter expressions are evaluated before expanding) and ...
14:00:42markunbertrik: almost there!
14:01:12kugelmarkun: are you still using the database.unignore stuff?
14:01:29*bertrik invites other meizu owners to help hacking
14:03:24bertrikmarkun, it seems there is a kind of emergency-reset circuit on the meizus, just hold play or enter for 20 seconds or so
14:03:52markunkugel: you want to remove it?
14:04:00kugelbasically, yes
14:04:11bertrikI guess (but have not tried yet) that discharging or disconnecting the battery is not necessary
14:04:19markunbecause it scans the whole tree?
14:04:52markunI think it's a useful feature, but also would prefer it not to scan everything
14:05:12gevaertsbertrik: I'd help, but someone took my M3 away! ;)
14:05:21kugelI don't think it's worth saving a few database.ignore files
14:06:09markunkugel: you could do a forum poll
14:06:20 Join Sajber^ [0] (
14:06:35kugelwell, I think you're about the only one using it
14:06:59gevaertsbertrik: normally discharging or disconnecting the battery isn't necessary, yes. The problem I had in Gent was getting my finger timing exactly right to get it to DFU mode after the reset. I managed once in a few dozens of tries
14:07:08kugeldoing a forum is probably not helpful, I expect most people do not understand the gain of removing it
14:07:20markunkugel: then explain that in the poll
14:07:35markunif they don't understand the gain, then they will probably not care if you remove it either
14:07:41kugelyou just yourself you prefer to not scan everything
14:08:23gevaertskugel: if you can't explain the gain of removing it in the forum poll in a way that makes forum people understand, you haven't properly understood it yourself ;)
14:08:25markunyes, I said that. But I also prefer to have the unignore feature work :)
14:08:46kugelYou can't prefer both
14:08:59gevaertsYou can. It's called a setting :)
14:09:24bertrikI haven't used the feature, but the double inverted logic of unignore just seems weird to me
14:11:03gevaertsI think there are four solutions. (a) just remove database.unignore, (b) stop at database.ignore if the setting "Use database.unignore" is set to off, (c) add database.ignorenonfinal (or some other name) that can be overridden with database.unignore and make database.ignore stop searching, and (d) do nothing
14:11:31gevaertsI tend to prefer (b) or (c)
14:11:39markunI like c!
14:12:21markunbetter than adding a setting
14:12:26markunbut maybe a different name
14:12:50gevaertsyes. That name is lousy
14:12:53kugeland there the trouble begins
14:13:02kugelI bet we aren't able to agree on a name
14:13:28markunkugel: you can't be serious..
14:14:01markunyou just want to remove the feature and are trying to come up with a good excuse
14:14:03gevaertsmarkun: "database.naturalsort" ;)
14:14:27kugelmarkun: no, just my experience
14:14:49markunkugel: which other feature had not been implemented or was removed because of the name?
14:15:11 Join teru [0] (
14:15:12kugelI haven't said that
14:15:22kugelI just said that I think that finding a name will be hard for us
14:15:52kugelnot that we should go for (a) just because of it
14:16:00markunok, good :)
14:16:01 Quit dash32 (Remote closed the connection)
14:16:08pixelmahow many people need to "unignore" a folder that's in another one they want to ignore? And in how many cases it is really only one folder in a lot of others that would need to be ignored if "unignore" wouldn't exist
14:16:39markunpixelma: who knows..
14:16:53*bertrik thinks this is all too complex
14:17:09gevaertspixelma: the main problem is that without unignore you can't ignore the root
14:17:14pixelmait is only really useful if you need to ignore a lot of others otherwise... why not put the subfolder elsewhere?
14:17:37kugelIMO 1 sounds like a good approximation :)
14:17:51pixelmagevaerts: you could also solve that by rearranging your data structure
14:18:11gevaertsIf you only use the DAP for listening to music, you don't need unignore I think. If you also use it to transport random files, I think unignore is useful
14:18:29markunpixelma: I like my tree, my music is under "audio" along with other things like "podcasts", "audiobooks" etc.
14:18:37kugelgevaerts: can you elaborate that?
14:18:53pixelmayou only use database.ignore (or unignore) if you want to use the database
14:19:16pixelmathen data structure becomes less important, I think
14:19:36kugelmarkun: and that's makes 2 .[un]ignores either way
14:19:47gevaertskugel: if you dump random stuff in the root of your filesystem, the database scan will pick it up. That's a pretty common case if you meet someone who wants to give you some files
14:19:59markunkugel: I have 1 .ignore in my root and 1 .unignore in my "music" folder
14:20:08kugelgevaerts: solved by creating a folder "/random_stuff"
14:20:17markunall the test tracks and other audio files gets ignored
14:20:31pixelmagevaerts: you can put the files into subfolders - you can even do so in Rockbox without a computer
14:20:49pixelma(can take some time in some cases)
14:20:52 Quit r0b- (Read error: 110 (Connection timed out))
14:21:03gevaertskugel: not really. You can't just say "put it on this one, there's enough room" anymore. If you use it to get files often enough, that gets annoying
14:21:17markunpixelma: so, how should I arrange my files? Make 2 folders in the root, and put the things I don't want ignored in one and the rest in the other?
14:21:22kugelgevaerts: huh?
14:21:37pixelmamarkun: as I said, put your test files in a "test" folder with an ignore
14:21:42kugelhow's the root different from /random_stuff in terms of "room"?
14:22:04markunkugel: my podcasts are not random stuff
14:22:04gevaertskugel: because you have to tell someone where to put it. You're not always the person sitting at the keyboard
14:22:40pixelmagevaerts: you can sort them later even on the player...
14:22:47gevaertsIf you only copy random files once a month, that's not a problem. If you do it seven times a day, it gets seriously annoying
14:23:06pixelmahow many people do this?
14:23:10kugelso you're using unignore also?
14:23:13markunanyway, I like the unignore feature, I can add and remove files and folders and everything keeps working as I want to with just these 2 files
14:23:28 Join stripwax [0] (
14:23:39gevaertspixelma: I have no data, but I'd think it's not very uncommon
14:23:39markunI wouldn't mind having a database.scan file in the root where I specify which folders I want to have scanned
14:23:44markunwould that solve the problem?
14:23:54gevaertskugel: I don't use the database
14:24:04pixelmagevaerts: and I would think if your folder is named properly like "put random stuff here" people could find out
14:24:09bertrikmarkun, database.scan was exactly what I was thinking too
14:24:21markunif database.scan exists, just use those folders, otherwise look at database.ignore and remove .unignore
14:24:33gevaertssounds good
14:24:37markunbertrik: it would also make the scanning faster
14:24:50bertrikyeah, because it doesn't even enter the ignored subdirs
14:25:07markunand also the other folders with no music
14:25:55gevaertsI'd keep looking for database.ignore anyway. If you meet one, stop
14:26:09bertrikwe could put a database.scan at lower sub-dirs too and make it empty to ignore even deeper subdirs
14:26:19markungevaerts: yes, that too
14:26:35bertrikso an empty database.scan has the same effect as a .ignore file
14:26:59markunbertrik: I thought you didn't want to make it complex!
14:27:02gevaertspixelma: maybe I find I'm using stuff differently than the author intended often enough to start getting annoyed by "Then do it this way!" on sight...
14:27:05kugelmarkun: go implement :)
14:27:10kugelsounds nice to me
14:27:28markunkugel: can't you do it for me? :)
14:27:41gevaertsbertrik: an emptu database.scan would be the same as not initialising the database :)
14:27:43kugelI'm not sdure
14:27:59markunkugel: ok, maybe I'll do it later
14:28:09kugelbertrik: I wouldn't do that
14:28:15markunhaven't even downloaded the source and installed a cross compiler here
14:28:15pixelmahaving both seems redundant to me (and is probably more complex :\ )
14:28:15bertrikmarkun, I think it's very easy, just one file type (database.scan) that can exist at different levels
14:28:26kugelpoint of database.ignore is that you don'T need to read it (it's empty)
14:28:44markunI say let's at least keep database.ignore as well
14:28:44gevaertsmarkun: there are some interesting cases to think about, like what if database.scan has a/ and a/b/c/ in it, and there's a/b/database.ignore. What happens?
14:28:47markunsince people use that
14:29:11kugelI think scan should override everything
14:29:31markungevaerts: then a gets scanned (excluding a/b) and a/b/c as well
14:29:35kugelit's the ultimative order for scanning
14:29:48markunsame as .ignore in a/b and .unignore in a/b/c
14:30:30markunbtw, should the folders in .scan be absolute or relative? If we make it relative, then you can still move the folder around
14:30:42markunor support both?
14:30:50bertrikI'd say relative
14:31:10gevaertsI'd support just a single database.scan file in .rockbox, or maybe in the root of each drive
14:31:36gevaertsIf they can be anywhere, then they should have relative paths
14:31:36markunif we put it in .rockbox we should not make it relative :)
14:31:46gevaertsexactly :)
14:31:56markununless we add ../ every time
14:32:03markunbut that's a bit pointless
14:33:45markunkugel: thanks for bringing this topic up. I think it's a win-win situation with both of the things I wanted (fast scan, "unignore" like power)
14:35:42pixelmaand there I hoped it could get simpler...
14:36:23bertrikI won't lose sleep over this because I don't use the database ignore/unignore features personally, but I would very much like the system to be "as simple as possible, as complex as necessary". In particular not have exceptions to exceptions to exceptions
14:43:53 Join n1s [0] (n=n1s@rockbox/developer/n1s)
14:45:31markunpixelma: I think it will get simpler
14:45:37markunyou can use .ignore if you want to
14:45:42markunor use a .scan file
14:45:50pixelmadoesn't sound like it
14:46:55terui found some files that svn:keywords isn't set in apps/plugins (i checked only files in apps/plugins).
14:46:59markunpixelma: if you are happy with .ignore just stick to that. It will not get more complex.
14:47:22markunand if will not scan beyond that file (like it does now)
14:47:33markunand if -> and it
14:47:50pixelmayou have twio files and filetypes to take care of (and the user needs to know about both), codewise you need to take care of both and need to handle "overlapping" and conflicts
14:48:00 Quit DarkDefender ("Leaving")
14:48:46pixelmathe user needs to if he or she wants to use it in a way ignore can not
14:49:19 Quit robin0800 (Read error: 110 (Connection timed out))
14:49:51pixelmait will be more complex codewise and to me .ignore is enough :/
14:50:04 Join robin0800 [0] (
14:50:23markunthey don't need to be aware of both.
14:50:28markunThey can stick to what they use
14:50:45markunI don't think anyone will create database.scan files by accident
14:50:56*gevaerts might :)
14:51:09markunok, maybe 1 of our users then :)
14:51:11pixelmanot if they use unignore and/or want to achieve something similar
14:51:40pixelmaerr... then they need to know I mean
14:51:45markunpixelma: if kugel is right then nobody uses .unignore but me ;)
14:52:20kugelI'm not sure if have multiple scan files is a good idea
14:52:25markunme neither
14:52:30kugelit always means: opening, parsing etc
14:52:37 Quit Thundercloud (Remote closed the connection)
14:52:38pixelmaor someone gives them "random stuff" and a database.scan is amongst it...
14:52:40gevaertsmarkun: last week someone asked here about it, so there are probably a few more
14:53:05kugelgevaerts: that was the first one since it was committed..
14:53:13kugeland he didn't even know about it
14:53:19 Join Thundercloud [0] (
14:53:31 Quit faemir (Read error: 110 (Connection timed out))
14:53:33gevaertskugel: that just means it's documented well and it has no issues
14:53:52markunpixelma: when there is a patch we can discuss it a bit more
14:55:11bertrikkugel, is opening/parsing such a problem? we're already scanning and it's just a bunch of lines, hardly any parsing needed
14:55:59kugelIn my idea, there would be 1 scan file, all folders in their (and the sub folders) would be scanned
14:56:09bertrikpixelma, I consider the case that someone copies a database.scan file by accident quite remote
14:56:11kugel1 scan file per volume/fs
14:56:33markunkugel: I think that's what normal people would do anyway
14:56:34***Saving seen data "./dancer.seen"
14:56:55gevaertsbertrik: depends. If there's one global scan file, yes. If there can be one in each directory, who knows?
14:56:56markunbut if someone like gevaerts wants to go crazy and make a big .scan .ignore mess, why not? :)
14:57:10pixelmaI really think that a database.ignore (maybe with a special casing for the root) is enough for everyone (except markun maybe)
14:57:48markunpixelma: if that's enough for most people then they should not make .scan files
14:58:02markunremoving .unignore will speed up scanning for everyone
14:58:12markunand .scan will speed it up even more if they decide to use it
14:58:12pixelmacool, but it's worth adding the code
14:59:15pixelmawith all the corner case ideas that came up now, I couldn't even follow what it will do or not
14:59:50 Join martian67_ [0] (n=martian6@about/linux/regular/martian67)
15:00:15 Join ucchan [0] (
15:00:53 Join r0b- [0] (
15:02:46 Quit n1s ("Lämnar")
15:08:37 Quit martian67 (Read error: 110 (Connection timed out))
15:09:46 Join n1s [0] (n=n1s@rockbox/developer/n1s)
15:10:08mcuelenaeregevaerts: should USB serial work upon defining USB_ENABLE_SERIAL in debug_menu.c or should I do something else?
15:10:35gevaertsmcuelenaere: depends
15:10:54gevaertsYou need to do that before plugging in, and you need to make sure you have enough endpoints for it
15:10:54 Quit ucchan (Read error: 104 (Connection reset by peer))
15:11:09mcuelenaereah, then I probably need to disable mass storage
15:12:24 Quit robin0800 (Read error: 60 (Operation timed out))
15:14:38CIA-71New commit by amiconn (r21803): * ARM asm DSP and codec/plugin functions: Use r12 scratch register properly ...
15:15:50Bagderhm, I think we could introduce a simple MSG command to the build server to tell clients about stuff, like when a build round ends, just for us who happen to check the client outputs etc
15:16:18gevaertsthat would be good
15:16:55*gevaerts also wants the client to be able to run an external script with that MSG data :)
15:17:14Bagderanother good idea
15:17:39 Join rasher [0] (
15:17:43Bagderthe server already runs optional external scripts on some events, so the client could just use the same concept
15:17:57amiconnBagder: Did you see the old build system is still hanging? (for about 14.5 hours now)
15:21:03 Join robin0800 [0] (
15:21:12Unhelpfula thought has occurred to me since last night... perhaps asmdefs ought to go in sooner rather than later? basically, commit the build system parts first, start converting individual bits of asm, since the whole point of asmdefs is to make asm abi-agnostic... and then think about building with EABI when it's done.
15:28:55stripwaxHow do I build the rockbox manual? I'd thought I'd just create a build dir with Manual instead of Normal or Sim, but that seems to have just built the same as Normal for me
15:29:18gevaertsmake manual
15:30:24stripwaxah! what's Manual as build type used for?
15:30:48pixelmaan old leftover
15:31:10pixelmaalso, ManualHowto should have told you about "make manual"
15:31:52gevaertsBagder: did you get my bootloader mail?
15:32:30stripwaxpixelma - I should probably add ManualHowto to the DocsIndex 'for developers' section, in that case. I didn't find ManualHowto.
15:32:45Bagdergevaerts: I did, but I've been too lazy to deal with it yet
15:32:52gevaertsOK :)
15:32:59pixelmastripwax: good idea :)
15:33:06BagderI'll work on it tonight
15:33:09gevaertsas long as it didn't get lost
15:35:11kugelgevaerts: ready for release?
15:35:32kugelDon't forget to tag them
15:35:59gevaertsindeed. I should do that...
15:36:14pixelmaI should really try the new bootloader on my c200... so I do need the bootloader, sansapatcher and the "manual install" description, right?
15:36:15BagderI find the path situation on the download server to be in a sorry state
15:36:25BagderI mean, this is bootloader 6.0 we say
15:36:31Bagderbut they're spread all over
15:36:45Bagderand some dirs contain mixed bootloaders (and versions)
15:36:46kugelthe most sorry thing is the 5.0 symlink :)
15:36:56Bagderno, that's one of the most sensible ones in there
15:37:08Bagdersince it'll make 5.0 links to remain functional
15:37:17gevaertsindeed. It actually points to 5.0 after all :)
15:37:24Bagderyes, and will continue doing so
15:37:47kugelit points to bootloader/sandisk-sansa
15:37:56Bagderit points to the dir where the 5.0 bootloaders are
15:38:11kugelbut well, I don't care :p
15:38:16Bagderas the 4.0 points to the 4.0-dir
15:39:37Unhelpfulaha! i've found the magic to produce stubs without the other eabi changes... i think. :)
15:39:39kugelok, I can accept that :)
15:39:53gevaertsBagder: these are all named 6.0. They probably aren't the 6th version of the bootloader for H10 and mr100, but the released ones for those don't have a proper version number anyway
15:40:13Bagderyes, but part of the reason why I didn't do this yet is the path situation
15:40:27BagderI didn't really see how it would be done the best way
15:40:45Bagderbut I'll just make up something fine
15:40:48 Quit evilnick_home1 (Read error: 113 (No route to host))
15:40:54gevaertswell, we didn't release them with 3.3 anyway, so there's no reason to hurry
15:41:01Unhelpfulgcc doesn't actually determine the relocation type used for calls. the assembler does that, based on the abi version and the instruction used. so building a non-eabi gcc that passes -meabi=4 to the assembler gets us long-call stubs :)
15:43:28kugelUnhelpful: i.e. that works with our current compilers?
15:43:46 Join Jaykay [0] (
15:44:37robin0800the bootloader version flashes by so quickly perhaps it could be added to rockbox info in the system menu?
15:44:46Unhelpfulkugel: it should. i've not tested with gcc-4.0.3
15:45:14Jaykaywhat is the purpose of r21795 and r21803?
15:46:00gevaertsJaykay: "Saves a bit of stack and execution time" not clear enough?
15:46:53Jaykaygevaerts: execution time of what? and thats for r21795
15:47:38gevaertsof ATA and LCD drivers?
15:47:48Unhelpfuli just tested with 4.3.3, though, and mp3 on e200 still breaks :/
15:49:05Jaykaygevaerts: and what do these drivers execute and are all of the functions of the drivers now faster?
15:49:28gevaertsthe drivers execute instructions
15:49:46mcuelenaeregevaerts: for some reason, when I enable the USB serial driver (no mass storage, no HID) dmesg gives me this: usb-storage: probe of 1-3:1.0 failed with error -5
15:49:48*GodEater wonders what the purpose of these questions are
15:50:03mcuelenaere( )
15:50:12kugel Unhelpful I think mp3 generally breaks with that compiler
15:50:30kugelI tried 4.4.0 RCsomething and it crashed as well
15:50:37gevaertsmcuelenaere: what does lsusb -v say for the device?
15:50:56JaykayGodEater: i only wanted to know what is better now with these two revisions, gevaerts short answers force me to ask further questions
15:51:55 Join antil33t [0] (n=Mudkips@
15:52:04GodEaterJaykay: but since you didn't understand the answer, I wonder what you're going to do with the information
15:52:13gevaertsmcuelenaere: weird. usb_storage shouldn't even look at it
15:52:40*gevaerts wonders if this is another one of those HAL features
15:53:08JaykayGodEater: i thought i could get answers i could understand, i was just curious...
15:53:19mcuelenaeregevaerts: that vid/pid combination used to do usb_storage, perhaps that's it?
15:53:38gevaertsmcuelenaere: yes, but it shouldn't be...
15:56:36mcuelenaerehmm I changed pid to a4a6 and now usb_storage backs off, but usbserial doesn't kick in..
15:57:09gevaertsdid you modprobe usbserial with the right vid/pid arguments?
15:59:32mcuelenaereah no :/ forgot about that, I just did modprobe usbserial
16:01:06mcuelenaeregevaerts: yep, that did it :)
16:02:31 Quit amiconn (Read error: 104 (Connection reset by peer))
16:02:31 Quit pixelma (Read error: 104 (Connection reset by peer))
16:02:49 Join pixelma [0] (i=quassel@rockbox/staff/pixelma)
16:02:50 Join amiconn [0] (i=quassel@rockbox/developer/amiconn)
16:04:19 Join matsl [0] (
16:04:29 Quit stripwax ("")
16:12:47CIA-71New commit by FlynDice (r21804): AMSSansa: Use single adc_read instead of multiple ascodec_reads to read voltage for display in View HW info.
16:21:20kugelUnhelpful: ping
16:21:26Unhelpfulkugel: do you have any idea where that breakage starts? :)
16:21:44kugelnope, I haven't looked into it further, as other codecs were slower too
16:22:21Unhelpfuli've just got the -meabi=4 asm patch into our current gcc version... going to try it now. if it still crashes i guess i should see if reverting binutils fixes it...
16:25:36kugelwe don't have ARRAY_LEN in plugins?
16:26:21kugelwe don't have this at all :S
16:27:14 Quit matsl (Read error: 60 (Operation timed out))
16:28:41 Join mt [0] (n=mt@
16:32:40CIA-71New commit by bertrik (r21805): Meizu: implement power driver (USB power detect / charging status / poweroff)
16:33:24kugelbertrik hardcore coding
16:34:16 Join PureDarkn [0] (
16:35:59 Part PureDarkn
16:36:08CIA-71New commit by mt (r21806): Add seeking support in cook codec.
16:37:09kugelARRAY_LEN(x) gives 2 in the one line (the reason it crashes), but when I printf it it says 1 :S
16:37:10 Quit ReKleSS (Read error: 104 (Connection reset by peer))
16:37:31 Join ReKleSS [0] (
16:37:55bertrikfunny, I see the GPIO attached to the piezo beeper change value when I pinch this meizu m3 :P
16:38:07gevaertsextra button!
16:38:43bertrikyay for combos
16:41:35kugelpixelma: I got the crash fixed
16:43:27CIA-71New commit by mt (r21807): Some fixes for the standalone test program.
16:44:09*gevaerts sees things happening when he plugs a USB device into his F20 :)
16:44:29teruis there any reason some files don't have svn:keywords? may i set svn:keywords and commit?
16:45:01Unhelpfulteru: probably an accident :)
16:46:03kugelcan I add ARRAY_LEN to system.h?
16:46:31kugelor misc.h
16:47:29kugeloh, we have ARRAYLEN
16:48:41 Join aaron424 [0] (
16:49:02n1skugel: i was about to say that :)
16:49:39kugeln1s: we also still have strncpy ;)
16:51:41*Unhelpful appears to have mp3 playback again (not putting on headphones)
16:53:03Unhelpfulhrm, but all tracks fail without error for vorbis... no error displayed, just an immediate skip to the next track.
16:53:10n1skugel: right...
16:53:13gevaertsOK. USB host on gigabeat F is definitely possible, and shouldn't have unexpected problems
16:53:51n1sgevaerts: great
16:54:01*bertrik waits for gevaerts' commit
16:54:39gevaertsbertrik: possible, not ready :)
16:55:23gevaertsAll I've really verified so far is that if I follow the init sequence from the spec, and I plug in a device, the "device plugged in" bit changes
16:56:21 Join petur [50] (n=petur@rockbox/developer/petur)
16:56:28kugelI just want to be able to use hid to control my samsung with my clip :D
16:56:38***Saving seen data "./dancer.seen"
16:56:48kugelthe clip is an awesome remote also :)
16:56:58n1syeah, clip remotes!
16:57:54gevaertskugel: the samsung isn't OHCI, so that will have to wait a bit (unless you feel bored)
16:58:09kugelI don't
16:59:38Unhelpfulusing this patch in place of the usual one for gcc should give you long-call stubs without the other EABI changes. the snapshot binutils will still be needed for this to work properly
17:00:00Unhelpfulalso, the generated object files are probably *very* non-standard and incompatible, as they claim to be EABI ;)
17:01:12CIA-71New commit by gevaerts (r21808): Remove OHCI registers from s3c2440.h and move them to their own header. Since s3c2440 seems to be very close to the OHCI spec, there's no reason not ...
17:02:06Unhelpfulobviously i need to link the patch:
17:02:43n1sUnhelpful: what's the benefit of not getting the full eabi?
17:03:06Unhelpfuln1s: not breaking ASM assumptions about structures
17:03:50kugelUnhelpful: are you too lazy to not fix them, or are they just not worth it?
17:03:55n1swill the patch likely work with future gcc version (or will it be neaded at all?)
17:04:26Unhelpfulif there are no strong objections to it soon, i'll commit asmdefs and start working on fixing asm structure assumptions
17:04:51 Quit n1s ("Lämnar")
17:04:59CIA-71New commit by teru (r21809): set svn:keywords property
17:05:56Unhelpfulalso building gcc<4.3 with full EABI breaks a bit with our current multilib settings. for some reason gcc and gas disagreed about whether ldrd is supported. changing that target to arm946e-s fixes it, and should be correct for all targets for which we use arm9e now.
17:08:22mtWeird, I can't reproduce the same warnings for e200 sim in the last commit
17:09:04Unhelpfuli'm inclined to think that it's worth fixing asm for EABI in the long run. it makes many structures smaller, though i don't *think* it ever forces unaligned accesses to do so. this results in a 25% memory savings, for example, for core jpeg's decode buffer, since the rgb888 pixel struct is not padded to 4 bytes.
17:09:52 Quit ReKleSS ("Leaving")
17:10:36kugelUnhelpful: what was the reason mp3 crashs?
17:11:27CIA-71New commit by mt (r21810): Silence warnings to fix yellows for now.
17:15:38 Join funman [0] (n=fun@rockbox/developer/funman)
17:21:28Unhelpfulkugel: i don't know. it happened even without EABI on 4.3.3
17:21:45Unhelpfulbut it doesn't happen with the new binutils alone.
17:22:45 Join kugel_ [0] (
17:22:55 Quit kugel (Nick collision from services.)
17:22:59 Nick kugel_ is now known as kugel (
17:24:34kugelUnhelpful: that's what I said
17:25:45kugelI think I didn't try 2.19.1 binutils
17:28:24 Join MrDuck [0] (
17:29:24funmanperhaps we can have a look at the sd/mmc linux framework to see how errors are handled (retry / card reset)
17:29:24CIA-71New commit by gevaerts (r21811): Add data structures
17:30:59 Join stoffel [0] (
17:32:40 Quit amiconn (Read error: 104 (Connection reset by peer))
17:32:40 Quit pixelma (Read error: 104 (Connection reset by peer))
17:33:12 Join pixelma [0] (i=quassel@rockbox/staff/pixelma)
17:33:14 Join amiconn [0] (i=quassel@rockbox/developer/amiconn)
17:33:49funmanDo you know if the CRC of data blocks are stored on SD cards, or calculated by the card before transmission ?
17:33:57 Quit Rob2223 ()
17:34:00CIA-71New commit by bertrik (r21812): Meizu M3: initial version of battery readout and (uncalibrated) charge/discharge curves
17:34:12 Join Rob2222 [0] (
17:34:15bertrikfunman, I think this is pure a transmission thing
17:35:29funmanThe SD spec mentions "data transfer integrity" and not "data storage integrity" so I believe you are right
17:35:47funmanSo I wonder why 10 transfers get corrupted in a row
17:38:07pixelmakugel: good you fixed the crash, care to upload a new patch or so?
17:38:24kugelyes, after I fixed the REPEAT thing
17:39:22funmanThe SD spec (4.6 Error Conditions) only mentions CRC checks for commands (and only say the card's state doesn't change if the CRC of a command sent by the host is wrong)
17:41:48funmanIn case of any error (CRC or Write Error) during Write Multiple Block operation, the host shall stop the
17:41:52funmandata transmission using CMD12
17:43:38*funman writes that on his TODO list
17:44:10 Join DarkDefender [0] (
17:45:21 Quit funman ("free(random());")
17:46:35 Quit aaron424 (Read error: 104 (Connection reset by peer))
17:49:31 Join amiconn_ [0] (
17:55:58 Join pixelma_ [50] (n=pixelma@rockbox/staff/pixelma)
17:56:34 Join tvelocity [0] (
17:57:41 Join Ubuntuxer [0] (
17:58:34kugelpixelma: on the tracker
17:59:01CIA-71New commit by Ubuntuxer (r21813): match the manual to the changes in
18:00:29Ubuntuxerrevision 21529
18:00:38 Quit linuxguy3 (Read error: 104 (Connection reset by peer))
18:01:21 Quit teru ("Quit")
18:03:08 Join matsl [0] (
18:04:24pixelma_Ubuntuxer: that should have been in the commit message, not here ;) But thanks :)
18:04:37 Join linuxguy3 [0] (
18:04:41pixelma_kugel: alright, will have a look later
18:04:50kugelnice, thanks
18:05:23 Quit pixelma (Read error: 110 (Connection timed out))
18:05:24 Quit amiconn (Read error: 110 (Connection timed out))
18:06:49 Nick amiconn_ is now known as amiconn (
18:06:52pixelma_kugel: it looks like that patch reverts the chopper changes
18:07:01pixelma_you just committed
18:07:06kugelit does?
18:07:37 Join n00b81 [0] (n=n00b81@unaffiliated/n00b81)
18:08:10kugelinteresting, git still has some surprises left :p
18:10:23 Nick pixelma_ is now known as pixelma (n=pixelma@rockbox/staff/pixelma)
18:10:55 Quit antil33t (Read error: 104 (Connection reset by peer))
18:11:09 Join antil33t [0] (n=Mudkips@
18:12:46kugelpixelma: updated
18:14:38amiconnmt: Is this brute-force seeking?
18:15:41 Quit Jaykay ("ChatZilla 0.9.85 [Firefox 3.5/20090624025744]")
18:17:41mtamiconn: what do you mean by that ? (btw it's briefly described here : in Milestone 3, last point)
18:18:18 Join n1s [0] (n=n1s@rockbox/developer/n1s)
18:19:10amiconnAh, so not entirely brute force, but skimming through the whole file (as opposed to a simple lseek())
18:20:55mtYes, skimming through the file. Although I'm still not sure what brute-force means in this context. :)
18:23:14amiconnDecoding from the beginning of the file until the seek point is reached
18:25:03mtAh ok. No decoding though, just searching for the nearest timestamp.
18:28:56kugelbertrik: any idea what the pop fifo is?
18:29:04kugelfor dbop
18:29:19 Quit robin0800 ("Leaving")
18:31:55 Join dys [0] (
18:33:26 Quit Thundercloud (Remote closed the connection)
18:36:59 Quit mt (Remote closed the connection)
18:41:06 Join efyx_ [0] (
18:46:29 Join mt [0] (n=mt@
18:47:26 Join wincent [0] (
18:48:28 Quit DarkDefender ("Leaving")
18:50:54bertrikkugel, I think there is only one FIFO, the CPU pushes data into it and the DBOP pops data from it at its own pace
18:51:43kugelI'm just wondering why the statusregister has bits for the pop, but no interrupts
18:53:44 Join robin0800 [0] (
18:54:40bertrikI think the pop status signals are for information only, it would be of no use to have two interrupt sources because they occur almost always at the same time
18:56:42***Saving seen data "./dancer.seen"
19:05:39kugelbertrik: hmm, success, sort of
19:06:11kugelbertrik: I did the lcd speed up thing using wakeup_signal
19:06:23Unhelpfulhmph! neither the new binutils nor the passing to it of EABI flags is responsible for breaking vorbis. if i enable -mlong-calls, so that long calls are used instead of short calls with linker-generated long-call stubs, the faux-eabi toolchain gives me a working vorbis codec again. :/
19:06:32kugelno blue bars (even though 99.5fps when boosted), but 10fps less when unboosted
19:06:45linuxstbbertrik: Do you know how similar the Meizu's S5L8700 is to the Nano2G's 8701?
19:06:57 Join stripwax [0] (
19:07:23bertrikI only have the datasheet to the 8700, but I think quite similar
19:07:25 Quit stripwax (Client Quit)
19:07:42 Join stripwax [0] (
19:08:10bertrikat least the peripheral ranges seem to be the same (although 8700 has no crypto engine as far as I know) and I expect the actual peripherals to also work the same
19:10:08linuxstbI'm looking here - - is the status table up to date? (e.g. it says neither the bootloader or Rockbox build?)
19:10:25bertrikno it's not up to date
19:11:09bertrikthe bootloader does compile now for example
19:12:36bertrikat least the meizu M3 bootloader does, the M6 bootloader should be compileable with little effort
19:13:43bertrikI was planning to clean up the MeizuM6Port, create a MeizuM3Port page and clean up MeizuReverseEngineering
19:14:07CIA-71New commit by alle (r21814): Unify semitone and cent macros and make the formula a bit more obvious
19:14:43 Join DarkDefender [0] (
19:16:43 Quit robin0800 ("Leaving")
19:17:01linuxstbbertrik: I was just wondering if, now that's it's possible to run code on a Nano2G, it's worth setting up a minimal bootloader target in Rockbox, based on the existing 8700 code...
19:17:16kugelbertrik: no idea how to implement the wakup_signal stuff in asm :)
19:18:33bertriklinuxstb, not that hard I would guess, just a linker script and a bootloader main and some way of packaging the bootloader up so it can be sent to an ipod
19:18:57bertrikthe big problem is that we don't have a driver for the NAND so we also don't have a file system
19:19:17linuxstbYou have lcd drivers for both the m3 and m6?
19:20:01 Join TheSeven [0] (
19:20:15bertrikno, only for the m3, there has been some research into the m6 display (init sequences, getting an LCD ID) earlier but it never worked so far AFAIK
19:21:27TheSevenok, so i heard you're discussing nano plans here?
19:21:46*bertrik prods gevaerts and markun to start working again on their M6's
19:22:48linuxstbTheSeven: I was wondering about getting a very minimal (i.e. doing nothing) Rockbox bootloader build working for the Nano 2G. Rockbox bootloaders are basically cut-down versions of Rockbox itself, using the Rockbox low-level drivers, but with a "main()" that simply loads the main Rockbox binary from disk.
19:23:16TheSevenok... what kind of hardware access do these need?
19:23:28TheSevenconcerning NAND: we definitely have some whimory thing in there
19:24:40TheSevenright now we can't properly install a bootloader on the firmware partition yet, but that will come as soon as we find the relevant code in the disassembly
19:24:44bertrikTheSeven, yes, it's in the Meizus too
19:24:44linuxstbI'm not sure what you mean by "what kind of hardware access do these need?" - a functioning bootloader needs buttons, lcd, disk, but I was just thinking of a "do nothing at all" bootloader that can be used for experimenting.
19:24:55linuxstbThis "bootloader" would be the code uploaded via iBuggerLoader
19:25:23linuxstbIt's mainly to make use of the Rockbox build system to allow people to easily hack with C
19:25:28TheSevenin fact i'm wondering whether i should upload an extended ibuggerloader through ibuggerloader :-P
19:25:36linuxstb(and to share any existing code with the Rockbox Meizu ports)
19:25:56n1sany success with other new ipods yet?
19:25:57TheSevenonce i get these bulk IN endpoints working, you'll have some kind of minimalistic debugger then
19:26:09n00b81n1s: not yet
19:26:49n00b81n1s: Unfortunately the notes files are not yet compatible with 3g/4g. What iPod do you have btw?
19:26:59n1si have a 4g nano
19:27:16n00b81You can always do some testing :P But there are a ton of files
19:27:42TheSevenibuggerloader provides the following functions right now: reset, reconnect, write (i.e. copy data from usb to ram), fastwrite (the same, but 4byte-aligned), and execute. so you can use it to poke at AHB etc. to study hardware, but it would be easier, if we also had a read command, which depends on that evil IN pipe
19:27:59bertrikTheSeven, on the Meizu I'm using some kind of built-in DFU mode that allows me to write my bootloader test program to an 1MB NOR flash
19:28:47TheSevenbertrik: the nano has such a thing, too, but it's nonstandard and encrypted, so you would need to encrypt the file on a nano before you can upload it through dfu
19:28:50linuxstbTheSeven: Ah, so currently you can upload data to the Nano via USB, but not the other way around?
19:29:00TheSevenlinuxstb: yes
19:29:28TheSevenbertrik: and as itunes is not able to talk to that dfu mode, restoring the original nor would be hard, so it's too risky
19:30:14TheSevenibuggerloader was written to provide a replacement for that, by just uploading arbitrary data to arbitraty memory locations (usually SRAM), and execute them from there
19:30:59 Join fml [0] (n=4fd3c6a6@gateway/web/cgi-irc/
19:31:22TheSeveni was planning to set up some trivial c environment myself in fact... i.e. init code, linker scripts, ...
19:31:47TheSevensounds like that's your intention, too
19:32:28fmlHrm... I'm looking into the pitch screen code and can't see what can be done to significantly reduce the bin size. One thing would be to rewrite it from scratch with that aspect in mind and hope that the result will be better than what we have now. But I wouldn't like to do that.
19:33:17bertrikTheSeven, I'm copying a lot from similar rockbox targets :P
19:33:22linuxstbTheSeven: Yes, that's the idea.
19:33:45kugelI can't reproduce the 80fps
19:34:05linuxstbTheSeven: Do you have any feedback methods to prove your code is running once uploaded?
19:34:32TheSevenwell, a little. by making it crash, lockup, or return to ibuggerloader
19:34:59TheSevenas long as you don't touch SDRAM, you can return into the debugger by jumping to the LR passed to the entry point
19:35:15bertrikmaybe we could help, e.g. identifying the backlight enable, or maybe an LCD init sequence
19:35:33TheSevenbertrik: we have a NOR dump now, so that should be doable...
19:35:36bertrikwe already have code to write text to the screen that could help a lot
19:36:02TheSevenwhat would help even more would be to get that IN pipe working, so we could just read back memory
19:37:11TheSevenmy long-term plans (at least until i managed to decrypt nor yesterday) were to write a bigger clone if ibuggerloader in c, which provides a lot of additional functionality, using the openiboot usb stack, as this seems to fit to that hardware 100%
19:38:07 Quit Ubuntuxer (Read error: 110 (Connection timed out))
19:40:16bertrikI can have a look tonight at that dump, it's usually not hard to figure out the hardware connections, just takes a lot of patience
19:40:28TheSevenwhat i forgot concerning feedback methods: we have uart up and running, so if you have the appropriate hardware, you can use that
19:40:45bertrikWhen I find anything interesting, I'll put the results on the rockbox wiki if you don't mind
19:41:16TheSevenI'm right now disassembling that dump, until now straightforward with the control flow, to get an idea of the structure
19:42:29bertrikCan I download it somewhere?
19:45:42 Quit stripwax ("")
19:46:25 Quit r0b- (Read error: 104 (Connection reset by peer))
19:48:35 Join darkip [0] (
19:49:21darkiphey, is there any way to customise the database browsing to show the album title underneath each track title (like on the ipod classic firmware)
19:49:24 Join r0b- [0] (
19:50:13 Part r0b-
19:50:13 Join r0b- [0] (
19:50:38 Join toffe82 [0] (
19:52:33pixelmaI wonder if other people have this problem too but: in timestretch mode, just lowering the pitch there is a certain range where I don't hear anything (at around 60%), going lower gives me noise again, higher is no problem at all. I'm trying with music mp3's at different bitrates usually above 128kb/s vbr and cbr
19:53:13pixelmasometimes when adjusting a bit around those value I get a burst of noise
19:53:27CIA-71New commit by peter (r21815): Patch by Wincent Balin: convert pdbox from app to viewer
19:53:45pixelmaoh, and I already observed it while trying a build before the new UI
19:53:59pixelmathis is on a c200
19:56:21pixelmadarkip: I don't know a way to make the display use two lines per entry but I think you could maybe let it show the album title in front or after the title (I am not sure though and couldn't tell how)
19:56:45darkiphmm, so there's no native feature for this...
19:56:59darkipmaybe I'll have a dabble in the code and see what I can do myself
19:58:11pixelmayou can customise the view a bit using tagnavi syntax (it's explained in the DataBase wiki) but as I said only the entry itself
19:58:28darkipok, thanks again :)
20:00:31fmlpixelma: yes, I also experience strange behaviour with time stretch sometimes. Sometimes playback is stopped and the playlist is corrupted.
20:01:36pixelmadarkip: and of course the virtual "folders" and sorting can also be customised
20:02:14pixelmafml: "nice"
20:04:27 Quit pixelma (" laters")
20:05:31 Join Zagor [242] (n=bjst@rockbox/developer/Zagor)
20:07:11BagderZagor: will you make the new build table use new links to the title pics and create a fresh set?
20:08:19Bagderpossibly we should make them with a Makefile or something
20:08:35Bagderto make it easier to generate new ones when the builds file changes
20:09:12ZagorI'm simply adding a 'next if (-f "$dir/pic-$id.png");' to the loop
20:09:53Bagdergood idea
20:11:02BagderI'll also move the build scores into the builds file, no point in having two separate ones
20:11:27*gevaerts thinks that the scores need to be tweaked. They seem to favour kugel's server ;)
20:11:35Zagorlooking forward to some late night sessions next week :)
20:11:51Bagderyeps, that should help us get this go live soon
20:12:32kugelgevaerts: hehe, they just favor the fastest one :>
20:12:44Zagorkugels server is bloody invincible
20:13:30kugelgevaerts: your combined only has 20k more than the fastest split one?
20:13:49 Join cmwslw [0] (
20:13:53gevaertskugel: don't forget that there's still an old-style builder in there as well
20:14:07kugelah, right
20:14:21gevaertsZagor: is the 40838 cpu seconds on cancelled builds correct?
20:14:33JdGordon11 hours!
20:14:37Zagorwho runs jakorasia-cg? it's very fast with that low bogomips.
20:14:53Zagorgevaerts: it does look a bit high
20:15:11cmwslwhey guys- I'm cmwslw from Linux4nano
20:15:11gevaertsHow many servers didn't finish anything?
20:15:17Bagderbogomips is really bogo...
20:15:34Bagderhey cmwslw
20:15:45ZagorBagder: yeah, I'd just like to know what machine it is that is sort of the inverse of Atom
20:15:53gevaertsBagder: not always. My 1193 bogomips client is really slow :)
20:15:55cmwslwdoes rockbox have any info on the iPods clickwheel?
20:16:09Bagdergevaerts: ;-)
20:16:33cmwslwI'm talking about pinouts etc.
20:16:37BagderZagor: I believe saratoga's machine is a quad-core...
20:17:51Zagorit would be interesting to see the mhz values too
20:18:03kugelgevaerts: split them again, you were doing more builds with it :(
20:18:06Zagorperhaps they are only slightly less bogus
20:18:24JdGordonI love this line from the bogomips wp page... "It is not usable for performance comparison between different CPUs.[4]" :D
20:18:28 Join BdN3504 [0] (n=5ce539e0@gateway/web/cgi-irc/
20:18:32Bagdercmwslw: I can't seem to find anything like that in our wiki
20:19:10ZagorJdGordon: yeah, but practically nothing is other than a custom benchmark
20:19:12BagderJdGordon: I figure we can all agree to that!
20:19:57Zagorperhaps we should indeed to a small test compile as benchmark when the client starts
20:20:16Bagderwould probably be better, yes
20:20:28ZagorI don't want to rely on the speed number too much, but having it semi-reliable would enable more detailed optimization
20:21:17gevaertsZagor: MHz is tricky. It can change (laptops), and my arm machine doesn't even have it in /proc/cpuimnfo
20:21:20Bagder27K 32K 34K 36.9K 40.8K
20:21:34Bagderthe number of "wasted cpu seconds" just goes up!
20:21:43Bagderthose are the last 5 rounds
20:21:45Zagorgevaerts: bogomips changes with mhz on the laptops
20:21:50BdN3504my gigabeat fails initializing the database. i tried it with a current build and the 3.3 release, but neither works. the initialization seems normal (as in rockbox gathers datat for all 15780 files) then tells me to reboot to enable the db. when i do so, i get the normal committing database at startup, but it doesn't finish, only commits 3/9 and then cancels. when i try to access database in the main menu, i get
20:21:51gevaertsBagder: that could point to a bug :)
20:22:03Zagorwe could busy-loop for a second before reading /proc/cpuinfo :)
20:22:25gevaertsZagor: you could. It would still all be bogus of course :)
20:22:37BdN3504i look through all the useful tools section of the wiki, hoping there was one which would take the database creation to a pc instead of rb
20:22:44 Quit HellDragon (Read error: 104 (Connection reset by peer))
20:23:16BagderBdN3504: tools/database/
20:23:35 Quit fml ("CGI:IRC 0.5.9 (2006/06/06)")
20:23:35BdN3504but i couldn't find one. does such kind of software exist? (i.e. one that looks at the device and then creates the respective database files on it)
20:23:36ZagorI'm thinking about run all makes "nice" by the client btw.
20:23:40gevaertsZagor: one thing you could do is just look at how long each build takes, and use the build score/build time ratio to sort clients
20:24:03 Join HellDragon [0] (
20:24:10Zagorgevaerts: yeah but as far as I understand it tends to vary quite a bit
20:24:22Bagderyes, build times vary too much
20:24:29BdN3504do i have to compile that myself? is it linux or pc? will i have to just "make" in tools/databse?
20:24:40 Quit amiconn (" .")
20:24:41BagderBdN3504: yes, yes, yes
20:24:53gevaertsZagor: I'd really leave nice values to the people who run it. Some of us use dedicated VMs, so renicing won't buy you anything
20:25:00BdN3504linux or pc-< yes?
20:25:01gevaertsBdN3504: most people run linux on a pc
20:25:10BdN3504i mean windos
20:25:11BagderBdN3504: exactly
20:25:12BdN3504:) sorry
20:25:21BdN3504go figue
20:25:40BdN3504well you know what i mean. so it's only linux?
20:25:43Zagorgevaerts: right, but it does make the mother prioritized over the compilers. the ping response times are worryingly long at times
20:25:45BdN3504is it documented?
20:26:04 Join _lifeless [0] (n=lifeless@
20:26:11Zagoralso I don't see how it would be negative
20:26:21gevaertsZagor: ah yes. I'm not sure if that's the cause though. At least for me, network is *slow* while things are uploading
20:26:56gevaertsuploading uses curl, right?
20:27:19Zagoryes I think that's the main reason too, I just figured niceing has several positive effects
20:27:26Zagorgevaerts: curl yes
20:27:50gevaertsZagor: you're right. niceing shouldn't hurt
20:28:20gevaertsMaybe add the −−limit-rate option to the settables?
20:28:33BdN3504could you explain, why the database doesn't work anymore? or is this my device specific problem? does it have something to do with the limitations that can be set within rb?
20:28:49gevaertsPeople with limited connection probably know how much they can spare
20:30:45 Part cmwslw
20:31:48BdN3504i get "cc1: error: unrecognized option `-Wno-pointer-sign'" when i try make in tools/database
20:32:05linuxstbBdN3504: Have you checked your disk for errors? That seems to be the most common cause of problems.
20:32:23BdN3504yeah, i just ran scandisk on it
20:32:33BdN3504and chkdsk afterwards, no errors
20:34:37BdN3504if i delete all database files form the .rockbox dir on the device (except for the databas.ignores) and try to rebuild it, that should create an immaculate database, shouldn't it?
20:35:29 Quit TheSeven ("ChatZilla 0.9.85 [Firefox 3.0.11/2009060215]")
20:37:48linuxstbbertrik: The Meizu uses the ARM in big-endian mode?
20:38:06*bertrik blushes
20:38:15bertrikI don't know to be honest
20:38:20Bagderit does
20:39:09gevaertsI think one of the first meizu-related commits was to
20:41:03bertrikaccording to the s5l8700 datasheets the CPU can be configured for LE or BE
20:41:17 Join Thundercloud [0] (
20:42:19ZagorBagder: heh, I forgot to zero $wastedtime :-)
20:42:37Bagderthat explains it a little ;-)
20:42:50 Quit flydutch ("/* empty */")
20:42:53gevaertsZagor: were you trying to suggest something about all this? :)
20:43:40 Join Jaykay [0] (
20:44:27 Join TheSeven [0] (
20:44:29CIA-71New commit by peter (r21816): More work on PDBox by Wincent Balin. The PDBox plug-in is now working with the pdpod_test.pd file from the archive
20:46:22ZagorI'm considering not handing out targets someone else is uploading
20:46:28bertriklinuxstb, why are you asking?
20:46:50CIA-71New commit by bagder (r21817): moved the "score" into the builds file, removed the separate score file and ...
20:46:52JdGordonZagor: that makes sense...
20:46:55Zagorit requires some extra complexity in the form of a timeout (if the uploading client exits before completion) but it could save some time
20:47:16Bagderwhy would it save time?
20:47:33gevaertsZagor: maybe just put them lower in the sorted list
20:47:46Bagderright, that could be a sort key
20:48:05ZagorBagder: becuase in the case of big >30s targets it is unlikely that anyone will compile+upload in the time it takes for someone to just upload. better then to start on a target that isn't already compiled.
20:48:27Bagderyes, so we should factor that in in the sorting
20:48:29Zagoryeah, I meant sorting it away
20:48:38Bagdermakes sense
20:48:49Bagderit's better to hand out one that hasn't even begun uploading
20:48:53*gevaerts thinks that all problems should be fixed by sorting differently
20:49:10Bagderyes, the sorting is the key to most
20:50:14 Join stripwax [0] (
20:50:16Zagorbut this also requires client changes to notify the server when compilation is done. currently it only suggest that by requesting a new build. but it doesn't say which of its' builds are compiled.
20:50:37gevaertsZagor: maybe use the log upload?
20:50:42gevaertsIsn't that first?
20:51:05ZagorI don't want to mix apache logs and the perl script...
20:51:45Zagorcurrently the client uploads the log and the zip before saying "completed". only then does the server look for the log.
20:52:21linuxstbbertrik: Because I'm adding a nano2g build target, based on the meizus...
20:52:37CIA-71New commit by peter (r21818): PDBox: One file with stuff is enough....
20:52:44gevaertshm, yes. To do it cleanly you do need a new command
20:52:53bertriklinuxstb, could you help create an for the meizus? I already made an attempt at it.
20:53:38linuxstbbertrik: Possibly. I want to familiarise myself with the 870x first, by looking at the bootloaders.
20:56:44***Saving seen data "./dancer.seen"
21:02:17 Join notlistening [0] (
21:04:57 Nick n00b81 is now known as n00b81|away (n=n00b81@unaffiliated/n00b81)
21:06:47*linuxstb has a basic nano2g target and wonders whether to commit it, given his previous record with new ports... ;)
21:07:05 Join BryanJacobs [0] (
21:07:30gevaertslinuxstb: this time it's hardware that several other people also own
21:07:51linuxstbgevaerts: You think it's more popular than the Elio?
21:08:04*linuxstb can't believe that
21:08:12gevaertslinuxstb: I think so. Maybe not as popular as the DAX though
21:09:35JdGordonon the topic of new targets... any ideas how to go about cracking the iriver e100's firmware?
21:11:44 Quit thegeek (Read error: 104 (Connection reset by peer))
21:11:59linuxstbTheSeven: How is memory mapped on the Nano2g? On the Meizus, it looks like (from the Rockbox linker scripts) to have 16MB RAM remapped from 0x80000000 to 0x0, and 256KB of SRAM at 0x22000000.
21:13:11TheSeven50K BR @0x20000000, 256K SRAM @0x22000000, 1M NOR @0x24000000, 32M SDRAM @0x08000000
21:13:20 Quit freqmod ("No Ping reply in 90 seconds.")
21:13:40TheSevenduring boot first BR, and later SRAM is mapped to 0x00000000, however i haven't spotted SDRAM being there yez
21:14:23linuxstbSo when code is uploaded via iBuggerLoader, SRAM is at 0x0 and SDRAM is at 0x08000000?
21:14:37TheSevenhowever, if that's better for any reason, mapping SDRAM there is a matter of some control reg writes, i've already tried that and it worked first shot
21:15:38 Join freqmod [0] (
21:15:58TheSevenlinuxstb: yes. ibuggerloader uploads to 0x22000000 and jumps to 0x22000020, which maps to 0x00000000 and 0x00000020 respectively
21:16:18TheSevenchanging these values is no problem at all, though, as you may have seen in the python script
21:16:37 Quit preglow (Remote closed the connection)
21:16:40 Join preglow [0] (
21:16:56linuxstbDo you mean that SRAM is accessible at both addresses?
21:17:25 Join thegeek [0] (
21:17:31 Join dash32 [0] (
21:18:47linuxstbTheSeven: I have no opinions yet about where RAM should be mapped to, it's just that I'm currently adapting the Rockbox Meizu code to compile/link for the Nano
21:19:01CIA-71New commit by peter (r21819): Another patch by Wincent Balin (from the FS #10416 series): get rid of some warnings. PDBox now builds without any error or warning.
21:20:16TheSevenyou could even map SDRAM to zero before your code gets run, through an ibugger script
21:20:25TheSevenjust to show you the possibilities
21:20:44TheSevenso if any issues arise from the different memory layout, we can just make it meizu-like
21:23:20 Quit matsl (Remote closed the connection)
21:26:25 Quit Sajber^ (Read error: 104 (Connection reset by peer))
21:26:55 Join Sajber^ [0] (
21:28:47 Quit stoffel (Remote closed the connection)
21:31:41 Quit QncOPFGl (Read error: 110 (Connection timed out))
21:32:29 Join QncOPFGl [0] (
21:36:29CIA-71New commit by bertrik (r21820): S5L8700/Meizu: miscellaneous minor fixes, stubs added, keywords set
21:36:37linuxstbbertrik, gevaerts: Do I need a special arm-elf-gcc to compile the meizu builds? It was a while ago when I last ran
21:36:56gevaertslinuxstb: maybe just try?
21:37:00bertriklinuxstb, I didn't do anything special
21:37:10linuxstbgevaerts: I did, but get errors about big/little endian mismatches...
21:37:20gevaertsin that case, I guess you do
21:37:30linuxstbUnless my changes have broken it...
21:38:02*linuxstb re-runs
21:38:14gevaertsThe changes were on 2008-04-28
21:38:14linuxstbI guess it needs a big-endian libgcc.a
21:39:51BryanJacobsahh, I think I've finally got the interspersing loader working
21:40:15BryanJacobsthat was more effort than I thought it would be -_-
21:40:29bertrikI'm completely sure that I have a rockboxdev setup later than that
21:40:38linuxstbHow is code run on the Meizu? Is it written to flash at 0x24000000, and uses IRAM at 0x22000000 for data?
21:40:59*BryanJacobs spent half an hour trying to figure out why Wavpack was saying his files were invalid, only to discover that he'd toggled a boolean so that the first chunk was from the second file x_x
21:41:14linuxstbBryanJacobs: Just in time for mid-term evaluation... (although I've already passed you!)
21:41:21BryanJacobslinuxstb: yay!
21:41:40*BryanJacobs is happy that he has something that could potentially be committed, finally
21:41:56linuxstbHow much new code is needed in the core?
21:42:00BryanJacobsthis code doesn't interfere with anything except for Wavpack hybrid files with readable correction files present
21:42:38BryanJacobsaround 20 lines in playback.c for the buffering implementation, and around 40 lines in buffering.c to handle codec-specific callbacks
21:43:03linuxstbIt would be nice to see the binsize difference - see "rockbox-info.txt" with and without your changes.
21:43:07BryanJacobsabout 5 lines in metadata/wavpack.c to handle adding the callback
21:43:22linuxstbBut it sounds about as unobtrusive as it could be...
21:43:37BryanJacobsit only takes effect when this particular type of file is played
21:43:56BryanJacobsthe only other harm it does is making struct memory_handle and struct mp3entry slightly bigger
21:44:07gevaertsBryanJacobs: you mean you spent weeks on writing 60 lines of code? ;)
21:44:20*gevaerts hides
21:44:24BryanJacobsgevaerts: this was only written since Thursdayish, actually
21:44:40BryanJacobsand there's also code in the wavpack codec... linuxstb only asked for "in the core"
21:45:00gevaertsI know :)
21:45:42BryanJacobsoh also there are a few extra arith ops in buffering.c since now widx != available
21:45:45linuxstbgevaerts: Can you answer my meizu question from about 4 minutes ago?
21:46:07gevaertslinuxstb: I don't really know...
21:46:38linuxstbWho would know? markun? bertrik?
21:46:55gevaertsI suspect markun knows this best
21:47:11bertriklinuxstb, I'm not sure, just look at the link file or .map
21:47:52linuxstbbertrik: Yes, that's what I'm doing. I just wanted someone to confirm my understanding. I can't look at the .map yet though (can't build it...)
21:48:06 Quit notlistening (Remote closed the connection)
21:48:07 Quit darkip (Read error: 110 (Connection timed out))
21:48:14 Join bluebrother [0] (n=dom@rockbox/developer/bluebrother)
21:48:33linuxstbOK, has finished, and the meizu builds fine now...
21:48:38gevaertslinuxstb: IRAM at 0x22000000 sounds familiar, but I have no idea where the code is
21:49:12linuxstbAccording to the map, code starts at 0x24000000, which the lds says is the flash
21:49:36bertrikyeah, as far as I can see, it runs from flash and uses only IRAM
21:49:37linuxstbcode, data, stack, bss are all in IRAM at 0x22000000
21:49:42*gevaerts slaps his forehead
21:50:02gevaertsbertrik: you *can* run directly from dfu, but you have to change the .lds then
21:50:13 Nick n00b81|away is now known as n00b81 (n=n00b81@unaffiliated/n00b81)
21:50:13bluebrotherpetur: just spotted filenames like rsqrt~.c in pdbox. Are those filenames correct? Looks quite strange
21:50:24peturthey are
21:50:35peturboth strange and correct ;)
21:50:40 Join BHSPitLappy [0] (n=BHSPitLa@unaffiliated/bhspitmonkey)
21:50:43bertrikgevaerts, you mean we could build our own "SST39VF800.dfu"?
21:50:45bluebrotherok. Weird, but well :)
21:51:17gevaertsbertrik: something similar anyway. I'm not sure if I'd use dfu for the second stage
21:51:50linuxstbSo currently you permanently flash a bootloader via dfu?
21:51:59bertrikyes :)
21:52:39 Quit mcuelenaere ()
21:54:46gevaertsbertrik: see firmware/target/arm/s5l8700/ for the DFULOADADDR bit
21:56:11CIA-71New commit by zagor (r21821): Added sub 'command' to reset ping timer each command. Added timer to measure and report ping response time.
21:56:12bertrikwell flashing the bootloader worked so far for me for getting the basic drivers implemented
21:56:47bertrikwe probably still need to think a bit about the whole rockbox bootloading process
21:58:29bertriklike, how to dualboot rockbox with the OF on the meizus
22:00:57bertriklinuxstb, the NOR image of the OF contains just basic setup code and a Rar archive that is extracted to DRAM. The Rar takes about 600k of the total 1 MB NOR, so we have some room here
22:04:16TheSevenlinuxstb: ipods are little-endian btw.
22:04:29TheSevenat least unless you switch it :-P
22:07:03linuxstbTheSeven: Yes, I assumed that (from your code)
22:07:19 Join notlistening [0] (
22:07:45TheSevenjust wanted to clarify this because you were talking about needing a bigendian libgcc
22:07:50linuxstbbertrik: Is the OF available unencrypted?
22:08:06linuxstbTheSeven: Thanks. But that was just so I could compile the meizu builds
22:09:00 Join JdGordon| [0] (i=43a02c5a@gateway/web/freenode/x-539d944851b9861f)
22:12:53notlisteningI have seen a mail on the Dpeech Dispatcher mailing list about a plugin to use speech dispatcher in rockbox. This must mean that the writer is ruunning rockbox in linux for this to work, is that right?
22:14:50Zagornotlistening: or is he talking about using the dispatcher to create voice files?
22:20:56 Join pixelma [50] (n=pixelma@rockbox/staff/pixelma)
22:21:01notlisteningyeah need to check with the guy it is not clear
22:22:11CIA-71New commit by zagor (r21822): Report to server when zip upload starts. Run make with nice.
22:22:34stripwaxDo 60GB ipod 5Gs exist with 400mAh batteries instead of 600mAh? Also is there such a thing as a 60GB 5.5G ?
22:23:10gevaertsstripwax: in the world of refurbished players, who knows?
22:23:39gevaertsEspecially 60GB 5.5G sounds reasonably likely
22:24:09stripwaxyeah. specifically I'm wondering about FS #8596 and whether the battery on my 60GB 5G is actually materially different to Uchida's
22:24:40stripwaxOr, indeed, whether my battery is somehow an outlier with better stats and Uchida has the standard 60GB 5G battery
22:25:00bertriklinuxstb, yes
22:25:37CIA-71New commit by zagor (r21823): Accept _UPLOADING response...
22:26:24stripwaxIf the 400mAh and 600mAh have different discharge curves, do that just mean we need to define BATTERY_TYPES_COUNT to 2 for ipodvideo and setup two curves?
22:28:24ZagorI doubt they have different curves. don't curves look pretty much the same for each battery chemistry?
22:30:01stripwaxPerhaps; his readings seem to be consistently 100mv or so below mine so maybe the specs of the cell (rather than the chemistry) is somehow different
22:30:22linuxstbbertrik: So you could modify that file to include a Rockbox bootloader?
22:30:50stripwaxnearly 200mv actually
22:30:59 Join Ubuntuxer [0] (
22:31:16Zagorstripwax: that's a lot. is the lower voltage the "400mah" battery?
22:32:34stripwaxZagor - that's something I would like to find out .. his runtime is about two-thirds of mine, and his discharge curve matches the current Rockbox ipod 30GB (400mAh) curve, so that is my suspicion.
22:32:50stripwax(and his has the lower voltage, yes)
22:32:51gevaertslinuxstb: the meizu OF is basically an 8K unrar tool followed by a rar archive that has the main binary
22:33:24 Quit bmbl ("Bye!")
22:33:47linuxstbgevaerts: Do you have a copy?
22:34:29linuxstbBut I guess there's not much point in implementing a "real" bootloader until you can read from the NAND...
22:34:42bertriklinuxstb, sadly so
22:35:05Zagorstripwax: these are single-cell batteries, aren't they? liion cells are 3.7V. starting out 0.2V lower than that looks like a malfunction to me. though I'm no battery guru.
22:35:15linuxstbbertrik: Is that going to be troublesome? e.g. is there an unknown FTL?
22:35:18*Unhelpful is going to go ahead and add the addtargetdir and asmdefs changes... not any of the crossdev changes yet, though.
22:35:29stripwaxWould be very useful to see other users'/dev's discharge curves for 30GB 5G models and 60GB 5G models.
22:35:41Zagorstripwax: indeed
22:35:55bertriklinuxstb, it uses the 'whimory' FTL
22:36:19gevaertslinuxstb: I have some OF copies, yes
22:36:28bertrikthe openiboot project has some read-only code for it apparently, but nobody took a serious look at it AFAIK
22:37:03stripwaxZagor - mmm.. so, my starting voltage is reported as something like 4.2V. His is starting out 4.1V ish. My ending voltage (where the knee really kicks in) is appx 3.8V; his is a shade over 3.6V.
22:37:07linuxstbbertrik: So that's used on all the new ipods?
22:37:27 Part Ubuntuxer
22:37:38bertriklinuxstb, I don't know for sure
22:37:45Bagderlinuxstb: I think that's the suspicion as its mentioned in the nano2g and used by the iphone
22:37:55stripwaxZagor - if the cell is 3.7V, I'd have thought the starting voltage would be 3.7V
22:38:21stripwaxI'm also no battery guru
22:38:22bertrikstripwax, most battery benches I've seen had their knee at about 3.6V
22:38:42kugelIIUC 4.1x to 3.7 applies to all li-ion
22:39:23Zagorkugel: so 3.7 is basically the floor voltage?
22:39:25gevaertslinuxstb: do you want some to have a look?
22:39:42linuxstbgevaerts: Yes, I'll have a quick look at one.
22:40:02kugelZagor: that's how I understood it, yes. I.e. you should shut down soonish of you go under 3.7V
22:40:13BdN3504i can't fix my database, can anyone help me? please, just say yes or no, because i won't stick around if nobody is willing...
22:40:15TheSevenLi-Ion/Li-Poly: Full: 4.15-4.25, Empty: 3.4-3.7
22:40:29linuxstbgevaerts: I'm guessing a similar technique to mktccboot will work for dual-boot.
22:41:01gevaertslinuxstb: I haven't looked at that one yet
22:42:32 Quit BdN3504 ("CGI:IRC (EOF)")
22:42:33Unhelpfulis there any objection to changing the substitution in functions.make so that the dep is on $BUILDDIR/lang/lang.h instead of $BUILDDIR/lang/lang_core.o? the former will be used anyway if make dep is run after those files are generated.
22:43:29ZagorUnhelpful: but make dep isn't run after those files are generated
22:43:34stripwaxTheSeven/kugel - cool, thanks. So having 3300 as 'battery level dangerous' and 'battery level shutoff' doesn't make sense?
22:43:44 Join amiconn [0] (i=quassel@rockbox/developer/amiconn)
22:43:44 Quit pixelma (Nick collision from services.)
22:43:44 Join pixelma_ [0] (i=quassel@rockbox/staff/pixelma)
22:44:02 Nick pixelma_ is now known as pixelma (i=quassel@rockbox/staff/pixelma)
22:44:12TheSevenstripwax: <3.3 will heavily damage it, so that's about the lowest you may ever go
22:44:13kugelstripwax: that sounds quite dangerious
22:44:19 Join pixelma_ [50] (n=pixelma@rockbox/staff/pixelma)
22:44:22 Quit mt (Read error: 104 (Connection reset by peer))
22:44:23UnhelpfulZagor: really? i tend to run it occasionally to make sure deps aren't out of date.
22:44:42 Quit pixelma_ (Client Quit)
22:44:49TheSevennormally you'll only go to ~3.6, but i've also seen devices going to 3.4 regularly
22:45:05TheSevenprobably depends on some manufacturing details whether the cells survive that or not
22:45:24TheSeven>4.25 is also an easy way to kill them
22:45:36 Quit BryanJacobs ("Java user signed off")
22:46:01stripwaxTheSeven - hm, but fully charged my battery reports >4.25 ..
22:46:06bertrikthis battery manufacturer shows curves down to 3.0V :
22:46:31TheSevenhowever, because of the discharge curve of li-ion, the runtime will not be a big difference between 3.4 and 3.6
22:47:04 Join ybit [0] (n=ybit@unaffiliated/ybit)
22:47:06bertrika lot of rockbox targets have their shutdown limit at 3.3V
22:47:36Unhelpfulalso, i simplified a bit locally... splits the whole input into files, and one regexp determines whether a file is a target (ends in :), a fully-specified source (starts with /) or if it needs a prefix added. this is in anticipation of asmdefs, which addes some more generated headers. also, it eliminates all of the special-case sed rules from mkdepfile, except for the lang headers that need a dir added.
22:47:43TheSeventhe manufacturers always max this out :-P
22:47:58TheSevenhowever, they definitely live longer if you don't go below ~3.4
22:48:07TheSeven3.0 should really be avoided
22:48:17ybitwhat could be causing rockbox to skip songs after about 1:30 into the song?
22:48:19bertrikso maybe we should increase this a little, say give a warning at 3.6V and shut down at 3.4V?
22:48:38ZagorUnhelpful: actually I don't remember why it looks like this. :) if it works, go ahead. simple is good.
22:48:49TheSevenwarning at 3.6 is a good idea, as it will very quickly go down from that point
22:49:26linuxstbybit: What kind of songs (what audio format?) and what device are you running Rockbox on? Does it always happen on every song/
22:49:40UnhelpfulZagor: i actually just realized that the special cases can be cleaner/safer if the sed is before addtargetdir... i'll fix that up and test it before commit. :)
22:50:05Unhelpfuli don't think addtargetdir needs a rename for this, seeing as that's still what it's really doing
22:50:59gevaertslinuxstb: since the meizu OF uses clearly separated loader and main binary, we can basically do whatever we like I think
22:51:15ZagorUnhelpful: I agree
22:51:16stripwaxTheSeven - mine would appear to go down very quickly after about 3.8, 3.6 probably critical, I doubt mine would even go as low as 3.4 judging by the battery bench
22:51:46TheSevenwell, that depends a lot more on the electronics than on the battery
22:51:54stripwax(although the other guy's battery in FS #8596 looks like it's fine to 3.6V, then fast from 3.6V to 3.4V)
22:52:14JdGordon|kugel: is there any reason to not commit the custom vp patch as it is, and fix up any redraw issues as they come up?
22:52:15TheSevenif shutoff limit is 3.3, it will shut off at 3.3, at least as long as the electronics are capable of operating at that voltage
22:52:15stripwaxI'm wondering what could cause the differences
22:52:38wincentbluebrother: The "~" sign at the end of the operator in Pure Data means that the operator works with sound signals (as opposed to control signals). More evident example is "osc~".
22:52:40kugelJdGordon|: not really, I actually have a quite committable version here
22:52:56TheSeveneither different manufacturer with slightly other discharge curve, or maybe just measuring issues?
22:53:03gevaertswincent: they use hungarian filenames?
22:54:17wincentgevaerts: Not that I am aware of. But "~" looks like a sine curve, does it?
22:54:56stripwaxTheSeven - I'll rerun a couple times and see if I can get same reproducible result. Battery is stock and I'm assuming the curve already in powermgt-ipod-pcf.c was also on stock battery. Could 200mv diffs be a tolerance issue alone?
22:55:14 Join gregzx [0] (
22:55:19gevaertswincent: it does a bit, yes :)
22:55:39UnhelpfulZagor: is there a reason i'm not understanding behind the used of $(1)_ as the output file in mkdepfile?
22:56:02Unhelpfulnevermind, i see now :/
22:56:29TheSevenstripwax: depends on the technology they use for measuring it. probably some kind of resistor-based voltage divider connected to an ADC. if these are 10% tolerance resistors, and one is 10% too high, and the other one 10% too low, that could be the issue
22:56:48***Saving seen data "./dancer.seen"
22:56:50ZagorUnhelpful: I guess a small comment would be in place...
22:56:53TheSevenif it's just a constant factor that your curve seems to be off, then you're probably having a measuring problem
22:57:09UnhelpfulZagor: i'll add that, too, then :)
22:57:54ybitlinuxstb: every song
22:58:00ybitipon nano
22:58:20stripwaxTheSeven - would it make more sense then for rockbox to use the actual min/max observed measured voltage rather than hardcoded readings that may differ due to tolerance?
22:58:21ybitwell, it doesn't skip @ the 1:30 mark every song
22:58:36ybitit might be the files, who knows. i'll monitor it a little longer
22:59:17stripwax(I realise that the answer may be 'yes but that hasn't been implemented anywhere' of course :-)
22:59:21linuxstbAnd what exactly do you mean by "skip"? It moves to the next track?
22:59:31TheSevenstripwax: depends on the hardware. if there is some kind of hardware shutoff at 3.3 or some such, one could try calibrating to that. however generally these measurements don't seem to be off that badly
23:00:04bertrikmaybe it is just a different battery chemistry, like LiIon vs. LiPolymer
23:00:32stripwaxbertrik - presumably both ipod batteries have the same chemistry? or is the hardware capable of charging both
23:02:37 Quit _zic (Remote closed the connection)
23:03:06TheSevenli-poly vs. li-ion is the same charging strategy
23:03:18TheSevenmaybe voltage levels tuned by 0.05V or some such
23:03:47stripwaxinteresting. judging by your numbers then my battery sounds more like a li-poly than a li-ion but I imagine I would have to dismantle the ipod to find out :(
23:04:34TheSevenli-poly usually have slightly higher voltages than li-ion (as far as i've observed)
23:04:39amiconnAll supported irivers, iaudios and ipods have LiPoly batteries. Only the archos FM/V2 Recorder has plain LiIon
23:04:40TheSevenhowever, they're very very similar
23:06:19stripwaxamiconn - not according to this (which could very well be totally wrong)
23:06:20amiconnNot only those, but most others, except the few with AA or AAA rechargeables or non-rechargeables
23:07:34amiconnI'm quite sure this list is incorrect if they write LiIon, although the two types are very similar wrt chemistry
23:07:45 Join dmb [0] (n=Dmb@unaffiliated/dmb)
23:08:07 Quit n1s (Remote closed the connection)
23:09:30amiconnLiIon has a bit lower voltage, and needs a solid metal case
23:09:48amiconnAll those foil wrapped "LiIon" batteries are in fact LiPoly
23:10:33stripwaxgreat info, thanks!
23:10:49 Join RalpH_himself [0] (
23:11:05 Part RalpH_himself
23:13:42 Quit dash32 (Read error: 110 (Connection timed out))
23:14:31 Join dash32 [0] (
23:19:11 Nick fxb is now known as fxb__ (
23:19:21 Quit Xerion (" ")
23:22:45 Nick fxb__ is now known as fxb (
23:23:52 Nick fxb is now known as fxb__ (
23:25:56CIA-71New commit by zagor (r21824): Down-sort builds that are being uploaded. Upsort zip builds without changing score values.
23:26:59 Quit JdGordon| ("Page closed")
23:27:03 Quit Jaykay ("ChatZilla 0.9.85 [Firefox 3.5/20090624025744]")
23:27:54 Quit Thundercloud (Remote closed the connection)
23:36:39 Join robin0800 [0] (
23:39:55kugelJdGordon: ping
23:42:45 Join robin0800_ [0] (
23:43:02 Quit robin0800 ("KVIrc Insomnia 4.0.0, revision: , sources date: 20090520, built on: 2009/06/08 19:18:46 UTC")
23:45:08 Part robin0800_ ("No matter how dark the night, somehow the Sun rises once again")
23:46:32 Join safetydan [0] (n=deverton@rockbox/developer/safetydan)
23:46:59JdGordonkugel: brup
23:47:30kugelI'm going to post a new customlist patch in a few minutes
23:48:50JdGordonim about to ruin my day by heading into work so ill look later
23:48:59scorche|shit is sunday!
23:49:11JdGordonI'd like to get in in soonish, we can work out the bugs later
23:49:26JdGordonI know.... but i want to take time off so need to make up sundays :(
23:49:43scorche|shand it is...almost 3
23:49:45*JdGordon dosnt have the willpower to work from home
23:50:27*scorche|sh is apparently watching a bunch of guys stand around on TV
23:51:31scorche|shspace shuttle going off today..
23:51:36kugelJdGordon: "we work out bugs later" as in you point to me when a report comes in? :p
23:51:47JdGordonsomething like that
23:51:49*gevaerts shoves people into the other channel
23:51:59*scorche|sh notices that he is in the wrong channel and blames JdGordon
23:52:06kugelJdGordon: btw, you broke themes with the statusbar work.
23:52:19*JdGordon hides
23:52:24kugelsuch as cabbiev2, it has "statusbar: on" in the settings
23:52:39JdGordonyeah, forgot to fix that.. ill do it now
23:54:49 Quit bluebrother ("leaving")
23:55:27JdGordonarg... perl :'(
23:57:22JdGordonwho knows
23:57:42kugelI touched it
23:58:00kugelyou need to look at WPSLIST though
23:58:14JdGordoni've modified both...

Previous day | Next day