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

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

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

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

#rockbox log for 2005-11-30

00:01:03 Join muesli- [0] (
00:01:26ioboundbadger/amiconn, thanks for help, i got it working by rewriting the assignment to an explicit call to a memcpy-ish method i wrote in my plugin
00:01:52amiconnBagder: Hmm, I don't seem to get it right :(
00:02:07andy_anyone get this when compiling from cvs?
00:02:08andy_dsp.c:473: error: can't find a register in class `ADDR_REGS' while reloading `asm'
00:02:15 Join webguest19 [0] (
00:02:44andy_gcc 3.4.3, binutils 2.16
00:03:14andy_only happens in dev build, normal builds fine
00:03:39preglowyes, dev build doesn't use -fomit-frame-pointer
00:03:42preglowand that routine needs it
00:03:55preglowdoes rockbox have a debug define?
00:04:24*Bagder runs off to bed
00:05:04 Join Wett [0] (
00:05:52andy_#ifdef BUILD_TYPE=DEVEL ?
00:07:09preglowthere's no reason to not -fomit-frame-pointer anyway
00:07:14preglowwe have no target debugger
00:07:17preglowexcept for the bdm :>
00:08:21andy_and rockbox crashdumps doesnt include stacktrace either :)
00:08:56preglownow there's an idea...
00:08:58andy_wee.. peakmeter for iriver recording
00:10:05andy_BBub: sure.. but i want to fix the bug with stop before i commit
00:10:44BBubyou havent looked into the file-splitting, also? :)
00:10:49preglownow, time for implemeting tick counter on nano
00:10:52Cassandraamicon, when you're doing fonts, do you usually include a column of white space pixels at the end, or does the renderer put whitespace in automatically?
00:11:10preglowno, now time for fixing the broken resampler...
00:11:19andy_BBub: I've read about that problem and I think I know how to fix it
00:11:31BBubah, that sounds good
00:11:31CassandraCommiting code with known bugs is a bad idea in general.
00:12:35amiconnBBub: What file-split problem?
00:12:45*amiconn curses Makefiles
00:13:54CassandraOh, never mind. I just looked at some bdf files, and it appears the whitespace is usually included.
00:14:08BBubamiconn: for me the recording stopped at 2 gb with the 2nd part only being 0 byte in size. the player told then told me the hdd was full when there were still 3gb left
00:14:29 Quit Jungti1234 ("bye")
00:14:50amiconnHmm, that's indeed strange
00:15:02*amiconn didn't try recording on iriver yet
00:15:22BBubyes, and any recording after that point only caused new 0byte files until i restarted the player
00:15:40 Join Membrillo [0] (
00:16:55 Quit Membrillo (Client Quit)
00:16:58andy_BBub: i think the problem is that the file split limit is 2 GB - 1 MB, but on irivier writes 32 MB at the time so before "recording.c" detects that it's time to split we are alread over the limit
00:17:11BBubah, i c
00:17:16amiconnThat sounds plausible...
00:17:21markunpreglow: yes, some of the update files are not encrypted
00:17:25BBubeven though i tested it on the h110 which has only 16 mb
00:17:51 Join ashridah [0] (
00:18:06markunBut the fwimg*.dat (they are in the romfs image and on the gigabeat's HD) are
00:19:23preglowmarkun: and there's no code you can see that decrypts it?
00:19:30preglowmarkun: if not so: good luck
00:19:31andy_amiconn: would it be ok to change MAX_FILE_SIZE (in recording.c) to 2 GB - 64 MB just to be safe?
00:20:01markunI don't know. I sent an email to dave hooper, maybe he wants to help out a bit
00:20:06preglowamiconn: so, you getting another player?
00:20:15preglowamiconn: and how come you don't like the clickwheel?
00:20:40amiconnThe wheel is just plain ugly. It's unusable without looking at the screen
00:20:40 Quit webguest07 ("CGI:IRC (EOF)")
00:20:51preglowamiconn: not with audible feedback
00:21:08amiconnI want plain buttons, and not on the sides please
00:21:15preglowamiconn: but no, anyway, i always look at the screen when using controls, so i haven't noticed that
00:21:39preglowand for controlling when you look at the screen, it's plain unbeatable
00:21:47preglowscrolling lists becomes a breeze
00:21:50amiconnWell, not really
00:21:57amiconn...for me
00:22:15preglowi get tired really quickly of just holding a button down and waiting for scrolling to finish
00:22:20preglowwith the clickwheel i have an active part in it
00:22:23amiconnYou can scroll relatively fast, but precise navigation is tricky
00:22:35preglowyeah, but so it is with a button and acceleration
00:22:52preglowanywho, it's a matter of preference, of course
00:22:56amiconnBut that's _only_ when scrolling fast
00:23:12amiconnWith a button you can scroll element by element
00:23:15 Join JdGordon [0] (
00:23:55amiconnAnother drawback of the wheel is that the wheel doubles as 4 buttons
00:24:07preglowdepends how you look on it
00:24:15preglowit's nice to not have to move the fingers too much
00:24:16amiconnAlmost impossible to press the buttons without accidentally triggering wheel events
00:24:19preglowespecially not to a side panel...
00:24:25preglowamiconn: oh? i've never managed that
00:24:26 Quit JdGordon (Read error: 104 (Connection reset by peer))
00:24:34 Join matsl [0] (
00:24:53preglowthe wheel should only trigger when there's a small difference in position, and ignore large changes
00:25:05 Quit muesli_- (Read error: 110 (Connection timed out))
00:25:07 Join JdGordon [0] (
00:26:21amiconnThe navigation in the apple firmware is strange to say the least.
00:26:23 Quit ender` (Connection reset by peer)
00:26:36preglowamiconn: in that we agree
00:26:37amiconnTook me quite a while to find out how to back out the sub menus
00:26:56preglowlucky me that i have no intention at all of using the apple firmware
00:27:30amiconnYeah, anyway, the only positive thing for me about the nano is that it is so thin
00:27:43preglowyeah, the thinness really amazed me when i first got it
00:28:09amiconnYou can stack 3 nanos and they are still thinner than the ondio
00:28:24preglowwow, that much thinner
00:29:25ioboundwell my plugin kinda worked but crashed after playing about 0.2 seconds of audio. any hints on how to debug that?
00:30:04JdGordonchop all songs into 0.2sec segmanats and call it working :D
00:30:07 Join hardeep [0] (i=hardeeps@norge.freeshell.ORG)
00:30:08amiconnandy_: Does num_recorded_bytes really increase in huge chunks on iriver?
00:30:37 Quit muesli- (Read error: 110 (Connection timed out))
00:30:46amiconnOn archos it increases gradually, as the data is read from the MAS
00:32:57andy_amiconn: that's true.. it increases in smaller steps
00:33:18 Join CheeseBurgerMan [0] (n=BurgerBo@
00:33:59ioboundwhat would it take to get pcm_play_data() from a plugin working in the ui simulator?
00:34:19ioboundi.e. what's missing, since it seems to work for codecs?
00:34:44 Join ghode|afk [0] (
00:34:59andy_amiconn: however, in pcmrec_new_file i flush the buffer before starting the new file
00:35:33 Join RotAtoR [0] (
00:35:44 Join DreamTactix291 [0] (
00:35:55 Quit _aLF ("^^")
00:36:35amiconnandy_: Imho this shouldn't matter
00:36:59amiconnThe pcm recording continually updates num_rec_bytes
00:37:16amiconnThis resembles file size + buffered data
00:37:28andy_amiconn: also true..
00:37:37 Join San [0] (n=Test@
00:37:44amiconnWhen the size approaches 2GB-1MB, the recording thread signals the recording engine to start a new file
00:38:29amiconn...and the engine flushes the buffered data (which was already included in num_recorded_bytes), then starts a new file
00:38:52amiconnThe only posible problem I can see would be too much latency
00:38:54ioboundwhat's with the "sector 3d" ?
00:39:07amiconniobound: It's the simulated config sector
00:39:25ioboundok, is sector 3d used as config in the real version too?
00:40:17amiconnThe real version uses (sector_number_of_first_partition - 2) to store the configuration
00:40:44amiconnThis is calculated at runtime; on most harddisks this is sector 0x3d == 61
00:41:06amiconnThe simulator always uses 0x3d, and simulates this in a file
00:41:10ioboundyeah, cause usually there's just the mbr and then 62 unused sectors, since partitions start on aligned offsets
00:41:16ioboundok, thanks for the info
00:41:28ioboundis there any risk of having a corrupted file system if rockbox hangs?
00:41:57ioboundare there many (cached/unflushed?) writes at any given time?
00:42:03amiconnFlash cards are usually formatted with 32 sectors/track, so on Ondio, the config sector is 0x1e
00:42:31amiconnThe is no big risk
00:42:38andy_amiconn: yep.. hum.. should read the problem description again
00:42:40andy_2nd: Rockbox tries to split the file after reaching the 2gb limit (i thought the limit was 4 gb for fat32 as stated in the wiki? 2 gb is the limit for old avi-containers afaik), but it doesn't succed doing that and prompts "Harddisk full - press OFF to continue" and leaves a 2nd file with 0 byte in size. When you try to record again you will only create more 0 byte files - yes, there was enough space left, about 3 gb.
00:43:07 Join LinusN [0] (
00:43:24LinusNandy_: do you know for sure that there was 3gb left?
00:44:01amiconnandy_: Technically the FAT limit is 4GB-1, but a number of programs runs into problems when presented with files >=2GB
00:44:10amiconnSo we're just playing safe here
00:44:33andy_LinusN: no, it's from the forum :) however, "harddisk full" might not be the real reason - it is shown when audio_status returns error..
00:45:01BBubLinusN: i posted that and yes i am sure :)
00:45:09LinusNyes, the only possible error on the archos was "disk full"
00:45:14amiconnI can't get my Makefle to behave :(
00:45:25LinusNBBub: how did you verify that?
00:46:00BBubi looked at the free space in the info screen and also let rockbox rescan the hdd to be sure
00:46:08LinusNok, good
00:46:28BBubit also showed the correct free space of 5 gb before recording
00:46:45andy_hm.. is_error is only set when open or write fails
00:47:26LinusNthere is a slight risk that the rockbox file system driver has issues with >2gb files
00:47:26 Quit arkascha (Read error: 104 (Connection reset by peer))
00:48:09amiconnandy_: Speaking about recording, apps/pcm_recording.c seems to be no longer needed...
00:48:09LinusNand that we never noticed on the archos
00:48:27andy_amiconn: yep.. i will remove it
00:49:08 Join DJDD_ [0] (
00:50:21andy_BBub: are you able to compile from source?
00:50:33 Join Wett_ [0] (
00:51:34andy_BBub: could try to change MAX_FILE_SIZE in apps/recording.c to say 64 MB (0x04000000) and see if that works?
00:53:57 Join muesli_- [0] (
00:54:55 Quit Wett (Read error: 110 (Connection timed out))
00:54:56 Quit xmixahlx ("blah blah blah")
00:55:01ioboundhey now this is pretty cool, i'm playing 2nd_pm.s3m in my dumb plugin in the win32 simulator with sound
00:55:32ioboundit's pretty choppy, 'cause the main play loop does a bunch of malloc()/free() calls, and I had to implement fake pow()/exp() methods so everything is out of tune
00:55:34 Quit San (Read error: 110 (Connection timed out))
00:55:45ioboundbut the tune is recognizable
00:56:03BBubandy_: i dont see a apps/recording.c in the bleeding edge source
00:56:48andy_BBub: sorry, apps\recorder\recording.c
00:58:25 Join TCK [0] (n=tckocr@
01:00:49 Join muesli- [0] (
01:01:41 Quit matsl ("ERC Version 5.0 (CVS) $Revision: 1.776 $ (IRC client for Emacs)")
01:01:45preglowiobound: malloc/free in _MAIN_ loop?=? ? ?? ??
01:01:52ioboundyeah, pretty shocking
01:02:09preglowshocking and completely insane
01:02:15preglowthis is in the standard distribution?
01:02:37ioboundyeah it's basically 0.9.3 with a couple of one-lines to get it to compile without memcpy()
01:03:24preglowyou can use memcpy if you just include a 'extern struct codec_api *rb' at the top of all the files, somewhere, preferably in a common header
01:03:26LinusNDocBrown: i have a test for you
01:04:01ioboundi don't really see how an exported "_rb" symbol helps with linking against a missing "_memcpy" symbol..
01:04:01 Part webguest19
01:04:19preglowiobound: ahh, you'd have to do some #define memcpy rb->memcpy as well, yeah
01:05:08ioboundpreglow, i did that for the various strcpy/fopen/realloc() calls it does, but the memcpy() in question is generated automatically by gcc (the code in question copyed a struct using "=" operator)
01:05:19preglowiobound: how nice
01:05:49ioboundi even implemented a poor man's realloc() since it seems to want to do that a lot during loading
01:06:32ioboundunfortunately it never reclaims freed space so it's bound to crash sooner or later
01:06:40***Saving seen data "./dancer.seen"
01:07:27preglowdo you know why it does realloc?
01:07:32BBubandy_: im testing it now - btw the byte counter is not updating while writing to the disk
01:07:39BBubthe timer is counting on though
01:07:54ioboundi skimmed the source, it seemed to have something to do with you don't know how long / how many patterns/rows/something you need until you're done, kind of
01:08:45ioboundi didn't look like a 5 minute job removing the realloc calls, to put it that way. but hey, i'm not an C expert in any way
01:09:13ioboundi'm just happy to have sound playback in the win32 simulator from my libdumb plugin, and being able to gdb and step in the dumb.rock plugin with debug symbols
01:10:05LinusNDocBrown: the new test boot loader outputs the button readings on the lcd
01:10:13 Join webguest60 [0] (
01:11:11LinusNDocBrown: btn: a b c d, where a=main ON, b=remote ON, c=main hold, d=remote hold
01:11:25LinusNplease try it and report the results in the wiki
01:11:31LinusNtime to sleep
01:11:33LinusNnite all
01:11:42 Part LinusN
01:11:45BBubandy_: the splitting works now
01:12:38ioboundyeehaw, just removed all choppyness by using a double-buffering scheme. the only issue now is the tuning is out of whack due to bad exp()/pow() implementation
01:14:51andy_BBub: i think it's more to it than that.. someone said in the forum that the 2 GB split had worked
01:15:08andy_BBub: so, sometimes it works but sometimes it doesnt :)
01:15:32BBubhmm, at least for me it didnt :)
01:15:36DocBrownLinusN: Saw it and will try... Nite...
01:15:44 Quit DocBrown ("CGI:IRC 0.5.7 (2005/06/19)")
01:16:03preglowiobound: we can fix that
01:16:07preglowiobound: try a simple lookup table
01:16:21preglowiobound: that's what we'll use anyway, it's way faster than any taylor approx we can do
01:16:32ioboundexp would do for lookup, but how to make a small enough lookup table for pow() which takes two doubles as parameters?
01:16:47preglowiobound: i'm willing to bet the base number is more or less always the same
01:17:02preglow2, i'd imagine
01:17:06ioboundoh, and there's log() too. i just return "1.0" for all of them, ha ha
01:18:13ioboundall pow calls either use "0.5", "DUMB_SEMITONE_BASE" or "DUMB_PITCH_BASE" for "x" in pow(x,y)
01:18:32preglowthen we'll be alright
01:18:58preglowa small interpolated lookup table should be precise enough
01:19:01ioboundlog() is only used once and only in the .mod loader
01:19:38ioboundexp() is also only used once, in some method called "it_filter()"
01:19:47preglowyeah, the impulse tracker filtering
01:19:55lostlogicpreglow: were you able to profile Tremor, or if not, how did you decide what functions were worth moving to IRAM?
01:19:58 Quit mordov ("")
01:20:10preglowlostlogic: i didn't, most the of the tremor work has not been done by me at all
01:20:28lostlogicoh, my mistake.
01:20:34ioboundok, i'm leaving for tonight, thanks for all the help so far
01:20:42preglowiobound: okies, nice work anyway
01:21:02ioboundnot sure if you'll say that again when/if you see the source. :)
01:21:12preglowiobound: i trust you'll clean that up ;)
01:21:26ioboundalright, bye anyway! :)
01:21:29preglowlostlogic: but no, the guy who did the optimising didn't do proper profiling either, he based himself on reading the source and an x86 profile
01:21:32 Quit iobound ("Leaving")
01:21:47 Quit muesli_- (Read error: 110 (Connection timed out))
01:22:09preglowlostlogic: and btw, lear is the one who did the iram function placement
01:22:29preglowhe's still around
01:22:46lostlogicwhich was why he knew about the tremor test decode app from xiph svn, makes sense.
01:23:38lostlogicpreglow: initially, I'm not even interested in actual time-in-functions but rather call frequency, to get proper time-in-functions, I'd have to create a profiling arch. for rockbox (unless there's one around that I've missed)
01:23:43 Join mordov [0] (
01:24:51ashridahi remember someone working on something like that back in the day. then he vanished
01:24:51preglowlostlogic: it's not that simple, you more or less have to do a proper profile, cache plays a huge part in coldfire performance
01:25:11preglowlostlogic: just by switching the placement of a couple of functions in libmad, i improved performance 25% once
01:26:02preglowlostlogic: the code cache is implemented in such a way that functions can collide by sharing the bottom 12 or so bits of an address
01:26:14lostlogicI see what you're getting at...
01:26:29lostlogicwhich would explain some of the weird results I was getting moving things around in IRAM yesterday.
01:26:39preglowlostlogic: iram is not cached
01:26:50lostlogichence the lightbulb
01:27:08preglowi was toying around with a CACHE_FUNC define some while ago, which would be used on all important functions, so that they would be placed right next to each other
01:27:18preglowin a section of their own
01:27:25preglowobject file section, that is
01:27:35 Quit andy_ ()
01:28:26preglowbut it never materialised, since we don't have a good way of measuring performance
01:28:47lostlogicneed profiling ;(
01:28:54preglowlooking at the boost ratio to find out if something just got faster is no bloody fun
01:29:56 Quit Febs ("CGI:IRC")
01:30:05 Join Febs [0] (
01:32:34 Join XavierGr [0] (
01:33:30lostlogicso how would we go about getting real profiling on coldfire? I guess I'll have to read more about how gcc's profiling symbols work, and see if there's a way for rockbox to generate them for a particular thread in a way that can be referenced against the object files...?
01:38:06preglowwell, it's not impossible that gprof's profile library is ported to m68k
01:38:19preglowit will take some porting anyway, but might be worth checking out
01:39:33preglowright now i'm thinking perhaps some profile function support in rockbox itself would be nice, like perhaps a rb->prof_start() call and a rb->prof_end() call, and then rockbox caches the results for you, pretty much like logf works
01:39:39lostlogicfrom reading about profiling on, what I'm getting is that the basic functionality should work on m68k... the problem is more with getting a profil() syscall into rockbox for the 'real' profiling (time in function)
01:39:43preglowthen rockbox could use timer.c for timing
01:39:53lostlogicyes, I was thinking along those lines as well
01:40:33preglowsimple, and quite effective
01:40:51preglowthe essence is the functions should be really simple to use
01:40:57preglowso profiling becomes easy
01:41:12preglowand afaik, it shouldn't really be much work
01:41:21preglowyou could even share the logging facility with logf
01:41:39preglowlogf and profiling could be one option
01:41:40lostlogicthat was the part that was most intimidating to me actually −− getting the info written out.
01:41:51preglowlostlogic: well, there you see, that functionality is already implemented
01:42:43amiconnI doubt that a plain timer gives even half-decent values, without disturbing execution too much
01:42:59amiconnThe timing would need to be really fine-grained, correct?
01:43:06preglowcan't timer.c deal with that?
01:43:12preglowbut yeah
01:43:15preglowpretty fine-grained
01:43:26lostlogicrockbox is preemptive, right?
01:43:28amiconntimer.c just offers an interval timer, like the tick
01:43:31preglowlostlogic: correct
01:43:33amiconnlostlogic: nope
01:43:37preglowehh, no
01:43:41preglowi mix terms
01:43:48lostlogicI see
01:43:52lostlogicthat makes profiling harder
01:44:00lostlogicbecause we have to use a userspace timer call.
01:44:19preglowamiconn: but one should be able to fix the timed value when you know a) the timer interval, and b) approximate number of cycles needed by the timer interrupt?
01:44:23lostlogicin preemptive systems, the timer interrupt would do it
01:44:30amiconnpreglow: The interval is variable, but the lower you set the interval, the more overhead the timer will produce
01:44:42amiconnTimer operation will also disturb caching
01:44:50preglownot if we iram it
01:44:58amiconn...and you need to boost all the time to get reliable timings
01:45:03preglowso be it
01:45:14lostlogicprofiling isn't for the weak of battery.
01:45:28amiconnlostlogic: There is no distinct 'user space' in rockbox
01:45:43amiconnThe CPU runs in supervisor mode all the time
01:45:48lostlogicamiconn: aye, the point being there's no kernel space timer because there's no kernel space ;)
01:46:14preglowlostlogic: i don't see why that matters
01:46:17amiconnBtw, there is a 'kernel space' timer
01:47:03amiconnIt might be possible to do some trickery with the timer, but that requires handling it outside of timer.c
01:47:12amiconn...or even inside, but with a special method
01:47:35lostlogicI obviously need to read way more of the rockbox system code before starting to tackle this.
01:48:20amiconnInstead of just using timer interrupts and setting the rate high, it might be possible to set the rate low, and use both the interrupt counter and the timer counter itself for calculating the deltas
01:48:55preglowrockbox id3 info viewer has no stereo/mono indicator?
01:49:48amiconnThis is somewhat tricky, because the profiling function needs to stop the timer, wait a couple of cycles in order to process an eventually pending interrupt, then read both interrupt counter and timer value, the let the timer continue
01:50:08amiconnThis would also account for the processing overhead in the profiling function...
01:50:13CassandraRockbox has a new font. Woo and yay! ;)
01:50:27amiconn...because the timer would be stopped while in that function
01:50:29preglowi find it strange that a mono file shows different peaks on the peak meter...
01:50:35preglowCassandra: what font?
01:50:50CassandraIt's called jackash - see my last checkin message.
01:51:00 Join AnInternetUser [0] (
01:51:08preglowanyone got any mono mp3s that can confirm this for me?
01:51:25CassandraFinally got around to converting my very old Linux console font to bdf properly.
01:51:46AnInternetUserehh is there a page that shows the latest progress for rockbox?
01:52:11preglowAnInternetUser: there, right on the front page
01:52:20Cassandra(It actually predates the time Linux started using bdf console fonts. Scary.)
01:52:35AnInternetUserhow about for a specific console
01:52:47preglowAnInternetUser: you mean target? ;)
01:52:54AnInternetUseryeah whatever
01:52:58AnInternetUserlike h300 which i have
01:52:59 Quit Kohlriba ("Leaving")
01:53:07preglowAnInternetUser: no, there's nothing like that, that would mean someone manually making a list
01:53:21preglowAnInternetUser: but the target is usually specified in the commit message
01:53:43AnInternetUseri thought i saw a list where all the stuff was checked off
01:53:49AnInternetUserbut i can't find a link to it anywhere
01:53:56preglowAnInternetUser: ahh, yeah, but that might be inaccurate
01:54:31amiconnpreglow: The only problem with that method I can see is that the value will become wrong if interrupts are disabled for too long...
01:54:45lostlogicamiconn: my goal really is to reuse gprof functionality rather than writing something for rockbox from the ground up.
01:55:05amiconn...but that would also happen with a short-intrval timer (even more likely)
01:55:21amiconnlostlogic: I have no idea how that works, I never used it
01:55:53preglowlostlogic: would indeed be great, but i foresee a host of problems
01:56:35lostlogicit works by 'simply' recording the PC every timer tick (on modern systems) and then counting the ticks at each PC in order to determine time in function
01:56:55 Quit AnInternetUser ()
01:56:56preglowand you need a fixup program
01:57:02preglowthat in our case would probably have to read map files
01:57:10preglowor nah, not really
01:57:29 Quit Paul_The_Nerd (Excess Flood)
01:58:04preglowbut the base binaries we want to test are stripped of everything after they've been linked and relocated to the plugin base address
01:58:10preglowso you need some trickery there
01:58:40amiconnI don't see a problem here
01:58:45preglowthe .elf files that still have symbols haven't yet been relocated to the base address
01:58:56preglowbut you can probably fix that quite easily
01:59:04amiconnYou can map PC->function with the .map files
01:59:04lostlogicnot a big problem... that's basic remapping
01:59:18preglowamiconn: yeah, but nice to have a chart with statistics be done automatically
01:59:22 Quit Febs ("CGI:IRC (EOF)")
01:59:27preglowamiconn: using map files is tedious work best done by a computer
01:59:35lostlogicpreglow: that's what gprof the application does any way
01:59:42preglowuses map files?
01:59:42amiconnHmm, except for static functions
01:59:51lostlogicpreglow: can
01:59:56preglowlostlogic: then hooray!
01:59:57 Quit XavierGr ("Trillian (")
02:00:02preglowamiconn: yeah, but that's a problem anyway
02:00:41amiconnSo the problem actually is how to record all the PC values, efficiently
02:00:50preglownot a very big problem...
02:01:21preglowthat shouldn't be a huge problem
02:01:21lostlogicthat and implimenting mcount functions that -pg makes every function call so that we don't have to write our own compiler in order to do this!
02:01:51 Join XavierGr [0] (
02:02:17lostlogicwell if we can do both of those things effectively...
02:02:45lostlogicis there a limit on how long a given thread can hold the CPU if it never yields?
02:02:58preglowit's not enforcable
02:03:22preglowour thread context only saves a subset of cpu state, so we can't just steal control of a thread
02:03:31pregloweven though it is possible by interrupt
02:04:27preglowwe save all registers that aren't scratch, plus stack pointer, that's that
02:04:32 Nick CheeseBurgerMan is now known as CBM-away (n=BurgerBo@
02:04:41XavierGrwow TiMiD is now idle for a 6th consecutive day! TiMiD: Were are you? We are missing you. :)
02:04:49lostlogicok, so can't we count execution time in functions by time between mcount calls less yields or something?
02:05:46preglowyou need to rephrase the last part of that so it's understandable
02:06:44lostlogicok, so the system is cooperative, which means that while the thread we are profiling has control, time spent in A is the time between when A calls mcount and when B calls mcount
02:06:57lostlogicso unlike preemptive system profiling we don't have to have two separate calls
02:07:11lostlogicunless of course I'm totally wrong.
02:07:44preglowsounds right, as long as nothing here actually calls yield, heh
02:07:57lostlogicso, can we just record RTC values in mcount, and at every yield and do the calculations later?
02:08:28preglowwell, if you exchange 'rtc' with 'some high res timer', then yes
02:08:57lostlogictime-mcount-B - (time-mcount-A - (time-resume - time-yield)) = time-in-A
02:09:34lostlogicwe have high res timers that don't suck, I hope?
02:09:41preglowask amiconn
02:11:38XavierGris it on purpose that the win32 sim always says backlight: on when I press a button? (I suppose that this is the backlight emulation?)
02:11:52lostlogicXavierGr: yesa
02:13:55 Join }{CBM}{ [0] (n=BurgerBo@
02:14:14 Quit CBM-away (Read error: 104 (Connection reset by peer))
02:14:29CassandraBacklight emulation doesn't seem to have much effect on X11.
02:15:26 Quit Wett_ (Read error: 110 (Connection timed out))
02:17:01XavierGrCassandra: What motley means? :x
02:18:38CassandraErm, literally it means the kind of clothing a a medieval jester wears.
02:18:56CassandraIn this context, it means sort of random and all over the place.
02:19:03XavierGrha so you mean old character set in the commit?
02:19:07 Quit }{CBM}{ (Read error: 104 (Connection reset by peer))
02:19:12XavierGrah ok then
02:19:23 Quit BBub (""Nonstop Belze." -")
02:19:47CassandraYep. The font was originally designed back before Linux had proper i18n support and used the standard VGA 850 codepage.
02:19:52 Join }{CBM}{ [0] (n=BurgerBo@
02:20:19preglowand lovely times they were
02:20:40CassandraI remember Linux 1.0 being announced.
02:20:46CassandraThat really dates me, doesn't it?
02:21:34preglow'fraid i can't remember that
02:21:42markunI started using linux in 1995. I'm not sure what the version of the kernel was
02:21:50XavierGrI must had been in liquid form back then... when Linux 1.0 was released?
02:21:51preglowbut i remember testing a really early one and being awe-struck at being able to multitask in textmode!
02:22:01CassandraI think I was about 19 at the time.
02:22:15preglowwhen was linux 1.0 released?
02:22:23XavierGryear? yes
02:22:46preglowthen i was around 13
02:22:51preglowno, 12
02:22:53XavierGrah ok then I was 9 years old!
02:23:11CassandraFortunately I don't remember DOS 1.0
02:23:21 Quit muesli- (Read error: 110 (Connection timed out))
02:23:29CassandraAlthough I do remember the ZX-81, which was released about the same time.
02:23:33preglowearliest dos i remember was 5.0
02:23:40preglowi used dos 6.22 a _lot_
02:23:41markunpreglow: what's your birthday? I'm from 1979-12-12
02:23:56preglowmarkun: ninth of june, 1981
02:24:17markunI always imagined you being older than me :)
02:24:22preglowthere you go
02:24:30Cassandrapreglow, congrats - you're the same age as the ZX81.
02:24:52CassandraHow's it feel to never have lived in a time before home computers?
02:24:53preglowCassandra: can't brag about ever having used one, though
02:25:11preglowCassandra: reassuring
02:25:23markunI'll have some sleep. See you all tomorrow
02:25:28preglowmarkun: i bet i will
02:25:40CassandraOr mobile phones, actually. Although in 81, mobile phones were the size of a car battery.
02:25:42markunyou too (in advance)
02:25:55preglowmarkun: wont be long
02:26:09preglowCassandra: there were some around, mobile phones spread pretty earily in norway
02:26:15preglowearly, even
02:26:35preglowwe even had our own network type, called nmt
02:28:24CassandraAnyway, I should get some sleep too. Night all.
02:29:03 Join Benacool [0] (
02:31:24lostlogicpreglow: just got a message that makes me very happy: /home/lostlogic/rockbox/apps/codecs/Tremor/vorbisfile.c:241: undefined reference to `mcount'
02:32:17lostlogicso if we define that as an extern in the headers of what to be profiled, and implement it in rockbox... things should magically work
02:33:07XavierGrlostlogic: What are you working?
02:33:47lostlogicXavierGr: crazy idea of building profiling ability into rockbox that preglow and I seem to be tossing about
02:34:05Benacoolprofiling of what ??
02:34:23Benacoolexemple ?
02:34:25lostlogicthe point is to make it possible to profile things in rockbox.
02:34:30lostlogicI'm working on it for the tremor codec
02:34:55XavierGrprofiling as to windows profiling .e.g?
02:35:16lostlogicwhat do you mean by windows profiling?
02:35:17Benacoolokay ....
02:35:29Benacoolthe windows accounts
02:35:32XavierGrdifferent users and stuff
02:35:39XavierGr(don't shoot)
02:36:13lostlogicprofiling is monitoring all of the operations of a program, generally for the purpose of performance tuning it, or finding a memory leak
02:36:37Benacooli see ^^
02:36:54XavierGrah in a plugin or a thread that will be set by an option?
02:37:29XavierGror have it always on?
02:37:43lostlogicit will be a compile time option, very performance damaging to do the actual profiling.
02:37:48 Join San [0] (
02:39:06XavierGrwell I made a neat battery performance tsr plugin the previous month, if you are interested on such things. (though it has nothing to do with your work, it is just a performance log on realitime usage of the battery)
02:39:38lostlogicnot at the moment, I'm really just interested in tremor performance tuning currently :)
02:40:30lostlogicpreglow: this is a lot easier if we require all code to be profiled to be run on one thread. ok, no ok?
02:42:01preglowlostlogic: completely ok by me
02:42:08preglowlostlogic: i'm only interested in codec code anyway
02:42:19preglowlostlogic: and dsp code
02:42:24preglowlostlogic: both of those are run in the codec thread
02:43:25preglowlostlogic: well, if you get this working, it'll most certainly be appreciated
02:43:30preglowso please do :)
02:43:43preglowi'll go to bed now
02:43:56lostlogicg'night, I'll try...
02:53:41XavierGrpreglow: what is better? Have a flag (diff_flag) to tell if the user made changes to a small (fmr file) or make a fn that will detect if there is a change (byte to byte)?
02:56:56 Quit San (Read error: 110 (Connection timed out))
03:00:32 Quit webguest60 ("CGI:IRC")
03:02:57 Join Febs [0] (
03:06:43***Saving seen data "./dancer.seen"
03:07:31 Nick }{CBM}{ is now known as CheeseBurgerMan (n=BurgerBo@
03:08:41JdGordonrandom q time.. any aussies here with optus digital?
03:09:12ashridahfoxtel only here, sorry.
03:10:07lostlogicman, gcc's profiling stuff is some slick code
03:20:27 Quit hardeep ("BitchX: all the things phone and hop won't include")
03:28:25lostlogicif profiling information is stored in normal RAM, it'll be mega-slow, but if it's stored in IRAM, it'll force things being profiled to use less than their fair share of IRAM.
03:32:31 Quit DreamTactix291 (Read error: 110 (Connection timed out))
04:02:48 Quit JoeBorn ("")
04:19:35 Quit Benacool (Read error: 110 (Connection timed out))
04:23:35JdGordonis the record button planned to do anything on the h300 in the wps sccreen?
04:25:05XavierGrcurrently record button, only starts recording in the recording screen nothing else... (in iHP-100)
04:25:53JdGordonye, but is anything planned for the button in the wps screen? coz id like to reimplement the shuffle/reply modes on the h300 like the origional fw
04:26:02JdGordonif hte button has nothing planned ofr it
04:26:53XavierGrnothing is planned for it. So you can play as you like with it.
04:27:21XavierGrBut I guess that button assignments will be talked a great deal before a release.
04:28:23JdGordongreat.. now to find the wps screen code :p
04:28:30XavierGrnot to mention that current button mappings needs to be changed. Buttons are way too inconvinient this way IMHO
04:29:13JdGordonthey really should be userconfigurable.. so u dont have to do stupid checks in the code for which keypad your dealing with
04:29:59XavierGrwell the code for wps resides in apps/gui/gwps.c and gwps-common.c
04:30:02XavierGrI think
04:31:05XavierGrgo where the buttons are called. Then call your button and add your code. Though I don't know if it is easy to add what you want.
04:31:47JdGordonwell im bored.. and i realised im missing that feaure too much :D
04:32:24XavierGrSometimes I make special builds for myself with changed buttons.
04:37:24 Join Paul_The_Nerd [0] (
04:37:45 Quit PaulJ (".")
04:37:58 Join San [0] (
04:47:32 Join tvelocity [0] (
04:58:09 Quit San (Read error: 110 (Connection timed out))
05:05:05 Join DreamTactix291 [0] (
05:05:57 Quit Paul_The_Nerd ("Chatzilla [Firefox 1.5/undefined]")
05:06:48***Saving seen data "./dancer.seen"
05:08:59 Quit CheeseBurgerMan (Read error: 104 (Connection reset by peer))
05:13:42 Join Paul_The_Nerd [0] (
05:13:55 Join elinenbe [0] (
05:14:15*ashridah idly wonders if the cpu speed cap will be iriver's RLD syndrome :)
05:15:15Paul_The_NerdRLD syndrome?
05:15:58ashridahRed LED of Death. an issue some archos's had where the unit would lock up and the hard drive LED would stay on.
05:16:01ashridahtook a while to solve or so
05:18:42JdGordonXavierGr: i think ive done it... was damn simple once i figured out how rb deals with its settings :p
05:22:54JdGordon:'( the status bar is fucking with my head... whats the repeat mode that has SHF but no rep symbol?
05:26:22JdGordonok, how do i get some debug info with the sim?
05:30:10Paul_The_NerdIsn't that "reshuffle the entire playlist upon reaching its end"?
05:32:03XavierGrashridah: Do you have any news on it? Or did you hear Linus saying that it is very difficult to solve?
05:33:20 Join CheeseBurgerMan [0] (n=BurgerBo@
05:33:59ashridahon the cpu speed issue/ i've heard him say it's not something he understands yet, that's about it
05:34:29XavierGrstrange indeed
05:34:57XavierGrwhat the progress on the H300. I think I have lost some episodes...
05:35:37JdGordonsound works, no radio/recording
05:35:38XavierGrCan it record? What about music? Is it the same with the H100 (except the CPU frequency)? Do plugins work allright (H100)?
05:36:01XavierGrsound works as in H100?
05:36:16JdGordoni dunno how it works in the h100 but it works fine in the 300
05:36:30ashridahsound works, but the cpu speed makes some of them skip
05:36:37JdGordonsounds perfect to my destoryed ears so thats all i care about
05:36:37XavierGrI mean all the codecs, without frequent crashes .e.t.c...
05:36:46ashridahmost plugins work, and many are currently being updated with colour and fullscreen support
05:36:53ashridahXavierGr: most of the codecs work untouched'
05:36:58ashridahsome just need optimisation
05:37:05JdGordonoh, hasnt crashed on me yet, i only use mp3z tho... oggs arent done fully yet
05:37:30XavierGrlucky "bastards" we waited so long before playback was implemented with all those codecs in H100 ;P
05:38:08XavierGrfor you it was just a week....
05:38:30XavierGron the other hand you waited so long to see a port so....
05:38:51DreamTactix291i look at it this way. the more platforms the better
05:39:10DreamTactix291i just have bragging rights that mine was the first non-archos one platform supported :)
05:39:59*XavierGr wonders when an official release for H100 will be out! :)
05:40:22DreamTactix291probably can't be all too long from now
05:40:35DreamTactix291just minor things now probably
05:40:40ashridahthere's a cycle of feature freeze to get through first
05:40:46DreamTactix291the firmware runs very stable nowadays
05:40:52XavierGrImagine it: With our installer and the pics on the supported models page! :D
05:40:53ashridahand i suspect they'd like to have recording working against least one codec
05:40:54 Join FingerSoup [0] (
05:41:11DreamTactix291they could do wavpack :P
05:41:36XavierGrwavpack seems nice, though I don't know how much of use it would be for me.
05:41:43DreamTactix291for me
05:41:50DreamTactix291my entire lossless collection is wavpack and has been for a long time
05:41:52 Join }{CBM}{ [0] (n=BurgerBo@
05:41:56DreamTactix291so i'm more than familiar with it
05:42:18lostlogicwho here thinks I'll actually succeed in making a profiling system for rockbox?
05:42:39XavierGrat least I managed to record my concert without those iriver damn clicks! :D
05:42:40DreamTactix291you can succeed at anything if you keep at it
05:43:18lostlogicDreamTactix291: thanks :) I've basically ripped out one good function from the GCC implementation of profiling and we'll see where it gets me.
05:43:26XavierGrI was very worried because 2 week ago recording was still at the debug menu
05:43:43XavierGrand very buggy.
05:44:02DreamTactix291once stuff gets up in priority it usually gets done fast
05:44:03XavierGrbut no it is lovely, only peak meter remains.
05:44:52Paul_The_NerdAnd encoders.
05:45:11XavierGrwell I will not hold my breath for them...
05:45:27DreamTactix291well here's the thing
05:45:35JdGordon:'( my box is fucked
05:45:38DreamTactix291fixed point mp3 encoders aren't of high quality
05:45:55DreamTactix291so if you really care about archiving a recording that's probably not the way to go
05:46:04JdGordondoes chanign tracks with repeat stall the player for any1 else?
05:46:05Paul_The_NerdWho said MP3 encoders?
05:46:12DreamTactix291well that's one encoder
05:46:28Paul_The_NerdThere's already a working lossless WV encoder on Rockbox, it's just not hooked into the recorder
05:46:38DreamTactix291i've used the wv encoder
05:46:43DreamTactix291it works fine
05:46:47Paul_The_NerdSame here
05:47:00XavierGrwhy it isn't comitted yet, any problems with it?
05:47:06Paul_The_NerdIt's comitted Xavier
05:47:14Paul_The_NerdIt's only useable as a viewer on .wav files
05:47:31XavierGrah so you can encode the wav files later?
05:48:00ashridahthere's just currently no framework for encoding files on the fly
05:48:07XavierGrwhat about an unnoficial mp3 plugin with a good encoder?
05:48:18DreamTactix291read what i said above
05:48:26DreamTactix291fixed point mp3 = like sine
05:48:30DreamTactix291not really of quality
05:48:55XavierGryes but we could find non-open source good wuality mp3 encoders
05:49:02XavierGrit is just that it would be illigal
05:49:11FingerSoupI'm looking at some of the plugin code, and I see a bunch of hex for the graphics. Is there a helper utility to make these graphics? if not, how do I go about encoding my own? (I've figured out how it works with minesweeper, but I don't know about sudoku...)
05:49:15XavierGrbut if it is unofficial hehe
05:49:43JdGordonFingerSoup: there is a tool to ocnvert bmp to hex, or just load them as bmps on the player
05:49:45DreamTactix291XavierGr: yeah but fixed point mp3 encoders still aren't very great
05:50:27Paul_The_NerdDreamTactix291: "great" isn't really necessary. If you want to preserve quality, you do lossless anyway. If you're using lossy on-site, you're *probably* just doing voice anyway.
05:50:36FingerSoupso opening a BMP in a hex editor should do it?
05:50:39XavierGrwell iriver has a good one. It is copyrighted though. And there comes the Unofficial!
05:50:51DreamTactix291iriver's mp3 encoder isn't that good
05:50:57Paul_The_NerdFingerSoup: There's a tool, bmp2rb or something. Don't quote me on the name, I don't remember. =/
05:50:59DreamTactix291it's just better than shine :P
05:51:02XavierGrI thought it was...
05:51:40FingerSoupahh, I see it - in the tools directory of rockbox-devel.... thanks
05:51:53XavierGrwell as I said wav is enough for me. I am happy with it anyway.
05:52:20DreamTactix291and you can losslessly compress on the unit anyway
05:54:06 Quit TCK (Read error: 110 (Connection timed out))
05:56:28lostlogiclil help: can someone unravel this line into english for me? pfdindex = ++pfd[0].link;
05:57:06Paul_The_NerdI think if it did realtime compression to WV or some other lossless format as it recorded, most people wouldn't care too much about lossy recording.
05:57:49XavierGrwhat about lossy wavpack! ;p
05:58:15 Quit CheeseBurgerMan (Read error: 110 (Connection timed out))
05:58:30DreamTactix291is lossy wavpack integer?
05:58:33DreamTactix291i know lossless is
05:58:59Paul_The_NerdI dunno
06:06:06 Quit tvelocity ("Leaving")
06:06:27 Nick }{CBM}{ is now known as CBM-away (n=BurgerBo@
06:11:36 Quit DMJC (Read error: 104 (Connection reset by peer))
06:16:45lostlogicin need of a rough estimate of the total number of caller->callee pairs in any given codec.
06:16:54lostlogic(function pairs)
06:23:11lostlogic1000 seems reasonable.
06:24:25FingerSoupIs there a functional simulator for either IRiver H300 or iPod?
06:25:47lostlogicthere is a sim for H300, keep in mind that the rockbox sims are just UI simulators, not full sims
06:27:07FingerSouptrue, but if I want to test a plugin for a connect 4 game i should test it in the sim before I ask to have it added to the CVS, right?
06:27:34FingerSoup(Also I want to test Color)
06:28:18lostlogicyes, test in sim.
06:28:22lostlogicnot sure about color
06:28:24Paul_The_NerdWorking on the sim in now way guarantees it'll work on the hardware (and the opposite is also true)
06:28:31Paul_The_Nerd*no way
06:28:43Paul_The_NerdBut it's still a good idea
06:29:34FingerSoupok. Does the h300 sim actually work with a color display (it didn't last build I did a week ago)
06:30:16JdGordonye, its a colour display
06:33:58 Join Lost-ash [0] (
06:36:53FingerSoupOk, I'm getting green screen with garbled data. Remote works, but I just get lines on the h300 sim main screen...
06:37:42 Quit ashridah (Nick collision from services.)
06:37:44 Nick Lost-ash is now known as ashridah (
06:38:08 Join San [0] (
06:38:52FingerSoupis this normal?
06:40:01Paul_The_NerdI'm *guessing* not.
06:40:33ashridahwhat were you doing at the time? the lcd driver is bound to have bugs
06:41:19JdGordonany1 know why the quick screenw as removed from h300?
06:41:22FingerSoupI just compiled it. ../tools/configure, make, make install...
06:41:52FingerSouplatest CVS
06:42:01ashridahoh, wait, Sim.
06:42:04*ashridah missed that bit
06:42:16FingerSoupis the sim broken?
06:42:23JdGordonnot here
06:42:27ashridahnot sure, i haven't used it in a while
06:42:33FingerSoupLinux or cygwin?
06:42:57FingerSoupwin target? not X11?
06:43:13JdGordonthey were syaing last nite the x1 sim needed work..
06:43:18JdGordoncould be what ur takling about?
06:43:43FingerSoupok... Win target didn't work for me previously on the H100, I assumed that the X11 was the best build for all.
06:43:56FingerSoupwill make clean and try again...
06:45:36JdGordonwhats the ucl pack that keeps killing the compile?
06:46:44ashridahJdGordon: quick compression utility
06:47:18JdGordoncan i disable it so the build works?
06:47:35ashridahit's being merged in and may cause random breakages atm because they need to get the rockbox firmware compressing for some of the older targets that have limits on how much data they can load on boot
06:47:44ashridahi have no idea, amiconn is the person to ask there
06:52:22Paul_The_NerdSee, if such a thing actually existed JdGordon, I'd be handing them out to everyone I know. (Including possibly myself, but the insecure webserver is only up when no the other computer is physically powered down anyway. :-P)
06:52:52ashridahaah, if only.
06:55:46Paul_The_NerdThe last two tick boxes apply to everyone I know in person.
06:58:22 Quit San (Read error: 110 (Connection timed out))
07:01:26 Quit RotAtoR ()
07:05:57lostlogicsleep on this: How can I hash the ~1 million possible PC values to the ~1000 profiling counters efficiently?
07:06:52***Saving seen data "./dancer.seen"
07:25:09 Join aliask [0] (
07:27:48 Join DMJC [0] (
07:28:07aliaskJdGordon: You still around?
07:28:16JdGordoni r
07:28:29solexx_preglow: forget it, I see you fixed the problem with mono files
07:28:29aliaskDid you get that WPS patch for the play mode running
07:29:08JdGordonits a dead simple patch, but i dont know if its the firmware in general, or my playaer or what but it easily locked up the player
07:29:28JdGordonand switching from shuffle -> sorted list (and vice verca) doesnt work so well
07:29:33aliaskLocked up your player?
07:29:55JdGordonye jumping to next track with any repeat mode set
07:30:07JdGordonso dunno if it was the patch or what causes it
07:30:43aliaskIf you hold the A-B button when nothing is playing, you can alter the shuffle, etc, but not while something is playing. That might have something to do with it.
07:31:30JdGordonOH.. i didnt know the music had to be stoped to do it
07:31:39*JdGordon has been wondering where that damn screen went
07:31:42Paul_The_NerdIt *shouldn't* have to be.
07:31:59aliaskLet me double check it, but I think that's how it went.
07:32:01Paul_The_NerdBut for some reason that screen isn't showing up during WPS on the H300.
07:32:35aliaskYeah, you have to stop the music.
07:32:39JdGordonye, try it.. music has to be stopped, not even paused to get the screen
07:34:08*JdGordon hates the damn key mapping... NAV should cancel the track nav and play should play the song.. not the other way arround...
07:34:19aliaskAlso, on a side note, do the BMP's in peoples h300 WPS screens flicker?
07:34:31JdGordonye a bit...
07:34:44aliaskOk good, just as long as I'm not the only one.
07:34:47Paul_The_NerdPlay is actually Resume.
07:34:52Paul_The_NerdIt plays what's already playing.
07:34:57Paul_The_NerdNav selects an object
07:35:06Paul_The_NerdIt makes sense from one perspective.
07:35:10JdGordonye, thats what i mean.. its confusing :p
07:35:17*Paul_The_Nerd shrugs.
07:35:31Paul_The_NerdOther people think it's perfectly logical. I thought it was confusing initially, but now think it's really more logical
07:35:42Paul_The_NerdYou launch *any* file by "selecting" it. .txt or .mp3 or otherwise
07:35:45aliaskThe transition from the iRiver firmware is a bit of a steep slope, because some things are the same, and then when something different pops up...
07:36:04 Join TCK [0] (
07:36:56aliaskIs anyone working on Rockboy any more?
07:37:14Paul_The_NerdNot at the moment.
07:37:16Paul_The_NerdIt's been in limbo for a while.
07:37:28JdGordonwhat button loads the quick menu on the h100?
07:38:12aliaskIt's a shame, it has such potential. And it's so close too. Pity my C skills fall a bit short. (I started learning yesterday, and have so far made extremely basic plugins)
07:38:39Paul_The_NerdThe menu that's A-B on the H300 is also A-B on the H100
07:39:11Paul_The_Nerdaliask: I looked at it myself, but I'm not that great with C either, since it's been forever, and the code was... well... confusing.
07:39:36aliaskDoes it run full speed at 120mhz?
07:40:04Paul_The_NerdI think the estimate was about 72% speed or something like that?
07:40:12*Paul_The_Nerd isn't sure.
07:40:40aliaskHrm, I could do a test to see what % it is at 90mhz.
07:42:18Bgrmorning :)
07:42:23Paul_The_NerdHonestly, I'd actually like to see an NES emulator. :-P I mean, if I had a color screen model.
07:43:33aliask35% by my calculations.
07:43:40 Nick aliask is now known as alias|afk (
07:46:09*Bgr tryed schip8 yesterday
07:46:12 Join Membrillo [0] (
07:46:58Membrillocan the win32 sim only be used by devs?
07:47:02Membrillofor h3xx btw
07:47:30BgrMembrillo: if you mean that u must build it on your own, then: yes
07:47:55ashridahyou could probably use it as a particularly unweildy music player :)
07:48:42Membrillowell, I just wanted to use it for testing out wps's. Saves sending them across and testing them
07:48:49 Join Kohlrabi [0] (
07:48:56ashridahit'd work for that
07:49:06ashridahbut you'd need to build it yourself.
07:49:13Membrilloah ok
07:49:20Membrillothats not an easy thing?
07:49:40 Quit crwl ("harry, you're a beast!")
07:50:02Paul_The_Nerd"easy" is a very subjective term...
07:50:47Paul_The_NerdI've never had much luck with the "devkit"
07:50:52Bgrhm, this won't work
07:50:52MembrilloBgr: thanks thats perfect
07:51:01BgrMembrillo: this won't work ...
07:51:03Membrilloit wont?
07:51:23BgrPaul_The_Nerd: i had luck with it ...
07:51:28MembrilloBgr: why is that?
07:51:44Bgrbut the last time i checked out cvs ... i had to edit some makefiles ...
07:51:54Bgr(because of the ucl not compiling)
07:51:58Paul_The_NerdBgr: Ah well. Things just don't like me.
07:52:56Bgranyway, i primary compile under linux, but i had work with premiere last days ...
07:53:34MembrilloBgr: this guide is very linux based... im on windows :\
07:53:43Membrilloa little confusing
07:53:56BgrMembrillo: cygwin is "simulation" of linux under windows
07:54:17Bgreven the DevKit uses cygwin
07:54:24Bgrbut this is just more hidden
07:54:56Membrilloi think ill just send them to my iriver :P im going to get very confused very soon
07:55:09 Quit FingerSoup ("Say What?")
07:56:15Membrilloi wonder how i could write a .wps viewer in VB
07:56:26Membrillold be useful
07:58:08BgrMembrillo: see
07:58:45Paul_The_NerdIt could use updating though
07:58:47 Join crwl [0] (
07:59:04Paul_The_NerdAnd it doesn't preview them
07:59:19Bgrhm, yes
07:59:24Membrillowell, i have no problem writing the wps
07:59:32Membrilloits just viewing it is a pain
07:59:34BgrMembrillo: ah, ok
08:00:21 Nick alias|afk is now known as aliask (
08:00:23Bgrmaybe someone should write an wps "viewer" (most preferably in C, using the rockbox's wps code as much as possible)
08:00:58Paul_The_NerdBgr: They're called "UI simulators" :-P
08:01:01MembrilloBgr: yeah i was thinking using the rb code would be very benefficial, but i only know VB, not C
08:01:12Membrillootherwise it wouldnt be too difficult
08:01:30Bgryes, and it won't have the same bug as rockbox's wps ;)
08:01:48Membrillowhats that?
08:01:57Bgrthe same bugs
08:02:11Membrillooh right gotcha :P
08:02:31Paul_The_NerdWhich bugs are those?
08:02:48BgrPaul_The_Nerd: aren't there any ?
08:03:09Bgri don't have anything particular in mind
08:03:30Paul_The_NerdWell, no the H300 bitmaps flicker, but thats not so much a bug in the WPS code as the display code I think. Wouldn't really know about that.
08:04:09Paul_The_NerdOtherwise, I'm pretty sure it's all "working as intended" at the moment.
08:04:33Bgri've seen only one bug in wps and it's the fact that when the statusbar is on and i load another wps ... the first line of the wps goes on the top of the statusbar
08:04:40Bgr(at least on the remote)
08:05:04Bgrsomeone confirmed this
08:05:05Paul_The_NerdDo your WPS have the %we or %wd tag in them?
08:05:21BgrPaul_The_Nerd: i haven't digged into WPSes
08:05:42Bgrjust loaded other wps than the default
08:05:44Paul_The_Nerd%we and %wd enable or disable the status bar.
08:06:01Paul_The_NerdI *think* they're necessary now for WPS to function properly
08:06:07Bgrand the first line (artist) didn't showed
08:06:35Bgrthen i loaded rockbox_default (is that the correct name)
08:06:54Paul_The_NerdBgr, was that one of the included WPSs or is it from the WpsGallery or elsewhere?
08:07:47Bgrand it showed on the top of statusbar (the artist name was long and i was able to see flickering statusbar/wps artist)
08:07:58Bgrafter reboot it was OK
08:08:07BgrPaul_The_Nerd: it was from the included ones
08:08:12Bgrboxes or iAmp
08:08:26Bgragain, all this was on the remote
08:08:50Bgrbtw, s/<Bgr> and the first line (artist) didn't showed/(the artist name was long and i was able to see flickering statusbar/wps artist)
08:09:13Paul_The_NerdI thought both of those had the status bar disabled...
08:09:22Paul_The_NerdThat is strange.
08:09:33 Join FingerSoup [0] (
08:09:35Bgri'll test again, just give me a sec
08:09:47Paul_The_NerdI'll test too.
08:09:51Paul_The_NerdAre you H300 or h100?
08:10:22Bgrh300 :)
08:10:25Paul_The_NerdDo you load it by "Browse themes"?
08:10:30Bgrno, i dont
08:10:41Bgrbtw, boxes disables the statusbar, correct
08:10:46Bgrbut it flickers on the remote
08:11:05FingerSoupok, I'm trying to build graphics for a rockbox plugin. the bmp2rb program is giving me erronious results.... ie: 4 digit hex values for an unsigned char... can I just split these, or will I end up with problems?
08:11:35BgrPaul_The_Nerd: ok, it's simple
08:11:40Bgrjust load rwps
08:11:43Bgrlike boxes
08:11:46Bgrthen rockbox_default
08:11:54Bgrand u'll see that the artist is not showing
08:12:07JdGordonFingerSoup: which player u converting it ot?
08:12:25Bgrafter that reboot and u'll see the artist line
08:12:30Bgrunder the statusbar
08:12:31FingerSoupI'm using target 0 - Mono
08:12:33Paul_The_NerdIt works fine on mine
08:12:41Paul_The_NerdAt least with "browse themes"
08:12:47Paul_The_NerdLemme try to manually load the rwps
08:12:51FingerSoupIs there a restriction on my bitmap sizes?
08:12:59JdGordondont tihnk so
08:13:12BgrFingerSoup: it's up to the screen size iirc
08:13:29MembrilloI just finished iPodVOL optimisation for h3xx
08:14:05Paul_The_NerdBgr: If you use Browse Themes it works fine.
08:14:18Paul_The_NerdIf you browse the .rwps it behaves as you describe.
08:14:21BgrPaul_The_Nerd: but if you use the browse .rwps ... :)
08:15:00FingerSoupWell, it's a small bitmap.... 22x19
08:15:11Paul_The_NerdBut is default the only one that does this Bgr?
08:15:14JdGordonhas any1 tried enabling the fm reciever on the h300 and seenn if it works without mod?
08:15:20BgrPaul_The_Nerd: no idea :)
08:15:22Paul_The_NerdThe default wps/rwps are actually handled differently than all the others
08:15:33Bgrbut if it's the default, it's in the rb code
08:15:50Bgrsee, i haven'
08:15:57Bgrt had time to dig into this
08:16:40FingerSoupI'm getting mixed 16 and 8 bit digits outputted. ie: 0x3c, 0xfcfc,0xfc, 0x50, 0x00,
08:16:49JdGordonthats fucked :D
08:17:01FingerSoupCan I split the 0xfcfc to 0xfc 0xfc?
08:17:11JdGordontry it and find out :p
08:17:23BgrFingerSoup: isn't 0x3c = 0x003c ?
08:17:25JdGordonworst thing that will happen is it wont display properly
08:18:06FingerSouptrue... If not, is there an *ick* manual way to calculate?
08:18:19FingerSoupie: plotting on graph paper?
08:19:29JdGordonwhich player? i tinhk thats doable on the archos...
08:19:42FingerSoupalso, is it normal for the output to end in a comma? ie:0x00, 0x00,};
08:20:08Bgrhm, the last isn't supposed to be normla
08:20:33FingerSoupSomething's funny in there...
08:21:43FingerSoupI am supposed to feed it a 1-bit BMP, right?
08:22:25BgrFingerSoup: r u using rb2bmp ?
08:22:33 Join Jungti1234 [0] (n=jungti12@
08:22:40Bgroh, yeah
08:23:18Bgriirc the wps loading code supports only 1 (and maybe 2) bit bitmaps for now
08:23:41FingerSoupIt's not for wps.... it's for a Connect 4 plugin...
08:24:17FingerSoupSo I need the bmp2rb to make my bitmap...
08:24:25Bgrthen u should be able to use 16bit bitmap ...
08:24:30Paul_The_NerdBgr: I suspect the internal default rwps needs to be updated.
08:24:48FingerSoupok, I'll give it a shot, changing my bit depth...
08:24:50BgrPaul_The_Nerd: i suspect the same
08:25:22BgrFingerSoup: why don't u take a look at how the rb logo is showed ?
08:27:00FingerSoupI've been looking at plugins such as minesweeper and bejeweled. I want the graphics to be self-contained.
08:28:13FingerSoupthey have something that is SIMILAR to what bmp2rb outputs, but it's always a 2 hex-digit number, never 4, and the array ends with a hex digit, not an empty spot before the curly brace....
08:28:41Bgrhm... doesn't the bejeweled use a 1bit bmp as mask
08:30:00Bgri think it does, because of the color defines
08:30:06Bgr#define COLOR_BLACK LCD_RGBPACK(0,0,0)
08:30:06Bgr#define COLOR_RED LCD_RGBPACK(255,0,0)
08:30:11Bgrand so on...
08:30:52FingerSoupstatic unsigned char jewel[8][66] = { {0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,...
08:30:58Membrillois there any way you can align text by pixel rather than left center or right in WPS's?
08:31:15FingerSoup 0x00, 0x00}
08:31:26BgrMembrillo: i don't think there is such way (for now)
08:31:32FingerSoupThat's the part I need to convert from BMP.....
08:31:45Bgr rb->lcd_mono_bitmap(jewel[0],
08:31:45Bgr j*TILE_WIDTH,
08:31:45Bgr (i-1)*TILE_HEIGHT+k-TILE_HEIGHT/8,
08:31:46***Alert Mode level 1
08:31:46Bgr /* draw new position */
08:32:11Bgr [bj->playboard[j][i].type]);
08:32:28JdGordonany1 know if enabling the radio and fixing errors will kill me h300???
08:32:33FingerSoupyes. that's not my problem code.... the problem is my "static unsigned char jewel[8][66] = { {0x00," is totally miunged....
08:32:44FingerSoupmunged even....
08:33:07FingerSoupA char is 8 bit... FCFC is 16 bit...
08:33:14BgrJdGordon: i don't think that u should try
08:33:32JdGordonits the same tuner as the h1x0's isnt it?
08:34:02Paul_The_NerdBgr: It only applies to the default, as far as I can tell
08:34:39BgrJdGordon: yes, it is, but you don't know whether it's on the same i2c address ...
08:34:54JdGordonwhats the worst that could happen?
08:34:54Paul_The_NerdBgr: Though poking around, I can't see why yet.
08:35:13BgrJdGordon: to write commands to different chip than the FM tuner
08:35:31JdGordonso what? crash? or crash and burn?
08:35:44Bgrmaybe crash ...
08:35:59Bgri don't know what chips are connected via i2c ...
08:36:14JdGordonit couldnt kill the hardware tho? right?
08:36:27Bgrthe power managment unit was connected also via i2c iirc
08:39:01 Join San [0] (
08:41:47***Alert Mode OFF
08:43:50JdGordonwell.. it didnt kill my player :D
08:44:13Bgris there radio
08:44:21 Quit Strath (Read error: 104 (Connection reset by peer))
08:44:40JdGordonno, the tuner screen works, but no sound.. and music wouldnt play once i quit it.. so needs some work
08:44:42BgrJdGordon: u just didn't realise it yet
08:44:45JdGordonbut it didnt crash or kill it
08:45:02 Join Strath [0] (
08:45:03Bgryour disk info is totally destroyed :P
08:45:03JdGordoni rebooted into iriver and sound works so im happy
08:45:16Bgrjust kidding
08:45:24*JdGordon needs to do safer things when bored :p
08:47:31Bgrhi, Jungti1234
08:49:02Membrillojungti, i told you i would linger :P you want me to test that wps now?
08:51:09 Join einhirn [0] (
08:56:05Membrillowhere can i get it
08:56:59JdGordonwell that was scary.. music wouldnt play in rb.. even without the radio compiled...
08:57:01JdGordonall fixed tho
08:57:56JdGordonso at least we know its "farily safe" to compile it in... untill more proper testing is done
08:59:12 Quit San (Read error: 110 (Connection timed out))
09:04:27BgrJdGordon heh
09:05:18 Join ender` [0] (i=ychat@
09:05:29*amiconn notices ashridah was joking ;)
09:05:57Membrillodo you think it would be possible for rockbox to stop USB connection without rebooting?
09:05:58amiconn[05:16:03] <ashridah> took a while to solve or so <== 'a while', ahahaha
09:06:21BgrMembrillo ?
09:06:27*Bgr searches the logs
09:06:54***Saving seen data "./dancer.seen"
09:07:16*Bgr hopes that ashridah will not be right
09:07:24Bgrisn't right, even
09:07:58BgrMembrillo: ; search for "05.14.15"
09:08:24Bgramiconn: get an h300, please :))
09:08:34Bgruf, s/an/a
09:08:34Jungti1234maybe soon possible.
09:09:07Jungti1234Membrillo, Where is WPS?
09:10:26 Join amiconn_ [0] (
09:11:41 Quit amiconn (Nick collision from services.)
09:11:41 Nick amiconn_ is now known as amiconn (
09:12:15amiconnThe 'while' was more than 2 years...
09:12:39Bgryes, that's why i hope u'll get a h300 :P
09:13:02amiconnBgr: I couldn't find one at any of the electronics shop I visited yesterday
09:16:26FingerSoupamiconn :
09:17:21amiconnYeah, I know about the eBay option, although I would look at german eBay
09:17:23FingerSoupnot mine, but I figured there has to be some on ebay
09:17:29FingerSoupahh, ok
09:18:09FingerSoup$31 is pretty cheap though (3 days left tho...anything could happen...)
09:18:12BgrBID ONLY IF YOU ARE IN THE US. Please.
09:18:41Bgrhm, this guy has worse english than mine ... :)
09:18:43FingerSouptrue... Read the info - ships only to US...
09:19:55amiconnGetting things from eBay outside the own country (or another EU country in case of the EU) might cause nasty surprises
09:20:18amiconnBelieve me, I ran into that when getting my H140 from Switzerland :(
09:20:45amiconnThis was not the seller's fault; customs and VAT, you know...
09:21:04 Join San [0] (
09:21:33Paul_The_NerdWow, $31 is incredibly cheap. I'm surprised it started that low.
09:22:05ashridahhas it hit a reserve?
09:22:40Paul_The_NerdThere's not a mention of one
09:26:14Jungti1234 : wow..
09:27:23Paul_The_NerdNew York City served them a cease and desist a little while back
09:29:49Paul_The_NerdThey're $400 at the official iriver store online.
09:30:05Paul_The_NerdI'd say $31 is a rather good deal. :-P
09:30:10Paul_The_NerdThese are USD.
09:30:36Jungti1234The Korea is $268.99.
09:31:19 Quit Sando ()
09:31:51 Join Sando [0] (
09:36:38Jungti1234Soon new firmware of H300 becomes release.
09:37:38ashridahor so it's claimed.
09:38:03*ashridah fondly remembers being burned by promises of gapless playback and OTF playlists
09:38:27Jungti1234It becomes release certainly.
09:39:08Jungti1234iRiver employee spoke to me.
09:40:43ashridahi've heard that before.
09:40:50Paul_The_NerdYeah, they just kinda gave up on the H120. Made promises, and never kept 'em.
09:40:53*ashridah believes only when he sees
09:41:02ashridahnot that it matters to me, since i don't have a H300
09:41:09ashridahi hope you're right
09:41:12ashridahbut i wouldn't bet on it :)
09:42:19Paul_The_NerdHey, new firmware doesn't mean new features anyway.
09:42:35Paul_The_NerdMaybe they decided to make the volume have twice as many increments.
09:42:45 Join Vladoman [0] (
09:42:47ashridahheh. or it could mean you get an option called 'gapless' like we did, which noticably did jack
09:43:23ashridah(if you don't already)
09:43:59 Join LinusN [0] (
09:44:38Bgrmorning, LinusN:)
09:44:47Bgrhahaha Jungti1234:)
09:45:33*Paul_The_Nerd mutters something
09:45:51Paul_The_NerdI upgraded to a newer version of my IRC client, and it doesn't display channel joins any more
09:46:29ashridahLinusN: SDL works across X11 just like any X11 client does
09:46:40ashridahLinusN: unless you're using it to get a GLX context.
09:46:53LinusNPaul_The_Nerd: it's propably fixed in the next daily build of your client ;-)
09:47:02LinusNashridah: good
09:47:04ashridahit'd probably vastly simplify input handling, drawing to the screen (the same code would work on windows and X11, pretty much)
09:47:18ashridahof course, it means devs would need an SDL library for cygwin
09:47:21ashridahand headers
09:47:29ashridahbut there should be instructions for that on the SDL website afaik.
09:47:31LinusNthat was my next question
09:47:40Paul_The_NerdI should really use a real client, rather than ChatZilla
09:47:52LinusNPaul_The_Nerd: i use xchat2
09:49:23ashridahLinusN: covers the various ways to build against sdl for windows
09:50:18perplexityqemu is compiled with SDL and that works very well cross platform
09:51:18*ashridah notes it's a pity he's about to fly 500k's away from his linux development machines in two days :(
09:51:44FingerSoupok, I'm still having issues with bmp2rb. I can't get good output from it... It keeps spitting out 16 bit digits for an unsigned char....
09:52:22FingerSoupare there any restrictions on the type of file I feed it?
09:52:41FingerSoupusing target 0 (Mono for most players)
09:54:19LinusNFingerSoup: oh
09:55:29FingerSoupOutput (whitespace removed) const unsigned char white[] = {0x80, 0x60, 0x10, 0x08, 0x04, 0x04, 0x02, 0x02, 0x01, 0x01, 0x01, 0x01, 0x01,0x01, 0x02, 0x02, 0x04, 0x04, 0x08, 0x10, 0x60, 0x80, 0x0f, 0x30, 0x40, 0x80, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x40b,0x00, 0x6423, 0x6665, 0x6e69, 0x2065, 0x4d42, 0x4850, 0x4945, 0x4847, 0x5f54, 0x6877, 0x7469, 0x2065, 0x3931, 0x230a, 0x6564, 0x6966, 0x656e, 0x4220, 0x504d, 0x4957,
09:56:17LinusNcan you dcc me the bmp?
09:57:20FingerSoupIt's 22x19. Does similar results in 1 bit and 24 bit...
09:57:35amiconnFingerSoup: This is a bug in bmp2rb, obviously introduced with colour support
09:58:11 Quit _Vladoman (Read error: 110 (Connection timed out))
09:59:36FingerSoupOk, so how do I calculate a proper bitmap? ie: can i sit down with graph paper and figure this out easily?
10:02:00amiconnOf course bmp2rb needs to be fixed, but until then you could check out a version from before colour support was added
10:02:56amiconnUnfortunately I won't have much time today...
10:03:10LinusNFingerSoup: i have fixed it
10:03:22LinusNjust committed it
10:03:50FingerSoupcool. thanks!
10:04:15LinusNhmm, did i fix it correctly, though...?
10:04:49LinusNno i didn't :-)
10:05:57amiconnI thought I added almost everything needed for colour support back when I added support for multiple input formats, but obviously something went wrong even in this small extension step...
10:05:59 Quit DJDD_ ("Trillian (")
10:06:17LinusNyes, i changed the type of the bitmap from char to short
10:06:32LinusNbut messed up in the malloc() and memset() parts
10:08:53 Join DJDD_ [0] (
10:10:43Jungti1234I that convert bdf to fnt myself succeed.
10:11:01*ashridah hands Jungti1234 the cookie of success
10:11:39ashridahwow, that seems to have translated well
10:11:43amiconnLinusN: Ugh, so the malloc was too small?
10:11:54LinusNyes (!)
10:11:57 Join webguest79 [0] (
10:12:03amiconnI wonder why it didn't segfault...
10:12:07LinusNme too
10:12:19Jungti1234I unused translation. :)
10:13:54amiconnI see that the 16bit output of bmp2rb uses the same simple but broken shift-right conversion...
10:14:58Jungti1234But, I don't know whether I must install Rockbox Development Kit how.
10:16:23LinusNamiconn: ?
10:17:13 Join B4gder [0] (
10:17:45amiconnLinusN: For 16bit output, the conversion of 8bit channel brightness to 5/6 bit is done by a simple right shift
10:17:51amiconnThis is cheap, but imprecise
10:18:18LinusNwhat do you suggest?
10:18:29amiconnTook me quite a while to figure out how it has to be done correctly.
10:18:44 Join linuxstb [0] (
10:19:21amiconnThe problem is that e.g a right shift by 3 maps 8 source values to one destination value (0..7 becomes 0, 8..15 becomes 1 etc),
10:19:31amiconnand this is wrong with graphics,
10:19:47amiconnbecause it should round to the _nearest_ value
10:20:06amiconnand that makes the ends of the whole interval half the width of the rest
10:20:33amiconnExample, disregarding scaling for the moment:
10:20:54*LinusN prepares his brain for a learning experience
10:21:56amiconnIf we map source intervals of 8 to one destination value, the first and the last source interval has to be cut in half
10:22:20amiconnSo 0..3 would map to 0, 4..11 would map to 1, ...
10:22:46amiconnThe inverse has to be done (and is done in bmp2rb) when expanding
10:23:07amiconn...otherwise white would never be white again
10:23:30 Join Zagor [0] (
10:23:35LinusNamiconn: i see
10:23:52amiconnThe right shift essentially does dv = sv / 8 which is equal to dv = (sv * 32) / 256
10:24:47amiconnWe need to do dv = ((sv + 127) * 31) / 255
10:25:37amiconnErm, wrong
10:25:49amiconndv = (sv * 31 + 127) / 255
10:27:09LinusNi'll fix that later today
10:27:40 Nick Lynx_awy is now known as Lynx_ (
10:29:37 Quit JdGordon (Read error: 104 (Connection reset by peer))
10:29:51 Join JdGordon [0] (
10:31:10amiconnOne more commit (error-free) and we'll have a stain-less build table :)
10:32:25JdGordonhey LinusN, linuxstb
10:32:57JdGordonis there a reason the quick menu can only be accessed when music is stopped? it worked fine earlier int he week didnt it?
10:33:25Jungti1234hey Membrillo
10:33:47linuxstbLinusN: Am I right in thinking the pcf50606 RTC doesn't have "spare" RAM for settings info?
10:35:40ghode|afkis there a website with more info on video playback for the h100?
10:39:29Membrillohey Jungti1234
10:40:03JdGordonis it possible to view the currently playing playlist?
10:40:24Membrilloi think you press a-b
10:41:05JdGordonhmm.. no its hold navi - playlist - vieww current.. but it locks up my player?
10:41:44JdGordonhmm... no, ye its a-b
10:42:01JdGordonworks fine in the sim.. but kills my h300
10:42:05 Quit DMJC (Connection timed out)
10:43:03 Join DMJC [0] (
10:43:58linuxstbJdGordon: Any progress on a Sudoku generator, or have you given up on that idea?
10:44:06JdGordonhmm... ok i tinhk i need to low level format the hdd and do a disk check
10:44:09JdGordongiven up :D
10:44:24JdGordonactually, no, lost interest...
10:46:52linuxstbJdGordon: I guessed that by the way you are searching Rockbox looking for something interesting to do :)
10:47:07JdGordonhmm.. ye
10:48:05JdGordonseen as the iriviers have a decent amount of ram.. is it possible to allow them to store the current playlist indicies twice? once for the sorted order and once for shuffled? that way we could jump between the 2 redicoulously quickyl...
10:48:45Paul_The_NerdI *do* keep accidetally generating shuffled playlists.
10:49:00Paul_The_NerdI'll add 4 or 5 folders, then check it before I save and it's all shuffled up because I forgot to turn it off.
10:49:25Paul_The_NerdOr at least the first bunch of songs are.
10:50:00JdGordonwhat id really like to do is put back the shuffling ability like the iriv fw.. but turning on/off shuffle stalled it..
10:50:05preglowlostlogic: you can't use iram for storing pc values, it's very tight there
10:50:35Paul_The_NerdWhat do you mean "like the iriver firmware?:
10:51:13JdGordonlong is 16bits ye?
10:53:01FingerSoupJdGordon - Just make an "All" Playlist, then use the "Shuffle Playlist" option....
10:53:19amiconnlong is 32bits
10:53:29amiconn(or more on 64bit CPUs)
10:53:55JdGordonno, i like quing up a directory or 3 and then have it on shuffle, but be able to turn shuffle of if i want to
10:54:31FingerSoupOops. How'd I screw that up....
10:54:37JdGordonamiconn: i told u guys yersterday my brain was asleep... i have nfi y i thought long was 16 :p what i meant to type was 64
10:55:12JdGordon... 64 = int*2...
10:55:14FingerSoupOh, Wait, it's 2:00 AM here....
10:56:07FingerSoupInt is 16 bits on 16 bit processors, 32 bits on 32 bit processors, and 64 on 64 bit processors, AFAIK...
10:56:41FingerSoupOr am I mixing that up with long? Aren't they synonyms?
10:57:22JdGordoni tinhk this is why we r tought in uni not to assume anything :D
10:58:31FingerSoupthere's a table for 32 bit processors in there.... 64 bit makes life more "Interesting"
11:00:03FingerSoupWhen in doubt, use sizeof()
11:00:26linuxstbThere's no reason not to always use sizeof()
11:00:53Paul_The_NerdClean, green build table. :)
11:01:07JdGordonso.. about the 2nd array idea? is it possibly doable? or stupid waste f ram?
11:01:15 Join TCK- [0] (
11:03:36FingerSoupIt's not supportable with the current architecture, I thought.... Rockbox is supposed to work like rockbox. Changing the shuffle might take a lot more work, and may break other platforms with low memory (Archos stuff probably)....
11:03:37preglowFingerSoup: not alwayas true, amd64 has a 32 bit int
11:04:24FingerSoupIs that built on 32 or 64 bit OS? Does it change, or is it always 32?
11:04:44FingerSoupie: AMD can run in 32, 64 or hybrid mode
11:05:05FingerSoupat least that is my understanding from the initial tech docs I saw...
11:05:47 Join mirak [0] (
11:06:25 Join lamed [0] (
11:06:57***Saving seen data "./dancer.seen"
11:08:27preglowFingerSoup: for both 32 bit and 64 bit oses
11:08:40preglowFingerSoup: it depends on the compiler, of course, but afaik, all current compilers enforce it
11:08:55preglowFingerSoup: the 'long' is 64 bit, though
11:09:45FingerSoupahh, that's where I get mixed up. It's the long that changes, not the int (unless you go lower than 32 bit)
11:09:52merbananFingerSoup: what is hybrid mode ?
11:10:15preglowFingerSoup: the int usually changes as well, but not for amd64
11:10:24preglowFingerSoup: guess they were afraid of lots of code breaking
11:10:46FingerSoupmerbanan - It's a predominantly 32 bit mode with access to some 64 bit registers...
11:11:10preglownah, i think it's the other way around, 64 bit mode with possibility to run 32 bit code, like current v86 mode
11:11:25FingerSoupok, could be...
11:11:36preglowit makes no sense for 32 bit code to use 64 bit registers
11:11:55FingerSouptrue enough...
11:12:01preglownot allowing it to happen, at least
11:12:36Membrillofinished my IpodVOL WPS. how do i submit it
11:13:13Membrilloi think its a very nice wps if i may say so IMHO
11:14:00 Quit TCK (Read error: 110 (Connection timed out))
11:15:45lamedlinusn, around?
11:17:29 Join wett [0] (
11:17:52JdGordonhow do u load a bmp from a file?
11:18:15lamedto a wps or just to view it?
11:18:25JdGordonin a plugin
11:18:47perplexityMembrillo: I wonder if anyone would ever say "I just spent 10 hours creating a WPS, but it's a load of tot.. never the less, where do I submit it" ?? ;)
11:20:25Membrilloperplexity: im very... very... confused
11:20:28t0masI can commit one...
11:20:43t0masbut I'll have to check the rules first... as I don't know what exactly are the rules for new wps-es
11:20:49perplexityIt was a friendly poke at this −−-> <Membrillo> i think its a very nice wps if i may say so IMHO. Please don't take it seriously
11:21:43Membrilloi found this
11:22:27lamedhow do i mount the player in cygwin?
11:22:55JdGordonlinuxstb: where is the b+w bitmap loader function? i cant find it
11:23:07*amiconn wonders whether the wps packager can be made to handle wps'es with the same name for different targets
11:23:19amiconnlamed: You don't need to explicitly mount it
11:23:38*JdGordon thinks the directory structure needs work :p
11:23:42LinusNamiconn: that would be nice indeed
11:23:43amiconncygwin provides all windows drives as /cygdrive/<drivelettr>/
11:23:44perplexityamiconn: Could the wps perhaps be extended with some form of metadata tag listings it's applicable targets?
11:23:50JdGordon.. but understands its too mcuh effort and not really needed
11:24:14Membrillowho wants to check out my ipoddy based WPS?
11:24:34LinusNJdGordon: yes, the directory structure was good back in the old archos days, but not anymore
11:24:46LinusNMembrillo: rar?
11:24:46Paul_The_Nerdamiconn: I think the WPS packager could do that if it read the "Width" as being equal to, rather than being the minimum required width.
11:25:15t0masMembrillo: do you have a screenshot?
11:25:15Membrillooh i can do it as zip if you want
11:25:16LinusNPaul_The_Nerd: nah, it should pick the highest
11:25:26LinusNhighest that fits
11:25:30Paul_The_NerdAh, the highest that fits.
11:25:46Paul_The_NerdMy brain is on its way to shutting down. That should've been obvious.
11:25:47amiconnThere's another problem in the wps packager that I noticed.
11:25:49JdGordonthe devs are eurpoeans or yanks?
11:26:00Paul_The_NerdThey're people. :-P
11:26:05JdGordoni mean, should i spell colour with the u or not?
11:26:13t0masmost of us
11:26:15Paul_The_NerdI'm from Texas and I spell it with a U.
11:26:17JdGordonnot that it really makes a diff
11:26:17preglowmost are europeans, it seems
11:26:26preglowsome are even scandinavians!
11:26:27amiconnNot important atm because there are no player wps'es, but the script says that it can'zt go its magic because no lcd dimensions are given for player
11:26:43amiconn*it can't do...
11:27:56amiconnJdGordon: I use a mixed approach. For actual code I use american spelling bacuase it's common to do so, but for comments I try to aviod it
11:28:16*amiconn can't type
11:28:26*t0mas is off to install linux :)
11:28:44markungood luck t0mas
11:28:48t0masmy sister infected her laptop with a load of spyware again
11:29:03LinusNi uz aol spk as oft as poz, less typing
11:29:22JdGordonamiconn: no1 reads comments anyway.. so ppl wont care if u cant speel
11:29:23t0masLinusN: lol
11:29:57t0masLinusN: wh4t ab0ut l33t spe4k?
11:30:00MembrilloY0 DuD35. 133+ 5p33|< i5 teh b0mb
11:30:10JdGordonthe bmp2rb can read colour bmps ye?
11:30:13Membrilloi created a leet generator haha
11:30:19LinusNJdGordon: yes
11:30:25 Quit markun ("Lost terminal")
11:30:28t0masMembrillo: lol
11:30:53t0masand then we descend even further.... we start using msn emoticons... (A)
11:31:17 Join markun [0] (
11:31:20Membrillooh with my WPS, it seems you need to keep the folder and file name of ipodVOL
11:31:26 Join Polo_o [0] (
11:31:43markunwow, screen just core dumped..
11:31:43Membrillowhy is that?
11:31:55t0masyou mean including the caps?
11:32:25Membrilloi renamed the file and folder to hPod and RB couldnt find it
11:33:04t0mashm... I don't know
11:33:06LinusNdid you change the theme .cfg file?
11:33:22linuxstbLinusN: Am I right in thinking the pcf50606 RTC doesn't have any "spare" RAM for settings info?
11:33:23Membrilloi didnt know you had to
11:33:26t0masbut I'm leaving now... said I would be at my sisters place at 11:30 :)
11:33:32LinusNMembrillo: you don't
11:33:39Membrillowell i didnt
11:33:48LinusNdid you browse for WPS?
11:33:53 Quit t0mas (" see ya")
11:33:57LinusNor browse themes?
11:33:59 Quit DMJC (Read error: 104 (Connection reset by peer))
11:34:04LinusNlinuxstb: correct
11:34:33linuxstbLinusN: Have you done any work on that? I'm assuming we're going to need to split the HAVE_RTC define somehow.
11:34:47LinusNyes, somehow
11:34:48linuxstb(the ipod has a similar chip)
11:35:01Membrillobrowse themes i think
11:35:04 Join Zak1392 [0] (
11:35:12JdGordon"Reads a monochrome BMP file and puts the data in rockbox format in *bitmap" <- that format is the same as the format used to hardcode bitmaps to be drawn right?
11:35:15Zak1392hello guys
11:35:16amiconnlinuxstb: A separate HAVE_RTC_RAM perhaps?
11:35:19Bgrlinuxstb: isn't it even nearly the same (pcf50605) ?
11:35:20LinusNthere is a slight problem with writing to the pcf50606 from a background thread
11:35:26Membrillohow do i get to browse WPS
11:35:33Zak1392any progress linus?
11:35:40LinusNZak1392: not really
11:35:41XavierGrlinuxstb, preglow: here is a nice site that may interests you.
11:35:51Membrilloand for some reason the battery image isnt coming up
11:35:51amiconnJdGordon: The 'monochrome' needs to be removed from the description
11:35:52XavierGroh and good morning all!
11:36:07LinusNJdGordon: yes
11:36:33LinusNMembrillo: i guess the file name for the battery image is wrong?
11:36:43LinusNMembrillo: run it in the simulator
11:36:51JdGordonso i have to ask... is there a reason the code from bmp2rb.c wasnt copied into bmp.c? am i missing something?
11:37:06amiconnOf course there is...
11:37:15LinusNmalloc() for one :-)
11:37:17Membrillowhere can i get the sim from?
11:37:22amiconnThe code in bmp2rb uses mallo and hence loads of RAM
11:37:31LinusNMembrillo: you build it
11:37:36Membrillois there an already compiled version i can grab somewhere (i know the legalities are bad)
11:37:54Membrilloim scared of building
11:38:00lamed<amiconn> lamed: You don't need to explicitly mount it | well then how do i access it from cygwin?
11:38:19LinusNMembrillo: dangerous stuff ;-)
11:38:25 Quit webguest79 ("CGI:IRC (Ping timeout)")
11:38:27 Quit mordov (Read error: 104 (Connection reset by peer))
11:38:36amiconnlamed: [11:23:44] <amiconn> cygwin provides all windows drives as /cygdrive/<drivelettr>/
11:38:44lamedlinusn: have you read yesturday's log?
11:39:05LinusNlamed: which part?
11:39:21Membrilloone of the battery images doesnt work
11:39:26Membrilloand im not sure why
11:39:30BgrXavierGr: morning
11:39:31Membrilloits exactly the same as the rest
11:39:45BgrXavierGr: Amanda Coneyworth is the girl ... $30 :)
11:40:10Bgrthe biggest donor from the list
11:40:18XavierGrah yes :)
11:40:23 Join mordov [0] (
11:40:31XavierGrnow I am downloading the video LOL
11:41:09Membrilloso where do i download the source of the sim
11:41:37Membrillodo i get all these things
11:42:05BgrMembrillo: via cvs
11:42:15 Join TCK [0] (
11:42:49Membrillowheres that
11:43:00lamedlinusn: that part i've said something about rockbox still compiling the firmware file outside of ./rockbox folder, (=in the root), but bootloader v6 checks the firmware directory for the file before of checking the root. that might be a problem if someone forgot a firmware file in /.rockbox and he's trying to update. the makefile should be changed to place the file inside the folder imho, it seems nicer.
11:43:02Bgrhm, do u have dev env ?
11:43:10Membrilloim getting that now
11:43:20LinusNlamed: didn't you read my answer in the log?
11:43:27lamedah, no.
11:43:33lamedi'll go c
11:43:54FingerSoupif you get the rockbox-devel tree as opposed to the rockbox tree, it comes with all the stuff - UISim, etc...
11:44:10LinusNlamed: you aren't supposed to move rockbox.iriver there yourself
11:44:40Jungti1234Membrillo: Is it completion?
11:45:09MembrilloJungti1234: i thought so but i just found a problem
11:46:03Membrilloim so confused
11:46:18lamed| when we finally change the distribution archives| - isn't that just patching the makefiles?
11:46:58lamedamiconn: tt
11:47:15*amiconn wonders what's still wrong with the windows installer builds
11:52:11 Quit linuxstb (Read error: 110 (Connection timed out))
11:52:32 Join linuxstb [0] (
11:53:19markunGigabeat is spinning up for every song, strage if you have 32MB of ram.
11:53:46 Quit Membrillo ()
11:54:25 Quit TCK- (Read error: 110 (Connection timed out))
11:55:11preglowexcellent programming
11:55:48XavierGrwasn't it the iriver fw that buffered only 3 songs regardless buffer?
11:56:49XavierGrI can't still understand how rockbox has inferior battery time on mp3 with that. (add to that the low battery level running)
11:58:14lamedthat smash my ipod movie is so bad.
11:58:33XavierGrbad in quality?
11:58:41XavierGri haven't seen it yet
11:58:49XavierGri am downloading it
11:59:31BgrXavierGr: just iriver runs CPU at less frequency for mp3 decoding than rb
11:59:36lamedboading like hell...
12:00:16lamedjust a nerd guy that smashed a brand new ipod. he did most of the mayham not in the store but at the street.
12:00:39XavierGrah I thought that he did it in the store, pitty
12:01:02lamedxaviergr: he did, only a part. he stomped on it for couple of times.
12:01:15*Cassandra gets 20 hours or so out of her H120.
12:01:19CassandraThis is enough.
12:01:21XavierGrbgr: Even if that's the case hard disk buffering takes more battery than cpu (if not so great in difference)
12:01:36lamedcassandra: how old is the battery? (the player?)
12:01:37XavierGr20hours with a normal battery?
12:02:00Cassandra(OK - I cheated - it's a high capacity iPod v1 battery with the terminals switched.)
12:02:07XavierGrso there
12:02:10BgrXavierGr: ok, but hard disk spins not much of the time while playing, and the cpu is working non-stop
12:02:19lamedxaviergr: could you specify battery size for me? cassandra - 2200 i assume?
12:02:31CassandraI think so, yes.
12:02:40lamedyou _think_....
12:02:40XavierGrI got 25 hours with a 1900mah battery at 50 volume and 128 bitrate mp3s
12:02:48XavierGrno EQ
12:02:51lamedwith iriver's fw?
12:03:06XavierGrfor any benchmarls of the battery please consider using my tsr plugin found on the patch tracker.
12:03:21CassandraIt's a while since I bought it, and no, with Rockbox.
12:03:22XavierGrIt does a very good job on logging the battery on a real stress test.
12:03:37lamedxaviergr: could you specify battery size for me?
12:03:40XavierGrlamed: 25 hours under rockbox.
12:03:49XavierGrfor you?
12:03:56XavierGryou mean what to advice you to buy?
12:04:05Jungti1234Membrillo: It's problem.
12:04:31Jungti1234My H300 became fool.
12:04:51lamedno, i mean i need to know the battery size, because there's some local store that can manufacture/order batteries, they might grab an 2800mah for me...
12:06:19lamedi've read a reply of yours over misticriver's site, so i've figured you must know.
12:07:04JdGordonhow do we do dynamic allocation in rb?
12:08:32CassandraI think that's the one I have.
12:08:55CassandraJdGordon, nope.
12:09:07CassandraThat is, we don't.
12:09:18JdGordonAH.. hmm..
12:09:28lamedno good, i have to know the dimentions.
12:09:58lamedxaviergr: i've ment the dimentions.
12:10:16 Quit TCK (Read error: 110 (Connection timed out))
12:10:54Jungti1234H300's wps is unstable.
12:12:20XavierGrlamed you need the physical dimensions of the battery?
12:12:48Jungti1234I made it.
12:13:21lamedxaviergr: exactly.
12:13:24JdGordontoo bad about the song... looks good
12:13:30JdGordonnice n simple
12:13:41XavierGroh come on Jungti1234 at least you could choose a bigger font for that. It is so empty
12:13:42lamedmy english sucks when i'm not around my translator...
12:14:04Jungti1234bigger font?
12:14:35Jungti1234I like it.
12:15:21XavierGrhere is my battery's specs
12:15:43Jungti1234I know there.
12:16:40XavierGrlamed: you were right the smashmyipod video wasn't so good. I am gonna watch part2 now. :)
12:17:00 Join Wett_ [0] (
12:17:05LinusNlamed: still there?
12:18:30LinusNyes, it's only a matter of changing the zip-creation script
12:18:30preglowJdGordon: you don't
12:18:45JdGordon... :'(
12:19:02*JdGordon is trying to be helpfull... but hurting brain
12:19:23LinusNbut if we started distributing rockbox zips with rockbox.iriver in .rockbox, people withj older bootloaders will have problems
12:19:38 Quit wett (Read error: 110 (Connection timed out))
12:19:44LinusNthat's why we'll wait until more people have upgraded
12:20:00amiconnLinusN: What's wrong with the installer builds?
12:20:07preglowJdGordon: there is a simple malloc for codecs, but its use is frowned upon, you should allocate statically all you need
12:20:11lamedah, you are so right about that.
12:20:13LinusNi have no idea, B4gder was supposed to check the logs
12:20:28JdGordonok, i think ive figured it out
12:20:28B4gderthey didn't give me much clues
12:20:32CassandraIs the installer going to get checked into CVS any time soon?
12:20:43lamedxaviergr: let me know if it's any better. thank for the link.
12:20:46B4gderWine exited with return code 1
12:20:46B4gder./installer/src/update-dailybuild: build-installer process failed: 31
12:20:49LinusNCassandra: that's a good idea
12:21:21Cassandra*nods* I think so.
12:21:37amiconnLinusN: Different matter - while I agree that exceeding the potential clipping limit for volume, I'm strongly against yet another option
12:21:44B4gderactually, most build system stuff should get checked in
12:21:50amiconnI know that you also want volume to be in dB
12:22:12LinusNyes i do
12:22:21XavierGrlamed: no, part 2 actually sucks big time. Don't watch it. I expected something more funny. pitty. I will probably head to HAHA
12:22:31amiconnI have an idea how to account for both, but I'm not sure about the best way
12:22:38LinusNamiconn: shoot
12:23:28amiconnWe could just define the maximum volume guaranteed to not clip as 0 dB
12:23:50amiconnThen negative values never clip, positive values clip potentially
12:24:21LinusNexactly how i want it
12:24:23amiconnThe problem with that is that the 0 dB point would move with varying bass/treble
12:24:44Cassandrawhich would be very confusing, I feel.
12:25:08amiconnWe could simply define 0dB as the "minimal maximum" i.e. the one which doesn't clip at maximum bass+treble
12:25:51amiconnIn fact the MAS3587/MAS3539 do it almost the same
12:25:52LinusNlike the 93% on archos recorder
12:26:16amiconnOur 92% equal 0dB, 100% is +12 dB
12:27:00amiconnSo we could define today's 83% as 0 dB on iriver
12:27:14amiconnIt would go up to +24 dB then...
12:28:42LinusNsounds fair
12:29:01*perplexity just spotted the Chip-8 emulator... wooooooooo!!
12:29:27perplexityI remember programming on an original homebrew machine with the chip-8 rom years ago :) Now that takes me back
12:29:34LinusNare there any no-boot-black-lcd h300 owners here?
12:29:36amiconnBtw, I think using the full 24dB range for bass boost is silly
12:29:58*Cassandra uses a value of 16dB.
12:30:14CassandraMy ER-4ps need the bass driven hard.
12:30:24LinusNwe've had this discussion before (not using the full potential of the dac) with bluechip, remember?
12:30:38 Quit Bagder (Read error: 113 (No route to host))
12:30:51LinusNthat's why he made that mas3687 plugin
12:32:34preglowi think defining 0db that way would be kind of silly
12:32:54preglowi've never seen a system that defined 0db as anything other than full scale output
12:33:02preglowand we shouldn't break with too many conventions
12:33:37B4gderwell, I don't like the idea of db as volume scale at all
12:33:59CassandraIt does tend to be rather confusing.
12:34:14preglowi don't really see the big point in making the volume control use decibels as a unit
12:34:18preglowit wont be accurate anyway
12:34:35preglownot unless we map the scale we use to the one the uda chip uses
12:35:06amiconnpreglow: That was the idea
12:35:17amiconn(mapping to the actual DAC scale)
12:35:21LinusNthe point of using dB would be that it's obvious that >0dB risks clipping
12:35:52LinusNi once had an idea of using >100% for the same reason, but i liked db better
12:35:55preglowLinusN: why? you assume that people know about how digital audio works. there are analogue products out there that can be driven to 6db and higher without clipping
12:36:15B4gderI don't think very many people will see the connection of 0db vs clipping
12:36:17preglowLinusN: i agree if you map 0db -> 100%, but not if you map 0db -> 83% or whatever
12:37:06lamedxaviergr: there was some debate on an oldish misticriver post, that the battery thick is 3.2 mm, not 3.8. it seems like a mistake, but are you sure with the data you gave me? - another question. is it totally squarish - i mean, are the coreners round off? is the battery tapered at the tips?
12:37:23preglowa bit tapered here and there, yes
12:37:44preglowwhy don't you open it up and see?
12:37:52XavierGrlamed I cant be sure about these details, sorry
12:39:02amiconnI' also in favour of a dB scale for volume
12:39:18amiconnThe volume scale is logarithmic, so % don't make much sense
12:39:32CassandraMy old one is square
12:39:44LinusNthe only problem is that Joe User isn't familiar with dB
12:39:49FebsThere would be much confusion about using a db scale. You've got people who just can't understand why 20 on the iriver doesn't equal 50 on Rockbox. How will they understand a volume setting with negative numbers.
12:40:04*Paul_The_Nerd votes for dB
12:40:04XavierGrIMHO volume on rockbox is already confusing
12:40:05LinusNon the other hand, who cares about the scale?
12:40:15preglowLinusN: does it matter? i thought we did this for internal reasons anyway
12:40:19XavierGr50 isn't half the volume...
12:40:21 Join Wett [0] (
12:40:31CassandraAnd approx 2mm thick.
12:40:31LinusNyou adjust the volume until it is right for your ears
12:40:36amiconn..and that > 0 dB mean potential clipping isn't my own idea... it's just how the newer MAS' do it
12:40:43Paul_The_NerdIf they're already confused, think that 1% is a set increment, then does it may even force them to realize it's not the expected scale.
12:40:54CassandraI believe that the replacement is thicker though.
12:40:57Paul_The_NerdWow, that sentence doesn't even read like I'm a native English speaker...
12:41:09XavierGrhaha Febs you got my ignorance
12:41:16B4gderPaul_The_Nerd: no, you can't be! ;-)
12:41:19 Quit Zak1392 ()
12:41:29Paul_The_NerdIn my defence it is 5:40 in the morning.
12:41:37preglowanyway, the display shouldn't be decibel anyway
12:41:45preglowanyway anyway anyway
12:41:47CassandraArbitrary units for volume are fine, since it's entirely subjective what the 'correct' volume is anyway.
12:42:26amiconnThere is quite a number of consumer devices with a dB scale... even using negative values for less than full volume
12:42:46XavierGryes I have seen that a lot.
12:42:47preglowwell, anything else would be plain wrong, so they should
12:42:50Paul_The_NerdMy home stereo does.
12:43:06CassandraMost peak meters do.
12:43:24CassandraBut then my rather nice stereo uses 0-100 just like Rockbox.
12:43:30ashridahand lets not forget every self-respecting equalizer uses +ve/-ve scales
12:43:35amiconnMy home stereo happens to use a scale of 0..100 though
12:43:54B4gderI would say that _most_ devices use 1-100, 0-100 or 1-10
12:43:59Paul_The_NerdMy cheap stereo uses a scale from 1 to 30 though.
12:44:15amiconn...but I also know a fairly old panasonic compact stereo that uses dB
12:44:28Paul_The_NerdI think you've got two decisions to make then
12:44:51preglowstill: why are we switching to db?
12:44:55Paul_The_NerdOne: Do you have some means of indicating where the "possible clipping may occur" point is (>100% or >0dB)
12:44:59ashridahgive people yet-another-option! yay!
12:45:01CassandraI just don't see the point in all these clipping options.
12:45:03Paul_The_NerdThen, if the answer to that is definitely YES, then what scale?
12:45:08preglowCassandra: why not?
12:45:30CassandraWell, the default has always seemed to work fine for me.
12:45:42preglowCassandra: well, yeah, but other people use rockbox as well :)
12:45:43 Quit merbanan (Read error: 110 (Connection timed out))
12:45:48Cassandra(And I use obscenely high boost settings and very quiet headphones.)
12:45:51FebsThe present default is probably THE single biggest complaint about Rockbox.
12:46:01LinusNFebs: yup
12:46:03preglowi think it's a decent option
12:46:12Cassandrapreglow, no other *important* people though. ;)
12:46:46preglowCassandra: of course not, but we've got to toss the other masses an option or two from time to time to give the impression we care about them
12:47:03XavierGrso if the volume scaling change: I will be able to raise volume in such way that clipping will occur?
12:47:19preglowXavierGr: yes
12:47:33XavierGrwithout an option to disable such thing?
12:47:33preglowthat is, not with the volume scale change itself
12:47:47preglowwe're talking about an option that will let you choose as well
12:47:48XavierGr(with bass,right)
12:48:00CassandraI say if we must do it leave the volume as is and add another option.
12:48:01XavierGrah then I am with it too.
12:48:23 Join Moos [0] (
12:48:23CassandraMuch though I hate adding more options.
12:48:40XavierGrtrully though I don't care I never use eq. ;p
12:48:42preglowbut yeah, are we changing to decibel scale to ease detection of possibility of clipping, or not?
12:48:44LinusNi dislike optionitis as well
12:48:50preglowXavierGr: me neither
12:49:10XavierGrbut why dislike too many options?
12:49:11LinusNpreglow: my idea is using db in both the interface and internally
12:49:26XavierGrI thought that this is the good part of Rockbox
12:49:27*Cassandra ponders the idea of moving advanced configuration (or extreme tweakability) to a plugin, thus keeping the main binary small and simple.
12:49:32XavierGrUser configurable
12:49:53LinusNCassandra: the main application still has to understand the settings
12:49:59CassandraXavierGr, confusing, and takes up more memory.
12:50:19CassandraLinusN, true, but you don't need to have all the menus for configuring it in the binary.
12:50:22XavierGrI see as problem only the second
12:50:27preglowthe menus aren't the space eater
12:50:54XavierGrit is better to have more happy users instead of some confused
12:50:59CassandraXavierGr, well, I'm actually concerned with making Rockbox more accessible to mainstream users, so both are a concern for me.
12:51:04XavierGrthat's why there are manuals you know
12:51:12FebsPeople don't read them.
12:51:24XavierGrtheir problem IMHO
12:51:29Paul_The_NerdI think that there's already a pretty significant division in users.
12:51:30preglowwell, yeah, a bit
12:51:39preglowbut we also know they don't, so have the possibility to adjust
12:51:43Paul_The_NerdThose who understand why the volume seems different than iRiver FW, and those who don't.
12:51:54CassandraXavierGr, yes and no. If it puts them off using Rockbox, it's not a good thing.
12:52:24preglowbut i also like having options, it's what sets rockbox apart from the other oses
12:52:25Paul_The_NerdAnd I think the former group will probably nod and think something like "those kooky developers" then shrug and go back to normal life with a change in the volume scale.
12:52:26XavierGrAll well known programs (that are famous and very Joe user friendly) are full of options.
12:52:31CassandraSince we rely entirely on word of mouth to 'market' Rockbox.
12:53:21CassandraRockbox is currently the Microsoft Word of MP3 players.
12:53:28XavierGrWell these options make rockbox to shine over all the other firmwares. This sweet taste of configurability ;p
12:53:39CassandraSure, it's neat to have all those options, but your average user doesn't need 90% of them.
12:53:42preglowhowever, we might want to hide some advanced options somehow
12:53:46preglowwithout cluttering things too much
12:54:03XavierGrwel that's an option too
12:54:36perplexityput all the *advanced* type configuration options into a plugin
12:54:37preglowi've got one leg in each camp, really, i think rockbox is a better os with more configurability, but i also see that lots of people never want to deal with them
12:54:43XavierGrbut setting an option limit to rockbox is not good. Rockbox should be limitless.
12:54:51preglowperplexity: why into a plugin? why not just a separate menu?
12:55:03 Quit Wett_ (Read error: 110 (Connection timed out))
12:55:07preglowthe menu code isn't what takes space, the code to handle the different options is what takes space
12:55:11perplexityJust thinking of binary size mainly
12:55:24perplexityOh well, then a separate menu
12:55:25preglowand that code needs to be in the main binary, mostly
12:55:50XavierGrThough still if you make a poll you should see that most of the users like all sorts of options.
12:56:05 Quit linuxstb (Remote closed the connection)
12:56:07preglowmost of our current users, yes
12:56:18preglowwe also want the ones we aren't getting because of our current profile
12:56:29perplexityand _that_ is the issue preglow
12:56:31XavierGrare there any?
12:56:33 Join linuxstb [0] (n=linuxstb@
12:56:46XavierGrWHO will not use Rockbox? I mean come on!
12:56:52FebsI've mentioned this before, but one thing that would help with useability would be to move the Playback setting to the main menu.
12:56:53preglowfools and idiots!
12:57:08perplexityMy father has a h320 and I have the h340.. I showed him rockbox yesterday.. he liked the 8 second startup to sound, but the menus and options just put him off..
12:57:22XavierGrDo you now such a person? That owns a rockbox compatible device and refuses to operate it.
12:57:26XavierGrI know only 1
12:57:38preglowXavierGr: nah, but i don't know too many people with rockbox capable devices
12:57:48preglowXavierGr: primarily myself and... that's that
12:58:06FebsSure, there are a number of people on Mistic River who have chosen to stick with the original FW. They're in the minority, but they certainly exist.
12:58:08XavierGrwell I meant from forums irc and stuff.
12:58:09preglowthat is, i know people with ipods, but those don't count yet
12:58:46XavierGrFebs: what percentage we are talking about. I know only Haku that didn;t want to use rockbox because of WMA
12:58:49linuxstbamiconn: Yes, a separate HAVE_RTC_RAM was what I was thinking of. Maybe we'll need a CONFIG_RTC as well, but I'll see when I try and implement it.
12:58:51preglowFebs: do you know what the most usual complaint is?
12:58:58XavierGr(the first was a question)
12:59:14MoosFebs: it's because they needed option don't working yet or just working (remote, rec...)
12:59:36FebsProbably the most important factor to them is WMA support, but I think that there are some who just don't want to be bothered.
12:59:48CassandraI suspect a lot of iPod users won't want to re-encode their iTunes libraries.
12:59:50 Join TCK [0] (
13:00:01XavierGrI should set a new poll at MR.
13:00:05FebsThe recent inclusion of themes and WPS in the rockbox build is a BIG step towards getting some of the latter people over to RB.
13:00:28preglowCassandra: the ipod is a pretty different matter altogether, we are in no way targeting apple firmware users the way we're currently targeting iriver firmware users
13:00:37XavierGrIt seems crazy to me that more than 5% of Husers (that know rockbox exists) would choose iriver fw
13:00:43preglowCassandra: we know from the start we probably wont be able to convert the apple users
13:00:55FebsXavierGR: I don't think that MR is representative of the general population.
13:00:59Paul_The_NerdI still suspect you *could* implement an on-target iTunes decryptor.
13:01:12linuxstbXavierGr: Those 5% are probably the people with large WMA collections.
13:01:22MoosXavierGr: when rockbox will be "full" why don't use it ;)
13:01:24XavierGrFebs we are talking about users that know that rockbox exists.
13:01:27preglowPaul_The_Nerd: yes, but i believe there are certain legal implications in certain countries...
13:01:35Cassandrathe Apple format uses DRM. Philosophically we should not be supporting DRM.
13:01:40XavierGrHow could an Huser that never gets to internet to hear about rockbox?
13:01:45preglowand i wouldn't want to either, no
13:01:54preglowi'd just shun any way of supporting it
13:01:59CassandraXavierGr, from their friend that uses Rockbox.
13:02:09preglowand rather at least do a good _try_ at educating people
13:02:16FebsThere is a huge different between a user who "never gets to the internet" and a user who doesn't frequent an "iriver enthusiast" forum.
13:02:27Paul_The_Nerdpreglow: I didn't say it should or anything. I'm just saying that if the right people get bored, reencoding libraries won't be a barrier. It wouldn't even have to strip the encryption.
13:02:43preglowPaul_The_Nerd: definitely possible, yes
13:03:15XavierGrCassandra: do you know many persons in your physical life to use rockbox. I know only 1. And that's the only user I know with an iriver.
13:03:22Paul_The_NerdNot me mind you. I've never had an AAC file, let alone an encrypted one. I'm a fan of CD-Roms.
13:03:52XavierGr(excuse "physical" I mean real)
13:04:03preglowwhat, this isn't my real life? ;)
13:04:15*preglow tries to wake up
13:04:17CassandraNope. But you asked me how they would hear about it.
13:04:19XavierGrehm you know off the internet
13:04:25CassandraThat's the only possible mechanism.
13:04:44MoosXavierGr: like preglow already said just few Rockbox devices owners
13:04:51Paul_The_NerdI suspect that once the iPod port is playing music on 4gs and/or Nanos, you could get another front page on Slashdot and generate a huge amount of interest in both the project, and the other ports (assuming a list of in-progress and current targets was part of it.) That might be a solid way to get 'word of mouth' out, as it's a high profile target.
13:04:52Moosbut soon lot news one
13:05:24CassandraPaul_The_Nerd, precisely why I was so happy to see Rockbox for iPod happening.
13:05:26XavierGrThe user must be at least a little techy to know about rockbox. Or keep up with news about his device.
13:05:26preglowPaul_The_Nerd: indeed, let's just hope we can get the port done in reasonable time
13:05:36B4gderand then we can snatch over some IPL users ;-)
13:05:41preglowi _think_ i'll have lots of rockbox time in the next two months
13:05:55XavierGrgreat :D
13:06:00CassandraOh, right.
13:06:10Moospreglow: good to hear :)
13:06:24CassandraI think we will. iPodlinux is a nice toy, but it's a bit of a silly approach, really.
13:06:44XavierGrsilly but fun I guess...
13:06:49Paul_The_NerdiPodLinux isn't really aimed as a music player anyway
13:06:49JdGordonwe r getting a new fw for the h300's
13:06:59***Saving seen data "./dancer.seen"
13:07:20XavierGrI really want to see a life of an iriver programmer
13:07:34Febsgapless gapless gapless gapless gapless
13:07:35XavierGrI want to know if they are aware of rockbox
13:07:55XavierGrand which firmware they use if they have a rockbox compatible target
13:08:07Paul_The_NerdXavierGr: You kidding. Watch ALAC, FLAC and WV support show up in *their* next firmware. :-P
13:08:08 Quit Febs ("CGI:IRC (EOF)")
13:08:10 Join Febs [0] (
13:08:27XavierGror see their faces when someone shows them rockbox
13:08:48XavierGrI would hide in a corner for the rest of my life if I were them.
13:10:03XavierGrFebs: How this "gapless gapless" thing started. I can't catch the joke. Is it about the fraud attemp of iriver to emulate gapless?
13:10:16Paul_The_NerdIn the future DAP makers should just port rockbox to their targets for you guys. It'll probably be cheaper than writing firmware from scratch for them.
13:10:36XavierGrThey will not do such thing Paul
13:10:44Paul_The_NerdYes, but they *should*
13:10:55XavierGrCompanies don't like to have a "best" player on the market.
13:11:08Paul_The_NerdTrue. They also want to be able to be compatible with music stores.
13:11:09XavierGrThen people would had no need to upgrade to another player.
13:11:25XavierGrThat's what I hate most.
13:11:29FebsThe gapless thing came from one of the many Mistic River threads hyping the allegedly forthcoming iriver FW update.
13:11:43XavierGrlink l)
13:11:54Paul_The_NerdIt'd also hurt them quite a bit if people realized that so many new features could be added to their players *without* them having to buy new hardware.
13:13:16lamedguys, rb isn't playing your files with better quality. it does not prolong your battery life for sure. -every player is sold with _it's own_ firmware. if you've installed rb on an iriver, i guess you know what a playlist is and you need that. why not buying an ipod? - because the h hardware was better. - it has line in/out and stuff. why not stick with the old firmware? - well, if it hadn't came to your interest, you probebly shouldn't! - I vote
13:14:13JdGordonhow do u accept the filename in the playlist save screen? (the filename chooser)
13:14:15Paul_The_NerdWhat's that lamed?
13:14:28lamedwhat's what
13:14:37Paul_The_NerdWas that statement your opinion?
13:14:43Paul_The_NerdOr copied/pasted from somewhere?
13:15:05JdGordonpress play.. DUH!
13:15:10 Join Benacool [0] (
13:15:40lamedouch. no that was my opinion. I'm sorry i'm not a native english speaker as you can see.
13:16:55XavierGrwell it sure sounds strange for a rockbox fun lamed.
13:17:07Paul_The_NerdAh. Well, in an effort to explain why I'd switch to the new firmware: I like lossless music formats. I like the option to have the information I want displayed on the screen while playing music. I like to play Sudoku and Bejeweled on my player. Yes, it doesn't guarantee better battery life. But it does offer a lot of other stuff.
13:17:20Paul_The_NerdBy "the new firmware" I mean Rockbox.
13:18:03Paul_The_NerdAh well
13:18:38XavierGrI like rockbox as being the most configurable and versatile firmware on the world. I will never buy another DAP without rockbox compatibility.
13:18:42Jungti1234I want to translate plugin in Korean.
13:19:03XavierGrPlugins aren't translated iirc
13:19:51JdGordonwhat top time in the runnig time debug screen?
13:20:54XavierGrtop time that the battery survived.
13:22:44Moosbut those datas don't update fine, need to charging in same time I believe
13:23:14markunI also vote for db
13:23:28MoosXavierGr: your plugin is surly better for this ;)
13:23:30markundamn, should scroll down before typing :)
13:23:51XavierGragain I will pester. If you want accurate battery results use my battery_benchmatk tsr plugin. ;p
13:24:04XavierGrlol thanks Moos :d
13:24:39Paul_The_NerdWell, at the very least a patch could be put into the tracker to either add an option, or just change it, so those who would like it can use it, and those who are unsure can try it and form an opinion?
13:33:08JdGordondoes any1 else like the idea of adding a get_setting and save_setting sort of function for plugins to use, and then "force" them to load key mappings from their setting file?
13:33:36JdGordonby force, i mean nag the plugin dev, not a software restrcition..
13:34:47JdGordonany1 still here?
13:35:21Paul_The_NerdI've been distracted by videogames.
13:35:30Paul_The_NerdTo what purpose would you do that?
13:35:51JdGordonso key mappings arent hard coded, coz they are stupid :D
13:36:10JdGordonalso, to get plugin devs to make more user configurable settings
13:36:22ashridahdoes anyone maintain a working build of rockbox with unicode support? or just the patch
13:36:31markunashridah: I do both
13:36:45ashridahmarkun: do you happen to have linkage to info about the build?
13:37:09markunWell, I just keep it working here locally.
13:37:22markunIt's not on a webpage if that's what you're asking for
13:37:29ashridahah, i meant a release other people could use. bummer.
13:37:40ashridahjust someone asking about it in MR that's all.
13:38:12Mooslet's hope it will be commited soon :)
13:39:09markunthere is not much stopping unicode now.
13:39:17markunamiconn: you here?
13:39:27ashridahjust resolving build process issues, aren't you?
13:40:27markunB4gder is the makefile wizard, just waiting for his work for compressed rockbox images.
13:41:14markunamiconn wanted to first test unicode some more, but I prefer to commit it soon and then fix the errors that users report.
13:41:18 Join NicoFR [0] (
13:41:41Moossounds a good plan
13:42:15markunIt's not an official release, and it works pretty well (I and other people have been using the patch for a long time)
13:42:50ashridahyeah, it'd just be nice if no-one had to actually compile it to get it to work :)
13:43:03Mooshehe :)
13:44:11JdGordondoes rb have atoi()?
13:44:46markunJdGordon: I don't know, but you can use 'grep' on the source to find out.
13:45:16*JdGordon is a windows person... little help?
13:45:45markunuse the windows search then, it can also look for text in files
13:46:13B4gderyes, there is an atoi()
13:46:26JdGordon:D ta, just found it
13:46:50markunB4gder: the lcd of the gigabeat it not visible without backlight afterall
13:46:57B4gderah, ok
13:47:31markunBut then backlight can be dimmed, looked like it was off
13:49:18markunthe chip used for dimming is called "606A 0512", but I couldn't find a datasheet
13:51:55NicoFRhi everybody
13:52:13NicoFRI'm having trouble building the simulator...
13:52:23NicoFRnormal build works fine
13:55:09 Quit Wett (Read error: 110 (Connection timed out))
13:56:10B4gderwhat happens?
13:56:31NicoFRit stops immediatly
13:57:14NicoFRthe build stops immediatly
13:57:20NicoFRafter I enter make
13:57:39NicoFRi get "CC button.c"
13:58:05NicoFRthen "gcc : installation probelm, cannot exec 'cc1'..."
13:58:22B4gderthen you have a bad gcc installation
13:58:27 Quit Paul_The_Nerd ("Chatzilla [Firefox 1.5/undefined]")
13:58:39NicoFRbut the normal build ssems to work fine
13:59:12B4gderyes, but that uses another gcc
13:59:19linuxstbmarkun: What are your plans for running your own code on the gigabeat?
13:59:22B4gderthe sim uses a native gcc
14:00:57NicoFRok thanks I'll look for that
14:11:20markunlinuxstb: I'm not sure yet where to start
14:13:01B4gderI have a suggestion markun
14:13:24B4gder1. make sure you can upgrade to one of those packages like the one on my x30 page
14:13:51B4gder2. then create a modified romfs dump
14:14:07B4gderadding your own little hack somewhere, doing something silly just for a test
14:14:27B4gderthen upgrade to that
14:15:14MoosB4gder: did you already choice your next device (color one I mean)? seems markun is alone currently on the Gigabeat
14:15:52markunB4gder: I could change the script
14:16:32B4gderyou could in fact possibly consider adding rockbox to the gigabeat as a linux application! ;-)
14:17:07B4gderMoos: so many fun devices, so little time... :-)
14:17:19Mooshehe :)
14:17:47B4gderrockbox isn't even my main open source project I spend time on ;-)
14:17:55 Join webguest93 [0] (
14:18:27ashridahso rockbox is the 'other woman'?!
14:18:42B4gderthe other other woman
14:18:47JdGordonlinuxstb: can i suggest 2 functions for the plugin struct? (coded and being tested)
14:18:48ashridahah, yes, dayjob :)
14:20:07 Quit webguest93 (Client Quit)
14:20:14 Join djgalak [0] (
14:21:41ashridahB4gder: so which other OSS project do you work on?
14:22:10Moossee his web home page
14:22:12B4gdercurl is my main baby
14:22:26B4gderI'm even gonna work full-time on curl during december
14:22:27ashridahaah, that's right
14:23:06ashridahhm.better hit the sack
14:23:10 Quit ashridah ("sleep")
14:24:25Jungti1234I must sleep, too.
14:24:42MoosGood night
14:24:42Jungti1234bye all
14:24:46Jungti1234good night
14:25:02Moosafternoon here in Paris ;)
14:25:09Jungti1234Have a nice day.
14:25:30Moosthanks, have a good Rockboc dreams :)
14:25:50 Quit Jungti1234 ("bye")
14:27:19 Join Jungti1234 [0] (n=jungti12@
14:28:16JdGordoncrap... forgot to kill the sim and was wondering why my comp was running like a dog
14:28:50 Quit djgalak ("CGI:IRC (Ping timeout)")
14:30:36 Quit Jungti1234 (Client Quit)
14:32:13 Part Polo_o
14:37:39 Join [IQ]Rob [0] (
14:37:46 Nick [IQ]Rob is now known as Rob2222 (
14:38:09Rob2222hi all
14:39:25Rob2222Really nice work with the Iriver H300. :)
14:40:20Rob2222Do anyone know how far is the USB progress on the H300? Didnt found that on the site.
14:43:38JdGordonhow do i debug my damn code? it crashes and DEBUGF isnt doing squat!
14:44:52 Quit NicoFR ()
14:44:58perplexityHrm.. I'm having a similar problem JdGordon.. a build with DEBUGF enabled just crashes, while a build without runs except for the sound problem I'm trying to debug (h1x0 for X11 Sim here)
14:45:40JdGordonhow does the backlight print to the debug window? (when it says backlight: on) ?
14:47:08perplexityuisimulator/common/stubs.c: DEBUGF("backlight: %s\n", (value > 0) ? "on" : "off");
14:47:08perplexityuisimulator/common/stubs.c: DEBUGF("remote backlight: %s\n", (value > 0) ? "on" : "off");
14:47:13JdGordonhmm.. ok
14:47:24JdGordonso my code crashes before the DEBUGF line :'(
14:48:20markunIs it possible to for objdump to recognize data from code? I used to use a dos program calles sourcer (I think) that could do that.
14:53:29JdGordongrr.. never code while asleep :D
14:53:53 Join webguest38 [0] (
14:53:58JdGordonfound my bug :'(
14:54:23LinusNRob2222: USB is on my list of prioritized stuff
14:57:10JdGordonoh crap.. i think my code works :D
14:57:17JdGordonnot bad for 1am :D
14:57:36 Join [IQ]Rob_ [0] (
14:59:10perplexityhrm.. something in the X11 sim is causing 100% cpu usage on my box and causing pcm buffer starvation. When that occurs the x11/sound.c removes the pcm callback and sounds stops.. grrrr
14:59:42JdGordondoesnt the sim hold 100% usage sort of purposly?
15:00:30perplexityPerhaps, but there is something that is causing the pcm buffer to underrun.. when I modify sound.c to just wait until more data becomes available it plays 1 second, pauses for 2, and cycles like this..
15:01:24JdGordonLinusN: can i suggest 2 more functions for the plugin api?
15:01:27perplexityThe low watermark callback in apps/pcmbuf.c is getting hit very frequently while it's not playing sound.. so something is interfering with the decode thread I guess..
15:01:34 Nick [IQ]Rob_ is now known as Rob2222_ (
15:02:43JdGordonatoi doesnt do hex :'(
15:02:54 Join Bger [0] (n=Bager@
15:03:59perplexityI'm struggling to think where you would have an ascii string (in hex) in rockbox that you need to convert.. a wps or config file perhaps ?
15:04:10JdGordonconfig file
15:04:20JdGordonno big deal
15:04:57JdGordonwhats the proper way to make a patch?
15:05:14LinusNJdGordon: cvs diff -u > patchfile
15:05:52perplexityDoes that take into account new files added ?
15:06:10JdGordoncvs diff: No CVSROOT specified! Please use the `-d' option
15:06:33perplexityI usually cp -al my CVS tree, do my mods and then do a diff -urN tree.old tree > patch
15:07:02***Saving seen data "./dancer.seen"
15:07:15LinusNJdGordon: did you use a .tar file?
15:07:29JdGordonno, i extracted it manually
15:10:05 Join Bagder [0] (
15:11:29 Quit aliask ("Chatzilla 0.9.67 [Firefox 1.0.7/20050919]")
15:13:49JdGordonLinusN: do u like the idea of having plugins store their settings in files instead of hardcoding them? (im mainly talking about key mapings, but any plugin setting really...)
15:14:43LinusNthere is a configfile API for plugins
15:14:51JdGordonthere is?
15:15:22markunTo include your new files you can also use the tool 'cvsdo add [filename]'. If you then use cvs diff -uN the new files will be included.
15:15:42*JdGordon shoots self
15:15:45JdGordongnite folks
15:16:01 Quit JdGordon (Read error: 104 (Connection reset by peer))
15:16:01 Quit lamed ("CGI:IRC (EOF)")
15:16:59 Quit Benacool (Read error: 110 (Connection timed out))
15:17:32 Quit Rob2222 (Read error: 110 (Connection timed out))
15:18:57 Quit Bgr (Read error: 110 (Connection timed out))
15:19:16LinusNpoor JdGordon spent hours reinventing the wheel :-(
15:19:50markunLinusN: Is it ok if I modify the scroll code to have a pause after scrolling the whole line?
15:20:11perplexityBut I bet he learned a lot about Rockbox internals while he was there..
15:20:37LinusNi wouldn't mind personally, but i'm sure someone will...
15:20:45LinusNsomeone always does
15:20:56perplexityheh.. make it configrable ;p)
15:21:04markunyes, like pedro not liking the 'center scrolling' feature
15:21:44LinusNscrolling looks awful on my h100
15:21:58LinusNi have a 150px step size and a long delay
15:22:04markunThe 'center-scrolling' now starts scrolling when it's 1 line from the bottom. It used to be when it was at 2/3 of the screen. Which behaviour do you prefer?
15:22:40LinusNmarkun: i think that depends on the font
15:22:45perplexityOooookay.. so the X11 Sim works fine on my 2.6Ghz Athlon with kernel HZ=1000, but is naff on my 1Ghz PIII-M with kernel HZ=250 :(
15:22:57markunLinusN: The delay is only for lines that are less than 2x the screen width. I want to add the delay also to lines that are longer..
15:23:48wubblaLinusN: I just wanted to inform you about the fact that i'm finally able to boot rockbox using the HOLD-trick
15:23:49LinusNnote to self: try lamed's patch
15:24:04LinusNwubbla: can you try a thing for me=?
15:24:12wubblaLinusN: if you don't remember me: i'm the one you wanted to send you my unit...
15:24:15wubblaLinusN: of course
15:24:33LinusNa test bootloader
15:25:08wubblaLinusN: well... i'll flash it after lunch, ok?
15:25:12Rob2222_huh? ^^ something for test``
15:26:13 Join dropandhop [0] (
15:26:31Rob2222_ah, test BL for the startup bug, heh?
15:26:49markunLinusN: can I add the scroll delay and then remove it if there are many complaints, of shall I first ask some more people what they think about it?
15:27:52Moosmarkun: you've hot my unnecessary vote :)
15:28:06markunthanks Moos :)
15:29:27Moossometimes I've got this problem too, and that make me a bit crazy
15:29:44Moosit doesn't stop scroll and restart
15:30:16markunAlso makes me crazy
15:31:16Moossure if you make it optional (I know people here doesn't like much too options) sure that will enjoy some users
15:31:24markuntucoz was also in favor of it
15:31:44markunMoos: people can just change the scroll delay, that's what it's for I think.
15:32:04 Quit ^BeN^ (Read error: 110 (Connection timed out))
15:32:24Moosbut when the line is very long
15:32:38Moosit doesn't go bidirectionaly
15:32:50Moosconstant anoying scroll
15:33:01markunNo, but I made it pause when it's at the beginning of the line again.
15:33:24Moosthat why I like your idea :)
15:34:01markunIt's not really important. People are probably more waiting for unicode than some scrolling feature :)
15:34:08LinusNi like it
15:34:10lostlogicam I reading the map file right that there is about 500k of executable space in rockbox at this time (on h300)?
15:34:32markunLinusN: I'll commit it and blame you for liking it then :)
15:35:06 Join djgalak [0] (
15:35:36 Join Amar [0] (
15:36:25markunOne thing I'm wondering about.. in the beginning it waits scroll_delay before scrolling, when reaching the end of the line it waits scoll_delay * 2 before going back..
15:36:44perplexitylostlogic: I have moved no further on the profiling..
15:38:38lostlogicperplexity: I'm knee deep in writing a profiling architecture for rockbox so we can profile on the unit, no ETA as of yet.
15:39:08perplexityDid you look at the link that LinusN posted yesterday for the rockbox profiling patch that has already been started?
15:39:18lostlogiceek, no, do you have it?
15:40:02perplexityIt's fairly simple, but it makes it easy to instrument selected pieces of code..
15:41:04perplexityI have given up on the profiling idea myself and gone off to learn about (i)mdct and (i)fft to see if I can a) become a little more useful/enlightened, and b) pick a more efficient algorithm for the mdct in Tremor..
15:41:31lostlogichehe cool
15:42:03perplexityTremor first.. video codec next ;p) (Said mildly tongue in cheek)
15:42:16lostlogicI'll probably use the PROFILING_WRITE function from this patch in my stuff, but I'm working to leverage the -pg option of GCC and more automagic profiling...
15:42:33perplexityanyway,I have to shoot off, keep me posted on how you get on :)
15:42:52lostlogicI shall
15:42:52 Quit djgalak ("CGI:IRC (EOF)")
15:47:12 Quit Rob2222_ ("see you @")
15:50:03lostlogicis there an extant 'jiffies-like' counter that holds the number of ticks of some defined size since bootup?
15:55:34 Quit Amar ("CGI:IRC (EOF)")
15:56:54 Quit DJDD_ ("Trillian (")
15:57:54B4gderthere's a tick counter
16:00:38amiconnBah, what's that! :(
16:00:38 Quit DreamTactix291 (Read error: 104 (Connection reset by peer))
16:00:57amiconnSomeone broke the nice round-scrolling :(
16:01:03 Join DreamTactix291 [0] (
16:02:14LinusNmarkun: there you go
16:02:37*amiconn doesn't like it at all
16:02:56amiconnThe pause makes sense when scrolling bidirectionally, but not at all when scrolling round
16:03:07B4gderI agree
16:03:17B4gderanyone aware of any problems with our FAT and writes?
16:03:20markundamn :)
16:03:21B4gderFAT code
16:03:44markunI can change it back and just use it myself.
16:04:06 Join merbanan [0] (
16:04:12LinusNB4gder: no
16:04:13amiconnThe reason for a scrolling line is that the text is longer than the line, so what's the use of it stopping arbitrarily in-between?
16:04:25amiconnThe text should be readable during scrolling of course
16:04:38B4gderLinusN: thanks, got something like a bug report on the u-boot mailing list... :-)
16:04:42B4gderI'll see what he means
16:04:43amiconn...which requires to adjust the scroll step & speed accordingly
16:05:20markunamiconn: arbitrarily in-between? I pause at the beginning of the line.
16:05:37LinusNi got an international h320 today
16:05:37amiconnYes. For me that's still arbitrary
16:05:58LinusNbut it didn't have the bootloader black-lcd issue :-(
16:06:02amiconnA useful thing would be to implement jump-scroll for bitmap targets, imho
16:06:23LinusNthat's good
16:06:39LinusNi can simulate it woth 150px step size and a slow scroll speed
16:06:58markunHow does just-scroll work?
16:07:48lostlogiccan anyone think of a good reason to use a separate counter for profiling ticks from the main tick counter?
16:07:50amiconnJump scroll is implemented for the player
16:08:21amiconnIt shows the whole line in 'chunks', each for a set amount of time
16:08:59Lynx_lostlogic: maybe for profiling the main tick counter? ;)
16:09:00amiconnThe chunk is the viewable width of the LCD, but adjusted so that the next chunk starts at the beginning of a word
16:09:04amiconn(if possible)
16:09:29markunsounds nice
16:09:51amiconnYou can select whether you want jump scroll, and how many times it should jump scroll until it resorts to normal scrolling
16:10:02lostlogicLynx_: that's the question −− is that a reasonable way to count ticks spent in each function, or do I need to create a separate tick counter
16:10:03amiconn(iirc 1x ... 5x or always)
16:10:36markunwhat's the use of switching to normal scrolling?
16:11:05amiconnWell, I have jump scroll enabled on my player, jump-scrolling once
16:11:06Lynx_lostlogic: i don't really know
16:11:19LinusNamiconn: that's how the iriver firmware does scrolling
16:11:24amiconnIt's good for getting a quick overview of the string
16:11:32 Join djgalak [0] (
16:11:35amiconnmarkun: Just build a player sim if you want to see it in action...
16:11:47markunSim building is broken on FreeBSD
16:12:08amiconnDo you have wine?
16:12:29amiconnThen you can try it with the windows sim...
16:13:12markunI used to do a lot with the simulator, was quite a useful tool. Maybe I should try to convert the sed code that broke it.
16:13:38LinusNyou should
16:13:57mirakwhat IDE do you use to develop on Linux ?
16:14:13LinusNno IDE here
16:14:43LinusNemacs + bash shell
16:15:25LinusNamiconn: have you looked at the single-bitmap wps patch?
16:16:26markunThe Gigabeat player in landscape mode works very nice. Would other rockbox targets also benefit from it?
16:16:41LinusNlandscape mode?
16:16:58markunYes, turn the interface by 90 degrees
16:17:14LinusNlet's not go there
16:17:19 Join t0mas [0] (n=Tomas@unaffiliated/t0mas)
16:17:29 Quit djgalak ("CGI:IRC (EOF)")
16:19:00LinusNwhat's the resolution on that sucker
16:19:01B4gderthe rotate option! ;-)
16:19:22markunWhat's it on the X5?
16:19:41t0masit's time for a pocketpc port of rockbox
16:19:47*t0mas hates windows media player
16:21:02preglowwho doesn't?
16:21:39t0masmaybe I can port a simulator to the pocket pc...
16:21:43markunrotation will complicate a lot of things so I understand you don't want it, but it's still nice on the Gigabeat
16:24:21markunamiconn: did you try rockbox with the scrolling delay or do you just assume it sucks?
16:24:52markunmaybe assume is not the correct wording
16:28:41LinusNmarkun: hmmm, i now discovered that i don't like it either :-)
16:28:54LinusNi'd like it better if it stopped at the end of the scroll instead
16:29:39markunok ok, I'll change it back :)
16:30:48 Join ep0ch| [0] (n=ep0ch|@
16:31:29ghode|afkcan anyone recommend some good 2.1 pc speakers?
16:31:53ep0ch|i like the way iriver have done scrolling to overcome the dodgy lcd - i.e. vertical scrolling
16:33:10t0masghode|afk: Create Itrigue
16:33:30ghode|afkg4h i was hoping you wern't going to say that :(
16:34:02ep0ch|2.1 thats stereo + subwoofer?
16:34:02t0masI've got both the Itrigue 3200 and 3400
16:34:10t0masyes, it's stereo + sub
16:34:14ghode|afkI've been look at these: the sat's seem too big :/
16:34:32ghode|afkthe 3400's :)
16:34:41t0masI've got both... 3200 are really small
16:35:04t0maswith sound like a full theatre setup
16:35:10t0mas3400 has even better sound...
16:35:16ghode|afkwhich do you prefer?
16:35:16t0masbut they're bugger
16:35:18 Quit lostlogicx ("Lost terminal")
16:35:25t0masI have the 3200 on my oiwn desk...
16:35:30t0masmy sister has the 3400's now
16:35:44t0mas3200 is cheaper... and the sound is just great...
16:35:55t0masand they fit nice next to my tft screen... cus they're so small
16:36:26ghode|afkhehe thats encouraging
16:36:41t0masand for both the remote control rocks... you get some cables with it to connect your irivier to them :D
16:36:50ghode|afki see you can plug headphones into the remote unit, does it also control the volume?
16:37:10t0masit controls volume + bass, and you can plugin headphones and iriver into it...
16:37:13ghode|afkyeah that is the main feature i am looking at
16:38:10ghode|afkhmm about 14 difference in costs
16:38:20 Join lamed [0] (
16:38:47ghode|afkwhat is the sub like on the 3200? does it provide good bass?
16:39:51t0masyou can read some reviews yourself... I expected it to be not that good...
16:40:06t0masbut in dance like music it's moving my keyboard around the desk...
16:40:34lamedbolt it
16:40:36t0masin graphs in reviews it has a peak around 80 hz iirc... and that's the "disco" frequency
16:40:38ghode|afkok one last thing, is the base of the sat stable? i mean i don't want it to be falling over all the time
16:40:45t0masso creative tried to make em sound good in that kind of music
16:41:11t0maswell... they never fall over... but I never toch them...
16:41:25t0masthe base is quite heavy...
16:41:51*t0mas shakes his desk...
16:41:59t0masnope... not turning over :)
16:42:13B4gder"VFS: Unable to mount root fs via NFS, trying floppy"
16:42:26B4gderthat's a neat error message
16:42:40t0masdiskless client?
16:43:03B4gderwell, it is during development
16:47:54FebsLinusN, you might be interested in this thread at M.R.:
16:48:12LinusNFebs: i've read it
16:48:46LinusNhe probably forgot to update the codecs
16:51:08TCKI haven't been here for a while
16:51:22TCKbut can I report an mp3 that always freezes RB here?
16:52:04TCKit just comes up with horizontal lines, and pressing play gives an error message
16:52:42ghode|afkt0mas: just ordered the 3400s, couldn't find anyone with the 3200s.
16:54:34LinusNTCK: can you dcc it to me?
16:55:55dropandhopSlasheri - you happen to be around?
16:56:43t0masghode|afk: you should've asked...
16:56:45t0maswhich country?
16:57:29 Quit lamed ("CGI:IRC (EOF)")
16:58:00Slasheridropandhop: hi, i just came home
17:03:56dropandhoplook at you- ring you up and he shall appear
17:04:13dropandhopdidnt know what irc schedule you were on
17:04:26dropandhopdidn't mean to bug you...just thanks for thinking about the tick-fix
17:04:37 Join tvelocity [0] (
17:07:03***Saving seen data "./dancer.seen"
17:08:24 Join djgalak [0] (
17:10:54TCKlinus: I lied, it's an ogg, not mp3
17:11:30TCKand sorry about the delay
17:11:34LinusNTCK: are you behind a firewall?
17:11:52TCKI don't believe so
17:11:58TCKI'll just check
17:12:26TCKtry now
17:12:45LinusNtry to dcc it again
17:13:03LinusNno go
17:13:06TCKno luck
17:13:24TCKI can probably upload it to an ftp if need be
17:13:27djgalakhello all
17:14:04DBUGEnqueued KICK djgalak
17:14:35LinusNTCK: got my /msg?
17:14:51KohlrabiTCK got ogginfo?
17:15:05Kohlrabiit checks ogg files
17:15:13KohlrabiAnd I think it can find errors
17:15:13 Quit djgalak ("CGI:IRC (EOF)")
17:15:36TCKit plays okay in Winamp, and it also plays okay in the default iRiver firmware
17:16:55ep0ch|TCK: how recent is your rockbox build?
17:17:06KohlrabiStill could be broken :)
17:17:22TCK5th of this month
17:17:28ep0ch|oh update
17:17:36TCKalthough it didn't work around october time too
17:17:48TCKI'll try an update though
17:17:54LinusNplease do
17:18:02ep0ch|there has been a fix for old vorbis files since then
17:18:14 Join bobwise [0] (
17:18:36LinusNTCK: always try a newer build before reporting a bug
17:19:29ep0ch|yeah it should be fixed from a build after the 26th
17:21:00TCKyep, works
17:21:03TCKsorry about that
17:21:14TCKI just assumed that it was a longstanding problem
17:21:39TCKbeen meaning to report it for a while, but I've had little to no access to the net
17:21:39LinusNit was, until the 26th :-)
17:22:00TCKwell, I'm glad it got fixed
17:22:08TCKwhat was the nature of the problem anyway?
17:23:55TCKI see
17:24:13TCKso it's an old encoded vorbis problem
17:24:18TCKmakes sense
17:24:27dropandhopSlasheri - fill me in when you have some more tick-fix ideas....i'm curious what cha thinkin!
17:24:32TCKthanks guys
17:24:42ep0ch|yeah i think it was pre vorbis1.0rc3
17:26:05LinusNdropandhop: have you opened your player?
17:30:23dropandhopi haven' a little afraid to!
17:30:50 Quit Febs ("CGI:IRC (Ping timeout)")
17:31:38 Join lostlogicx [0] (
17:33:38LinusNi just wanted to check a thing on a ticking player
17:33:59markunLinusN: can you help me identify this IC? "606A 0512". On all units it's connected next to the backlight connector and I think it has 16 pins:
17:34:26LinusNthere is a decoupling capacitor between the digital and analog ground, that may not be present on the ticking units
17:34:49dropandhoplinus- do you want me to post in the forum for someone to look at this? and report back?
17:34:58dropandhopor maybe just to jump on here and contact u?
17:35:19LinusNdropandhop: whatever, please do
17:36:48 Join frederic [0] (
17:37:13 Nick frederic is now known as frederic_ (
17:38:44LinusNmarkun: have you tried to trace the connections?
17:39:17markunno, I haven't. From the CPU to the other components?
17:39:34LinusNall of them
17:39:53LinusNthat might give you an idea what it is for
17:41:06markuncould it be a dc/dc convertor?
17:41:27LinusNmaybe, or an EL backlight driver?
17:41:48markunyes, maybe. Is it possible to dim EL backlights?
17:43:18wubblaLinusN: I'm flashing the test-fw now...
17:43:37LinusNmarkun: i think i found it
17:43:39markunwell, the gigabeat's backlight can be dimmed
17:43:43lostlogicxlong = 4, int = 4, short = 2, char = 1 bytes in rockbox?
17:43:56markunLinusN: really? What is it?
17:44:19wubblaLinusN: btn: 1 0 0 0
17:44:35wubblaLinusN: using the HOLD trick
17:45:34LinusNmarkun: sc606 - Low Noise Smart LED Driver with Serial Interface
17:45:54wubblaLinusN: maybe the HOLD button state is just "swapped", e.g "on" actually means "off"
17:46:31LinusNwubbla: did it boot?
17:46:42wubblaLinusN: yes, it booted
17:46:50wubblaLinusN: (with the HOLD trick)
17:46:53LinusNbut only with the hold trick?
17:47:01LinusNwhat happens without the trick?
17:47:07wubblaLinusN: i'll try without it
17:47:15wubblaLinusN: yes
17:47:21wubblaLinusN: just with the hold-trick
17:47:35LinusNdoes it display anything without hold?
17:48:16wubblano, nothing at all
17:48:53wubblaLinusN: what have you changed in this firmware?
17:49:16LinusNi removed a check for the ON button, and added the btn debug output
17:49:39wubblaLinusN: hasn't it something to do with the HOLD button?
17:49:50mirakwhen the music's over
17:50:21LinusNwubbla: i'm not so sure about that, but it seems to help to start with the hold button active
17:50:48wubblaLinusN: indeed :-)
17:50:50markunLinusN: thanks!
17:50:59LinusNdoes the disk start spinning when it fails to boot
17:51:05wubblaLinusN: yes
17:51:58 Quit frederic_ ("Chatzilla [Firefox 1.5/undefined]")
17:52:38markunalso found the datasheet:
17:53:10LinusNmarkun: great
17:54:04wubblaLinusN: the button driver is initialized after the LCD driver, right?
17:54:42 Join leftright [0] (
17:54:45LinusNhmmm, no
17:54:51leftrightHi there
17:54:55LinusNoh, yes it is
17:55:24wubblaLinusN: hmm...
17:55:54leftrightGapless mp3 isn't as seamless as it used to be, doesn't sound good at all now
17:56:01wubblaLinusN: maybe these functions need to be fine-tuned:
17:56:02wubbla or_l(0x00100062, &GPIO1_FUNCTION);
17:56:03wubbla and_l(~0x00100062, &GPIO1_ENABLE);
17:56:03mirakwhen the music's over
17:56:11wubbladata = GPIO1_READ;
17:56:20LinusNwubbla: fine-tuned?
17:56:38wubblaLinusN: you're using this to check the button state, right?
17:56:46mirakhem mp3 do floating point calculs ?
17:56:53LinusNmirak: no
17:56:57Maximeis there a 'foobar2k addict' here? :x
17:57:02wubblaLinusN: and this somehow seems to be the point where everything breaks...
17:57:11mirakLinusN: most video codecs does ?
17:57:13leftrightpreglow: did you fidlle with gapless ? :-)
17:57:40LinusNmirak: i don't know much about video codecs
17:58:08mirakI heard some said that lack of FPU was a major issue
17:58:13MaximeI know this is off topic but, if s-o know fb2k well, can him tell me how to keep only the 'tray icon'?
17:58:37leftrightand.. crossfade shuflle mode doesn't deselect if shuffle mode is disabled
17:58:40LinusNwubbla: do you have a remote?
17:58:50LinusNlcd remote
17:58:52markunLinusN: I'm pretty sure it's the right IC. Page 15 of the datasheet shows the markings should be "606A yyww" (yyww being a datecode)
17:59:03wubblaLinusN: sadly i haven't got one... :'-(
17:59:03LinusNmarkun: bingo
18:00:58wubblaLinusN: you might log some debug data to a file on the disk, can't you?
18:01:20LinusNit seems to die long before the disk is initialized
18:01:53wubblaLinusN: hum. :-/
18:02:28ep0ch|Maxime: afaik you can only have foobar in the tray when its 'minimized'
18:02:41Maximenot cool
18:02:50wubblaLinusN: and why not completely left out the button checking and instead directly jump to:
18:03:02MaximeI like mpc for that, becuz it does not hide when u click in the "Desktop" shortcut
18:03:08wubblaLinusN: just for testing purposes, to see if it works
18:03:15 Quit TiMiD ("leaving")
18:03:26LinusNwubbla: it's the ide_power_enable(true) that makes the disk start spinning
18:03:31Maximeand as i put foobar in the other screen, i would like it to be always visible
18:03:33LinusNbut i'll try
18:03:46wubblaLinusN: then we'll know for sure that the bug is button-driver related...
18:04:51LinusNok, new zip,
18:04:57ep0ch|Maxime: i'm confused.
18:05:20leftrightSlasheri: you around ?
18:05:20ep0ch|maxime: what behaviour do you want exactly?
18:05:55wubblaLinusN: 2.1MB?
18:06:03Maximeep0ch|: that foobar will not hide when I press "desktop" key, so, that he don't appear in taskbar, only in traybar :x
18:06:08LinusNwubbla: yes
18:06:20wubblaLinusN: why is it that small?
18:06:44wubblaah ;-)
18:06:58ep0ch|Maxime: what "desktop" key?
18:07:08Maximein the quick launch
18:07:20ep0ch|you mean "show desktop"?
18:07:32LinusNwubbla: it worries me that the backlight isn't turned on
18:07:33Maxime(i'm french, so don't know the english callings)
18:07:44ep0ch|oh you mean "always on top"?
18:08:05ep0ch|oh no
18:08:24Maximeep0ch|: if an app is only on systray, it won't hide when "show desktop"
18:09:12CtcpIgnored 1 channel CTCP requests in 0 seconds at the last flood
18:09:12*wubbla is flashing...
18:09:20wubblahopefully this does work...
18:09:39ep0ch|i think i understand, well i use columns ui and there is a tickbox so you don;t have to have "minimize to systray" on
18:10:41Maximeyeah but.. the app will hide when "show dekstop" :s
18:11:17ghode|afkdoesn't that defeat the purpose of show desktop
18:11:35Maximeghode|afk: yeah, but mpc does it well
18:11:38 Quit CBM-away (Read error: 104 (Connection reset by peer))
18:11:45 Join CheeseBurgerMan [0] (n=BurgerBo@
18:12:28ep0ch|i dont think i can get mpc to behave like that
18:12:33Maximeghode|afk: it's because I extended my desktop to another screen, so i would like to have foobar always on the other screen
18:12:34wubblaLinusN: uh oh...
18:12:46Maximeep0ch|: in mpc, tell him to have only traybar
18:12:50Maximeand try ^^
18:13:00wubblaLinusN: i think that wasn't a good idea...
18:13:10ghode|afkif you find out how, let me know. i have just started to use duel screens and this is annoying
18:13:27LinusNwubbla: tell me
18:13:35wubblaLinusN: same thing egain
18:13:35Maximeghode|afk: ok ^^
18:13:49wubblaLinusN: except the fact that the HOLD trick doesn't work anymore
18:14:05Maximeep0ch|: you think we can 'modify' the source of fb2k not to show the traskbar thing?
18:14:57ep0ch|no, i think you should post something in the relavant forum.
18:15:38Maximefound it
18:15:41Maximefoo_notaskbar ^^
18:15:52 Join ^BeN^ [0] (
18:16:22Maximeit works
18:16:22wubblaLinusN: ok...nothing that i can do...
18:16:34Maximeghode|afk: found it, on the link I pasted
18:16:39wubblaLinusN: except sending it to you
18:17:04Maximeep0ch|: thanks, i found ^^
18:17:06wubblaLinusN: finally i really have to send it...
18:17:58LinusNwubbla: does the disk spin up?
18:18:03wubblaLinusN: yes
18:18:12wubblasame thing again..
18:18:26LinusNcan you see if there is any sign of a backlight?
18:19:10mirakare you unemployed ?
18:19:23wubblaLinusN: nothing...
18:19:24mirakor in vacation
18:19:30mirakor at work lol
18:19:56LinusNwubbla: new zip, same place
18:20:08wubblaLinusN: i cannot flash it anymore
18:20:15LinusNmirak: who?
18:20:19wubblaLinusN: because i cannot even boot the iriver fw
18:20:32mirakLinusN: people often here
18:20:44LinusNwubbla: i think you can
18:21:01LinusNhold ON and double click the reset
18:21:18mirakLinusN: I see you often here, that's why I am wondering
18:21:32LinusNmirak: i work from home
18:21:51mirakLinusN: I can't do that, I can't keep focused on my job
18:21:52LinusNbut i join the channel when i'm at the office as well
18:22:03mirakLinusN: when I work at home
18:22:22mirakto many things around
18:23:35wubblaLinusN: yay! iriver fw boots...
18:23:46LinusNi know... :-)
18:24:13LinusNthat's one of the safety nets i implemented in the boot loader
18:24:57 Join _FireFly_ [0] (
18:25:12wubblaLinusN: but i had to try it ~15 times to finally boot it ;-)
18:25:42 Part leftright
18:25:44wubblaLinusN: yes
18:25:45 Join rockriver140 [0] (
18:25:46mirakLinusN: with the current bootloader for H300 the main risk is to miss the flash ?
18:25:51wubblaLinusN: it was not that easy
18:26:19mirakI want to try, but I am afraid ...
18:26:46LinusNmirak: then wait until we get it more stable
18:26:46mirakI don't want you to say me go ahead, just to know what are the exact risks actually
18:27:08rockriver140so how do I get video working on my H140 !:-) wooo!
18:27:13LinusNthe biggest risk is to "miss the flash"
18:27:19mirakLinusN: mmm by stable you mean a lower brick ratio ?
18:27:36LinusNmirak: the brick ratio is so far 0
18:27:42LinusNno bricks yet
18:27:45mirakLinusN: lol
18:27:50mirakI feel it's my turn
18:28:00mirakI am kidding
18:28:09mirakLinusN: where do you live ?
18:28:15mirakin sweeden right ?
18:28:23mirakok I am in france
18:28:39mirakI wanted to know if you repair bricked flash with your bdm stuff
18:28:51mirakor happened to do it
18:29:02 Join webguest63 [0] (
18:29:15wubblaLinusN: YAY!
18:29:19wubblaLinusN: IT BOOTS!
18:29:29wubblaLinusN: without any glitches at all
18:29:40wubblaLinusN: what did you change?
18:29:43 Quit Moos (Read error: 104 (Connection reset by peer))
18:29:51wubblaLinusN: what was it?
18:29:57LinusNthe ide power
18:30:05wubblaLinusN: hooray!
18:30:15wubblaLinusN: checking if the iriver fw also boots...
18:30:19LinusNi'm not equally happy
18:30:54wubblaLinusN: hm?
18:31:09wubblairiver fw starts too...
18:31:28wubblaLinusN: you're the king :-D
18:31:54wubblaLinusN: i think it's time for bootloader v3, right? ;-)
18:32:12 Join TCK- [0] (
18:32:26mirakLinusN: so you advise to send iriver to iriver in case of flash failure ?
18:32:29LinusNwubbla: this might mean that there are two versions of the h300
18:32:36LinusNmirak: no, to me
18:32:49mirakLinusN: ok
18:33:03rockriver140sorry to repeat myself, but could someone kindly explain how to get the video on the iriver h100 series working before it gets commited? or just point me in the right direction to instructions? thanks ;)
18:33:04*LinusN collects dead players :-)
18:33:04wubblaLinusN: hmm...
18:33:16LinusNrockriver140: i have no idea
18:33:37mirakLinusN: I sent a power supply to antec RMA in Holland, it coasted me 20 euros but it's heavier
18:33:57linuxstbrockriver140: One person is working on video, for the h100, but I don't think he's released anything to anyone else yet.
18:34:03LinusNmirak: you shouldn't be that worried
18:34:04mirakrockriver140: what kind of video it can read ?
18:34:07wubblaLinusN: it actually could mean that there are two versions...
18:34:11mirakLinusN: ok :)
18:34:23rockriver140its rockbox video, .rbv
18:34:27wubblaLinusN: but why care about this, if it actually works? ;-)
18:34:34mirakrockriver140: what frame rate and resolution ?
18:34:41mirakrockriver140: is it compressed ?
18:34:48LinusNmirak: to my knowledge, nobody has ever bricked a player yet, not h1x0 or h3x0
18:34:50rockriver140i think its uncompressed AVI, 8 fps and with mp3 audio
18:34:51 Join iobound [0] (
18:34:56wubblaLinusN: i've played with rockbox for one day now and i haven't really noticed any major bugs...
18:35:18mirakLinusN: I have heard a guy crying on a french forum. I don't know what happened exacltly
18:35:22mirakafter that
18:35:29rockriver140so the files will be huge lol
18:35:35mirakrockriver140: That's Wett stuffs ?
18:35:44wubblaLinusN: just small ones, e.g. flickering of the LCD when using WPS's with bitmaps...
18:36:25mirakrockriver140: he said me internally it's 70fps. He use that high rate for creating a gray scale with latency
18:36:59webguest63linus: will it be possible to change the main meny background in h3xx rockbox?
18:37:44rockriver140ah right, the video showed 8fps, but what do i know. heres the link wett/capt-video2.avi">
18:38:00wubblaLinusN: you should probably write a small plugin that writes all relevant hardware info into a file so that you can analyze the differences between different players...
18:38:12_FireFly_had a main-dev got some time to look at the new wps-file-loader-patch ??
18:39:22rockriver140I have to say, im getting quite envious over the rockboxs quick port to the H300 :) I want a colour screen now :(
18:40:18LinusNwebguest63: yes, but not yet
18:40:57mirakrockriver140: I have it
18:41:13mirakrockriver140: I have H300
18:41:27mirakrockriver140: I mean I have seen that videp
18:42:25webguest63but youre working on it now, right linus?
18:43:06 Quit t0mas (" tot zo")
18:44:29LinusNwebguest63: not at all
18:46:17webguest63yeah but its nice to hear that youre planning to:D
18:46:25NixsosLinusN: when will you release an updated fw patcher for H300? i heard you fixed the black screen bug
18:46:26webguest63good work with rockbox, by the way !
18:46:39mirakLinusN: is rockbox skinable or something or every graphics are hard coded ?
18:46:46mirakI mean non modular graphics
18:46:55LinusNNixsos: i haven't fixed it yet
18:47:29mirakLinusN: when you play a song is it the same thread that play sound and pictures the screen ?
18:47:33 Quit einhirn ("Miranda IM! Smaller, Faster, Easier.")
18:47:39_FireFly_mirak: no
18:47:41 Quit TCK (Read error: 110 (Connection timed out))
18:47:43 Join einhirn [0] (
18:47:46_FireFly_they are differents threads
18:47:47mirak_FireFly_: that's cool
18:47:49XavierGrLinusN: I have a ticking H140. If you want I can open it.
18:47:58LinusNXavierGr: great
18:48:03XavierGrWhat info do you want exactly?
18:48:11_FireFly_due this you can navigate through the file-tree while the song is playing
18:48:23ep0ch|i'll open my ihp-120 up too if it helps
18:48:44mirakLinusN: you said there was probably two different H300 versions, what's up with that ?
18:49:34 Quit rockriver140 ("CGI:IRC")
18:49:39ioboundfyi, bootloader v2 works as advertised on my international h340 with 1.28k firmware
18:52:32mirakiobound: I will try it I guess
18:55:36LinusNmirak: it might mean that the IDE power is controlled differently on some models
18:56:58XavierGrLinusN: I will leave for the next hour or so. If you want tell me now what you want me to do, else if you are here after an hour we can talk this later.
18:57:28wubblaLinusN: btw. why do you actually need this extra ide_power_enable(false); call anyways?
18:58:10LinusNbecause i want to make sure that the drive doesn't spin up if you start it with the hold button enabled
18:58:29wubblaLinusN: hmm...
18:58:51wubblaLinusN: so you just removed this single call in the last test-fw?
18:59:36 Part bobwise
18:59:37LinusNno, i removed the ide_power_enable(true) call
18:59:56wubblaLinusN: huh?!
19:00:17wubblaLinusN: so theoretically the harddisk should never be activated?!
19:00:33LinusNit is, in ata_init()
19:01:05LinusNthe early ide_power_enable() call is to spin up the disk earlier to speed up the boot
19:01:14 Join tim66 [0] (n=tim@
19:01:45wubblaLinusN: i see..
19:01:51LinusNbut it seems that the h300 has problems with that
19:02:10LinusNperhaps the voltage regulators haven't stabilized...
19:02:22 Quit ep0ch| ("gone")
19:02:40LinusNin any case, we can remove that call and live with a slightly slower boot for now
19:02:49LinusNi have to go now
19:02:55LinusNcu later
19:02:57wubblaLinusN: slow :-)
19:03:02wubblaLinusN: cu!
19:03:23 Part LinusN
19:04:23 Quit webguest63 ("CGI:IRC (EOF)")
19:05:20XavierGrah! Linus left without saying me a word about the H140. Well I guss that the operation will have to wait. ;p
19:06:25mirakLinusN: when the iriver is off and hold enabled, rockbox is not active isn't it ?
19:07:07***Saving seen data "./dancer.seen"
19:07:33mirakthat's why I wonder he bother with that
19:08:11B4gderyou read the code?
19:08:18mirakB4gder: not yet
19:08:40mirakB4gder: I will do that soon. I need to read some stuffs for work
19:08:42B4gderI suggest you do that first before asking about code details
19:08:53mirakB4gder: sorry
19:09:10B4gderno worries
19:09:34mirakB4gder: where do you advise me start reading the code ?
19:09:53B4gderif you want bootloader details, check the bootloader code
19:13:42 Part linuxstb ("Leaving")
19:14:02mirakB4gder: I am more insterested by the kernel
19:14:08mirakI guess there is one
19:17:58markunB4gder: linus identified the backlight driver. It can dim the backlight LED in 32 steps.
19:18:17Slasherimirak: yes, the kernel is in kernel.c and thread.c
19:18:52Slasheriprobably you should start reading from apps/main.c. That is where the code starts (or it start from crt0.S in fact)
19:19:07mirakSlasheri: I know not much about OS architecture and fuctioning, that's will be enriching
19:19:34mirakSlasheri: what EID do you use to browse and code ?
19:19:45 Quit FingerSoup ("On the other hand, you have different fingers.")
19:20:03Slasherimirak: I don't use any IDE if you mean that. Just kate editor in kde and console tools (editors etc.)
19:20:23mirakSlasheri: ok
19:21:03 Join leftright [0] (
19:21:48SlasheriBut yes, the kernel architecture should be quite typical found on embedded systems so that should be good to read and understand
19:23:39leftrightHi Slasheri
19:23:45Slasherihi leftright :)
19:23:57leftrightgot a bug for you :)
19:24:01leftrightcrossfade shuflle mode doesn't deselect if shuffle mode is disabled
19:24:03Slasherihehe, nice :P
19:24:19SlasheriHmm, interesting
19:25:21SlasheriOr do you mean it will crossfade always when manually selecting a song if crossfade is activated?
19:25:34leftrighter no
19:25:51Slasheriok, i will try.. just a moment :)
19:26:23leftrightIf the crossfade "shuffle" mode is selected, it should then only crossfade if shuffle is enabled
19:26:31ioboundrock & roll, i'm playing 2nd_pm.s3m PERFECTLY in the win32 ui sim with my own plugin.
19:27:09ioboundi got the necessary exp(), log(), floor() methods from freebsd's libmsun, and found out that pow() could be expressed as exp(log(x)*y)
19:27:09leftrightbut if shuffle is deselected it should then disable the crossfading, or do I have it wrong
19:27:16ioboundwonder if it compiles for m68k too
19:28:06Slasherileftright: Hmm.. it disables it if i deselect the shuffle mode.. Then it will not crossfade when skipping tracks
19:28:36leftrightbut it continues to crossfade, is this correct ?
19:29:07Slasherileftright: if it's already crossfading when you disable the shuffle mode, then it has no effect
19:29:09_FireFly_crossfade doesn't have an mode shuffle only repeat has an option/mode shuffle
19:29:31_FireFly_or i'm misunderstanding something
19:29:31 Quit leftright ("CGI:IRC (EOF)")
19:30:01 Join leftright [0] (
19:30:16leftrightsorry about that, fingers and thumbs today
19:31:07_FireFly_leftright: do you realy mean that crossfade has an shuffle mode ??
19:31:24leftrightok, so if I only want to enable crossfading if I shuffle songs I dont need to select crossfade on ?
19:31:27_FireFly_i can't see such an mode in the crossfade option
19:31:33ioboundwhat does it mean when i get a green screen with I04:IllInstr 1000115A
19:31:35leftrightlast option
19:31:47ioboundi guess it means an illegal instruction?
19:31:48Slasherileftright: no, then you should select the shuffle mode
19:32:03_FireFly_i have only shuffle or mix as an option
19:32:16_FireFly_in fade out mode
19:32:20Slasheri_FireFly_: Hmm, there should be on, off and shuffle
19:32:32SlasheriThat is the first option
19:32:45_FireFly_ah ok ;)
19:32:56_FireFly_never saw it :)
19:33:18leftrightSlasheri, I do that but it still crossfades if shuffle is deselected, I dont understand
19:33:24_FireFly_i hadn't enabled crossfade after the settings has changed
19:33:44Slasherileftright: that is weird..
19:33:51 Nick Lynx_ is now known as Lynx_awy (
19:35:12leftrightyep it is weird, I dont think I'm imaging it, no mushrooms today :->
19:35:44 Quit Maxime ()
19:36:32iobounddoes anyone know how I can find out where in my .rock file the illegal instruction at 1000115A is?
19:36:43ioboundi assume i have to use m68k-objdump −−disassemble?
19:37:01Slasheriiobound: if you have compiled rockbox, you can look from the .map files
19:37:20ioboundbut the map files all have addresses like 0x32f4047e
19:37:29ioboundi guess rockbox loads the .rock file somewhere else in memory or something?
19:37:36SlasheriHmm, 0x1... is on iram
19:38:07leftrightgapless also seems to be broken, it's not as seamless as it used to be (crossfade off ofcourse)
19:38:07ioboundokay? do I have to do something special to get my code on iram? it probably in my own code
19:38:15Slasheriyep, the .rock files are separate
19:38:29ioboundi just need to find out what instruction is illegal
19:38:37Slasheriin generally yes
19:38:37ioboundand it played so well in the win32 simulator :(
19:38:47Slasherinormally code is not in iram
19:39:11ioboundis that those IATTR things i've seen specified on some functions and data structures?
19:39:30Slasheriyep, excactly
19:39:33 Join Maxime [0] (
19:39:47ioboundok, so that means the error was not in my code? that's too weird.
19:40:04Slasheriit doesn't mean that.. =)
19:40:20iobounddo you know how I can find out exactly which instruction it is? i'm not afraid of objdump and 68000 assembly, although it's been a while since i did anything with it
19:41:04_FireFly_Slasheri: i have just tested it i can't here that the player crossfades if crossfade is set to shuffle and shuffle is off
19:41:10Slasherihmm, you could try the m68k-objdump but that is not the easiest way to figure it out..
19:41:36Slasheriiobound: probably you should check the plugins map-file
19:41:52ioboundyeah but nothing in the map file looks anything remotely like "1000115a"
19:41:58Slasheri(apps/plugins/ in the compiled directory)
19:42:13ioboundyeah i'm looking at it
19:42:17leftrightFirefly: try it with 3 sec fadeout and fade-in only
19:42:21SlasheriHmm.. then it could be something like incorrect pointer etc.
19:42:27ioboundit's all "0x32f41362" and so on, nothing like "0x10001234"
19:42:29_FireFly_leftright: i had it set to 6sec each
19:43:00Slasheriiobound: just try debugging where the problem occurs.. use logf for example
19:43:03ioboundany debugging hints in general? should I just add a lot of rb->splash calls? :)
19:43:05leftright6 secs ought to do it :-)
19:43:18ioboundwhere does logf log to?
19:43:28Slasherilogf is generally better, because it goes to the remote lcd
19:43:29_FireFly_the only thing which i have noticed is that the played-time-display stops ca. the amount of choosen fade-out time
19:43:42ioboundoh, ok. i guess i have to dig up my remote, i'm not even sure if it has an lcd
19:43:45Slasheriand you can see many logf messages at once
19:44:10ioboundtoo bad one can't debug across usb :)
19:44:26ioboundthis is gonna take all night, rebooting into iriver firmware, uploading a new .rock file, rebooting back to rbx etc
19:45:16_FireFly_iobound: why booting into iriver fw ??
19:45:24Slasherireally? that's not so hard.. i have a script so that i just need to plug usb cable, run make && ./ and unplug :)
19:45:25_FireFly_the bootloader has an usb-mode
19:45:43ioboundfirefly it does? i thought it wasn't implemented for h300
19:45:50_FireFly_oh ok
19:45:54_FireFly_that's true
19:45:54Slasheriah, h300 might be different..
19:46:02ioboundoh, sorry i forgot to mention that :)
19:46:59 Quit B4gder ("time to say moo")
19:47:03ioboundanyway, the .map file says .text starts at 0x324f0000, maybe that get mapped to 0x10000000 ? in which case i should look for 0x324f115a
19:47:33Slasheribut there might be some variables in iram or something like that
19:48:05ioboundit would make sense, "duh_render" is at 0x32f41142, just below the crash address, which makes sense because that's the methods called when it crashed
19:49:08Slasheriah.. that could be possible, i am not really sure what the address codes means that rockbox prints out
19:49:47ioboundcalculating 115a-1142 gives 18, and objdump on libdumb.a shows duh_render starting at 0x1a in its compile unit, so addin 18 to 1a yields 0x32, and at that address, "clrl %d0"
19:50:12ioboundthat just clears register d0, doesn't it?
19:50:20ioboundthat couldn't possibly be an illegal instruction?
19:50:25ioboundi guess i'm on the wrong track after all
19:50:27Slasheriyes, interesting
19:51:31Slasheristill i would say that logf or splash works better to locate which part of the code crashes
19:52:31leftrightDark Side Of The Moon is not a nice listening experiance with the gap(less) modified
19:52:47DreamTactix291nor is any other gapless album
19:53:07DreamTactix291The Human Equation sounds really bad with added gaps
19:53:24ioboundhm, other module files just makes it loop
19:53:40ioboundit kind of sounds like it just repeats over and over a small audio buffer
19:53:44_FireFly_why use mp3 when ogg has gapless "onboard" ;)
19:53:55leftrightok, who broke it no beer for yhat person
19:54:11DreamTactix291well even the iriver firmware gapped vorbis :P
19:54:11 Quit dropandhop ()
19:54:27_FireFly_the iriver fw != rockbox ;)
19:54:28DreamTactix291i've been enjoying gapless for a good long while now so i'm happy
19:54:33DreamTactix291i know that
19:54:39DreamTactix291i've been using rockbox for many months
19:54:41leftrightit was damn good a few days ago
19:54:53DreamTactix291leftright: they'll fix it
19:55:11_FireFly_afaik does gapless-playback of mp3s only work when encoded with LAME
19:55:26DreamTactix291they have proper delay and padding values in the LAME tag
19:55:33leftrightyep, mu whole collection is LAME@aps
19:55:42DreamTactix291then you're ok
19:55:47iobounduhm, does pcm_play_data() work at all from plugins?
19:55:48 Join Febs [0] (
19:55:56 Join JonSenior [0] (
19:55:57ioboundit seems rockboy is the only thing using it, and even there, the code is #if 0'ed out
19:56:02DreamTactix291mp3s for other people that i make are LAME 3.97b2 -V2 −−vbr-new
19:56:28leftrighthaven't tried 3.97 yet, is it much faster
20:00:03 Quit CheeseBurgerMan (Read error: 110 (Connection timed out))
20:03:34ioboundh300 people, is there any way to exit USB mode without having to unplug the usb cable?
20:06:25 Join iobound_ [0] (
20:06:36 Join dpassen1 [0] (n=dpassen1@resnet-233-61.resnet.UMBC.EDU)
20:07:43Sanprobaly not
20:10:27 Join JoeyJoeJo [0] (
20:11:46JoeyJoeJodoes anyone here know of any good DIY mp3 players? I found out about the echomp3 player, but I was wondering if there were any others
20:11:58Slasherihmm.. now i found many bugs from the playback when i tried gapless ogg files which previously worked fine
20:12:28 Join iobound__ [0] (
20:13:05 Join t0mas [0] (
20:14:02 Quit einhirn ("Miranda IM! Smaller, Faster, Easier.")
20:14:16 Join lamed [0] (
20:15:43 Quit JoeyJoeJo ("CGI:IRC (EOF)")
20:17:15 Join Wett [0] (
20:19:10 Join frederic_ [0] (
20:20:04 Quit lamed ("CGI:IRC (EOF)")
20:22:47iobound__so, does anyone know if using pcm_play_data() in plugins is OK on the h300?
20:24:26 Join webguest43 [0] (
20:25:40iobound__or if there is any limits on the size of a dma transfer?
20:26:16t0masI don't know the limits, but I'm sure there are some
20:26:45iobound__rbsound.c has a #define DMA_PORTION 1024 and seems to limit itself to that.
20:26:50iobound__maybe that's why things are crashing for me
20:27:33leftrightSlasheri, preglow mumbled something about chopping off 500 milliseconds for something or other when implementing pitch control
20:28:25 Quit iobound (Read error: 110 (Connection timed out))
20:30:16Slasherileftright: Hmm, that could be the reason why gapless no longer seems to work..
20:30:55 Quit iobound_ (Read error: 110 (Connection timed out))
20:32:55 Quit frederic_ ("Chatzilla [Firefox 1.5/undefined]")
20:35:19 Join Gibbed [0] (
20:35:20 Quit Rick (Nick collision from services.)
20:35:30 Nick Gibbed is now known as Rick (
20:37:00 Quit webguest43 ("CGI:IRC (EOF)")
20:49:41mirakiobound__: no there is no way
20:49:59iobound__for stopping usb, you mean?
20:50:08 Nick iobound__ is now known as iobound (
20:50:12mirakthis sucks
20:52:21mirakiobound: also it doesn't charge battery when in usb mode
20:52:36mirakit's a pity since in this mode the hard drive always run
20:53:26ioboundit is.
20:56:54 Join Maxime` [0] (
20:56:54 Quit Maxime (Read error: 104 (Connection reset by peer))
21:01:00 Join Bgr [0] (n=Bager@
21:02:05 Join RotAtoR [0] (
21:02:51 Join solexx [0] (
21:06:00 Quit iobound ("Leaving")
21:07:11***Saving seen data "./dancer.seen"
21:07:55 Join San||Away [0] (
21:15:24 Quit Bger (Read error: 110 (Connection timed out))
21:17:14 Quit Wett (Read error: 110 (Connection timed out))
21:18:23 Quit solexx_ (Read error: 110 (Connection timed out))
21:23:41 Join Bger [0] (n=Bager@
21:26:39 Quit San (Read error: 110 (Connection timed out))
21:27:59 Quit San||Away (Read error: 110 (Connection timed out))
21:32:45 Part leftright
21:32:45 Join iobound [0] (
21:33:48ioboundslasheri, you know, i just looked in, and 1000115c is "DMA0" - the interrupt handler which gets called after each dma transfer. 1000115c is 2 bytes higher than 1000115a (my crash location)
21:36:07 Join iobound_ [0] (
21:36:18iobound_1000115a is "rts", the end of lcd_remote_bitmap_part()..
21:36:35 Join linuxstb [0] (
21:37:38linuxstbiobound: Have you tried disabling the audio output in your plugin?
21:38:07iobound_no, what would the point in that be?
21:38:25linuxstbFor testing. You're saying it crashes.
21:38:42linuxstbI'm just asking if it works without audio output.
21:39:00iobound_well, it runs until play_pcm_data() and a few callbacks to the get_more() callback
21:39:18iobound_also, it works with audio output in the simulator
21:39:45iobound_and the "I04: Illegal Instruction at 1000115a" seems to point to the "DMA0" interrupt handler
21:39:47linuxstbOK. So you've already narrowed the problem down to the audio output.
21:40:06iobound_well, hm.
21:41:03iobound_i think i'm going to try to write a main loop that just renders into a buffer and does nothing with it, as you suggested.
21:41:35linuxstbIf I was you, I would write the data to a file and check that the decoding is OK.
21:41:45iobound_that's a great idea, thanks
21:41:53 Join webguest43 [0] (
21:42:12 Quit Bgr (Read error: 110 (Connection timed out))
21:42:23iobound_i'll try it later tonight
21:43:25linuxstbDo you know if dumb is endian-safe? The iriver is big-endian, the win32 simulator is little-endian.
21:44:13iobound_i'm not sure. i'm aware iriver is big-endian. i ported (well, copied and pasted, more like) exp(),floor(),log() from freebsd's math library and made sure to modify #if ENDIANNESS == BIG_ENDIAN to #ifdef ROCKBOX_BIG_ENDIAN
21:44:15Slasheriiobound: what are you doing inside the get_more() function? You should not call _any_ system calls from there
21:44:56Slasheribut now i have to go, cu tomorrow
21:44:58iobound_i'm calling duh_render() into my char[] buffer
21:45:04iobound_ok, seeya, thanks for you help so far
21:45:38iobound_i'll try rendering to a file, if that works then I'll know it's not endianess-trouble or anything in dumb
21:45:44Slasheriyou shouldn't do that
21:45:53Slasheriit probably causes it to crash
21:46:12Slasheriyou have to render the file outside and buffer it somewhere
21:46:53iobound_you think it causes the file system to be corrupted?
21:47:16Slasherino, rockbox crashes because you are making system calls inside ISR handler (get_more())
21:47:32Slasherino sleep, disk access or anything like that is allowed inside that function
21:47:55iobound_ok. but would it cause an "illegal instruction"?
21:48:02iobound_is it because rbx is not re-entrant, or something?
21:48:07Slasheriit can cause anything
21:48:14iobound_why? :)
21:48:24Slasherino, isr functions can never have system calls
21:48:32Slasheriyou have to implement buffering
21:48:35Slasheribut now time to go :) ->
21:48:39iobound_ok, thanks again
21:49:47 Join Rob2222 [0] (
21:51:08 Join Wett [0] (
21:51:10linuxstbiobound_: Do you know how much memory dumb is typically using?
21:51:30 Join iobound__ [0] (
21:52:36iobound__hi, i'm sorry, my dsl connection is really unstsable
21:52:45 Join TCK [0] (
21:53:02iobound__i'm not sure how much memory it uses but I can find it out pretty fast, I just need to re-enable the debug statements in my malloc()/realloc() implementation
21:55:46perplexityOk, another silly newbie question.. when compiling for the simulator (X11) how can I easily debug some data to the console? DEBUGF is not defined.. if I enable logf the sim segfaults, and I can't access printf from within a codec.. (it's a codec I'm debugging)
21:56:08 Quit iobound (Read error: 110 (Connection timed out))
21:56:18 Quit linuxstb ("CGI:IRC (Ping timeout)")
21:56:28iobound__hm, DEBUGF works fine for me in the win32 sim
21:56:31 Nick iobound__ is now known as atomic (
21:56:33perplexityheh.. nevermind.. I have been working on it for an hour.. I ask the question and 10 seconds later I get it working..
21:56:35 Nick atomic is now known as iobound (
21:56:41 Join Maxime [0] (
21:57:03perplexityDEBUGF is not working for me.. but I get compile warnings on printf, however it resolves at runtime and works fine..
21:58:24ioboundare you working on a plugin or what?
21:58:44perplexityyes.. playing with different mdct algorithms for Tremor
22:00:58merbananperplexity: integer ones ?
22:01:00 Join Bgr [0] (n=Bager@
22:01:36 Quit Kohlrabi (Nick collision from services.)
22:01:41 Join Kohlriba [0] (
22:01:52 Join linuxstb [0] (
22:02:15perplexityyes merbanan
22:02:24ioboundlinuxstb, i just changed my plugin to write to disk and not use the pcm_() calls
22:02:39ioboundit renders perfectly (although everything is byteswapped from the file written by the simulator)
22:03:21 Join DangerousDan [0] (
22:04:16ioboundlinuxstb, endianness in the output files considered, the win32 and the h300 produces bit-identical files
22:04:39 Quit TCK- (Read error: 110 (Connection timed out))
22:05:27merbananperplexity: ffmpeg ac3enc.c has a fixedpoint mdct, have you tried that one ?
22:06:03 Join Moos [0] (
22:06:43ioboundi'm amazed that freebsd's math library worked without nearly any tweaking at all on the h300
22:07:00linuxstbiobound: That's a good sign. How fast does it appear to be? i.e. is it realtime?
22:07:01perplexityNot yet merbanan, at the moment I'm working on shoehorning the ifft from libfaad in for testing..
22:07:27ioboundno, it appears to be very slow, but that might be because i'm writing 1kb blocks at a time to the disk?
22:07:52ioboundi think it spent like 10 seconds to generate a .raw file with about 2.5 sec of audio
22:08:00linuxstbNo, disk output doesn't slow decoding down that much.
22:08:08ioboundnot even when you write small blocks?
22:08:12 Quit iobound_ (Read error: 110 (Connection timed out))
22:08:39linuxstbI would be surprised. Try commenting out the writes and seeing how long it takes.
22:08:58linuxstbMy guess would be that it is almost the same time.
22:09:10ioboundis there a time() type function i can use
22:10:03linuxstbrb->current_tick can be used for that.
22:10:24ioboundok, and there's HZ ticks in a second?
22:10:55linuxstbi.e. starttime=*rb->currenttick;
22:11:11ioboundyeah, i've got it coded up already :)
22:11:13linuxstbYes - HZ is 100 on all platforms.
22:11:51ioboundbooting iriver firmware for usb mode..
22:12:33iobounddisconnecting usb..
22:12:49ioboundrebooting to rbx
22:12:58^BeN^How i can open sudoku board on h300 rockbox ?
22:13:14 Quit Maxime` (Read error: 110 (Connection timed out))
22:13:44ioboundaaand i must have done something wrong, it said 0 ticks
22:14:49ioboundyeah, i should have paid more attention to linuxstb, i didn't notice current_tick was a pointer
22:15:07merbananperplexity: are you sire it is a ifft you need ?
22:16:13 Join Massa [0] (
22:17:13merbananperplexity: I made a fixed point mdct but it's using 64bits ints, can be found here
22:17:25 Quit Bger (Read error: 110 (Connection timed out))
22:20:07perplexitythanks merbanan, I'll add it to my list of things to look at.. the Tremor inverse mdct is fixed point in any case, but I'm looking at any extra gains I can make with algorithmic changes
22:20:46ioboundthe results are in: #ticks spent generating 2.972 seconds of 16bit 44.1khz stereo audio: 2nd_pm.s3m: 600ticks snowb.xm (small chiptune) 3200ticks (!) adrenaline.mod 900ticks
22:21:12ioboundso it seems it's somewhere between 3% and 50% of realtime
22:21:26linuxstbThat's normal for a first version of a codec.
22:21:40ioboundit really seems to depend on the tune
22:22:23ioboundalso, this is with default resampling/interpolation.
22:22:51ioboundi think it uses "best" by default, and resampling uses a lot of floats, i think.
22:23:02ioboundalso i really need to get rid of malloc() calls in the render loop....
22:23:22 Quit Bgr (Read error: 110 (Connection timed out))
22:25:05 Join ashridah [0] (
22:28:28preglowiobound: you got it running on taget?
22:28:45preglowresampling currently uses tons of floats
22:28:47ioboundbut not with audio output, only disk writing
22:29:51ioboundit crashes after the dma thing or whatever it is calls get_more() a couple of times
22:30:03ioboundbut with disk writing, it produces an identical file as the win32 simulator
22:31:21preglowwhich is good news
22:31:30preglowthe use of floats will probably slow stuff down a lot now
22:32:38preglownot "probably"
22:32:42preglowit does, and probably by tons
22:32:45ioboundespecially the pow(), exp() and log() calls
22:32:57preglowyou've implemented those properly?
22:33:06ioboundyeah, i took them from freebsd's libm
22:33:10preglowgood god
22:33:13preglowthey should be Very slow
22:33:23ioboundthey were dead easy to port
22:33:31preglowwhat you've done is great
22:33:46preglowi might want to work on optimising it
22:34:29ioboundme too but i'm getting really tired of the compile/run/debug cycle, with all this rebooting and unplugging and replugging of usb cables :)
22:34:45 Quit thegeek_ (Read error: 104 (Connection reset by peer))
22:34:48preglowhaha, you don't say
22:34:50preglowi've gotten used to it
22:35:21ioboundi can send you the source if you want to take a hack on it
22:35:26preglownot now
22:35:31preglowi'm busy preparing for a longish trip
22:35:56preglowi'm going to another city for a couple of months to work
22:36:02preglowwhich might also mean a lot of rockbox coding
22:36:16ioboundnot all bad then :)
22:36:19linuxstbpreglow: I was worrying there for a moment...
22:36:28linuxstbtrips normally mean no coding.
22:36:34preglowthis wont will probably mean more
22:36:41preglowthere's nothing to do up there but drink beer and code
22:36:47 Join thegeek [0] (
22:36:55ashridahas long as there's net access
22:36:57preglowand read, that is
22:37:01preglowashridah: there is, 100mbit
22:37:01 Join Bger [0] (n=Bager@
22:37:08linuxstbThose two things don't normally mix very well though.
22:37:09ashridahcan i come? :)
22:37:10preglowi'll have my own office and will probably be camping there a ton
22:37:11solexxpreglow: thanks for fixing the issue with mono files that quick
22:37:11 Quit webguest43 ("CGI:IRC (EOF)")
22:37:22preglowsolexx: no problem, it was a quick one
22:37:42preglowlinuxstb: good luck i wont try mixing them, then, heh
22:37:45solexxnevertheless - I didn't even need to talk to you :)
22:37:56preglowsolexx: nah, amiconn made me aware of it
22:38:41preglowbut no, i'll be dsp coding for work as well
22:38:47preglowlet's hope i don't grow tired
22:39:54perplexitywoooooooooooooooooooooooooooooooo have an inverse fft working in Tremor.. now to try it on the device :)
22:40:07preglowperplexity: excellent!
22:40:23_FireFly_what is a inverse fft ??
22:40:38perplexitywell, it's what you get if you stand on your head and look at an fft ;)
22:40:41preglowperplexity: you mean inverce mdct, right?
22:40:58preglowan inverse fft wont do you much good by itself
22:41:11_FireFly_perplexity: lol and now waht is a fft or mdct
22:41:14perplexityWell.. it's an inverse mdct implemented using an ifft.. but yes..
22:41:22preglowperplexity: that's what i wanted to know, great
22:41:29preglowperplexity: truly excellent
22:41:31perplexityIn any case I have intelligable music coming out my speakers :)
22:41:46perplexityOh, it's an ugly, ugly hack.. but we'll see if I can't make it work
22:41:47preglowwill be fun to see which way boost rate swings now
22:42:10preglow_FireFly_: mdct is the transform function used in almost all lossy codecs
22:42:25perplexityI'm going down to my car to grab my iriver now.. then I'll probably fall into bed.. 10 pints of Stella and 2am are neither conducive to writing good code :)
22:42:47preglow10 pints of stella would have me tired of lager for a week
22:44:01ashridahmmm. stella
22:44:15_FireFly_preglow: if i understand it correctly then with this code it would be possible to do encoding in rockbox ??
22:44:39 Quit DreamTactix291 ()
22:44:47preglow_FireFly_: no
22:45:09_FireFly_ups yes have just found what an imdct is
22:45:53preglowused to drink tons of stella when in england, but then i discovered bitters
22:46:18ashridahtraitor! :)
22:46:47perplexityIt was all they had at that pub that was drinkable on tap.. the only other alternative was Fosters and I'd rather drink cats urine.. (Which is in fact part of the Fosters brewing process I'm sure)
22:47:00preglowthe entire beer market in norway is based on lagers, so i tend to drink anything else when i can
22:47:23preglowperplexity: it is in fact fermented cat's urine
22:47:32preglowwith slight hopping
22:47:41 Join DreamTactix291 [0] (
22:47:47perplexityI thought as much.. we take the finest tomcat, apply a light press and squeeze .....
22:48:06perplexityyou will never find an Australian with any taste that drinks that stuff
22:48:20ashridahperplexity: hahaha. i still can't believe the rest of the world takes our cat urine as a brand of beer.
22:48:27ashridahyou foreigners are crazy :)
22:48:31preglowi don't find stella that much better, but hey...
22:48:34perplexityI know.. it's called marketing :)
22:48:53ashridahno one, and i mean *no-one* ever drinks fosters here
22:48:58 Quit Cassandra (Read error: 104 (Connection reset by peer))
22:49:00perplexityI find stella a little rough.. personally I'll go a Kilkenny if I have to drink beer, otherwise it's onto the scotch
22:49:08perplexityI know ashridah.. it's awful stuff
22:49:39ashridahkilkenny's more of an ale than a beer, isn't it?
22:49:56perplexityMore of a Stout actually.. I like beer and red wine that you have to eat with a fork
22:50:19perplexityWynns Coonawarra 1998 Cab Sav is also a fav of mine
22:50:21ashridahhmm. didn't seem strong enough to be a stout, ah well.
22:50:31preglowstouts are nice
22:50:33*ashridah shudders
22:50:44preglowbut you can't get much better than an ipa
22:50:46preglowi like my hops
22:51:55preglowi believe kilkenny is categorised as an irish red ale
22:51:56 Quit _FireFly_ ("Leaving")
22:52:02perplexityOk, just replaced my vorbis.codec with my ugly version.. it skips on stuff the other codec did not skip with, but it plays.. it sounds good and it has _no_ optimisation at all..
22:52:24preglowperplexity: point is that an ifft is easier to optimise than the imdct in vorbis now
22:52:40perplexityIndeed.. I think I just made a mess in my trousers..
22:52:55preglowa healthy sound of excitement, now go change
22:53:07preglowi suck
22:53:18perplexitylol.. time for this little black duck to go to bed.. but I can go to bed with a smile on my dial :)
22:53:30preglowdo that, good night
22:53:36perplexityta for the idea on that one preglow.. lets hope it pays off :0
22:53:41 Join akaidiot [0] (
22:59:40ioboundpreglow, should i send my patches somewhere (after cleanup)? the devel mailinglist, for example?
22:59:43ashridahIs it just me, or does saving a playlist not actually show up in the filesystem tree if you have directory cache on?
22:59:59ashridahat least, not immediately.
23:00:10ashridahiobound: there's a patch tracker on sourceforge
23:01:20ioboundshould i just send a cvs diff -u? what about new and removed files? should i use regular diff? are there instructions somewhere?
23:02:20ashridahi remember someone telling me about a funky way to get cvs to allow you to 'add' new files even if you don't have commit access, but damned if i can remember what it is
23:03:09ashridahmight even have been Bagder
23:05:27ioboundunfortunately, i couldn't makefile myself out of a paper bag, so i've put almost all the code in one huge .c file
23:05:53ioboundwell, not that huge, 700 lines, but it would be nice to put the maths stuff in a separate file
23:06:25linuxstbYou could add it to libdumb itself
23:06:55ioboundand the malloc/realloc stuff
23:07:10ioboundi could, but these things would probably be nice to have for other plugins too
23:07:14***Saving seen data "./dancer.seen"
23:07:21ioboundmaybe they could go into a "nano-libc for plugins" thing
23:07:44preglowiobound: nah, please don't put it in a separate file, it'd be nice to not have to delete files when i/someone ports it to fixed point
23:07:59ioboundok then
23:08:07 Join San [0] (
23:08:32ioboundi have one dumb.c in plugin/ and a "dumbsupp.h" (dumb support methods) there as well, which is included from dunb.h in codec/ (ugly, i know)
23:09:12 Join hellsaviour [0] (
23:11:00preglowwhat support functions do you currently have?
23:11:17 Part JonSenior
23:11:47hellsaviourI am on Mac OS X and a few days ago I loaded rockbox with a bootloaded firmware from Bger and then had to get the .rockbox folder and such onto my h340 on my PC, because OsX recognises it as a hidden folder. Well, just now I found a way of showing all the hidden files/folders on OsX so that mac user won't have to use a PC to update RB. Its listed here on this website:
23:11:57ioboundyou can see it here, and
23:12:10hellsaviourI thought that the webmasters for might find this helpfull
23:13:02preglowiobound: dumb uses _DOUBLE_ floating point precision???
23:13:20ioboundi think so. at least it uses pow() and exp() and those work with doubles
23:13:24preglowdamn, small wonder it's slow
23:13:31preglowahh, right
23:13:33preglowof course
23:13:35preglowit doesn't use powf
23:13:48preglowwell, it might still just use floats
23:14:20ioboundthe implementation of exp() and stuff is pure voodoo it seems, full of strange magic hex constants and stuff
23:14:29preglowyeah, i'd expect so
23:16:08 Join leftright [0] (
23:16:21 Quit hellsaviour (" - for mac users")
23:17:23leftrightpreglow, I'm busy listening to DSOTM right now, and its not gapless as it used to be
23:17:53leftrightthere is definately a difference
23:18:36 Quit ashridah ("uni, must reenrol.")
23:23:54 Join DMJC [0] (
23:24:05preglowleftright: from which build?
23:24:09preglowwhich was the last that worked?
23:24:37 Quit Massa ("ChatZilla 0.9.61 [Mozilla rv:1.7.7/20050414]")
23:28:19leftrightpreglow, latest build
23:28:55leftrighterr, I'll get you a build which works, one sec
23:29:13 Quit RotAtoR (Read error: 113 (No route to host))
23:29:47 Join aliask [0] (
23:31:05 Quit webguest38 ("CGI:IRC (EOF)")
23:50:59leftrighthmm, preglow, I went back a month and those builds also exhibited artifacts at the track transitions, but maybe my memory serves me incorrectly but I do remember gapless being more seamless
23:51:28preglowi did do a change relating to gapelss perhaps a month back
23:51:48preglowbut it shouldn't have affected it destructively, just made more files work
23:52:23leftrightI remember us all being ecstatic a while back that it was damn near perfect
23:53:06preglowwell, i have no time to fix it now
23:53:09leftrightnow its distracting
23:53:21preglowand i don't even know what's wrong
23:54:14leftrightpair of scissors to remove the gap :)
23:55:53ioboundwho's the clever dude accessing /temp and /home on my webserver ;)
23:56:37 Join webguest03 [0] (
23:57:11 Quit drumRBoy320 (Read error: 104 (Connection reset by peer))
23:58:22leftrightwell thanks for all the hard work, later
23:58:27 Part leftright

Previous day | Next day