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 2009-07-14

00:00:45amiconnA piezo works best if you feed it its resonant frequency. Of course just for a couple of cycles if you want a click and not a beep
00:01:16 Quit petur ("Zzzzz")
00:02:03JdGordon|ah ok
00:02:07kugeldoes anyone see where configure touches /tmp ?
00:02:36kugelit seems to be sed failing
00:02:52Zagorkugel: would you mind testing before committing next time?
00:02:59bertrikthe OF does indeed have a more tone-like quality to it
00:03:28kugeltesting this way was easier
00:03:31 Quit martian67 (Read error: 60 (Operation timed out))
00:03:51kugelit's a good change anyway, no matter if it fixes my issue
00:04:14gevaertskugel: if /tmp/ is too small for these here-documents, your system is seriously broken
00:04:41kugeli.e. I tested if I didn't break configure itself. My problem really matter for te commit
00:04:59gevaertskugel: what's /bin/sh? bash?
00:05:01 Quit stripwax ("")
00:05:17bluebrotherit might make sense to make the build client run with LANG=C
00:05:30Zagorbluebrother: yes, the next version will do that
00:05:41Zagorand check for "No space left on device"
00:05:49 Join martian67 [0] (n=martian6@about/linux/regular/martian67)
00:05:51kugelgevaerts: yes
00:06:17bluebrotherconverting the whole Rockbox site to utf-8 might be a good thing too. Sounds like a task for MrSomeone :)
00:06:36 Join stripwax [0] (
00:07:47 Quit bluebrother ("zzz")
00:10:55kugelgevaerts: sed -u seems to help, although cat makes problems then
00:11:18kugelbut I also noticed the 1MB tmp was full also because of a failed earlier
00:11:54gevaertsbash seems to be broken for here-documents and $TMPDIR. That's only a few K though...
00:12:18 Quit advcomp2019 (Read error: 113 (No route to host))
00:12:50kugelalright, I just freed 0.99MB, it should be fine now :p
00:14:40kugelgevaerts: using < and << should normally mean to not create a file at all, doesn't it?
00:14:50Zagorkugel: why don't you want a normal /tmp dir?
00:14:58gevaertskugel: here documents need temp files
00:15:38kugelZagor: I want, but I don't know what the ownder planned with it
00:15:47kugelowner even
00:15:55Zagorbut you run a vm?
00:16:03kugelit may have a reason, and I want to try to not break the vm already
00:16:13kugelyes, but that runs multiple users too
00:16:16gevaertskugel: it comes pre-broken
00:19:17CIA-71New commit by zagor (r21850): Added MESSAGE command. Set LC_ALL to C. Wait before exiting when stalled connection. Kill builds with INT instead of KILL, to allow cleanup.
00:19:29kugel1MB is apprently plenty, I just didn't see that didn't cleanup the failed download
00:20:21JdGordon|Zagor: and check for "No space left on device" <- and do what when that happens?
00:21:40Zagoralmost. the server ignores that client until it reconnects. the build is not marked as completed and is thus done by someone else.
00:21:59Zagorit also sends a message to the client console
00:23:54 Quit advcomp2019_ (Read error: 113 (No route to host))
00:25:51stripwaxDoes anyone know how well USEC_TIMER works over clock speed changes on pp? I'm suspicious that it doesn't give consistent timings, causing curious jerkiness in pictureflow, possibly causing jerkiness in list viewers in general. Another reason I suspect this is due to the 'experimental' gui cpu boost patch making pictureflow jerkiness worse/weirder
00:27:28stripwaxThe ipod video button driver uses USEC_TIMER and I'd like to experiment with using some alternative (if there is one). I'll also see if video behaves nicer with the zero-wait-state patch
00:28:13 Quit flydutch ("/* empty */")
00:30:25amiconnThe PP usec timer isn't affected by cpu clock changes at all (well, except when switching to the 32kHz hibernation clock, which we don't do yet)
00:31:07amiconnSame applies to the two programmable timers. They're all running from the base clock and need no adjustment for clock changes
00:31:58stripwaxamiconn - Thanks as always! So bang goes that unfounded theory.
00:33:09amiconn(for those who are curious - the hibernation clock turns the usec timer into a sec timer)
00:36:43stripwaxcute. out of interest, what's stopping switching to the hibernation clock (a lot of driver rework needed or is there not a lot to it?).
00:37:43 Quit jgarvey ("Leaving")
00:38:57amiconnAll the hardware management threads need to learn how to go to hibernation mode and wake up again
00:39:52amiconnOther than that it's fairly easy on iPod G1 and G2, I already did it months ago using experimental code
00:40:37TheSevenamiconn: how does this clock switch work on the older ipods? is it a gipo?
00:40:39amiconnI have no idea what additional measures are needed on the newer ipods which have the PCF50606 PMU
00:41:17TheSevenexactly that switch prevents us from properly resetting the 2g nanos
00:41:56amiconnThe old ones need to make sure backlight is off, shut down the codec via i2c, the put the sdram into self-refresh and finally switch to the 32kHz source and switch off the 24MHz base clock
00:42:04TheSeven(bootrom tries to lock a PLL at 11GHz and wonders why that doesn't work, if you don't switch to 32kHz before resetting)
00:42:56amiconnThe hibernation code needs to be in iram because of this. SDRAM cannot be accessed in self-refresh mode
00:43:32 Quit dfkt ("-= SysReset 2.53=- Ph'nglui mglw'nafh Cthulhu R'lyeh wgah'nagl fhtagn.")
00:44:16amiconnIt's not gpio, but several bits in the PP clock and ram controllers
00:45:27 Quit r0b- (Read error: 110 (Connection timed out))
00:46:31TheSevenamiconn: i mean just switching the cpu clock source, not the other steps
00:46:31amiconnYes, that's done by selecting another clock source in the clock controller
00:47:01TheSevenok... on the nano2, that's done by something external :-/
00:47:35amiconnWell, it's an entirely different SoC
00:47:36TheSevenand when booting, they lock the PLL at 200MHz based off the 32kHz source
00:47:51kugelpixelma: did you get anywhere with testing the pla rework yet?
00:47:54TheSevenjust hoped they could have done that on the old ones, too
00:48:25amiconnPP has two base clocks, 32kHz and 24MHz. The programmable PLL uses the 24MHz source as its base
00:49:36amiconnThe 24MHz source can be disabled. I think that actually disables the oscillator
00:49:48 Join evilnick_home [0] (
00:50:27ZagorJdGordon|: do you still have a build-NNNN dir left?
00:51:02 Nick fxb is now known as fxb__ (
00:51:10JdGordon|nice welcome message btw
00:51:46Zagorok so INT doesn't work. I'll try the TMPDIR fix instead.
00:51:55kugelJdGordon|: are you going to have time tonight to skim through the customlist patch?
00:51:59JdGordon|Zagor: the folder with the unlink errors are a different number than the remaing build- dir
00:52:28JdGordon|yeah, that dir is from 3 days ago!
00:52:42 Join fdinel [0] (
00:52:44AlexPZagor: When does the score update? Before the most recent build, I got your score is x, you are fast. No builds were completed, and I just got the same score and message again.
00:54:07JdGordon|kugel: i dunno... i'll have time alsmost for sure, but motivationj and energy :(
00:54:20 Quit shotofadds ("Leaving")
00:54:21ZagorAlexP: the score updates at the end of each round.
00:54:58AlexPZagor: So I take it that it should have changed for me, having failed to do any builds
00:55:35ZagorAlexP: I would have thought so, yes...
00:56:15Zagorah, bug. it doesn't. fixing...
00:57:16***Saving seen data "./dancer.seen"
00:57:35kugelgah, another bug
01:08:27 Join coney [0] (n=james@2001:470:8:8b:203:47ff:fedf:1267)
01:08:36 Quit jfc^2 (Remote closed the connection)
01:09:05 Join n00b81 [0] (n=n00b81@unaffiliated/n00b81)
01:09:23 Join martian67_ [0] (n=martian6@about/linux/regular/martian67)
01:09:36 Join jfc^2 [0] (
01:10:11 Quit martian67_ (SendQ exceeded)
01:11:01 Join martian67_ [0] (n=martian6@about/linux/regular/martian67)
01:13:51CIA-71New commit by zagor (r21851): Don't let children inherit mother's SIGINT handler.
01:14:11amiconnZagor: Is there a reason why you used SIGINT and not SIGTERM?
01:14:24 Quit mcuelenaere ()
01:14:45Zagoramiconn: lack of brains :-) TERM is what I really meant
01:15:32amiconnNow the client (probably the build child) always prints "received interrupt"
01:15:54Zagoryes, it's a mistake.
01:16:44CIA-71New commit by zagor (r21852): Kill children with TERM instead of INT.
01:18:47 Quit martian67 (Connection timed out)
01:19:12 Join r0b- [0] (
01:21:40saratogaargh cook undefs debugf
01:23:31 Quit JdGordon| ("Miranda IM! Smaller, Faster, Easier.")
01:24:55 Quit agaffney (Read error: 110 (Connection timed out))
01:25:52 Quit Thundercloud (Remote closed the connection)
01:28:11 Quit stripwax ("")
01:31:20saratogaif I include mdct_lookup.h in two different .c files in the same codec, do the lookup tables get allocated twice? or is the linker smart enough to use the same table for both .c files?
01:35:07amiconnCode in a .h file? Eww. In this case you'll most probably have them allocated twice
01:36:04kugelsaratoga: move it in a c file maybe?
01:36:29 Join JdGordon| [0] (i=46077f47@gateway/web/freenode/x-bc6fba0f5c4d2ab0)
01:36:34amiconnThese tables are static - the linker doesn't know they're there
01:37:12saratogawell thats annoying
01:37:15amiconnIf they wouldn't be static, you would get a duplicate symbol error instead
01:37:27saratogacook uses the same trig tables as the mdct lib
01:37:34kugelWhat's the problem with the C file?
01:37:38saratogaso i'd like to be able to delete cooks and somehow include the mdcts
01:38:05saratogawithout having them allocated twice
01:38:23kugelmdct_lookup.c, then have extern lookup_table[] in mdct_lookup.h
01:38:25amiconnMove the tables to a .c file, un-static them, and keep the declarations in the .h
01:39:01saratogadoes unstaticing them have any impact on other codecs that use the mdct library?
01:39:38amiconnNot unless they're also doing such a weird thing as duplicating them
01:39:46saratogaok good
01:40:13saratogaalso, I notice cook uses a lot of 16 bit constants for trig functions
01:40:13 Join Blue_Dude [0] (
01:40:31 Join Lazatar [0] (
01:40:33saratogathats got to be slower then a full 32 bit mul right since it has to mask them?
01:40:55amiconnIiuc this mdct lib is our own anyway, so we're not forced to keep wrong constructs for potential upstream syncs
01:41:15amiconnIt depends on the architecture
01:42:57amiconnIt also depends on how the constants are loaded
01:43:41 Quit Zagor ("Leaving")
01:44:17kugelsaratoga: if it's signed then it has to do sign extension I think
01:44:43saratogaugh cook is such a mess . . .
01:45:14Blue_DudeHey, can someone have a look at my patch at FS #10199? It's ready for a hard look now. Thanks.
01:48:20 Quit martian67_ (Connection timed out)
01:48:23kugelBlue_Dude: I think you're better off writing to the -dev list with a somewhat detailed description what it does
01:50:00 Quit kugel (Remote closed the connection)
01:50:08 Quit Lazatar (Remote closed the connection)
01:50:30 Join Lazatar [0] (
01:52:49Blue_DudeI wrote an email three days ago and have heard squat from the -dev list. The functional description is in FS where it's supposed to be.
01:56:29 Quit Lazatar (Client Quit)
01:56:38 Join martian67 [0] (n=martian6@about/linux/regular/martian67)
01:57:43 Quit martian67 (SendQ exceeded)
01:59:16 Join martian67 [0] (n=martian6@about/linux/regular/martian67)
02:03:34amiconnUnhelpful: Regarding the offset autogeneration mechanism - there are differences between architectures
02:04:09amiconnARM and MIPS use .word, m68k and SH1 use .long
02:04:27amiconnAnd there's the _ prefix for SH1
02:05:12 Join bzed_ [0] (
02:05:20 Quit martian67 (SendQ exceeded)
02:07:12 Quit JdGordon| (Ping timeout: 180 seconds)
02:07:30 Quit bzed_ (Read error: 104 (Connection reset by peer))
02:08:12 Join bzed_ [0] (
02:08:25 Quit bzed (Read error: 104 (Connection reset by peer))
02:08:28 Quit graey (Read error: 110 (Connection timed out))
02:08:29 Nick bzed_ is now known as bzed (
02:09:53 Join martian67 [0] (n=martian6@about/linux/regular/martian67)
02:10:45 Quit martian67 (SendQ exceeded)
02:11:10 Join martian67 [0] (n=martian6@about/linux/regular/martian67)
02:13:20 Quit Blue_Dude ("ChatZilla 0.9.85 [Firefox 3.0.11/2009060215]")
02:14:27 Join agaffney [0] (n=agaffney@gentoo/developer/agaffney)
02:24:29 Join martian67_ [0] (n=martian6@about/linux/regular/martian67)
02:29:53 Part toffe82
02:34:10 Quit martian67 (Read error: 110 (Connection timed out))
02:35:19 Join advcomp2019 [0] (n=advcomp2@unaffiliated/advcomp2019)
02:49:24 Quit martian67_ ("out")
02:49:28 Join martian67_ [0] (i=martian6@2001:470:b:356:0:0:0:beef)
02:49:46 Quit martian67_ (Remote closed the connection)
02:49:56 Join martian67 [0] (i=martian6@about/linux/regular/martian67)
02:51:00 Quit martian67 (SendQ exceeded)
02:52:24 Join martian67 [0] (i=martian6@about/linux/regular/martian67)
02:54:21 Quit pixelma (Read error: 60 (Operation timed out))
02:57:18***Saving seen data "./dancer.seen"
03:03:16 Part wincent ("Kopete 0.12.7 :")
03:08:12saratogaI cannot believe how hard its proving for me to get rid of the stupid duplicate trig table in cook
03:08:56 Quit amiconn (Read error: 110 (Connection timed out))
03:10:03JdGordonsaratoga: get your butt into #rockbox-commuinty :p
03:10:11JdGordonalso... dont be afraid of the del key :D
03:19:45 Join Rob2222 [0] (
03:36:30 Join robin0800_ [0] (
03:36:54 Quit scorche (Read error: 104 (Connection reset by peer))
03:37:01 Quit Rob2223 (Read error: 110 (Connection timed out))
03:44:51 Quit martian67 ("out")
03:44:56 Join martian67 [0] (i=martian6@about/linux/regular/martian67)
03:46:51 Quit robin0800 (Read error: 110 (Connection timed out))
03:49:01 Join bzed_ [0] (
03:49:59 Quit bzed (Read error: 104 (Connection reset by peer))
03:50:06 Nick bzed_ is now known as bzed (
03:56:03saratogapeople on the mailing list are calling each other autistic
03:56:08safetydanis there anyway to filter all messages that mention top or bottom posting?
03:56:17saratogamaybe its time we consider getting rid of the mailing list
03:56:33 Join BHSPitMonkey [0] (n=stephen@unaffiliated/bhspitmonkey)
03:56:39 Quit martian67 (Read error: 104 (Connection reset by peer))
03:56:58 Join martian67 [0] (i=martian6@about/linux/regular/martian67)
03:58:06 Quit robin0800_ ("DMDirc exiting")
04:06:34 Quit TheSeven (Nick collision from services.)
04:06:49 Join The_Seven [0] (
04:06:53 Nick The_Seven is now known as TheSeven (
04:07:18 Quit Sajber^ (Read error: 104 (Connection reset by peer))
04:17:24 Join robin0800 [0] (
04:18:07 Quit robin0800 (Remote closed the connection)
04:28:41 Quit n00b81 ("Leaving")
04:49:37kkurbjunI agree the user mailing list has gotten out of control on this topic - can we start kicking people that continue to discuss that thread?
04:57:22***Saving seen data "./dancer.seen"
05:05:29Unhelpfulhuh... make seems not to be running the commands for a rule in their correct order
05:07:13 Quit efyx (Remote closed the connection)
05:09:24 Join Horscht [0] (n=Horscht2@xbmc/user/horscht)
05:13:42 Quit Horscht (Client Quit)
05:22:07 Join lanny` [0] (
05:23:50 Part lanny`
05:24:04 Quit JdGordon ("Leaving.")
05:26:12 Join JdGordon [0] (
05:36:10 Quit fdinel ("Miranda IM! Smaller, Faster, Easier.")
05:43:38 Join Horscht [0] (n=Horscht2@xbmc/user/horscht)
05:49:40 Quit daurn (Read error: 110 (Connection timed out))
05:53:20Lloreankkurbjun: Well, it's a somewhat important question - are the list rules on topic for discussion on the list?
05:53:32LloreanNow certain threads of it are certainly wandering pretty far astray.
05:54:44 Part Llorean
05:55:00 Join Llorean [0] (
06:04:59 Quit JdGordon (Read error: 104 (Connection reset by peer))
06:07:38 Join JdGordon [0] (n=jonno@rockbox/developer/JdGordon)
06:09:18 Join Horschti [0] (n=Horscht2@xbmc/user/horscht)
06:23:40 Quit dash32 ("Verlassend")
06:27:40 Quit Horscht (Read error: 110 (Connection timed out))
06:49:35 Join scorche [50] (n=scorche@rockbox/administrator/scorche)
06:54:18saratogaapparently mpg123 has an mp3 decoder with arm optimizations now
06:55:52saratogahmm though it looks pretty rudimentary
06:56:10saratogastill its actively developed unlike libmad
06:57:24***Saving seen data "./dancer.seen"
07:13:02 Nick ayahuasca is now known as fortuna (
07:13:58 Quit scorche (Read error: 60 (Operation timed out))
07:17:00 Join Horscht86 [0] (
07:20:07 Join Leakydick [0] (
07:22:44Leakydickanyone feel funkiness between the 2.x builds and the 3.3 in the sansa e200 series? or is it just me? I hadn't been keeping up on a regular basis and feel like i am missing context menu, OR some keys are mismapped.
07:26:26scorche|shhuh?...there werent any 2.x builds for the sansa e200 series
07:26:52Leakydicki guess i was referring to timeframes more than official releases
07:26:56Leakydickmaybe 1.5 yr back
07:27:20Leakydickopressing down now acts like the power button in the 3.3
07:27:27Leakydickreturns to main menu always
07:27:33scorche|shi dont know about 'funkiness', but a LOT has changed since then, yes...i believe we did change the keymaps a long while ago
07:28:07scorche|shif you are a bit confused how things work now, perhaps the manual can tell you a bit more about what each button now does?
07:28:11Leakydickthanks for your recollection and the help. im in the middle of installing cyg on this machine, just copy and paste didnt seem to work out too well for me
07:28:16 Join daurn [0] (n=daurnima@unaffiliated/daurnimator)
07:28:30scorche|shcopy and paste?
07:29:06Leakydicki used to build RB on a 32 bit machine and had cyg installed there. i copied the dir and added paths, but it didnt take well on this 64b machine
07:29:38scorche|shit is easy to use svn anyway...
07:29:56Leakydicksvn is what i'll use
07:30:06Leakydickit's the build enviornment that is my first issue
07:30:22Leakydickthought cyg would be portable app, like copy and paste to new machien and set paths
07:32:24Leakydickthis team has done some excellent work, my most favorite feature additions are the plugin architecture, the SDHC, and the unicode support
07:32:40Leakydicknow that USB charging is done, i have no need for the sansa FW
07:33:15 Join scorche [50] (n=scorche@rockbox/administrator/scorche)
07:35:00 Quit Horschti (Read error: 110 (Connection timed out))
07:38:19 Join einhirn [0] (
07:42:29 Join stoffel [0] (
07:42:54Leakydickback in business... cyg is a clusterf*, it was trying to be smart and without uninstall I had to purge everything I copied over and all the registry entries that had "CYG" except a couple that matched cyg cuz it was PoliCYG...
07:44:24scorcheLeakydick: as a note, this channel is strictly for the discussion of rockbox related chat only...all social and offtopic chat should go somewhere else as said in the topic of this channel
07:49:08Leakydicki was trying to be helpful for the logs
07:49:28 Join flydutch [0] (
07:49:34Leakydickbattling cygwin for rockbox build environment
08:09:21 Quit r0b- (Read error: 110 (Connection timed out))
08:18:24 Join graey [0] (
08:38:29 Part safetydan ("Leaving.")
08:44:12GodEateranyone know if the libgphoto2 bug is likely to affect sansas running in recovery mode too ?
08:44:23 Quit martian67 (Read error: 110 (Connection timed out))
08:44:55 Join amiconn [0] (i=quassel@rockbox/developer/amiconn)
08:48:33 Quit daurn (Read error: 110 (Connection timed out))
08:48:35 Join Rob2223 [0] (
08:51:30 Join n1s [0] (n=n1s@rockbox/developer/n1s)
08:54:20LloreanGodEater: Depends entirely on if those PID/VIDs are in the list
08:57:28***Saving seen data "./dancer.seen"
08:58:16bertrikI spotted a cheap Samsung YP-S3, it seems to use a Samsung S5L8700 SoC too
09:00:35 Quit n1s (Remote closed the connection)
09:01:04GodEaterLlorean: I realise that, I just can't reconcile the values in the xml file with what lsusb tells me
09:01:42bertrikThe OF of YP-S3 seems to be based on VxWorks
09:01:47LloreanGodEater: Aaah. Dunno
09:02:02GodEaterand I have no sansa based player with me to just try it :)
09:03:42 Quit amiconn (Remote closed the connection)
09:03:50 Join pixelma [0] (i=quassel@rockbox/staff/pixelma)
09:03:50 Join amiconn [0] (i=quassel@rockbox/developer/amiconn)
09:05:33 Quit scorche (" rawr...that is all...rawr")
09:05:51 Quit Rob2222 (Read error: 110 (Connection timed out))
09:08:05 Join petur [50] (n=petur@rockbox/developer/petur)
09:08:56 Join n1s [0] (n=n1s@rockbox/developer/n1s)
09:14:23tmztGodEater: does it not mount as storage?
09:15:00GodEater< GodEater> and I have no sansa based player with me to just try it :)
09:17:22tmztthat was you
09:21:15 Join miyodea [0] (n=msellers@
09:23:16 Part miyodea
09:24:51 Join Thundercloud [0] (
09:24:53 Quit TheSeven ("ChatZilla 0.9.85 [Firefox 3.0.11/2009060215]")
09:30:19 Quit graey (Read error: 110 (Connection timed out))
09:31:36 Quit BHSPitMonkey ("Ex-Chat")
09:43:01 Quit Mikachu (Read error: 104 (Connection reset by peer))
09:44:42 Join gartral [0] (
09:45:40 Quit Thundercloud (Remote closed the connection)
09:45:42 Quit MrDuck (Read error: 113 (No route to host))
09:46:16gartralIf one has put but a bitmap and jpeg AA in a folder as cover.bmp and cover.jpg, which will be the one used?
09:46:47LloreanCouldn't just just test? It'd take fifteen seconds.
09:49:26gartrali assume the only real way to test is put two different picures in, ona as a bmp and one as a jpg, right?
09:49:43Lloreanthat would be the easiest, yeah
09:52:15tmztor a jpg with gemoetric shapes
09:53:23*amiconn thinks that looking at the source is easier
09:54:00CIA-71New commit by learman (r21853): Remove a function accidentally added back in r21841.
09:54:08amiconnBut in fact that is a valid question. It should be added to the manual if it isn't already
09:55:05LloreanI thought it was on the albumart wiki page.
10:00:03gartralalbumart wiki does not appear too mention which is "looked for" first
10:01:53 Join Mikachu [0] (
10:03:08LloreanWell why don't you fix it after your test?
10:03:48gartralhappily... as soon as my slow computer finishes "flattening' the jpeg i made
10:08:18amiconnHmm, weird yellow
10:08:40amiconnOnly in the new build system, and only for one sim, although other sims were also built using gcc 4.4.0
10:10:08tmztwhat does the build system for rockboxutil look like on windows?
10:10:10 Quit n1s ("Lmnar")
10:13:27CIA-71New commit by Ubuntuxer (r21854): try to make buttons in jewels more unified (part of FS #10193)
10:36:42 Join n1s [0] (
10:36:55rasheramiconn: could be some sort of vendor patch *gag*
10:39:37 Join TheSeven [0] (
10:44:47 Join svqyqb [0] (
10:44:55 Quit svqyqb (Read error: 104 (Connection reset by peer))
10:50:14 Join MrDuck [0] (
10:51:37linuxstbgartral: The manual answers your question -
10:54:30 Join Casainho [0] (
10:54:35gartrallinuxstb: how does it say which format is looked for first? the *only* remote indication i can see is that "Jpeg, jpg, bmp" options in the examples, which I would like to note; isn't there on the e200's manual
10:55:08linuxstbgartral: "a list in braces means that those file extensions are tried in that order"
10:56:18gartralim sorry *not in album art wiki page*
10:57:31***Saving seen data "./dancer.seen"
10:59:06gartralis it ok just to merge the info from the manual too the wiki page?
10:59:51 Join Sajber^ [0] (
11:01:14linuxstbMaybe it would be better to delete any duplocate information from the wiki, and just refer people to "The album art appendix" in the manual.
11:01:43gevaertsgartral: not really. We want the manual to be complete, and have the wiki for missing bits, but no duplicate information
11:04:01CIA-71New commit by Ubuntuxer (r21855): Adapt manual to the changes in jewels (part of FS #10193)
11:09:40gartralwell thank you all
11:12:02 Join Ubuntuxer [0] (
11:13:32CIA-71New commit by Ubuntuxer (r21856): Change buttons in pegbox for Fuze (part of FS #10193)
11:14:51amiconnpetur: Regarding the gcc 4.4.0 warning - the code does indeed look fishy
11:15:13peturamiconn: I didn't really check the code
11:15:24amiconnIt looks like the ! should be outside the first inner parentheses
11:15:59amiconnif ((!s2->lines[i]->flags & ML_TWOSIDED) || (s2->lines[i]->backsector == s1)) ...
11:16:16amiconnShould most probably read if (!(s2->lines[i]->flags & ML_TWOSIDED) || (s2->lines[i]->backsector == s1))
11:17:01amiconnChecking flags and then turn the result into a boolean
11:20:30 Quit Casainho ("ChatZilla 0.9.85 [Firefox 3.5/20090624025744]")
11:21:56linuxstbUbuntuxer: I'm not sure I understand your commit message "try to make button in jewels more unified" - isn't your patch mainly a change for IPOD_3G_PAD? Also, why remove JEWELS_MENU for the h100/h300, I would have thought it would be better to add that mapping to more targets?
11:28:48 Join Lynx_ [0] (
11:29:11peturamiconn: yes, it looks like that
11:31:20 Join teru [0] (
11:38:32 Join DarkDefender [0] (
11:41:01 Quit Sajber^ (Read error: 104 (Connection reset by peer))
11:41:29Ubuntuxer@linuxstb the patch is a part of FS #10193, so actually it makes just sense with the all the rest changes
11:43:27linuxstbUbuntuxer: Yes, I read FS #10193, and it doesn't help. I'm just saying that commit messages should describe what the change is, and I don't think that did.
11:44:27amiconnpetur: Hmm, I just looked it up. ! takes precedence over &, so the current code is probably buggy. But then, prboom doesn't seem to have fixed it in their trunk
11:44:38linuxstbUbuntuxer: So why remove the ability for h100/h300 users to press MODE to enter the menus?
11:45:49amiconnI wonder when EV_DoDonut() is called - the timedemo in the shareware version doesn't
11:46:04Ubuntuxerbecause FS #10193 removes the ingame menu and makes the menu altogether
11:46:36AlexPWhy remove the ingame menu?
11:48:15 Join _lifeless [0] (n=lifeless@
11:48:20Ubuntuxerthe aim is to unified all menus, so it's clearer and easier for a user
11:48:22linuxstbUbuntuxer: Reading the code, JEWELS_MENU was just an alternative way to access the same menu that is on "long select".
11:48:56Ubuntuxeri know, but JEWELS_CANCEL replace it
11:49:21 Quit _lifeless (Client Quit)
11:50:44Ubuntuxerplease test my silly huge patch, in which I included everything and you'll understand
11:51:22linuxstbUbuntuxer: I shouldn't need to - commits should make sense by themselves (the commit message in combination with the code)
11:53:37UbuntuxerI should divide my patch..., sorry
11:56:33 Join evilnick_home1 [0] (
11:56:53UbuntuxerWhat should I do know?
12:00:32linuxstbI think maybe you just split your patch too much - I don't think it made sense to commit those simple keymap changes by themselves, without committing the menu rework.
12:01:25linuxstbSo probably carry on with what you're doing, but be aware that each commit should make sense (and also try and describe them more fully in the commit message).
12:03:06 Quit __lifeless (Read error: 113 (No route to host))
12:12:53 Join kugel [0] (n=kugel@rockbox/developer/kugel)
12:14:36 Join Sajber^ [0] (
12:16:01CIA-71New commit by amiconn (r21857): Silence the gcc 4.4.0 warning by making this code look like it was probably meant to be. I'm not 100% sure this is correct - latest prboom doesn't ...
12:16:20 Quit evilnick_home (Read error: 113 (No route to host))
12:17:45 Join graey [0] (
12:24:12 Join _zic [0] (
12:25:22teruUbuntuxer: do you mind my committing FS #10099?
12:28:55Ubuntuxeryes, I'll commit it soon, but please wait,
12:32:42 Quit TheSeven (Nick collision from services.)
12:33:00 Join The_Seven [0] (
12:33:04 Nick The_Seven is now known as TheSeven (
12:33:44 Join eekfonky1 [0] (
12:34:18eekfonky1Is there any ETA for rockbox to work on 2nd gen Ipod Nano?
12:34:32markunof course not :)
12:35:38linuxstbkugel: What exactly are you asking for comments about in the dev list? I can't see a relevant patch in FS #9873, just the initial one...
12:38:58eekfonky1markun: are you answering me?
12:40:09GodEatereekfonky1: we NEVER give ETAs. Ever.
12:40:41markuneekfonky1: well, the port has just started
12:40:56GodEatermarkun: it has? who's working on it ? :)
12:40:56markunwe can't permanently change the firmware yet
12:41:03markunGodEater: linuxstb
12:41:37GodEaterlinuxstb: another port ? :)
12:41:52eekfonky1are there plans to release a version that will work on a 2nd gen ipod?
12:42:17AlexPAs has been said
12:42:18GodEatereekfonky1: be careful with your naming convention. We already have a port for the 2nd gen ipod.
12:42:19markuneekfonky1: there are hopes, but don't count on it any time soon
12:42:29AlexPeekfonky1: There are no etas, no plans
12:42:30linuxstbeekfonky1: No explicit plans, no. If it happens, then it happens. It depends on luck and people contributing.
12:42:58eekfonky1can you direct me to the port to see if it works on my ipod?
12:43:05AlexPeekfonky1: There is no port
12:43:08markuneekfonky1: which ipod do you have?
12:43:11AlexPFor the 2nd gen nano
12:43:19eekfonky12nd gen nano
12:43:20GodEatereekfonky1: I think you missed the point of my warning =/
12:43:35markuneekfonky1: really, there is nothing to try yet
12:43:48eekfonky1thank you for your help
12:43:56AlexPeekfonky1: There is no port at all for the 2nd gen nano. The 2nd gen ipod has a port, but that is the 2nd ipod that was ever released and is many years old
12:44:51 Quit kugel (Read error: 113 (No route to host))
12:45:13GodEaterlinuxstb: so is markun right? or is he making it up ?
12:45:28 Part eekfonky1 ("Leaving.")
12:46:09linuxstbGodEater: Now that eekfonky1 has left.... I've added the Nano 2G to the build system, yes. It's using a similar SoC to the Meizu M3/M6, so it should be able to share that code.
12:46:29linuxstbBut I haven't actually tried running the resulting binary yet...
12:46:52markunlinuxstb: but the bin2note util works, right?
12:47:03*linuxstb still wants to understand if the code to put the SoC into big-endian mode in crt0.S for the Meizu is needed.
12:47:29GodEaterare we going to add the classic too ? As I understand the same exploit for the 2G Nano works on that too ?
12:47:53linuxstbmarkun: Yes, I wrote that for use with TheSeven's "iBuggerLoader" (the loader for iBugger - his planned on-target debugger for the Nano2G). My program produces the same note as he released.
12:48:26linuxstbGodEater: I don't think it works fully - that report turned out to be false...
12:48:34GodEaterah ha
12:49:27*GodEater wonders if it's worth trying to find an ipod classic from somewhere to see
12:51:43GodEatergood god
12:51:51GodEaterthey cost a bloody fortune on ebat
12:53:06GodEaterand it appears everyone is calling the entire bloody range "classic" now.
12:53:15linuxstbI saw some for 140GBP - is that what you mean by a fortune?
12:55:31GodEateryes, I was expecting the price to have dropped a bit more than that now
12:55:48GodEaterI can see some for <100 now
12:55:54GodEaterbut still a few hours to go
12:57:33***Saving seen data "./dancer.seen"
12:59:34n1slinuxstb: any idea how to try to adapt the exploit to the 4g nano?
13:04:42 Quit MrDuck (Read error: 104 (Connection reset by peer))
13:04:54 Join MrDuck [0] (
13:09:09 Join mcuelenaere [0] (
13:12:12CIA-71New commit by Ubuntuxer (r21858): Combine the two menus, improve highscore list, automatically save game, improve help text (buttons are missing yet) and clean up the code in jewels ...
13:14:31linuxstbn1s: No, I admit to not fully understanding it. You could ask in #linux4nano-dev....
13:15:16n1sok, thanks
13:21:32 Join efyx [0] (
13:22:52 Join kugel [0] (n=kugel@rockbox/developer/kugel)
13:33:38 Join adibis [0] (n=aditya@
13:34:21adibishey, I was wondering, is it possible to install rockbox on your own MP3 player. I mean, a player designed using audio codec ICs and microcontroller at home.
13:34:47adibisif yes, then what are the constraints on hardware design. If there are any
13:34:53n1shmm, I'm syncing the strlcpy patch and the PA plugin introduces a ton of strncpy calls, would it be better to move strncpy to the plugin lib and leave imported code using it?
13:35:42n1sadibis: you would have to port rockbox to it
13:35:46Torneadibis: yes, it's possible, but you'd have to write a port
13:35:52Tornethere are very few constraints on design
13:36:04Tornebut some design choices may make the porting more or less effort
13:36:10Tornedepending on similarity to platforms we already support
13:36:42Tornea microcontroller is unlikely to be sufficient, though
13:36:52Torneyou need a "real" processor, for some value of real
13:39:24CIA-71New commit by unhelpful (r21859): Handle sh and m68k compiler's output correctly in asmdefs2file
13:40:29 Join wincent [0] (
13:40:29 Quit Leakydick (Read error: 110 (Connection timed out))
13:41:41 Quit adibis (Read error: 60 (Operation timed out))
13:41:43n1seh, not PA, Pure Data, I think...
13:42:21n1sso PD maybe, anyways, any oppinions?
13:44:15 Quit Ubuntuxer ("Leaving.")
13:46:44n1swincent: any oppinions? (you are the pdbox coder right?)
13:47:12wincentnls: Wait, I will check the logs
13:48:33n1swincent: i'm syncing a patch that converts rockbox to use strlcpy instead of strncpy, i saw that strncpy is used a lot in pdbox so i thought i might be better to move strncpy to the pluginlib so imported code can continue to use it
13:49:03n1simported code in plugins that is
13:50:13wincentnls: Is strncpy not already in plugin.h?
13:53:05n1swincent: yes, but this patch removes it as all users besides pdbox are converted to use strlcpy instead
13:53:31wincentnls: What does strlcpy?
13:54:33Tornewincent: safer version of strncpy. guarantees zero termination and returns the length copied so you can detect truncation
13:54:45n1sit's similar to strncpy but guarantees null termination and returns the length of the source so it's basically considered to be a bit safer and less prone to cause bugs
13:55:42n1swincent: so what i'm asking is if you think pdbox sholud use it too or if it should keep using strncpy?
13:55:57 Join aditya [0] (n=aditya@
13:56:03wincentWhat if you change the macro strncpy in wrappers.h to use rb->strlcpy?
13:56:44n1sthat is possible but the code might need small adjustments in some places
13:57:30n1sand for clarity i think using a wrapper like that would be kind of confusing
13:57:46Tornethat will copy one less byte sometimes
13:57:57Tornewhich shouldn't affect code which is using strncpy correctly :)
13:58:01Tornebut you never know
13:58:39n1swincent: how likely are future merges with upstream code into pdbox?
13:58:43kugeln1s: leaving it in the pluginlib might be a good idea
13:58:49Unhelpfulwincent: wrappers.h is not intended for that
13:59:06gevaertsTorne: I have some code (not in rockbox though) that depends on the exact strncopy behaviour
13:59:09kugelI think we generally want to make importing code as easy as possible
13:59:23wincentThen move it to pluginlib.
13:59:31gevaertsso you really have to check all cases, not just do a blind global search and replace
13:59:57n1swincent: ok
13:59:58Unhelpfulwrappers.h is intended for code imported from core
14:00:05Tornegevaerts: well, ok, you can do that if you are clever. but most people aren't :)
14:00:05wincentnls: Future merges are likely.
14:00:37n1sgevaerts: yeah i found a few such places in rockbox, one i code i'd written myself :)
14:00:43kugelTorne: that's just a case where gevaerts used strncopy instead of memcpy
14:00:53gevaertskugel: no :)
14:01:16n1skugel: you might also rely on the zero padding of strncpy which strlcpy doesn't do
14:01:55kugelfor strings?
14:02:13Torneyah, you may be writing fixed length output data
14:02:21Torneand thus need full zero padding
14:02:31Tornefor various possible reasons (checksums and the like?)
14:02:58n1skugel: keybox does for the buffer that is used to generate the encryption/decryption key
14:03:27n1smemset+strlcpy does the trick though and is more obvious
14:07:56CIA-71New commit by teru (r21860): add a way to quit without saving game. ...
14:13:05n1si wonder if i should remove my changes to doom too then
14:14:37Torneanything that's likely to get updated from external sources should probably be left alone, i'd think
14:14:55amiconnmemset + strlcpy are two calls, strncpy is just one. And memset()ting the whole buffer is even more wasteful
14:15:03Torneglibc still doesn't have strlcpy, does it?
14:15:11Tornethey objected to the idea
14:15:20Torneso external code is not gonna want it
14:15:52n1samiconn: yes, i know but it is clearer that you depend on the zeroing of the buffer, this example is not speed critical in any way
14:16:07amiconnClearer in what way?
14:16:49n1swell, IMO of course, since you explicitly zero it
14:17:18amiconnstrncpy() also zeros explicitly - it is documented behaviour
14:17:46linuxstbYes, it's documented, but perhaps not well-known. So if code depends on that, a short comment to that effect wouldn't hurt.
14:18:07n1samiconn: yes i know it's documented but if a string is strncpy'ed the zero padding might just be a side effect
14:18:14gevaertsZagor: (logs) I just noticed that one of my clients had stopped with a "Terminated" message. I think this is the same issue as others that stopped with "Killed" earlier, only the signal changed. I suspect that the pid is sometimes wrong, or maybe the process group handling is different than we expect. This is not in a VM by the way
14:19:23gevaertsn1s: by the same reasoning someone could call strncpy(a,"",20), and depend on the zeroing, while the string copy is just a side effect.
14:19:56gevaertsstrncpy() is defined as it is, with all parts equally important
14:20:10n1sgevaerts: my point is that since the zeroing always happens it is not clear if the code depends on it
14:20:54gevaertsn1s: in that case you assume it does until proven otherwise
14:22:31n1syes but IMO a memset+strlcpy or memcpy or whtever is quite clear in that the buffer should be zeroed. Not that this is a major point for this patch in any way, just what i think and this particular example turned up only once
14:23:19gevaertsit's clear, yes. It's also inefficient
14:23:25n1si agree
14:24:13n1swell except when you first memcpy and then memset the remaining part of the buffer
14:25:15gevaertsand depend on people to get the numbers right? I thought you were trying to avoid bugs?
14:26:04n1si'm just saying i don't think anyone should do that unless the code is very time critical
14:26:20n1sinsert , where appropriate
14:26:43n1sor ;
14:27:13linuxstbn1s: You're suggesting we get rid of strncpy completely?
14:27:16n1sanyway i think the patch is ready soon and will probably commit it later today
14:27:45*gevaerts disagrees. If there's a well defined function for something that does exactly what you need, it's wrong to use two other functions for that
14:27:54n1slinuxstb: that was how i understood the original plan but as it is now i moved it to the pluginlib
14:30:17linuxstbIt does seem odd to get rid of a standard existing function, just because some people don't know what it does...
14:31:33adityaone more question, where can I find information about decoding ogg? That is, I don't want to use the ready made OGG-Vorbis decoder IC for decoding the music. Maybe code it on the controller or use an FPGA/CPLD...
14:31:34*amiconn fully agrees with linuxstb here
14:32:00 Nick fxb__ is now known as fxb (
14:33:54n1slinuxstb: ok, so what do you think should be done then? as the patch is now only plugins use strncpy. Should i just leave strncpy in the core anyway? or should the core calls not be converted at all?
14:35:29kugelit's on MrSomeonesTODOList. The items there *should* be out of question
14:35:52n1skugel: noone complains until you are about to commit
14:36:03markunaditya: what are you trying to do?
14:36:08linuxstbn1s: Ah, so you're saying that every single use of strncpy is better replaced by strlcpy?
14:36:14kugelWe're not implementing the complete standard, that's no reason to keep strncpy in
14:36:29linuxstbn1s: Or have you replaced some by memset+strlcpy?
14:36:44adityamarkun: there are ogg decoder ics available. I want to know if the decoding can be done on the microcontroller itself and play ogg.
14:37:01adityamarkun: once that is done, rockbox porting on that player will follow
14:37:04kugeln1s: I don't think he's complaining (yet).
14:37:04linuxstbn1s: Or simpler - do you have a link to the patch? ;)
14:37:10markunaditya: which player?
14:37:24kugellinuxstb: a few have been replaced with memcpy IIRC, the patch is on the tracker
14:37:53markunaditya: did you read the specs?
14:38:22n1slinuxstb: i'll post a fresh diff
14:38:35adityamarkun: err.. I am thinking of building a player.
14:39:48markunaditya: and you want to do all the decoding inside an FPGA/CPLD?
14:40:29n1slinuxstb: fs#10080
14:41:36adityamarkun: or the controller itself.
14:42:27Tornemarkun: he wants to use a uC rather than a higher-end processor like most o four targets, i believe :)
14:42:38aditya^^ yes
14:42:49Tornethis is unlikely to be realistic, in my opinion
14:42:55adityaTorne: can be an ARM controller also.
14:42:57n1sbasically the majority of the strncpys were replaced with strlcpy, a few with memcpy and *one* with memset (implicit, buffer is initialized with 0) + strlcpy
14:42:57markunit's not impossible
14:43:00Tornejust grab a cheap ARM9 SoC
14:43:17Tornesomething powerful enough to run our codecs on the processor
14:47:30 Quit gevaerts (Read error: 60 (Operation timed out))
14:49:56 Join gevaerts [0] (n=fg@rockbox/developer/gevaerts)
14:50:16kkurbjunaditya: I think decoding an ogg on an FPGA/CPLD woudl be a difficult task if you are not pretty familiar with digital logic design
14:50:32GodEaterscorche|sh: any idea how much longer your "friend" wants to sit on his twitpic hole ?
14:50:34kkurbjunyou can also put microprocessors on an FPGA though which might be more realistic
14:50:42kkurbjunsoft microprocessors that is
14:50:42GodEateroops, wrong channel
14:51:36markunkkurbjun: and maybe some separate DSP routines?
14:52:13kkurbjunwith a Spartan 3 family device and a microblaze processor I think you could do quite a bit completely self contained in the FPGA, or if you wanted to go an even more open route you coulod try an opencore, but that would probably be considerably more difficult
14:52:36kkurbjunmarkun: you could use seperate dsp routines, but I think the microblaze would be fast enough to run rockbox comfortably
14:52:47kkurbjunwithout any special dsp routines
14:53:00 Quit gartral ("Why must we be chicken or sheep when we have the blood of the Fox, Wolf?")
14:53:01kkurbjunas for power consumption, an FPGA might not be the best bet though :-D
14:53:38 Quit Ridayah_ ("The Rise and Fall of the Heavens themselves is dependant upon Humanity's belief and disbelief.")
14:53:46kkurbjunand I dont' think you could get much out of a CPLD, there's not too much logic available in the ones that I've worked with
14:54:17 Join Ridayah [0] (n=ridayah@
14:54:34linuxstbn1s: Patch looks good to me. I was just confused by your conversation with gevaerts... So the intention is to keep it for plugins, and not use it there?
14:54:36adityakkurbjun: hehe
14:55:03adityakkurbjun: so, the second option. Is it possible to have the decoder on the processor?
14:55:23adityaI mean, use a software decoder. Save it to flash. Load it in ram at startup. Use it.
14:55:36adityasave it to the same flash as the one with the music files.
14:55:39markunaditya: that's how rockbox does it for all targets
14:55:59markun(all flash targets which play Ogg Vorbis)
14:56:01adityamarkun: so, a seperate decoder IC is not needed?
14:56:11kkurbjunaditya: yes, I think that a soft procesor would be fast enought to do the decoding
14:56:14markunit's all done in software
14:56:24adityaokay, where can I get more information about that?
14:56:43adityaI mean, the codec and how is it loaded for the playing of the file...
14:56:51n1slinuxstb: thanks for looking, sorry fopr any confusion, yes keep it for plugins and use it in imported ones, like doom and pdbox
14:56:56*gevaerts finds some sillyness in the patch. It was already there before though
14:57:12n1sgevaerts: please tell
14:57:15kkurbjunaditya: It's in the rockbox source
14:57:36***Saving seen data "./dancer.seen"
14:57:53 Join ucchan [0] (
14:57:57gevaertsn1s: have a look at apps/plugins/rockboy/menu.c, line 242. It does memset, and then strncpy of a constant string
14:57:59 Join daurn [0] (n=daurnima@unaffiliated/daurnimator)
14:58:09n1sgevaerts: aha
14:59:22n1syeah, that's probably out of the scope for this patch
15:01:36Torneaditya: the minimum 'spec' for the ports of rockbox at the moment is something like "a processor fast enough to decode mp3 in software without being fully utilised, 4-8MB of ram but ideally a lot more". We support a couple of players with hardware mp3 decoder chips which are lower spec than that, but they are very old and can't decode anything but mp3.
15:01:56Torneyou could port to less hardware, if you wer ecunning enough about it, but it would be a lot of work to do, most likely.
15:02:10gevaertsn1s: that depends on what you consider to be the scope of this patch. If it is "introduce strlcpy", it isn't. If it is "clean up strncpy usage", it is
15:02:52n1sok, i'll clean that up a bit then ;)
15:03:20CIA-71New commit by teru (r21861): use lib display text to display help messages (FS #10099).
15:03:50adityaTorne: okay, so, say ARM11 processor with a ram chip of 8MBs connected to it...
15:03:54kkurbjunaditya: Another option if you are interested in messing around with hardware/fpga's and cplds is to find a board that has something like a virtex2 pro that has an embedded powerpc on it also
15:04:00adityaTorne: will it be able to run rockbox?
15:04:02kkurbjunthat would give you a ton of processing power
15:04:06Torneaditya: Are you intending to play files from flash or HDD
15:04:11Torneif you are going to use a hdd you want a lot more ram
15:04:14Tornebut from flash, 8mb is plenty
15:04:21adityaTorne: flash. SD card.
15:04:21Torneand an ARM9 is sufficient, an ARM11 is loads of power
15:04:33GodEateris it worth pointing out to aditya that someone is already builing their own player specifically to run Rockbox ?
15:04:38Torneyah. an ARM9 SoC with 8MB ram and an SD interface would be a reasonable rockbox target
15:04:48Tornean ARM11 would be more fun but probably not necessary
15:04:49TorneOh, er
15:04:54markunGodEater: I think so :)
15:04:54Torneyeah, ther eis that. See the lyre project
15:05:28GodEaterI mean, not to discourage or anything, if you really want to go your own way and stuff. But they've done a lot of work already.
15:05:43 Join LambdaCalculus37 [0] (i=44a0430d@rockbox/staff/LambdaCalculus37)
15:05:51adityaGodEater: I just want to learn...
15:06:16Torne is the people doing it already. You might want to check it out anyway even if you do want to make your own thing
15:06:19GodEaterjust wanted to make sure you weren't about to repeat a lot of hard work someone else had already done ;)
15:06:25Torneas many of their discussions/etc may inform your own decisions
15:06:39Tornewe're not about to discourage people from doing things for fun/educational purposes :)
15:06:44Tornebut lyre's website might still be a good resource
15:07:24adityaGodEater: nah.. it's a one man project. Just something to do in final year of engineering
15:07:38GodEateraditya: excellent choice of project :)
15:07:53TorneThat sounds pretty cool, yes :)
15:07:56gevaertsn1s: there are some more points where I disagree wth both the old and new usage, but that's probably just me (copying a constant into a fixed-size buffer)
15:08:02adityaGodEater: the college project is different. This is fun/hobby project
15:08:48adityafor college, I have to make an object tracker. Collect information about it's movement in space using accelerometers and gyroscopes.
15:09:05n1sgevaerts: yes i've seen that in some places
15:11:37kkurbjunaditya: if you are looking for something to mess around with using something like this might be a good starting point:|66%3A2|39%3A1|72%3A1205|293%3A1|294%3A50 It really depends on whether you want to make something that could be provided to others t
15:12:20CIA-71New commit by lenzone10 (r21862): Updated italian translation.
15:12:25gevaertsn1s: I think strncpy should remain available though, especially to plugins and codecs (because they tend to inherit external code). If you feel strong about it, you could ue a mcrothat als prints a #warning
15:13:52gevaerts*could use a macro that also prints a #warning
15:13:52linuxstbIt's probably worth a mention in docs/CONTRIBUTING - that we prefer strlcpy to strncpy
15:14:00linuxstb(for new code)
15:14:01adityakkurbjun: thanks
15:14:04*gevaerts wonders what happened there
15:14:11n1scan codecs use the pluginlib?
15:14:29linuxstbNo, but they have a codeclib... I don't see why codecs need strncpy though.
15:14:46n1slinuxstb: currently none do
15:15:01linuxstbThen I say ignore the codecs.
15:15:10adityaI have to go to my music class... will come back for more details later...
15:15:13 Nick aditya is now known as adi|away (n=aditya@
15:15:27n1sso moving it to the plugin lib is fine with you?
15:15:39n1sor should it remain in core?
15:16:12linuxstbI don't feel strongly... Although I'm generally in favour of changing imported code, so would be tempted to remove uses of strncpy if they can all safely be replaced with strlcpy...
15:16:13 Join funman [0] (n=fun@rockbox/developer/funman)
15:16:24gevaertslinuxstb: they could want it in metadata reading, but I guess that's one of the areas that need lots of changes anyway
15:16:56linuxstbgevaerts: metadata reading is in the core anyway.
15:17:18linuxstbAnd yes, our metadata parsers are all written specifically for Rockbox.
15:17:27gevaertslinuxstb: that's why I suggested the #warning. You can then use it while porting so you can concentrate on the difficult bits, but you have to get rid of it before committing
15:18:04 Quit TheSeven ("ChatZilla 0.9.85 [Firefox 3.5/20090624025744]")
15:18:09linuxstbI don't think strncpy is even used much in metadata parsers though, as the text often needs to go through character-encoding conversion.
15:18:24ucchanThere is a question. for flyspray, how can I add the task in "This task depends upon"?
15:19:32linuxstbucchan: Is this a task you created yourself?
15:21:26gevaertsn1s: core isn't going to use imported code without heavy changes anyway, so if strncpy is strongly deprecated, I see no reason to allow it there
15:21:29 Join LiamAether [0] (
15:22:07linuxstbn1s: Which plugins will be left which use strncpy?
15:22:15n1sgevaerts: ok, so pluginlib it is
15:22:45n1slinuxstb: pdbox is a heavy user and i left doom too
15:27:05linuxstbn1s: Yes, looking at them now, it would seem a bit intrusive to change those...
15:27:07ucchanHi, linuxstb. I would like to add to my task
15:28:16linuxstbucchan: When you click on "Related tasks", do you see a "Add new related task" box?
15:29:44ucchanNo "Add new related task".
15:30:04linuxstbThen I guess you don't have the rights to do it. What task do you want to edit? (I can do it)
15:30:12ucchan"Remove" button only.
15:31:25ucchanBecause the authority which I have is not enough probably will be?
15:31:53LiamAetherI just recently updated my rockbox to 3.3. Now my custom theme is broken. Did the syntax of the wps change?
15:32:42linuxstbLiamAether: When did you previously update?
15:33:08linuxstbucchan: I can make the change if you tell me what you want to do.
15:33:49LiamAetherI occasionally manually dropped the build of the daily zip in to the .rockbox directory, today I used the automated installer and that updated the bootloader too
15:37:18linuxstbLiamAether: But yes, there have been various changes to the WPS syntax over the past year or so, but I can't remember anything in the last few months...
15:37:43LiamAetherI didn't think it would be a problem, since I used a simple text theme
15:37:45linuxstbDid your previous updates definitely work? i.e. did you see any changes in Rockbox?
15:38:10LiamAetherI haven't noticed anything in the UI
15:39:10linuxstbMaybe your manual updates didn't work then, and only when you used rbutil did the update actually happen... The most common problem is that the WPS parser is now stricter, so it will reject your WPS when it sees some errors, rather than just ignoring them.
15:39:47 Quit CaptainKewl (Remote closed the connection)
15:40:32LiamAetherany way to cscan the wps for errors, so I can correct them? Would be a bit tedious to compare it manually to the wiki
15:41:21ucchanFor FS #10433, I would like to add to FS #10422, FS #10423. For other my patches (codec's patches, text viewer's patches, etc.), I want to add tasks.
15:45:29 Quit _zic ("Ухожу")
15:48:40linuxstbucchan: OK, I have changed 10433
15:51:28ucchanlinuxstb thanks.
15:52:11ucchanAnd I want to edit my wiki page (YoshihisaUchida), Whom should I ask?
15:53:27 Join Lynx [0] (
15:53:30 Join aaron424 [0] (
15:53:54 Nick Lynx is now known as Guest62708 (
15:55:50ucchanOn the Wiki page, I want to manage my tasks (patch, bug, etc.).
15:56:57AlexPLiamAether: The sim with −−debugwps will tell you about errors. There is also a checkwps utility in svn
15:57:49 Quit Lynx_ (Read error: 60 (Operation timed out))
15:57:49CIA-71New commit by nls (r21863): FS #10080 ...
15:57:49 Nick Guest62708 is now known as Lynx_ (
15:59:12 Join evilnick [0] (i=0c140464@gateway/web/freenode/x-54a3aecc9e00aa88)
15:59:16linuxstbucchan: You can ask me! I have just given you rights to edit the wiki.
15:59:16 Join TheSeven [0] (
15:59:18LiamAetherAlexP: Thank you. But I don't seem to be able to find that svn.
15:59:51*linuxstb proposes MrSomeone writes a "debugwps" plugin
16:00:52LiamAethera gui based wps editor with preview would be a dream
16:03:52linuxstbLiamAether: Do you want to post your wps somewhere? Maybe someone can help out and see the mistake immediately...
16:04:29ucchanOh! thanks. I can edit my wiki page. My page is edited in an idle moment.
16:04:53LiamAetherlinuxstb: Alright. I wasn't sure if it was allowed.
16:05:11LiamAetherI've found checkwps, but apparently it's sourcecode for linux
16:05:13linuxstbDon't paste the whole thing here - just a link to it...
16:05:23linuxstbNo, it will compile fine for windows...
16:05:29linuxstb(and OS X, and ....)
16:05:33LiamAetherof course a link
16:05:59LiamAetherwho in their right mind would post an entire file to an irc channel?
16:06:18linuxstbNot everyone is in their right mind...
16:09:57 Quit evilnick ("Page closed")
16:10:44kugeln1s: \o/
16:11:28kugellinuxstb: porting checkwps shouldn't be a too tough job
16:11:48*n1s fixes red
16:13:01LiamAetheralright, here's my wps file:
16:13:30 Quit LiamAether (Read error: 104 (Connection reset by peer))
16:13:35linuxstbkugel: BTW, what is your dev-list post about? What are you proposing to commit - the patch in FS #9873?
16:14:08*linuxstb thinks that perhaps FS #9873 and allowing "spare" buttons to be customised is a question for the RSB
16:15:35 Join LiamAether [0] (
16:15:35CIA-71New commit by nls (r21864): Missed one strncpy call
16:15:48LiamAetherI'm sorry, Firefox crashed
16:16:14 Quit MrDuck (Success)
16:16:19AlexPLiamAether: You could just paste the wps code into e.g.
16:16:38 Join evilnick [0] (i=0c140464@gateway/web/freenode/x-8cd53085919a46cf)
16:16:48LiamAetheroh alright I try that
16:17:03AlexPIt makes it easier for people to look at :)
16:17:16LiamAethersorry, I didn't know about that
16:23:07funmanmake: Das Target »all« wurde wegen Fehlern nicht aktualisiert.
16:23:32linuxstbHmm, seems checkwps is broken again...
16:23:35funmanamiconn, would it make sense to set LANG variable to "C" on your build client ?
16:25:57 Join AndyIL [0] (i=AndyI@
16:26:33amiconnThe new build system already does this. Do you think it's worth doing this for the old system?
16:33:09*linuxstb blames for breaking checkwps, but can't believe it's been broken for 7 months...
16:35:45 Part teru ("Leaving...")
16:35:58*gevaerts is a bit surprised by the strlcpy() delta
16:36:47 Join toffe82 [0] (n=chatzill@
16:36:57amiconnlinuxstb: Why not? Do you know anyone who uses it?
16:37:06 Quit AndyI (Read error: 110 (Connection timed out))
16:37:08 Join jgarvey [0] (
16:37:36AlexPThe theme site uses it doesn't it?
16:37:48LiamAetherit's funny how I don't get half of what's discussed in here
16:37:50linuxstbAlexP: Yes, it should do...
16:39:03amiconngevaerts: Our strncpy() was optimised, strlcpy() isn't :\
16:39:08linuxstbLiamAether: I think the problem with your WPS is the standalone "|" character. Maybe replace that with "%|"
16:39:28LiamAetheralright I try that
16:39:45aaron424I still cannot manage to install the rockbox bootloader on my m250 v4
16:40:00amiconnThat's why the delta is much smaller on SH1 - it builds using -Os, meaning __OPTIMIZE_SIZE__ is defined
16:40:19linuxstbrasher, scorche|sh: Do you know when checkwps was last updated on the themes site?
16:41:46LiamAetherwow something that simple...
16:42:23kugellinuxstb: I'm going to implement it for all (if possible)
16:42:38kugeland it's not costumizable, it's fixed
16:42:49linuxstb"it" is what exactly?
16:43:08kugelthe function the patch adds
16:43:36*linuxstb gives up...
16:43:49kugellinuxstb: i.e. playlist viewer from wps (and insert-into-playlist in the filebrowser)
16:44:10linuxstbSo two new features?
16:44:18kugelbut I only intend to commit the first part for now
16:44:40kugelis the description on the task really that unclear?
16:45:31linuxstbNo, the task is clear, but your repeated references to "it" and "the first part" are not.
16:46:12linuxstbAnd the fact that the mailing list thread you replied to was a long discussion about many things, getting nowhere...
16:48:12 Quit ucchan ("Leaving...")
16:51:40rasherlinuxstb: ages ago - it doesn't compile these days
16:51:57rasherlinuxstb: I can check when the last time was
16:52:05linuxstbrasher: Yes, that would be useful...
16:52:26linuxstbI'm guessing it's r19357, which brings the entire settings struct into the wps parser...
16:53:42rasherlinuxstb: some on 2009-04-26, some on 2009-05-13
16:57:37***Saving seen data "./dancer.seen"
16:58:44 Quit kugel (Read error: 60 (Operation timed out))
16:58:47*linuxstb sees r20000 builds (almost), so that's not the problem...
17:00:50 Join Sajber^1 [0] (
17:02:23 Join Grahack [0] (
17:02:29mcuelenaerewasn't a checkwps build going to be added to the new build system?
17:02:56LiamAetherlinuxstb: If you aren't too busy, would you mind looking at another wps?
17:04:43 Join Lear [0] (i=chatzill@rockbox/developer/lear)
17:06:19AlexPmcuelenaere: It seems a good idea
17:10:33*linuxstb bisects and now blames r20977
17:11:27linuxstbLiamAether: I think the problem there will be (at least) the progress bar - that changed...
17:11:30 Join dash32 [0] (n=dash32@
17:13:34 Join BryanJacobs [0] (n=bryanjac@
17:14:22linuxstbLiamAether: Yes, the %P tag doesn't exist any more - use %pb (I think - I'm not a WPS expert...)
17:14:44LiamAetherI just look %pb up on the wiki
17:15:57 Quit jfc^2 (Read error: 104 (Connection reset by peer))
17:16:19 Join jfc^2 [0] (
17:17:56 Quit Sajber^ (Success)
17:18:13LiamAetherlinuxstb: Thank you. That really was the only problem.
17:20:26 Join itcheg [0] (i=4117734b@gateway/web/freenode/x-5ccff72a3f24d033)
17:24:06 Part itcheg
17:24:15 Join itcheg [0] (i=4117734b@gateway/web/freenode/x-5ccff72a3f24d033)
17:25:18saratogaodd that you just noticed that now, the status bar changed a year ago
17:25:58linuxstbsaratoga: I think LiamAether thought he was updating, but in fact his updates didn't work, until he used rbutil for the first time...
17:26:35 Join FOAD_ [0] (
17:28:28 Quit Zarggg ()
17:30:47 Join pyro_maniac1 [0] (
17:30:48gevaertsn1s: did you see the "Achievements of Mr Someone" bit on the wiki?
17:31:10pyro_maniac1funman: ping
17:32:47pyro_maniac1funman: i know its not the right way to ask but will there be a chance that you will have a look at the sound of the YH-920 ?
17:32:50n1sgevaerts: can't say that i did
17:33:43funmanpyro_maniac1: could be, but i don't see this happening in a near future, also because i know next to nothing about PortalPlayer chips
17:33:57 Quit jfc^2 (Read error: 104 (Connection reset by peer))
17:34:04gevaertsn1s: near the bottom on the MrSomeonesTodoList page. Usually his achievements are moved there instead of being deleted
17:34:17n1sgevaerts: ah, ok, will add it
17:34:19 Join jfc^2 [0] (
17:34:49pyro_maniac1funman: so low_light need a YH-920 to fix that?
17:35:16n1sa bit weird to write "kudos <yourself>" :)
17:35:43funmanpyro_maniac1: i don't know if he's interested in yh920
17:36:06gevaertsn1s: you could let MrSomeone take all the credit :)
17:37:59 Join Zarggg [0] (
17:38:51pyro_maniac1funman: as far as i know he owns the other devices. and sound is the last missing thing to get these devices "supported" thought
17:40:36funmanaren't plugins also needed for these to be supported?
17:40:47funmanand battery charging
17:40:58n1sand a manual
17:41:03n1sand rbutil support?
17:41:21pyro_maniac1wasn't there a change on the devcon?
17:41:44gevaertsthere was
17:43:08gevaertsFor basic "supported" status (I don't remember if we found a good name for it), the port needs to be usable for normal day to day use. That means that if it doesn't eat your files and it usually can play them, it's probably good enough
17:43:59gevaertsIIRC, basically the porters decide if it's good enough
17:45:35linuxstbgevaerts: Isn't a way to install it also useful?
17:45:58*linuxstb wonders if the beast will ever enter "supported" status...
17:46:30gevaertslinuxstb: the DevConEuro2009 page says "fairly stable and documented install mechanisms are required. A target would need to be useful for daily use for music listening"
17:46:43gevaertsso yes, you need to be able to install
17:47:03linuxstbgevaerts: Do you think that contradicts what's been on the TargetStatus page for a while?
17:47:11 Quit FOAD (Read error: 110 (Connection timed out))
17:47:12 Nick FOAD_ is now known as FOAD (
17:47:52 Quit einhirn ("Miranda IM! Smaller, Faster, Easier.")
17:48:07 Join einhirn [0] (
17:48:20amiconnThere's one point on MrSomeonesTodoList which I did months ago...
17:49:25gevaertslinuxstb: I'm not sure about the "three major operating systems" bit, I'd have to listen to the meeting recording again for that. The rest seems to agree
17:49:28GodEaterMr Someone won't be happy. You're stealing his thunder.
17:51:27*amiconn fixed the wiki. r19950, Feb 2009
17:51:45 Join r0b- [0] (
17:51:47amiconnNow just someone needs to make all those screendumps
17:52:27linuxstbgevaerts: Well, rules can be flexible... I would expect that in most cases install tools compile easily on all three (something like mkamsboot), in which case there is no excuse to not release for all three. But if an OS X port of something more complex like beastpatcher is the only thing not done, then I wouldn't want that to prevent supported status.
17:52:54 Quit einhirn (Client Quit)
17:52:57gevaertslinuxstb: exactly
17:53:49*amiconn wants beastpatcher on Amiga ;)
17:54:04 Join robin0800 [0] (
17:54:28linuxstbIs your amiga running linux?
17:54:58 Quit flydutch ("/* empty */")
17:55:03amiconnNo, it's running AmigaOS
17:56:30 Nick fxb is now known as fxb__ (
17:56:50 Quit petur ("work->home->movies")
17:56:55 Join vedlith [0] (
17:57:28 Join flydutch [0] (
18:00:57 Quit jfc^2 (Read error: 54 (Connection reset by peer))
18:01:18 Join jfc^2 [0] (
18:01:47 Quit jfc^2 (Read error: 104 (Connection reset by peer))
18:02:08 Join jfc^2 [0] (
18:02:39 Quit jfc^2 (Read error: 104 (Connection reset by peer))
18:02:51 Join JdGordon| [0] (i=ae91cd87@gateway/web/freenode/x-ed43fb1bbb241c10)
18:03:00 Join jfc^2 [0] (
18:03:05gevaertslinuxstb: I'd expect that about half of the in-progress ports listed on TargetStatus could be moved to "suppported, but not released"
18:03:27 Quit jfc^2 (Read error: 104 (Connection reset by peer))
18:03:28gevaertsWith footnotes for most of them of course
18:03:38JdGordon|does that mean we are changing the meaning of unsupported?
18:03:48 Join jfc^2 [0] (
18:04:07linuxstbJdGordon|: I don't think it's ever had an official meaning...
18:04:08*GodEater still thinks we need a better definition of the term "supported"
18:04:17 Quit jfc^2 (Read error: 104 (Connection reset by peer))
18:04:22linuxstbGodEater: Isn't that what you defined at devcon?
18:04:38 Join jfc^2 [0] (
18:04:39GodEaterlinuxstb: it's what I tried to define
18:04:59gevaertslinuxstb: my impression at devcon was that we all agreed on what we meant, but we had no idea which words to use
18:05:05 Quit jfc^2 (Read error: 104 (Connection reset by peer))
18:05:13GodEaterwe've used the term in the past to mean both "will run rockbox" and "will run rockbox and we'll provide help for it"
18:05:22linuxstbgevaerts: Didn't anyone point to the text on TargetStatus?
18:05:25GodEaterit's not clear in some cases which is which though
18:05:32 Join jfc^2 [0] (
18:05:36GodEaterlinuxstb: didn't you listen to the meeting ? :)
18:05:39gevaertslinuxstb: maybe
18:05:41linuxstbGodEater: No...
18:05:59 Quit jfc^2 (Read error: 54 (Connection reset by peer))
18:06:00GodEaterand after petur went to all that trouble to record and edit it. tut tut.
18:06:11linuxstbDid he ever release a two-channel version?
18:06:20 Join jfc^2 [0] (
18:06:24GodEatercan we set a ban on this jfc^2 chap until he sorts his life out ?
18:06:36GodEaterI'm getting fed up with all the joins and parts
18:06:46*funman ignores joins/parts/quits on #rockbox
18:06:47 Quit jfc^2 (Read error: 54 (Connection reset by peer))
18:07:06funmanjust like the irclogs, this provides better readability IMO
18:07:08 Join jfc^2 [0] (
18:07:37 Quit jfc^2 (Read error: 54 (Connection reset by peer))
18:07:57 Quit Lear ("ChatZilla 0.9.85 [Firefox 3.5/20090624025744]")
18:07:58 Join jfc^2 [0] (
18:07:59gevaertsAnyway, we decided to basically have a big set of "supported" targets, meaning "will run rockbox, but not all features might be there yet (although it will play music)", and a subset of those (name unclear) being "This will work well, most likely you won't want to use the OF for anything"
18:08:01JdGordon|GodEater: anyone with a rockbox cloak can do it... I dont know the temp ban command
18:08:11GodEaterthere isn't a temporary one
18:08:20JdGordon|or any actually
18:08:41 Quit jfc^2 (Read error: 54 (Connection reset by peer))
18:08:58*gevaerts may misrepresent the conclusion a bit due to bad memory, but it shouldn't be far off
18:09:02 Join jfc^2 [0] (
18:09:16JdGordon|gevaerts: we can go all marketting and call them silver and gold status targets
18:09:18jfc^2I seem to have all of my Freenode channels dropping and reconnecting.
18:09:23JdGordon|or poo for anything else
18:09:28jfc^2I'll drop off until I this resolves.
18:09:33linuxstbjfc^2: Thanks.
18:09:36 Part jfc^2 ("I wonder what this button over here does..")
18:10:01GodEaterthat sounds about right to me gevaerts
18:10:42GodEaterJdGordon's suggestion isn't entirely without merit either
18:10:53gevaertsit was proposed I think
18:10:55GodEaterthough it pains me to say it ;)
18:11:38*gevaerts thinks he'll spend some time tonight on actually make some devcon decisions a reality
18:16:09 Part pyro_maniac1 ("Leaving.")
18:17:55 Quit robin0800 (Remote closed the connection)
18:18:20funmanJdGordon|: is battery life a real problem for supported status?
18:18:32funmanthe Sansa AMS have much less battery life than the OF, but still are very usable
18:18:39amiconnLicensing and GPL compatibility is such a grey area :(
18:18:51JdGordon|funman: not really
18:19:11JdGordon|but having 1/3 the OF is never nice
18:19:24funmani have some ideas but wait for SD driver stability to work on it
18:19:41gevaertsI'd say that battery life can be important for gold, but never for silver
18:20:25JdGordon|yeah, I
18:20:27JdGordon|'d agree
18:20:30funmaninstead of using precious metals, what about using other elements? (noble gases?)
18:20:53JdGordon|ignoble alkaloides!
18:21:17funmanHelium (light) for started ports, up to the heaviest element for the best status?
18:21:50BryanJacobsjust don't use things that begin with uu
18:21:50JdGordon|no, we should use arbitrary numbers instead
18:22:05JdGordon|or, maybe the first few greek letters?!
18:22:18BryanJacobs"newbie: the beast port is current only unununium status, don't use it"
18:22:47JdGordon|sholdnt you be drawing picturs BryanJacobs ?
18:23:17BryanJacobsJdGordon: aren't I supposed to be "part of the community", playing with all the cool kids here in #rockbox?
18:23:40*BryanJacobs is on target for being done by the pencils down date
18:23:51JdGordon|yay :)
18:24:04BryanJacobsnot with the pictures, those will be done sooner ^_^
18:28:34linuxstbamiconn: You have a GPL problem?
18:29:26amiconnI would still like to port XSkat to rockbox, but its license is not GPL compatible.
18:29:45amiconnThe question is whether loading a plugin can be considered static or dynamic linking
18:30:01*BryanJacobs remembers the long discussion on this topic about codecs earlier
18:30:03linuxstbWe've always considered it to be...
18:30:47funmanlinuxstb: to be, static or dynamic?
18:30:48amiconnYeah, just bringing it back up because I'm reading up a bit on the linux kernel
18:31:20linuxstbMy view at least is that I want all code in Rockbox (including plugins/codecs and even "data" like themes and fonts) to be GPL'd, or a similar "libre" license...
18:31:25JdGordon|i have a mp3 which causes my clip to stop playback every time i think... is it useful?
18:31:46funmanamiconn: what is the licence of xskat?
18:31:47evilnickJdGordon|: Is it specific to the clip?
18:31:51linuxstbfunman: To be whatever forces plugins to be GPL'd.
18:31:52JdGordon|hmm.. maybe not
18:31:58funmanJdGordon|: no
18:32:20amiconnfunman: It's a homebrew license that would be gpl compatible except for a single term
18:32:56amiconnThe GPL doesn't allow further restrictions apart from those it imposes itself. That homebrew license doesn't allow to make XSkat clones and call them XSkat
18:33:26BryanJacobsrrrrgh, like BSD-4. That's obnoxious
18:34:13BryanJacobswhy do people DO that? Copyright law protects you from other people stealing your logo/name, you don't need a license!
18:34:33JdGordon|people are stupid
18:34:48 Nick adi|away is now known as aditya (n=aditya@
18:35:23amiconnlinuxstb: Yeah, but that's exactly the grey area. If loading a plugin equals dynamic linking, it is debatable whether the plugin should be considered derived work, especially if it wasn't originally written for rockbox
18:35:26 Join Ubuntuxer [0] (
18:36:33amiconnThere's also the possibility of having an LGPL'd glue layer - but if that is possible, what makes that solution special?
18:36:48linuxstbamiconn: Well, what I'm saying is that whilst the legalities are a grey area, there's also a desire (at least from me, and I hope others) to keep all of Rockbox under the GPL, or something compatible.
18:38:52linuxstbBryanJacobs: I didn't think you could copyright a name, just trademark it?
18:39:15BryanJacobsif you produce an intellectual work, you automatically hold copyright: that includes logos
18:39:31linuxstbYes, but a name isn't an "intellectual work"...
18:39:33BryanJacobsthe name could be a work but is most likely a trade name, you're right
18:39:50BryanJacobsluckily, those are also automatically protected the moment you use them to do business :-)
18:40:16BryanJacobsyou don't NEED to register a tradement; the date of registration is just kind of like the date of a patent request, it shows when you had the idea
18:40:32adityakkurbjun: hey you there?
18:40:33BryanJacobsand also puts the burden of diligence of the infringer
18:40:40BryanJacobs*on the
18:42:28 Join kugel [0] (n=kugel@rockbox/developer/kugel)
18:42:34linuxstbThe xskat license appears to be intended to stop people releasing derived versions of xskat not clearly distinguished from the original. The GPL also has similar clauses, so I don't see it as incompatible in spirit...
18:42:58evilnickBryanJacobs: It's quite a murky situation. Given that different countries have different requirements.
18:43:38BryanJacobslinuxstb: which clauses of the GPL are similar?
18:43:44 Quit JdGordon| (Ping timeout: 180 seconds)
18:43:45 Quit r0b- (Read error: 110 (Connection timed out))
18:43:47linuxstbBryanJacobs: 2a)
18:44:03 Quit aaron424 ("ChatZilla 0.9.85 [Firefox 3.0.11/2009060308]")
18:44:20 Quit Ubuntuxer ("Leaving.")
18:44:33 Join r0b- [0] (
18:44:50 Quit gevaerts (Nick collision from services.)
18:45:00 Join gevaerts [0] (n=fg@rockbox/developer/gevaerts)
18:45:42BryanJacobsbut the GPL would still allow you to call the modified version "xskat"
18:46:54 Join robin0800 [0] (
18:46:58linuxstbTrue. I'm really just talking about my personal opinion, not whether they're legally compatible license, which I would say they're probably not.
18:47:28BryanJacobsI love clause 2c and how worthless it is
18:48:20BryanJacobs"you must print out a copyright notice when the program starts and a disclaimer of warranty. Unless you don't feel like it, in which case you don't have to."
18:48:38 Quit LiamAether ("ChatZilla 0.9.85 [Firefox 3.0.11/2009060215]")
18:48:58linuxstbamiconn: Have you ever contacts the xskat author?
18:49:42 Join JdGordon| [0] (n=Miranda@nat/microsoft/x-21e3542d0e1a6051)
18:51:03linuxstbamiconn: What did he say?
18:52:21CIA-71New commit by Toni (r21865): Collect encoder data chunks (usually < 1kB) before writing to disk. The reduced number of ci->write() calls give a nice write speedup to the mp3 ...
18:52:43amiconnHe doesn't want to relicense XSkat under the GPL. What is strange though is that he puts the card images under the GPL
18:53:00bertrikmarkun, anyone, I saw an inexpensive Samsung YP-S3 for sale and downloaded the firmware upgrade file to have a look at it. It contains big-endian ARM code, there's a string saying "8700" and it seems to access the same addresses for peripherals as mentioned in the s5l8700 datasheet!
18:53:20BryanJacobsamiconn: you could ask for a Rockbox-specific inspection
18:53:42amiconnNope, as that wouldn't be GPL compatible
18:54:05linuxstbOr a stripped-down version of the core code licensed under the GPL (basically, dual-licensing the core)
18:54:14gevaertsBryanJacobs: it's the rockbox license that's the blocker here
18:54:35saratogayou could ask him for permission to call your derived version under the same name, in which case you would be reduced to GPL compatibility by voiding that extra restriction
18:54:36*linuxstb has had these kind of conversations with people in the past, and they've never been fruitful - people who use home-brew licenses don't seem to understand them...
18:54:36BryanJacobsyeah, I suppose saying "you can only use this in Rockbox" is an extra restriction :-P
18:54:52amiconnThe core code is what he put under this homebrew license. The X11 gui parts would need to be replaced
18:55:34 Quit Sajber^1 (Connection timed out)
18:55:41saratogait sounds like the underlying problem here is that the guy is too stupid to understand his own licensing, so might as well ask him the simplist possible question
18:55:54 Join jfc [0] (
18:56:19amiconnNo he isn't
18:56:30amiconnHe says that his license is GPL incompatible on purpose
18:56:44amiconnI still have the whole conversation (in German)
18:56:48BryanJacobsbut if he just gave *us* the ability to call it "xskat", then it wouldn't be OK anyhow because being part of Rockbox is an extra restriction above the GPL
18:57:38***Saving seen data "./dancer.seen"
18:59:04linuxstbamiconn: It sounds like there's no solution. Any kind of "hack" to Rockbox to get around the GPL would (I guess...) open up closed-source plugins...
18:59:30JdGordon|no it wont
18:59:48JdGordon|closed source plugins wont happen because the api changes too often
19:00:03linuxstbWhat about releases?
19:00:07amiconnIt's a real pity because (1) there seems to be no other open source skat engine and (2) XSkat *is* open source - just not gpl compatible
19:01:44funmanhow complex is xskat ?
19:01:48linuxstbamiconn: Yes, I agree. There's also the entire emulator world, most of which has the "no commercial use" clause. Fine for us in practice, but not GPL compatible...
19:02:54linuxstbamiconn: You just need to get the xskat author interested in Rockbox...
19:03:02gevaertslinuxstb: I'd consider "no commercial use" a deal breaker. That's not *just* GPL-incompatible
19:03:33BryanJacobsit might be nice to see a hardware player which comes with Rockbox stock
19:03:51BryanJacobsand that would be excluded by "no commercial use"
19:03:57gevaertsexactly. We don't want to forbid that
19:04:20gevaertsAlso, how about getting paid to record something?
19:04:32linuxstbgevaerts: That's what I meant by "in practice". I really don't see anyone using Rockbox commercially...
19:04:36BryanJacobsor using a Rockbox-powered device as a DJ
19:05:01amiconnHmm, the XSkat author violates the GPL himself
19:05:08gevaertslinuxstb: really? "commercial" doesn't imply big enterprises...
19:06:19funmanselling a rockboxed player on ebay for example?
19:06:24amiconnHmm, or not. The cardset is licensed under the GPL, but it's his own work, so he can dual license it
19:06:45BryanJacobsamiconn: you actually can't violate the GPL for something you own the copyright to
19:07:19gevaertsamiconn: also, whether using a cardset is linking in the GPL sense is another one of those nice grey areas
19:07:32BryanJacobsit's a shame Stallman didn't make a pluggable license (I understand why he didn't, I just don't think the GPL has been the unifier he hoped for)
19:07:36amiconnIt's built-in, as C arrays
19:07:54linuxstbBryanJacobs: Why can't you violate the GPL on your own software?
19:08:11rasherlinuxstb: because you don't need the gpl as the owner
19:08:11bzedreal pity because (1) there seems to be no other open source skat engine and (2) XSkat *is* open source - just not gpl compatible
19:08:14BryanJacobslinuxstb: because you own it :-) you can use it how you like OUTSIDE of the GPL
19:08:17bzed19:01:44 < funman> how complex is xskat ?
19:08:42gevaertsamiconn: ah, interesting. Sounds non-distributable then...
19:09:15amiconnWell, in this case it's a dual license. Homebrew in built-in form, and GPL's in gif form
19:09:19linuxstbBryanJacobs: Yes, but if I offer you some software under the GPL, but then refuse to give you the source, aren't I violating the GPL?
19:09:53gevaertslinuxstb: no. You're violating a promise, but not the GPL as such
19:10:16BryanJacobslinuxstb: the GPL says you must accompany redistribution with the source - not that the copy you receive MUST HAVE COME with the source code
19:10:33BryanJacobstechnically, the person on the giving end doesn't agree to the GPL - only the receiver does
19:10:43 Quit stoffel (Read error: 113 (No route to host))
19:12:39 Quit r0b- (Read error: 110 (Connection timed out))
19:12:51BryanJacobsfor instance, I could give you some software with a license that says "you must stab yourself in the chest in order to redistribute this code", but I don't have to stab myself to give you that license
19:12:57*gevaerts reads the xskat version and ends up confused
19:13:37 Quit n1s ("Lmnar")
19:14:59 Quit itcheg (Ping timeout: 180 seconds)
19:15:26 Join Sajber^ [0] (
19:15:40*linuxstb moves to #rockbox-community
19:15:53linuxstbBryanJacobs: Anyway, how goes wavpack?
19:16:39BryanJacobslinuxstb: working on buffering drawings right now
19:17:07 Join Xerion [0] (
19:17:40BryanJacobshave you tried FS #10434
19:18:11BryanJacobsit works... not perfectly, but it works on target
19:20:19 Join Ubuntuxer [0] (
19:20:38linuxstbBryanJacobs: No, I'll try and have a play tonight. Did you ever measure the binsize increase?
19:21:08BryanJacobslinuxstb: no, take a look at the disclaimers on that patch :-)
19:21:17BryanJacobsI have two HUGE buffers in there
19:21:35BryanJacobsthis patch isn't designed for real use... it's a proof of concept
19:21:58BryanJacobsin fact, I think this will only work on targets with 64MB of RAM
19:22:45linuxstbHow big is huge?
19:23:16linuxstbBTW, your patch contains vorbis.c for no good reason...
19:23:36BryanJacobslinuxstb: oops
19:23:46BryanJacobsno lines added/removed there I hope?
19:23:50 Join dys` [0] (
19:23:54linuxstbJust a whitespace change...
19:23:57BryanJacobsI probably touched some whitespace before doing an svn diff
19:24:09BryanJacobsone sec while I check how big the buffers are
19:24:27funmanSomeone wants to contribute a Fuzev2 to an interested developer (in case you have not noticed in on the Sansa AMS forum thread)
19:24:59linuxstbIs adding secpath[] to mp3entry really needed? (there are a few copies of that in RAM, so it's a waste).
19:24:59BryanJacobstwo 15MB buffers
19:25:16saratogafunman: take it for your own
19:25:27BryanJacobslinuxstb: no, I suppose not, it could be done just by having a bool there for "secondary file present"
19:25:39funmansaratoga: if i accept the offer i don't think i'm going to look at it before months
19:25:48BryanJacobstwo 15MB buffers were large enough that I could play the "Bad Horse Chorus", my small Wavpack test file
19:25:51 Join daurnimator [0] (n=daurnima@unaffiliated/daurnimator)
19:26:05kugelI would take it, but I'm also unsure if I had the time and energy to look at it soon'ish
19:26:11linuxstbBryanJacobs: Where are those buffers added? They can't be in the codec itself...
19:26:12kugelwell, I most probably will have time, actually
19:26:20BryanJacobslinuxstb: they are
19:26:21funmani have your fuze, ej0rge's c200v2, soap's clipv1 and all of them still have medium/major bugs i should fix
19:26:33 Nick fortuna is now known as karma (
19:26:39BryanJacobs"static char blockbufs [2][BLOCKBUFSIZE];"
19:26:43linuxstbBryanJacobs: So you've increased the codec buffer size from 1MB?
19:26:59linuxstbHave you actually tested this on a real target?
19:26:59BryanJacobsuhm, no
19:27:09BryanJacobsit works on the sim
19:27:10linuxstbsim != real target ;)
19:27:26*BryanJacobs needs to move the buffers into playback.c then
19:27:38kugelfunman: hasn't there been some work?
19:27:47BryanJacobs1MB will only be enough if I make them rings and rewrite libwavpack/unpack.c to handle smaller bits
19:27:49linuxstbBryanJacobs: Or get rid of them - that doesn't look like a feasible solution...
19:27:50funmankugel: work on what ?
19:28:09kugelfuzev2 (mkamsboot for example)
19:28:11BryanJacobsthe right solution is to make the hybrid-lossless code use little bits of audio instead of a whole block in a go
19:28:11kugelwe should probably take it until he changes his mind :)
19:28:16linuxstbBryanJacobs: I thought the intention was to interleave the data in the main audio buffer?
19:28:26BryanJacobslinuxstb: it is interleaved in the main buffer
19:28:30 Join stoffel [0] (
19:28:34kugelwith a big note *we can't promise anything*
19:28:45BryanJacobsthe issue is that there's no way to know how long it'll be before we can advance the buffer
19:28:51BryanJacobsmaybe I should just keep two ridx values
19:29:30BryanJacobssee, it could be that we don't need a WVC chunk until after reading the next 100 WV chunks
19:29:48linuxstbSo how does the interleaving work? Does it pay any attention to the file itself, or just buffers X bytes from file one, then Y bytes from file two?
19:29:49BryanJacobs"chunks" here being BUFFERING_DEFAULT_CHUNKSIZE bytes
19:29:57BryanJacobsthe latter
19:30:01BryanJacobs32k of one, 32k of the other
19:30:24linuxstbHmm, I would have expected you to look inside the file to interleave the data in the way it needs to be consumed by the codec.
19:30:48BryanJacobsbut see, that's the thing - the WVC blocks can be huge
19:30:52linuxstbThis is what I had in mind by "codec-specific" buffering.
19:30:54BryanJacobsI did that research back at the beginning
19:31:04linuxstbSo you're saying that's not feasible?
19:31:11BryanJacobswell, it will be for large buffers
19:31:24funmankugel: yes there is already dual boot support for fuzev2, in an external git repository
19:31:26BryanJacobsbut for small ones or large files we'll get a buffer full of WV
19:31:34BryanJacobsor WVC, with none of the other file having room
19:31:48kugelwhy is it not in SVN? :(
19:32:28linuxstbSo do you need random access to the entire WVC block, or is it processed from start to end?
19:32:39funmanMrSomeone didn't do it yet / it didn't reach the top of my TODO list ;)
19:32:39BryanJacobsthe latter is true in the compact decoder
19:32:54BryanJacobsbut I haven't rewritten the hybrid-lossless code to behave that way yet
19:33:09BryanJacobsno random access is required but currently it does the whole block and CRC check in one loop
19:33:53BryanJacobsif it's OK to have 1MB chunks in the buffer then I can interleave the blocks
19:34:54kugelfunman: I'm going to write him
19:35:09amiconnIsn't there code that converts a lossy .wv + .wvc into a lossless .wv?
19:35:30BryanJacobsyes, there is
19:35:50amiconnThat code could be used to do the buffering, so that the in-memory stream is plain lossless .wv
19:35:51BryanJacobsit decodes the wv+wvc into PCM, then reencodes it as a lossless file
19:36:09amiconnOh, I thought it could be done without going through pcm
19:36:22BryanJacobsthe methodologies used by the hybrid encoder are incompatibly different from those used by the normal mode
19:36:53 Quit dys (No route to host)
19:37:46linuxstbBryanJacobs: What do you mean by "1MB chunks in the buffer" ?
19:38:06amiconnHmm, that leaves interleaved buffering. I guess both the .wv and the .wvc uses packets which relate to a fixed number of samples?
19:38:11BryanJacobsa Wavpack block can be 1MB in size
19:38:30BryanJacobsamiconn: that's a "block", yes
19:38:39BryanJacobsbut the size of the block is not predictable
19:38:42linuxstbAre blocks independent?
19:38:46BryanJacobsyes, they are
19:38:48amiconnSo you could just alternate between the .wv and .wvc blocks
19:39:00BryanJacobsamiconn: thus, up to 1MB interleaving
19:39:10BryanJacobsI thought that would be too big - if not that makes the code simpler
19:39:10 Nick Horscht86 is now known as Horscht (
19:39:17BryanJacobsa block can be 1MB in size
19:39:32linuxstbBryanJacobs: The difference (I think) is that your loader is not parsing the data to interleave at block boundaries?
19:39:37BryanJacobsthat's right
19:39:37amiconnAh, hmm.
19:39:56BryanJacobsI'll go do that, shouldn't take long
19:40:01amiconnThen I think lowmem targets need to be excluded from hybrid wavpack
19:40:02BryanJacobsthe wavpack block header format is really simple
19:40:12BryanJacobsamiconn: well, if we do it this way
19:40:22amiconnOr maybe they could use a different method. Just decode on load, and buffer pcm
19:40:27JdGordon|kugel: I didnt get around to looking at 8799 last night :/ nothing major has changed since the last one though yeah?
19:40:32BryanJacobsamiconn: whoa!
19:40:43linuxstbBryanJacobs: But I think 1MB is still too big... Can you decode (e.g.) half a block at a time?
19:40:48linuxstb(or even less)?
19:40:51amiconnLess efficient than buffering the compressed data, but not that much of an issue on flash targets
19:40:52CIA-71New commit by Ubuntuxer (r21866): Make the progress bar in game mode puzzle shows the level process (part of FS #10139)
19:41:02BryanJacobslinuxstb: yes, I need at most 4k worth of readahead
19:41:17BryanJacobsbut, like I've been saying, I have to change the unpack.c code to do that
19:41:21BryanJacobsit's on my to-do list
19:41:21amiconnBryanJacobs: Hmm, then you could interleave partial blocks
19:41:30BryanJacobsamiconn: you mean like I'm doing now?
19:41:47BryanJacobsthere is a thought process in here, really there is...
19:41:50linuxstbBryanJacobs: Then I think that's the way forward... 1MB is just too big.
19:42:01BryanJacobswell, I'm glad we've all arrived at the same conclusion
19:42:16amiconnI don't know how you're doing it atm, but using 15MB static buffers doesn't sound like interleaved buffering in the main buffer
19:42:30kugelJdGordon|: uhm, nothing major, yes, but I found another bug
19:42:38JdGordon|bah :p
19:42:54BryanJacobsamiconn: it's interleaved in the main buffer but I need to copy some stuff out to be able to do bufadvance() in a timely fashion
19:42:57amiconnWhat's the size limit of a .wvc block? I guess less than 1MB?
19:43:05BryanJacobsstill 1MB
19:43:13BryanJacobswvc files are bigger than .wv
19:43:31amiconnErr, I meant the lossy (smaller) one of course
19:43:39BryanJacobs"The blocks may contain any number of samples (well, up to 131072), either stereo or mono. Obviously, putting more samples in each block is more efficient because of reduced header overhead, but they are reasonably efficient down to even a thousand samples. I have set the max size to 1 MB for the whole block, but this is arbitrary"
19:43:50linuxstbThat sounds like APE blocks - they can be very large, so Rockbox decodes them in small chunks.
19:44:37BryanJacobsI think I'm going to hit those drawings first because if the main buffering system supports multiple files this whole thing is not needed
19:44:59BryanJacobsbut I will make the codec do partial blocks before the summer's over
19:45:19BryanJacobsoh, and mono/seeking support of course ;-)
19:45:27amiconnThe loader needs to interlave the data in the order the decoder needs it, using rather small chunks. Otherwise this scheme won't work on lowmem targets
19:45:28linuxstbMaking the codec do partial blocks seems like something it should do anyway...
19:45:54BryanJacobslinuxstb: yes, should I do that first then? I'll work on whatever, all the tasks are fun
19:45:57JdGordon|kugel: what bug you find?
19:45:57kugelJdGordon|: it basically breaks font loading...:(
19:46:04*kugel was faster
19:46:11JdGordon|no you wasnt...
19:46:13BryanJacobsamiconn: you don't know the order the decoder needs it
19:46:23JdGordon|why does t break font loading?
19:46:57 Join r0b- [0] (
19:47:29 Quit mcuelenaere (Remote closed the connection)
19:48:02kugelno idea yet, didn't look after it. but one a ui viewport setting is in the config, the font is ignored and the font setting removed from config.cfg
19:50:23 Join mcuelenaere [0] (
19:53:53amiconnBryanJacobs: I think that should be possible once the decoder can handle partial blocks. Maybe it needs at least partial decoding during buffering to decide
19:55:36BryanJacobsno, we know what size a block is from the header
19:55:50BryanJacobswe just don't know in what order we'll need the stuff from the block
19:56:02BryanJacobsI mean, WITHIN a block it's linear
19:56:23BryanJacobsbut we can't say "if WV is A, and WVC is B, then we'll need bytes in order AABBAAAABAABBBB"
19:56:59amiconnWe can say that if we do partial decoding
19:57:22 Join itcheg [0] (i=4117734b@gateway/web/freenode/x-d4c07c19ad02d9b4)
19:57:26BryanJacobswell, no: just because we HAVE accessed things in a certain way doesn't mean we'll continue to do so
19:57:57BryanJacobslet me put it this way.
19:58:07 Join Zagor [242] (n=bjst@rockbox/developer/Zagor)
19:58:08amiconnI mean decoding the whole block, just not all the way to pcm. Just enough to find out in what order the decoder actually requests the data
19:58:09BryanJacobsin an ideal world, the byte we need next is always next on the buffer, right?
19:58:20 Quit stoffel (Remote closed the connection)
19:58:20BryanJacobsamiconn: seriously? that sounds like a bad idea
19:58:33BryanJacobswe do two whole passes?
19:58:43CIA-71New commit by Ubuntuxer (r21867): Replace bitmap menu with a standard menu, clean up the code and remove corrupt save game
19:58:46BryanJacobsone of which happens at the time of buffering?
19:58:52BryanJacobsand remember, the block is 1MB large
19:59:07BryanJacobsso either we buffer it and then memcpy to rearrange it
19:59:11BryanJacobsor we read it from the disk twice
19:59:45amiconnNot two *whole* passes. Decoding certainly consists of several stages. I would think that the first stage is enough to find out the order in which data is consumed
20:00:05BryanJacobsno, it doesn't quite work that way - it's a multipass filter
20:00:20kugelUbuntuxer: why does it use the doubled main menu again? I thought teru had a way to make a menu be dynamic
20:00:21BryanJacobsthe output of later stages get fed back into earlier ones, let me get you a flow chart
20:01:13linuxstbBryanJacobs: I was thinking along the same lines as amiconn - that the buffering would do some parsing and re-arranging of the data, so that it is in the order the codec needs it. The codec then wouldn't need to do that parsing again.
20:01:21BryanJacobs page 2
20:01:26 Join stripwax [0] (
20:01:41 Quit stripwax (Client Quit)
20:02:19BryanJacobsthat parsing would be the decoding process
20:02:20amiconnPage 2 shows the encoder, not the decoder
20:02:28BryanJacobsthe decoder is that in reverse
20:02:33BryanJacobs... almost
20:02:34 Join n00b81 [0] (n=n00b81@unaffiliated/n00b81)
20:02:55BryanJacobswe do multiple passes through a variadic function
20:03:08BryanJacobsand then apply some correction weights
20:03:14BryanJacobsand the correction weights change over time
20:03:32 Quit itcheg (Ping timeout: 180 seconds)
20:05:55 Quit tomcat_ha ("If visible identification is not possible, the pathologist may be able to take fingerprints from the body until decay sets in)
20:06:56BryanJacobsI think we'd have to go most of the way through the decoding process in order to get things sorted out correctly - and remember this would all take place in the buffering thread blocking disk reads!
20:07:34BryanJacobswe'd be losing the advantages of having a non-PCM buffer
20:07:54funmankugel: so do you have any idea about the locks/corruption problems on Sansa AMS ?
20:08:18funmanI see that FlynDice started reading the code, i hope he will notice something ..
20:08:19bertrikfunman, has anyone looked if logf turns anything up?
20:08:21kugelI yet to experience *any* of the reported problems
20:08:38funmanbertrik: what do you think we should logf() ?
20:08:39bertrikfunman, I assume you're talking about the lowmem ams sansas
20:08:46funmanbertrik: no i am not
20:09:18funmani am talking about the filesystem corruption and deadlocks which happens when using a uSD card
20:09:23kugelmy fuze just seems rock solid
20:09:42funmankugel: do you use uSD cards a lot?
20:09:51kugelalmost exclusivly
20:10:23funmani don't know much about these, but i was told there is different categories which relate to speed
20:10:32funmanperhaps we can confine the problem to specific card categories?
20:11:38BryanJacobsbye all
20:11:46kugelmy uSD is a pretty standard sandisk one, I'm not even surprised it works so welll
20:11:50 Quit BryanJacobs ("Java user signed off")
20:11:52funmani only have one card, and rockbox debug menu reports 25Mbits/s just like the internal storage
20:12:29funmankugel: do you mean you have (heard about) problems with not-sandisk cards?
20:12:49funmanalso, can you tell what is a "non-standard" card?
20:14:42kugelfunman: it's a cheap sandisk one. and sandisk "invented" SD, so I expect them to respect the standard more than others
20:15:07saratogafunman: i would take his fuzev2 and then work on it if you get time
20:15:16saratogaif not you can always forward it to another interested developer later
20:16:24funmani don't want to be a post relay
20:16:31amiconnBryanJacobs (logs), linuxstb: The rockbox wavpack codec already handles partial block decoding afaics.
20:16:44kugelsaratoga: heh, that's what I thought too (I already messaged him)
20:16:50funmani think kugel wants to work on his fuzev2 (at least i understood so)
20:17:30amiconnAnd it also looks like there are several decoding steps. The one that actually consumes the input stream(s) seems to be words.c: get_words()
20:17:40kugelI told him that we cannot promise anything (he's obviously aware of this anyway) and that we share the stuff between each other if one runs out of luck/time/energy
20:19:19amiconnCall chain is wavpack.c: codec_main() -> wputils.c: WavpackUnpackSamples() -> unpack.c: unpack_samples() -> words.c: get_words()
20:24:39 Join Thundercloud [0] (
20:25:19*gevaerts could probably just have repeated that...
20:26:46Zagorgevaerts: great. how?
20:27:54 Quit mcuelenaere (Remote closed the connection)
20:28:54gevaertsYou mean, how did it happen?
20:30:05Zagoryeah, or how you go repeat it.
20:30:14amiconnZagor: Btw, is it on purpose that the server disconnects the clients after each round?
20:30:25Zagoramiconn: no. I'm fixing that now.
20:32:00gevaertsI don't know how to make it happen. The last line before "Terminated" was "2009-07-14 13:47:22 Completed build h300"
20:32:28*amiconn wonders how he could monitor the build client once it's running in the background
20:32:52gevaertsThis was for revision 21859 on client hex
20:32:59gevaertsamiconn: screen
20:33:02cg_Zagor: same thing happened here... (jakorasia)
20:33:47cg_2009-07-14 11:23:34 Completed build h10 <−− after that, terminated
20:34:11amiconnI don't use screen, and I would like to run it completely in the background, so that it starts running after boot without me logging on
20:34:23amiconnI.e. from an init script or cronjob
20:34:23gevaertsCould it be a race where it tries to kill the process just after it has stopped by itself?
20:34:51 Quit DaCapn (Read error: 60 (Operation timed out))
20:35:08gevaertsMy case was very near the end of a round (it got Server message: 3 builds left 22 seconds earlier if I read times right)
20:35:11 Quit r0b- (Read error: 110 (Connection timed out))
20:36:01linuxstbamiconn: Does the output of the build client get written to a file?
20:36:55gevaertsamiconn: I'd run "./ > rbclient.out 2>&1 </dev/null" from an @reboot cronjob then
20:37:29FlynDicefunman: Yes I've been reading the sd code but mainly to troubleshoot the card init problems caused by voltage scaling.
20:37:42amiconnOf course I would log the output to a file. That still leaves the question how to monitor whether it has crashed/exited
20:37:56gevaertsZagor: what will "my $pid = $builds{$id}{pid};
20:38:09gevaerts" put in $pid if $builds{$id} has been cleared?
20:38:25Zagorgevaerts: 'undef'
20:38:35Zagorsimilar to null
20:38:43gevaertsZagor: and if you give that to kill? Could that then send the signal to itself?
20:39:34Zagorgevaerts: no idea. testing it now.
20:39:44*gevaerts suspects that this might be the problem
20:40:28Zagoryup, killing undef indeeds kills yourself
20:40:32FlynDicefunman: re: sd v1 and 1<<23 for ocr register in SD_APP_OP_COND, do you have any Idea where that comes from besides ata-sd-pp.c?
20:40:35Zagorgood catch!
20:41:51*gevaerts thinks that the perl documentation should have warned about this
20:43:20gevaertsmost people probably don't try to kill undef
20:44:15kugelyea, blame perl
20:44:28saratogakugel: so are you going to take that V2 fuze?
20:44:47gevaertskugel: actually, in this particular case, I'm tempted :)
20:46:04kugelsaratoga: yes, it seems. I wrote him, that once he answers he'll be getting my address
20:46:40 Join r0b- [0] (
20:46:42kugeland he just did
20:50:43 Quit jfc (Read error: 104 (Connection reset by peer))
20:51:05 Join jfc [0] (
20:55:07CIA-71New commit by zagor (r21868): Don't kill undef $pid.
20:57:41***Saving seen data "./dancer.seen"
20:58:44 Join webguest38 [0] (n=562d61f0@gateway/web/cgi-irc/
20:59:34 Join raphi [0] (
21:00:33 Quit raphi (Client Quit)
21:03:33 Quit Grahack ("Leaving.")
21:04:10 Join scorche [50] (n=scorche@rockbox/administrator/scorche)
21:09:33 Join WPS [0] (n=5ce22570@gateway/web/cgi-irc/
21:10:34kugellinuxstb: re fs#9873, I uploaded the one I intend to commit on the tracker
21:11:03WPShey, how do you simulate "charger plugged in" in the h10 20gb simulator? Can charging be simulated? i didn't find it on the UIsimulator wikipage.
21:12:00kugelU simulates usb insertion, I assume if you hold the special button to charge-only it will simulate charg-only in the sim too
21:12:35linuxstbscorche|sh: Do you have any intention to carry on with the AV300? Seeing it in kugel's patch, I'm wondering if that port can be killed...
21:13:08kugelyea, pretty annoying to make up keycombos for dead ports
21:14:07 Nick n00b81 is now known as n00b81|away (n=n00b81@unaffiliated/n00b81)
21:14:07funmanFlynDice: bit 23 set in SD_APP_OP_COND argument comes from OF disassembly
21:14:08WPSi already tried that key combination but it doesn't work...
21:14:27kugeldoes it simulate USB at least?
21:14:43funmanFlynDice: it could well be some SanDisk code reused in their PP and AMS players ..
21:15:14WPSyes, in a way meaning it only shows the usb sign
21:16:29FlynDicefunman: I have checked sandisk v1 spec and the linux source code and they seem to not use the 1<< 23
21:16:50FlynDicefor v1 sd
21:17:04funmanFlynDice: i think this is controller-specific (voltages supported by the controller)
21:17:07kugelthe internal memory might needed it?
21:17:27funmankugel: i have seen no distinction between internal storage and microSD
21:17:32FlynDicedo you mean the as3525 or the sd controller?
21:17:59kugelfunman: I mean, maybe the internal does need it, but the microsd doesn't care so they just didn't special case it
21:18:06funmanthe sd controller embedded in the as3525, and the (identical, pl180) sd controller added by SanDisk in their players
21:18:25funmankugel: what's "it" ?
21:18:32kugelthe 1<<23 thing
21:18:50FlynDicelet me look again but I found that the as3525 could supply 2.0 - 3.6
21:18:53funmanplease check what it means exactly
21:19:08funmanwe have to give the cards the list of voltages we support
21:19:37funmanfor sdv2 cards we just say we support all voltages, and for sdv1 we only say we support the lowest voltage (bit 23 is the last bit in the voltage array)
21:20:06funmanFlynDice: we can't make assumption for the sd controller attached to the internal storage since it's not in the original as3525 design
21:20:08kugelcould that be the problem of some clips?
21:20:11FlynDicethe 1<<23 says we support only 3.5-3.6
21:20:21funmanah, so it is the highest voltage
21:20:26amiconnWPS, kugel: The sim automatically simulates battery discharge and charging cycles
21:21:52 Join shotofadds [0] (n=rob@rockbox/developer/shotofadds)
21:21:55 Quit robin0800 (Remote closed the connection)
21:22:17 Nick n00b81|away is now known as n00b81 (n=n00b81@unaffiliated/n00b81)
21:24:09JdGordon|kugel: I'm at work now so cant look at the patch, what was the final outcome for 9873 regarding button choice?
21:24:18JdGordon|or there is no choice?
21:24:35gevaertsDo people remember if the "gold" targets require rbutil support?
21:25:23JdGordon|rbutil and manual is for platinum targets
21:25:53*gevaerts decides to listen to the recording to make sure
21:26:23JdGordon|IIRC the descision was to not have those two as blockers for releases
21:26:55kugelJdGordon|: it never was about choice, not for me at least
21:27:30JdGordon|then I dont see the point of adding it
21:27:58kugelI think we should have rather a tight definition of silver and gold (whatever the names will be)
21:28:07FlynDicefunman: sorry, not the as3525 but the pl180 docs say we can supply 2.0-3.6
21:28:09kugelbut I do
21:28:46gevaertskugel: I'm working on a mail to -dev that has what I remember as the consensus in it, with a list of possible candidates for addition
21:28:49funmanFlynDice: makes sense then, did you test it on your uSD card(s) ?
21:29:23FlynDiceyes, but I don't have a v1 card and that would be all that it would affect...
21:29:34*JdGordon| once again complains about our stupid policy of abosultly no customzability in the buttons and menus
21:29:43funmanFlynDice: afaik the internal storage is v1
21:29:49 Quit jfc (
21:29:49 Quit Ubuntuxer (
21:29:49 Quit at0m (
21:29:49 Quit yosafbridge (
21:29:49 Quit karma (
21:29:49 Quit sitwon (
21:29:52NJoinkarma [0] (
21:29:55NJoinat0m [0] (
21:30:04NJoinsitwon [0] (
21:30:06NJoinjfc [0] (
21:30:07NJoinyosafbridge [0] (
21:30:29NJoinUbuntuxer [0] (
21:30:34shotofaddsgevaerts: the D2 with SD support must be at least copper by now ;-)
21:31:08FlynDicefunman: the internal storage worked fine... but maybe that's why voltage scaling was borking the uSd cards, they were using lower voltage...
21:31:55gevaertsshotofadds: I'd say silver :)
21:31:55FlynDicewhile the internal was still using 3.5-3.6
21:32:16funmanwhat about ?
21:32:36shotofaddsgevaerts: we "just" need to get multi-driver in SVN first...
21:33:47FlynDicefunman: Is that the same patch I asked you to look at yesterday?
21:33:48shotofaddsspeaking of which, could you give fs#10415 a quick once-over for sanity? I'm not entirely sure which drive number to use for things like fat_mount() during hotswap...
21:34:01 Join Sajber^1 [0] (
21:34:04shotofadds^that was @gevaerts
21:34:18funmanFlynDice: possibly ;)
21:35:44funmanyours have more comments though ;)
21:35:44FlynDicefunman: That patch works fine for me but I didn't try it with voltage scaling yet. I gotta head off to work now but i'll try experimenting with this info later.
21:36:52gevaertsshotofadds: that sounds like tricky question...
21:36:52FlynDiceyes, I need them to keep my place, happens when you get older... they say it's the second thing to go;)
21:37:27shotofaddsgevaerts: the current behaviour appears to work flawlessly, but I'm not sure that's by design or by fluke
21:37:40*gevaerts hasn't touched multidriver for too long, so he doesn't know for sure
21:38:17funmanFlynDice: comments can only make good
21:38:58gevaertsshotofadds: SD as the main drive will also be fun for rbutil, since it sounds perfectly reasonable to install or update a build with a card reader :)
21:40:19 Join robin0800 [0] (
21:41:00WPSamiconn that's not what i was asking for... i know the battery level changes constantly in the sim, but i want to simulate a charger being plugged in, so i can see if my %bp tag works right
21:41:23shotofaddsgevaerts: I like to present as many options as possible ;-)
21:41:35shotofaddsalso it will be possible to still boot from NAND if there's no SD present
21:42:45gevaertsshotofadds: I suspect the multidriver work misses something : the storage system asks the drivers how many drives they have, but it should also tell them what the drive numbers will be
21:43:55shotofaddsyes, I figured there might be a step missing and the current behaviour is "by chance". i'm not sure how that might work, though.
21:44:06kugelis it really the plan to run rockbox off the SD for the long term?
21:44:23shotofaddsit's not as crazy as you might think
21:44:45shotofaddsand it depends what you call "long term"... until someone writes a read/write driver for the TCC FTL, yes.
21:46:04 Join steven [0] (
21:46:07kugelshotofadds: no it's not crazy, I booted off the microsd on my fuze also (though, just for the fun of it mainly)
21:46:13 Quit Ubuntuxer ("Leaving.")
21:46:18stevenis there an easy way for me to get my album covers on rockbox?
21:46:46shotofaddsI can't see any alternative way of providing persistent settings
21:47:07 Quit robin0800 ("I didn't set up UberScript properly!")
21:47:11evilnicksteven: It depends how they are currently stored. Are they part of the tags or in separate image files?
21:47:25 Join robin0800 [0] (
21:47:52stevenevilnick: more than likely stored in a local folder where rythmbox puts them
21:48:09gevaertsshotofadds: if we make *num_drives accept an parameter (e.g. int sd_num_drives(int first_drive)), and storage_init() passes num_drives to all drivers, it should work I think
21:48:39 Quit robin0800 (Client Quit)
21:48:58 Join robin0800 [0] (
21:49:08shotofaddsgevaerts: that sounds logical, yes.
21:49:23evilnicksteven: If they're in separate files then copy them to the same folder as the music on the target and name them "folder.jpg" - that does assume that your music on the player is in a suitable folder structure. What player do you have?
21:49:31kugelsaratoga, funman: alright, he's sending it to me. Should take a month or so to arrive
21:49:49evilnicksteven: For more info:
21:51:49stevenevilnick: ipod 5.5
21:53:10 Quit robin0800 (Client Quit)
21:53:28 Join robin0800 [0] (
21:53:34 Join Sajber^2 [0] (
21:53:42evilnicksteven: Do you still use the OF for listening to music? If so, then that may cause some issues, as the OF obfuscates the filenames/folder names
21:54:00 Join Windlord [0] (
21:54:07stevenI just copied the music folder with proper structering over
21:54:19 Part Windlord
21:54:20evilnickIn that case, you should be fine with that Wiki page
21:54:23stevenie ; /The Beatles/Help!/Song.mp3
21:54:50stevendo I have to copy each picture for each album manually evilnick ?
21:56:02evilnicksteven: There are a few options. I like to keep the same folder structure on my computer so that I can copy over each "folder.jpg" file and be sure that they appear in the correct folder.
21:56:11evilnick(using a sync tool)
21:56:46stevenwhy use a sync tool when the file manager can just skip duplicates if the folder structure is set up right?
21:57:56evilnickBecause I keep .flac at home, but transcode to a lossy format for a portable player. Therefore I only want to copy the "folder.jpg" files
21:58:06stevenah I see
21:58:25 Quit Sajber^ (Success)
21:59:23 Join Sajber^ [0] (
21:59:53 Quit WPS ("CGI:IRC (EOF)")
22:00:02 Quit LambdaCalculus37 ()
22:00:05 Join Windlord [0] (
22:00:14robin0800steven, I keep master copy on computer and add or subtract change tags and add replay gain then use a sync program to match this to the player
22:01:00Windlordquick question: Is there any channel where I can get help regarding the creation of flash skins for the cowon d2?
22:01:08Windlordsorry I couldn't find any and had to ask here
22:01:27shotofaddsWindlord: not that I know of. your best bet is the forum at
22:01:46Windlordthank you
22:02:39FlynDicefunman: One last question before I head out, do you think internal sd card is v1 because we need bank switching?
22:03:01funmanFlynDice: perhaps it was cheaper to produce than sdhc cards
22:03:57kugelthey probably didn't even produce, but just emptying the stocks
22:03:58FlynDicefunman: but do you _know_ it is somehow or have you deduced this info?
22:04:29CIA-71New commit by rob (r21869): Guard the Timeout API declarations with the correct #ifdef
22:04:44funmanjust deduced, i have no informations from SanDisk
22:05:17FlynDicekugel: that makes sense, gotta go "drive the bus"...
22:05:32 Quit FlynDice (Remote closed the connection)
22:10:23 Join Sajber^3 [0] (
22:10:23***Alert Mode level 1
22:10:23DBUGSent KICK Sajber^1 to server
22:10:23DBUGSent KICK Sajber^2 to server
22:10:23***Alert Mode level 2
22:10:23DBUGsent MODE #rockbox +b *!*n=Sajber@*
22:10:23DBUGSent KICK Sajber^ to server
22:10:23DBUGEnqueued KICK Sajber^3
22:10:23***Alert Mode level 3
22:10:23Kick(#rockbox Sajber^1 :*bang* too many joined users) by logbot!n=bjst@rockbox/bot/logbot
22:10:23***Alert Mode level 4
22:10:23DBUGQ-Sent KICK Sajber^3 to server
22:10:23Kick(#rockbox Sajber^2 :*bang* too many joined users) by logbot!n=bjst@rockbox/bot/logbot
22:10:23***Alert Mode level 5
22:10:23Mode"#rockbox +b *!*n=Sajber@* " by logbot (n=bjst@rockbox/bot/logbot)
22:10:23Kick(#rockbox Sajber^ :*bang* too many joined users) by logbot!n=bjst@rockbox/bot/logbot
22:10:23***Alert Mode level 6
22:10:24Kick(#rockbox Sajber^3 :*bang* too many joined users) by logbot!n=bjst@rockbox/bot/logbot
22:10:24***Alert Mode level 7
22:14:36kugelfunman: That was the one I meant with "the microsd one"
22:14:49 Quit webguest38 ("CGI:IRC")
22:16:27kugelI believe in ku-ku, it's some sort hardware problem. Not that the hardware is broken, but that it operates at the very border (or even out of) the sd specs
22:16:51kugelhigh-speed class 6 sounds already suspicious, although I'm not familiar with these cards in general
22:17:54kugelI wonder if it would helpful to log what his cards responds to our commands
22:20:16funmanHow can we know the "class" of µSD cards by software ?
22:20:23 Join AndyI [0] (i=AndyI@
22:20:25***Alert Mode OFF
22:23:53bertrikfunman, isn't it in the CSD (at least for v2 SD cards)?
22:25:20funmanthe CCC (Card Command Classes) ? The spec shows a fixed value for it
22:25:48funmanit only defines which subsets of commands are available, not speed
22:28:01funmanTRAN_SPEED (max. data transfer rate) only has 2 possible values (25Mbit/s and 50Mbit/s)
22:28:16bertrikamiconn, the codec in the iriver Hxxx (the uda1380) can also handle other sampling rates, like 12, 24, 48 kHz
22:28:30funmanit is only 50Mbit/s for High-Speed
22:28:46amiconnYes, but only if the necessary clock signal is available
22:28:48kugelI don't have highspeed I think
22:29:09bertrikI ran into this when trying to re-use the uda1380 codec driver for the meizus
22:30:16bertrikthe meizus have a separate clock signal going to the codec (MCLK), so you don't have to use the PLL in the codec to reconstruct it
22:30:22amiconnAll our coldfire targets can only clock the i2s bus for multiples or fractions of 44.1kHz
22:30:28bertrikare we sure the irivers don't have this too?
22:30:46amiconnYes, MCLK is locked to the xtal
22:31:19bertrikhm, as far as I understood, it's locked to the LRCK signal, not to a crystal
22:32:39 Quit AndyIL (Read error: 110 (Connection timed out))
22:34:50 Join _lifeless [0] (n=lifeless@
22:34:55 Quit _lifeless (Read error: 54 (Connection reset by peer))
22:35:26 Nick karma is now known as punyhumanright (
22:35:51bertrikamiconn, I'll have a another closer look at the uda1380 driver and the datasheet with that in mind
22:35:55 Join AndyIL [0] (i=AndyI@
22:37:01kugelfunman: should I worry if I get "SD bank error" once in a while (randomly and not reproducable)?
22:37:28kugelJdGordon|: btw, what was the reason to not use home to go to the wps in menus/filetree?
22:37:38funmankugel: yes, did you get the status printed?
22:37:50JdGordon|kugel: home goes to menu
22:38:02kugelfunman: 0x10
22:38:07funmankugel: and the bank number?
22:38:08JdGordon|seen as that *is* the home
22:38:08amiconnbertrik: On an i2s bus, one of the devices needs to be the clock master. In theory both the codec and the cpu can be clock master, but in the coldfire targets the codec doesn't have its own clock source, so the cpu needs to be i2s clock master
22:38:25funmankugel: do you have a 4GB Fuze?
22:38:28kugelJdGordon|: IIUC it could go to the main menu from the wps, and to the wps from trees
22:38:48JdGordon|home+select is wps everywhere
22:38:55kugelfunman: no. although I just did get the message on my clip
22:39:31kugelisn't there only 1 "goto wps" action?
22:39:36amiconnAnd the MCLK is locked to the xtal, just offering a few discrete dividers
22:39:49bertrikamiconn, I think it's the same on the meizus, where the CPU sends the i2s bit clock AND a separate MCLK
22:40:03funmankugel: which capacity is your clip ?
22:40:05bertrikamiconn, so the MCLK is always the same frequency?
22:40:12bertrik(on the irivers I mean)
22:40:18kugel2GB, like my fuze
22:40:33amiconnSo you can either build a coldfire target that handles 11.025/22.05/44.1/88.2kHz *or* one that handles 8/12/16/24/32/48/64/96kHz
22:40:37funmankugel: so you got this message at startup, or while operating?
22:40:56kugelJdGordon|: as far as I see it, what I'm thinking should be doable (that's how the OF behaves too)
22:41:01amiconnChanging between these groups would require changing the main xtal
22:41:03kugelat startup
22:41:32kugelI wouldn't trust my clip if I didn't have it once on my fuze also
22:41:34CtcpIgnored 1 channel CTCP requests in 0 seconds at the last flood
22:41:34*shotofadds managed to get red on the new build servers only ... looks like the ipodcolor build took place in a 'dirty' build folder.
22:41:43amiconnbertrik: Not the same, but derived from the same base clock just by simple dividers. No PLL for that
22:42:08JdGordon|kugel: it could do it, of course, but thats not good... the menu is a much more important quick goto screen than the WPS..
22:43:16*gevaerts investigates
22:43:23amiconnZagor: "1247605338 cpu seconds" ??
22:43:29kugeloh dear, my clip is so broken, it plays completely wrong radio stations
22:43:40bertrikamiconn, the meizu generates the MCLK simply by dividing from the 200 MHz main clock, so it's never exact (but very close)
22:43:52Zagoramiconn: hehe
22:44:04kugelJdGordon|: really? I'm thinking the opposite is true :)
22:44:09gevaertsZagor: I have some left over build directories for the last few days
22:44:16JdGordon|that doesnt surprise me... you're wrong!
22:44:19bertrikthe I2S bit clock is a divided from the MCLK clock and the LRCK is divided from the bit clock again (/32)
22:44:26kugelJdGordon|: you apparently also failed to adapt the radio screen :p
22:44:31 Join B4gder [241] (n=daniel@rockbox/developer/bagder)
22:44:34gevaertsMaybe also rm -rf a new build directory in case it was still there?
22:47:00JdGordon|Zagor: is there any thoughts about having the client say its slow instead of having the server tell the client which group its in?
22:47:11JdGordon|my laptop should not be in the fast group
22:47:34*kugel doesn't think that's a good idea
22:47:40ZagorJdGordon|: no, the idea is that the server decides based on past performance
22:47:45B4gderslow has to be a relative thing
22:47:46bertrikamiconn, what approximate frequency is the MCLK on the coldfire targets?
22:47:57kugela) the border is hard to tell, b) the border is likely going to move
22:48:16bertrikas far as I understood the default on the s5l8700 is 256 times the sample frequency
22:48:25ZagorJdGordon|: you are right on the edge. a few more builds and you'll drop down to the slow category again.
22:49:12Zagorsome clients will likely be moving between the categories
22:50:28bertrik11.289 MHz for 44.1 kHz audio playback for example
22:50:37 Join fyrestorm [0] (
22:50:58Zagorgevaerts: yeah the killing has been a bit buggy lately, leaving leftover dirs
22:51:39gevaertsI suspect that it's the cause for this ipod color red. It's not a very likely occurence (1 in 5000 or so?), but I don't see any other explanation
22:52:54 Join agaffney_ [0] (n=agaffney@gentoo/developer/agaffney)
22:53:13 Quit agaffney (Read error: 60 (Operation timed out))
22:53:22 Quit r0b- ("( :: NoNameScript 4.02 :: )")
22:53:47 Quit AndyI (Connection timed out)
22:54:32kugelfunman: arg, again
22:54:58kugelsd bank -1 error: 0x2
22:55:16funman-1 means at startup
22:55:25funman0x2 is datacrcfail
22:55:33kugelstartup as in boot or sd_enable?
22:55:36funman0x10 is txunderrun .. ?
22:55:42kugelthat certainly wasn't at boot
22:55:42 Quit Thundercloud (Remote closed the connection)
22:56:19funmankugel: that's the only call path to sd_select_bank(-1)
22:56:20kugelfunman: but really wouldn't give too much to what my clip does. I'll report on my rock solid fuze rather
22:56:32kugelit wasn't at boot, believe me or not
22:56:34funmanoh right you mentioned your clip is broken
22:56:46funmankugel: i believe you
22:57:04funmando you believe me as well ?
22:57:37kugelit seems mutually exclusive :)
22:57:45***Saving seen data "./dancer.seen"
22:59:40amiconnbertrik: The coldfire base clock is 11.2896MHz, MCLK is half of that iirc
23:00:01soapfunman, you have my Clip v1 until the end of time.
23:00:10*amiconn might have found an unnecessary colplexity in the coldfire pcm driver
23:00:27soapof should I say "You have your Clip v1"
23:00:41funmansoap: thanks!
23:02:22amiconnUnhelpful: Did you see my remarks regarding asm offset generation on the various architectures?
23:02:23bertrikamiconn, looking at the uda1380 code, it does seem to enable an internal PLL to reconstruct the clock signal rather than use an external clock
23:02:50bertrikit sets bit DAC_CLK in register 0 to enable the "WSPLL"
23:02:55amiconnAnd what would that help us?
23:03:16Unhelpfulamiconn: yes, i "sort of" fixed it, in that it will unconditionally accept either _AD_ or AD_ as the symbol prefix, and .word or .long for the data. i don't think that ought to be a problem.
23:03:29amiconnIt only helps for supporting different sample rates if we can feed it an *external* clock, i.e. not generated by the cpu
23:03:41 Quit steven (Remote closed the connection)
23:04:29amiconnIn fact it does help this way - the irivers and iaudios can only play back 11/22/44/88kHz and record the same sample rates (except 88 for the UDA) from analog inputs, but the H1x0 is able to record all standard sample rates from s/pdif
23:04:45Unhelpfuli think the small ambiguity should be OK, provided devs don't decide they want to name symbols _AD_ or use a different type and expect them hidden
23:05:56bertrikI think we should strife to use direct clocks instead of reconstructing them, unless not possible in another way of course
23:06:17amiconnYou cannot reconstruct a clock that isn't there in some way
23:06:19bertriks/pdif sample rates include 44.1 etc?
23:06:45bertrikI don't understand what you mean to say
23:07:11amiconnThe iaudios support all standard rates from 8 to 96kHz for s/pdif *input*. S/pdif output is limited the same way as analog input and output
23:07:40amiconnThe only exception is loop-through monitoring during s/pdif recording
23:08:20amiconnThis does just output the same s/pdif signal present on the input, without any processing. The analog monitor sounds a bit funny in this case
23:09:07funmanWe can get the card class from the status register (sent on the data line after SD_SEND_STATUS command)
23:11:15amiconnUnhelpful: I'll check what gcc spits out for 'char', 'short' and 'long long'
23:12:47Unhelpfulamiconn: do we really care? if we tell people to use "int", that's good enough, i'd think. the value is the point, and i doubt there's much need for >32-bit asmdef values, especially as pretty much nothing can use them as immediates.
23:13:22amiconnJust curiosity
23:13:59 Quit dash32 (Read error: 110 (Connection timed out))
23:14:08 Join r0b- [0] (
23:15:55bertrikthe current uda1380 code seems to use the MCLK only for recording, it uses the WSPLL for playback
23:16:19 Join dfkt [0] (i=dfkt@unaffiliated/dfkt)
23:18:08kugelfunman: playing music for 30min w/o problems on my clip now
23:21:30amiconnUnhelpful: All use 'byte' for char. There is potential confusion if you match both .word and .long, as coldfire uses 'word' for short.
23:22:07amiconnlong long ends up as 2*.word or 2*.long depending on architecture
23:24:24CIA-71New commit by zagor (r21870): Analyze log and store errors and warnings in the db.
23:24:46Unhelpfulamiconn: i think this is mostly handled if we say "don't use AD_* names if you don't want it exported, don't use _AD* at all"
23:25:10gevaertsB4gder: will you be running the RSB vote again this year?
23:25:18amiconnAnd don't use anything else than 'int' or 'long'
23:25:47Unhelpfulright, which is what the existing asmdefs file says to use, anyway
23:26:31B4gdergevaerts: I guess I can, I'm just a little "off" during this week and the next, while I'm vacationing away from home
23:26:39CIA-71New commit by zagor (r21871): Only build non-existing pics.
23:28:26gevaertsB4gder: Anyone can do it of course, but there's a lot of confidentiality involved, so I guess it should be you, Linus or Zagor
23:30:53CIA-71New commit by zagor (r21872): Now grabs all data from database for much faster execution.
23:30:58 Quit evilnick_home1 (Read error: 113 (No route to host))
23:32:45CIA-71New commit by zagor (r21873): Integrated head+tail printout.
23:34:30 Join CaptainKwel [0] (i=2669ecc2@gateway/web/freenode/x-785634cece0ec72e)
23:35:50Unhelpfulrebuilding again now, but i'm pretty confident that stubs alone, or some other effect of passing the -meabi=4 flag to the assembler, is sufficient to break vorbis.
23:35:56CIA-71New commit by zagor (r21874): Check each build log on completion.
23:37:11CIA-71New commit by zagor (r21875): now does its' own head+tail printout. Added
23:39:29 Quit scorche (" rawr...that is all...rawr")
23:39:31 Quit Lynx_ (" HydraIRC -> <- Go on, try it!")
23:40:14amiconnUnhelpful: Is that a true eabi build, or just trciking the assembler? I would expect the latter to cause problems
23:41:16 Join safetydan [0] (n=deverton@rockbox/developer/safetydan)
23:41:42Unhelpfulamiconn: it doesn't cause any *other* problems that i can find. the only thing that *should* be changing in the object files is the headers, and the relocation types
23:43:52Unhelpfuli've saved .elfs of both and i'm comparing... so far only addresses are different.
23:46:10amiconnIt probably just exposes a hidden bug
23:48:11 Join MrDuck [0] (
23:52:25CIA-71New commit by unhelpful (r21876): Small fix for asmdefs, make sure $var is cleared the line after it is read, so that value *must* follow label immediately.
23:54:49 Join mirak [0] (
23:55:43 Join Stephen_ [0] (

Previous day | Next day