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 2007-07-02

00:00:29 Join ShadowSponge_ [0] (
00:00:29 Quit Quelsaruk (Read error: 104 (Connection reset by peer))
00:01:02Lloreanamiconn: So it's a log of a fraction, with one unknown parameter?
00:02:08 Part chrisjs169 ("Konversation terminated!")
00:03:05 Nick swolbhsub is now known as bushblows (n=bushblow@
00:03:36 Quit ShadowSponge (Nick collision from services.)
00:03:44 Nick ShadowSponge_ is now known as ShadowSponge (
00:04:35 Quit davina (Remote closed the connection)
00:05:02 Join relaxed [0] (n=relaxed@unaffiliated/relaxed)
00:06:31 Join linuxstb [0] (i=5343d4aa@rockbox/developer/linuxstb)
00:07:25linuxstbsaratoga: Around?
00:07:28 Quit merbanan ("Leaving")
00:07:54scorcheBagder: about ready to go to sleep?
00:08:28Bagderyes indeed...
00:08:56 Nick bushblows is now known as swolbhsub (n=bushblow@unaffiliated/bushblows)
00:09:05*scorche makes a note to talk to Bagder tomorrow
00:10:04 Join MrKeuner [0] (
00:10:11MrKeunerhi, there.
00:10:23MrKeunerWHat files should I update when upgrading rockbox?
00:10:39scorchejust unzip the folder to the root
00:10:48MrKeunerI do not really want to do all the settings again
00:10:53scorcheyou wont
00:10:54DerPapstdionoea: ping
00:11:03scorcheif you want to make sure you wont, save teh settings to a file
00:11:20MrKeunerscorche: also there are wps files etc
00:11:28 Quit ShadowSponge (Read error: 104 (Connection reset by peer))
00:11:44DerPapstdionoea: i have made a small patch for reversi to optimize it's usage (mostly on 3G) iPods
00:11:49MrKeunercan't I just copy rockbox.ipod
00:11:57*amiconn guesses the function caluclates ln2 of a fraction
00:12:26bluebrotherMrKeuner: no. You need to replace the other files too
00:12:49 Join \sic [0] (
00:13:02 Part Eimann
00:13:23MrKeunerso no way. I have to recopy the wps and related files and the settings files?
00:13:30saratogalinuxstb: I'm here
00:13:37bluebrotherbut since the settings are stored in a file (/.rockbox/config.cfg) you can simply store that somewhere else
00:13:49DerPapstMrKeuner: you're settings and wpses won't get overwritten
00:13:55scorcheas we said, they wont....yeah
00:13:59bluebrotherwhy don't you just extract the new build over the old folder and overwrite all files?
00:14:09bluebrotherthat's the way you usually are intended to update.
00:14:31MrKeunerI didn't know that unzip did not remove all the contents in a folder and unzip on them
00:14:35MrKeunerOK I'll do that
00:14:40dionoeaDerPapst: why is that 3G specific ?
00:15:09DerPapstdionoea: well it's not entirely 3G specific ;)
00:15:19linuxstbsaratoga: I've been doing some more asf work today - I've more or less decided to abandon libasf itself, and write a simple parser instead. libasf works by reading everything into memory structures first, and then parsing them. Whereas, it's better for Rockbox to parse as it goes along.
00:15:24dionoeashouldn't you just change the ifdef to "all ipods" ?
00:15:35DerPapstdionoea: it makes the layout a bit more consistent to all iPods
00:15:39saratogalinuxstb: that sounds like a lot work
00:15:50linuxstbsaratoga: But I've got the basics of a "get_metadata" function working already.
00:16:04DerPapstdionoea: that isn't nessesary imo.
00:16:16linuxstbsaratoga: Not really, asf is in fact a very simple format, and there are lots of objects we can simply ignore and skip.
00:16:22dionoeawhy are you changing it only for the 3G then ? (the wrap mode)
00:16:35saratogaI'm working on integerizing what you gave me
00:16:42saratogahow much of that is likely to change?
00:16:46DerPapstdionoea: the settings are ok for all other iPods except 3G that's why i limited to 3G only
00:17:05linuxstbsaratoga: You should assume all the asf stuff will change - best if you just stick to the wma decoder itself.
00:17:06dionoeaah ok ... if you say so :)
00:17:11dionoeaDerPapst: i'll apply the patch
00:17:25DerPapstdionoea: but you can change that to all iPods. don't make a big difference to me :)
00:17:32saratogawhat about the parts of the decoder that touch libasf?
00:17:34*dionoea needs a clean rockbox checkout
00:17:35saratogastuff like wma_decode_init()
00:17:36DerPapstdionoea: thanks ;)
00:17:54saratogacan I assume you'll still be passing in the same arguments?
00:17:59linuxstbsaratoga: Well, that's trivial to change.
00:18:12linuxstbI'll need to give you that information, so yes.
00:18:29linuxstbI'll probably keep that same wfx struct.
00:18:31saratogamostly i'm trying to sync the stuff I had with the older version of wmadec.c
00:18:40saratogathere was a fair amount of drift
00:20:04 Quit MrKeuner ("Ex-Chat")
00:21:20 Quit jhMikeS (Nick collision from services.)
00:21:32 Join jhMikeS [0] (n=jethead7@rockbox/developer/jhMikeS)
00:21:42dionoeaDerPapst: done.
00:21:49DerPapstdionoea: the 3G iPods has an entirely different button layout compared to all the other iPods. so one can't really use rew, menu, play, fwd as left, up, right, down.
00:22:00DerPapstthanks :)
00:22:41dionoeaI should be the one thanking you :) the keymap was horrible on my ipod too
00:22:50saratogalinuxstb: is there some linux app that people use to reformat the layout of code?
00:23:04linuxstbProbably, but I don't know it...
00:23:04saratogaits annoying to merge changes when teh lines, tabs and brackets are all moved around
00:23:09Bagdergnu indent
00:23:19DerPapstdionoea: hehe
00:23:28 Quit petur (Read error: 104 (Connection reset by peer))
00:23:37 Join petur [0] (n=petur@rockbox/developer/petur)
00:23:45Bagder(for source code (re-)formatting I mean)
00:26:54micklebyn00b, again alas: would like ide to run off and find definition for me, e.g. with "int fd=rb->creat("someFile");" find file & definition of creat(); Was spoiled by years using VisC++; should I expect Eclipse or xEmacs to pamper me so?? Do folks use vi? Thou gracious, saged masters
00:28:26 Join thegeek_ [0] (
00:28:36dionoeaDerPapst: you forgot the | BUTTON_REPEAT for the ALT keys :/
00:28:50dionoeaDerPapst: or was that on purpose ?
00:28:58saratogaBagder: thanks indent worked great
00:29:59 Join Robin0800 [0] (
00:30:05dionoeayay! i score 1520 points!
00:32:04markundionoea: ha, that's nothing..
00:32:19markunalthough I don't remember my high score
00:32:46dionoeanext time i'll remember to compile before commiting ... not after.
00:33:38 Quit x1jmp (Remote closed the connection)
00:37:32micklebyvery well, my betters. I shall grep like the n00b I am
00:41:37DerPapstdionoea: yes. you're right ;) sorry
00:41:51dionoeashould be fixed
00:42:03amiconnwow!! :D
00:42:27amiconnThe mini g1 boosts and unboosts playing aac, without hanging!
00:42:30*bluebrother awaits some good news from amiconn
00:42:35bluebrothercool :)
00:42:49amiconnSo now this needs testing on other PP5020s
00:43:02dionoeaGood night ppl (/me can't wait to read tomorrow's commits from amiconn)
00:43:12amiconnIt's a bit hackish atm
00:43:30amiconn(I'll redefine the CPU for the targets which are actually PP5022)
00:44:04amiconnPart #2 also isn't done yet (saving more power), I just applied part #1 (the clock magic for PP5020, and relock wait for PP5022)
00:44:15amiconnI'll test the relock thing next
00:44:36dionoeawould part 1 save power ? or is it stability issues only ?
00:44:43DerPapstdionoea: good night :D
00:44:54 Quit \sic (Read error: 110 (Connection timed out))
00:44:58amiconnIt will save power by the fact that cpu scaling works now
00:45:36amiconnCPU scalin gis disabled in svn for H10, and I had to disable it for mini g1 locally to make it even boot
00:46:18dionoeaah ok.
00:46:27amiconnThe relock thing might also save a bit of powe
00:46:48bluebrotherI can test on mini 2G if that helps ...
00:46:59amiconnThat's 5022, and I have one myself
00:47:07amiconnI would be most interested in H10 testers
00:47:21bluebrotheryeah −− just in case you're interested in more testers.
00:49:09amiconnlinuxstb: I even removed the TIMER kludge for PP5020, and it works _stable_
00:49:14*DerPapst could only donnate a PP5002 and a PP5022C
00:49:51DerPapstamiconn: nice work :)
00:50:02amiconnI guess that we can even run at 80MHz boosted with proper voltage handling
00:51:26amiconnHmm, I even have a suspicion what the PLL status magic actually does...
00:51:59*amiconn added PLL_CONTROL and PLL_STATUS display to 'view i/o ports'
00:52:59amiconnThe thing boosts/unboosts roughly once per second. No hang so far. Will listen to the whole album...
00:53:32 Quit mirak (Remote closed the connection)
00:53:37 Join \sic [0] (
00:55:53 Quit Entasis (Read error: 104 (Connection reset by peer))
00:58:51 Join webguest17 [0] (i=cf761d9c@gateway/web/cgi-irc/
00:59:07amiconnlinuxstb: pingggggg
01:00:07 Quit webguest17 (Client Quit)
01:02:42 Quit lazka (Remote closed the connection)
01:03:04saratogaamiconn: so you're sure that rockbox wasn't handling voltage correctly on the PP players?
01:03:16***Saving seen data "./dancer.seen"
01:03:17amiconnIt still isn't....
01:04:47 Quit pontape ()
01:05:47 Quit lee-qid ("aufwiederbyebientotsayonara")
01:06:09 Join lee-qid [0] (
01:07:03 Quit lee-qid (Client Quit)
01:12:08linuxstbamiconn: Do you have a patch?
01:12:35amiconnHmmm :/
01:12:50amiconnHalf an hour ago I would have had one
01:13:13jhMikeSamiconn: any change in the behavior of core locking? did you happen to pick up anything about the memory controller such as defining noncacheable areas or controlling what get flushed?
01:13:24amiconnNow I'm in the process of converting the appropriate targets to use a PP5022 define, and my tree probably doesn't compile
01:13:56amiconnjhMikeS: I didn't look at the core locking at all anymore
01:14:11amiconnI just want to solve the clock & power issues
01:14:49amiconnpcm-pp.c is strange :\
01:14:50jhMikeSby the TIMER kludge you were speaking of the |= when scaling?
01:15:10amiconnMy album is still playing btw
01:15:12*jhMikeS is redoing that a bit right now
01:15:14linuxstbWill you be treating the PP5021/22 the same?
01:15:22amiconnThere is no 5021
01:15:32amiconnThere is only a PP5022B and a PP5022C
01:15:55amiconnThe case print might be different, but that's not significant for the code
01:16:45amiconnjhMikeS: Didn't you do some work in pcm-pp.c?
01:16:57amiconnI don't understand the ifdefing
01:17:21jhMikeSI'm redoing the pcm level API a bit so that it is well locked again the interrupts handler for all players and the states are proper
01:17:47jhMikeSwhich #ifdefing in particular
01:17:47amiconnI'm wondering of about the ifdef above/around the asm blob
01:18:00amiconnDoes that mean Sansa uses the C version?
01:18:18amiconnI would think the C version is only there for reference
01:18:32jhMikeSno, sansa uses the asm
01:18:49jhMikeSsansa uses 16 bit IIS mode and the iPods are still using 32-bit
01:19:06amiconnOK, strange thing is:
01:19:14amiconnline 55: #ifdef CPU_PP
01:19:34amiconnline 157: #else /* !(CONFIG_CPU == PP5020 || CONFIG_CPU == PP5002) */
01:19:51amiconn(1) Shouldn't the commect read /* !CPU_PP */
01:20:04amiconn(2) Why not just simply #if 1 / #ese
01:20:53jhMikeSI don't know. the comment just hasn't been updated?
01:21:55jhMikeSthe C version of the fiq handler doesn't work anyway because of 1) gcc bugs 2) FIFO_FREE_COUNT is wrong. There's a count field for each of TX and RX fifos.
01:22:56jhMikeSMy changes will straigten that all out though
01:23:06*amiconn introduced an new CPU #define: CPU_PP502x
01:23:16 Quit \sic (Read error: 110 (Connection timed out))
01:23:44 Quit Febs (Read error: 110 (Connection timed out))
01:24:09 Quit bluebrother ("leaving")
01:25:04 Quit Robin0800 (" HydraIRC -> <- Go on, try it!")
01:26:57 Quit bdgraue (Remote closed the connection)
01:34:21 Join HellDragon_ [0] (
01:35:39 Quit HellDragon (Nick collision from services.)
01:36:09 Nick HellDragon_ is now known as HellDragon (
01:38:40saratogalinuxstb: you still around?
01:39:08DerPapstseems like the build servers haven't built the last commit...
01:39:53amiconnI hope it compiles; just compiling myself
01:42:18 Quit Xerion (" ")
01:42:53 Join maffe [0] (
01:49:04 Quit inversions (Read error: 113 (No route to host))
01:50:36linuxstbsaratoga: Yes
01:50:41linuxstbamiconn: OK, trying it now.
01:51:50 Join Febs [0] (
01:53:02saratogalinuxstb: are you writing the parser in fixed or floating precision? or does it not need either?
01:54:00linuxstbThere are no floats involved in the parser
01:54:03*jhMikeS knows the linuxstb is testing your patch on a 5020 device feeling ;)
01:56:55 Quit mickleby ("CGI:IRC (Ping timeout)")
01:57:44*amiconn is testing on PP5022 now
01:57:58amiconnNeed to make sure the relock wait is working properly
01:58:13*linuxstb is copying some music over - his ipod was empty
01:58:24amiconnThe color?
01:58:28jhMikeSlinuxstb: just make sure you actually applied the patch :P
01:58:49linuxstbamiconn: Yes
01:58:56linuxstbjhMikeS: I did...
01:59:03amiconnMy patch doesn't enable CPU scaling for H10 yet
02:00:58linuxstbamiconn: It boots, which is a good start ;)
02:01:23amiconnMake sure you play some music where it needs to boost/unboost a lot
02:01:36*amiconn copied an aac album exactly for this purpose
02:01:37 Join webguest35 [0] (i=9052c506@gateway/web/cgi-irc/
02:01:54linuxstbYes, I'm playing some mp3 and keeping an eye on the audio thread
02:02:04webguest35hello how can i overide loading? i want to load the original ipod fw
02:02:13amiconnjhMikeS: Hmm, what's the problem with CPU scaling on the sansa btw?
02:02:20amiconnIt's disabled in the config-*.h
02:02:39webguest35ipod is not recognized because it doesnt go to usb mode
02:02:47webguest35it just charges on rockbox mode
02:02:56crashmatrixwebguest35, turn if off, turn it on, and directly after turning on set it to hold
02:03:22crashmatrixand the original fw will magically load
02:03:33peturIs there anybody here that knows the fat code?
02:03:38webguest35thanks a lot , it worked :)
02:03:41linuxstbWhich ipod do you have?
02:04:32jhMikeSamiconn: It supposedly worked but the clocking setup was wrong and taking its toll on performance in general and partly causing lcd issues
02:04:53linuxstbamiconn: It's boosting/unboosting frequently (boost ratio about 68%), and still running. So looking good. But in the past, when Rockbox has been running well, it can take a couple of hours for it to freeze.
02:04:54*petur wonders if there's a reason update_short_entry() pokes around with char/long pointers iso mapping a fat_direntry on it...
02:05:03amiconnIs the PP5024 again different in regard of clock setup?
02:05:28amiconnlinuxstb: On the mini g1 I have here, it already locked up at boot in 50% of the cases
02:05:39jhMikeSI think so or at least the sansa is setup differently
02:05:50amiconnIn the other cases I was able to browse a bit. No chance of playing music
02:06:08linuxstbamiconn: The frequency of lockups seems to be random - some builds seemed to be fine, some locked up at boot.
02:06:34amiconnI will let the mini g1 play with repeat enabled (and charger connected) overnight
02:08:55*amiconn should do this with the mini g2 as well
02:09:03amiconnOnly one can have the charger though
02:12:19webguest35also i installed some themes but i dont see those fancy icons i see on the screenshots, solution?
02:12:54peturthey are installed separate
02:15:31webguest35thanks petur
02:18:46 Join toffe82 [0] (n=chatzill@
02:19:18webguest35and hopefully last do i enable the clicking that the normal ipod fw has?
02:21:12 Part pixelma
02:22:55scorchewebguest35: you dont, but there is a patch in the tracker if you want it
02:23:17 Quit Febs (Read error: 110 (Connection timed out))
02:23:43crashmatrixI just got this iPod to rockbox it
02:23:55crashmatrixand... batterylife completely sucks donky dick :/
02:24:15scorcheno need to use expletives like that...
02:24:58crashmatrixwell, yes, and no
02:25:28 Join p3tur [0] (n=petur@rockbox/developer/petur)
02:25:28 Quit petur (Read error: 104 (Connection reset by peer))
02:25:40saratogabuying an ipod for rockbox wasn't the best idea
02:25:45scorcheno...just no
02:25:48 Nick p3tur is now known as petur (n=petur@rockbox/developer/petur)
02:26:12webguest35if not ipod then what
02:26:16*petur leaves a non-working patch at for a fat32 guru to look at
02:26:20jhMikeSanything but
02:26:23crashmatrixthat figures :/ not dissatisfied, but man, 6 hours of playtime
02:26:40linuxstbWhich ipod do you have?
02:26:57crashmatrix5g 30 GB
02:27:08jhMikeSthose of us that actually do the work in our spare time on undocumented hardware feel very motivated by such comments
02:27:46peturamiconn: am I correct that Zagor is the person to talk to for fat.c?
02:29:20 Quit webguest35 ("CGI:IRC (EOF)")
02:30:30linuxstbamiconn: Bad news, my ipod froze...
02:30:34crashmatrixjhMikeS, "those ungrateful bastards" ? nothing of the sort
02:30:40amiconnYeah, the mini g1 too
02:30:53 Join JdGordon [0] (n=jonno@rockbox/developer/JdGordon)
02:31:02*linuxstb welcomes amiconn to the world of the pp5020
02:31:16*amiconn wonders why - first it plays the whole album w/o a single freeze, and in the next round it freezes after a while
02:31:26*jhMikeS one of of us
02:31:27*amiconn suspects that it has to do with the voltage stuff
02:32:32*jhMikeS is confident all the pp documentation will soon be publicly released :P
02:32:59linuxstbOf course - nvidia have a good record in that respect...
02:33:01*scorche just discovered that broadcom has an office a few min away...
02:33:42jhMikeSany day now ...
02:35:13DerPapstscorche: you maybe want to visit them? ;)
02:36:07scorcheDerPapst: well, perhaps i should sit there and not leave until they do as i wish
02:36:20 Quit petur ("Zzzzz")
02:36:21jhMikeSa sit-in?
02:36:31*scorche could use a few hundred other people for it to work though...
02:37:01LloreanA Rockbox sit-in at PortalPlayer HQ?
02:37:03jhMikeSmake them an offer they can't refuse
02:37:14scorcheLlorean: broadcom
02:37:21Lloreanscorche: PP would be more useful
02:37:22scorcheim sure nvidia is around here though
02:37:26DerPapstjhMikeS: a rockboxed iPod? :P
02:38:05*jhMikeS had more Godfatherly thoughts
02:38:12 Join Nico_P [0] (
02:38:51scorcheack...nope...nvidia isnt anywhere near
02:39:29scorchejhMikeS: you get the goat and "prepare" it
02:39:49 Nick HellDragon is now known as Ryuu (i=JD@unaffiliated/helldragon)
02:40:26*jhMikeS just ran out of goats and is requesting rockbox sponsor some more
02:40:49scorchebeen using them a lot lately?
02:41:26jhMikeSthey seem good for finding a new reg value or two
02:43:24*jhMikeS unfortunately lost his copy of the necronomicon
02:43:37*amiconn will look into this ln function and PP power control tomorrow
02:43:48amiconnNow, sleep
02:43:49JdGordonNico_P: gday
02:43:58Nico_PJdGordon: hi again :)
02:44:07JdGordonhad any luck?
02:44:20Nico_PI've made a bit of progress but spent the evening out
02:44:32Nico_Pit's 2:45 AM here :)
02:44:49amiconnlinuxstb: What do you think, does the patch make things better or worse?
02:44:57amiconnOn the mini g1 it makes things better
02:44:59JdGordon10.45am here... send me what you have, and ill play today unless i become social :p
02:45:00Nico_PI didn't get to speak to Bagder about the SVN
02:45:06*jhMikeS finds playback.c doesn't know a manual skip from it's own backside
02:45:20scorchejhMikeS: you can use my spare if you want
02:45:42jhMikeSok, fedex it overnight
02:45:52scorcheit would be easier to dcc it
02:45:55Nico_PI was planning on having another look for a bit... did you by any chance figure the problem with your SVN out ?
02:45:57scorcheit is a .666 file though
02:46:13jhMikeSdcc is broken here and nothing makes it work :(
02:46:30JdGordonNico_P: i literally got out of bed 10min ago... so not yet :p
02:48:18 Quit DerPapst (Nick collision from services.)
02:52:10 Quit saratoga ("CGI:IRC")
02:52:24linuxstbamiconn: It's hard to say. It works well, but then so have other builds in the past, seemingly at random.
02:52:57linuxstbamiconn: But it's close to what the OF does, then it's definitely a step in the right direction.
02:56:17 Join Hammer89 [0] (
02:57:16Hammer89Does anyone here know if the e200 running rockbox will work with a 4gb SD-card?
02:58:34Llorean4gb are SDHC, not SD
02:58:58Hammer89ah... guess not then ;)
02:59:39Hammer89is it a hardware issue? or will it be possible to make drivers to support a 4gb card? (I'm somewhat ignorant on this)
03:00:12LloreanI don't think anyone knows for certain whether or not it will be possible yet.
03:00:21Hammer89okay, thanks :)
03:03:18***Saving seen data "./dancer.seen"
03:04:00 Quit maffe (Remote closed the connection)
03:06:43JdGordonNico_P: you havnt fallen asleep yet have you?
03:06:58Nico_Pno, still working on the thing :)
03:07:21Nico_Pwant it now ?
03:07:44JdGordonno, just dont want you to leave before uploading it somewhere :D
03:07:54Nico_Pdon't worry ;)
03:08:19Nico_Pif you want I can try having a look at your svn if you give me an ssh account
03:10:39 Join reezeh [0] (
03:11:26 Quit linuxstb ("CGI:IRC")
03:32:44 Quit Thundercloud (Remote closed the connection)
03:33:17 Quit thegeek_ (Read error: 110 (Connection timed out))
04:14:04 Join toffe82_ [0] (n=chatzill@
04:20:23 Join webguest53 [0] (i=42b03e95@gateway/web/cgi-irc/
04:20:36 Part Hammer89
04:21:33 Join Teleosus [0] (i=42b03e95@gateway/web/cgi-irc/
04:21:33 Quit webguest53 (Client Quit)
04:21:51 Quit toffe82 (Read error: 110 (Connection timed out))
04:22:32TeleosusHello all, anyone willing to answer a quick question about future rockbox developments?
04:22:59LloreanTeleosus: It's somewhat unlikely anyone can answer a question about the future.
04:25:01TeleosusWell, I just bought an E280 and after quite a bit of web research it seems that the possiblity of rockbox helping in the ares of SDHC support is waning, anyone else feel the same?
04:25:28TeleosusNot that I would know how to do it
04:25:32JdGordonI dont belive any of the devs own a sdhc card, so we have no way of knowing if it works
04:26:23Teleosusnot that I would know anything about doing it....
04:27:07Nico_PLlorean: have you seen JdGordon mod on the BufferingAPIProposal wiki page ?
04:27:17 Quit Teleosus ("CGI:IRC")
04:27:18Nico_Pwe've set up an SVN repo with our work
04:27:18LloreanNico_P: I'll go look
04:27:21 Join Teleosus [0] (i=42b03e95@gateway/web/cgi-irc/
04:27:48LloreanNico_P: Ah
04:27:54Nico_PI've been making a bit of progress just now
04:28:02LloreanIs the SVN holding work on a stand alone, or is it a modified Rockbox?
04:28:13Nico_Pit's standalone
04:28:28Teleosushad network problems , sorry didI miss something?
04:29:05LloreanI'll check it out soon then.
04:29:19LloreanTeleosus: Nothing really. Generally speaking, nothing happens unless people who own the hardware are interested in it
04:29:44Nico_PLlorean: if you see Bagder around, could you ask him about hosting a parallel SVN or git on the rockbox server for this kind of work ?
04:30:10LloreanNico_P: Will do, but I think it's already been suggested.
04:30:19LloreanGit would certainly help with the SoC stuff
04:30:21TeleosusWell all I have really been able to deduce is that noone is really sure if it will be possible without a hardware update
04:30:23Nico_Pby me :)
04:30:31Nico_Por maybe someone else suggested it too
04:30:33LloreanI think others may have mentioned it too. ;)
04:30:46LloreanI seem to recall it's listed on the pro-git wiki page, though I can't remember the page's name
04:30:52LloreanBut I agree
04:31:56Nico_Pmaking a wiki page to list my ideas is on my list... I'll probably do it tomorrow
04:32:39Nico_Pyep... reading the previous sentence shows I need sleep now :)
04:33:21LloreanSleep is probably a good idea then
04:35:04TeleosusWell I own it but as far as finding the technical infomation require to hack into the card slot interface, I am at a loss
04:35:22 Quit Nico_P ("off to bed, bye :)")
04:40:17TeleosusAnyway, it is still a fair device, so I'll keep it anyway, 10 gb is enough, sorry about the off topic question, have a good work week
04:45:32 Join Febs [0] (
04:46:56 Part Teleosus
04:53:21 Quit billytwowilly (Remote closed the connection)
04:53:48 Join billytwowilly [0] (
05:03:21***Saving seen data "./dancer.seen"
05:18:15 Join Bossier [0] (
05:19:37Bossieranyone here I can talk to about screens.c in apps?
05:20:34 Join webguest03 [0] (i=47ace917@gateway/web/cgi-irc/
05:20:43webguest03i have no clue what to do
05:20:51webguest03to install rockbox on my ipod video 30gb
05:22:30krazykitread the manual
05:22:51webguest03k found it
05:22:54webguest03i was looking for it
05:22:58webguest03in the files
05:23:02webguest03but its actually in the site
05:23:11BossierI'll go ahead then, in line 893 or so of screens.c the #ifdef HAVE_LCD_BITMAP was changed from the if else with lcd_setmargins to a simple if else with statusbar_height, the thing is immediately after that statusbar_height is set to 0. I'm thinking that statusbar_height = 0 part should have come before the #ifdef so it didn't negate it
05:23:38 Quit webguest03 (Client Quit)
05:24:14BossierReason I ask is I was just syncing my copy of the scrolling margins patch and it concerns that, I went ahead and did the change on my own code and am noticing no ill effects
05:24:49LloreanBossier: I don't quite see your question
05:25:54LloreanIf HAVE_LCD_BITMAP is not defined, statusbar_height is always 0
05:26:23LloreanIf it IS defined, then if global_settings.statusbar is true, the height is set to STATUSBAR_HEIGHT, otherwise it is set to 0
05:26:48Bossieryes but immediately after that statusbar_height was set to 0, would that not negate the ifdef, I was thinking setting it to 0 should come before checking for havelcdbitmap, not before
05:26:52 Quit AdmiralJ ("Quitting")
05:27:08Bossierafter not before i mean
05:27:15LloreanI think you're reading it wrong
05:27:18Bossierbefore not after I mean, long day
05:27:30LloreanIf the define is true, it reads "else set it to 0" not simply "set it to 0"
05:28:06LloreanThat line is contained within the if-else statement if the define is true, it doesn't negate the setting.
05:28:09BossierI don't have else set it to 0 in my define, I have it set to 0 outside the define immediately after it
05:28:33 Quit Seed (Nick collision from services.)
05:28:41 Join Seed [0] (
05:28:47LloreanBut the else is there if the define is true
05:28:59BossierSo set it to 0 should have been in the else part of the define?
05:29:01Lloreanso you get "else set it to 0", if the define is true
05:29:19LloreanIf the define is true, remove the #ifdef and the #endif lines entirely, and pretend like they were never there
05:29:33LloreanIf you do that, you'll see that you get a full if-else
05:29:47LloreanJust delete those two lines, the ones starting with #, to see what it would be like if the define is true
05:30:21BossierI don't think you're getting what I'm saying, in my define I have an empty else statement
05:30:26LloreanI know
05:30:28LloreanI'm looking at the file
05:30:33LloreanDELETE the lines starting with #
05:30:36LloreanJust do that for me please
05:30:44LloreanYou'll notice that with them gone, the else statement isn't empty any more, is it?
05:30:53LloreanYou have to remove the whole line, not just replace it with a blank line.
05:31:51BossierI understand what you're saying, what I did with moving statusbar_height = 0 to above the ifdef and deleting the empty else is working for me, I was gonna go with that, but as you advise I'll do it your way
05:32:07LloreanDo you understand how preprocessor directives work?
05:32:31LloreanThen you know that if the define is true, what's handed to the C compiler doesn't contain either of the # lines
05:32:39LloreanInstead you get a fully functional, proper if-else statement
05:32:49LloreanWith the statusbar_height = 0; line being included in the else part of it
05:33:05LloreanIf you don't believe me, run it through a preprocessor with the proper define set and check the output.
05:33:55Bossiergotcha, I just don't like the way the ordering looks, I do know what you are saying, just a cosmetic thing really
05:34:17LloreanWell earlier you were talking about it negating the statement.
05:34:31LloreanJust wanted to make it clear that it actually doesn't, it's just a bit unclear what is happening there.
05:34:44Bossierbrain fart, I sometimes also forget that this is all in C and I can't get my brain out of object oriented mode
05:34:58LloreanRemoving the "else" entirely, and moving the =0 before the #ifdef would create functionally equivalent code, but it would require one extra assignment in cases where there is a statusbar
05:35:23Bossierthanks, feel kinda stupid now, but I'm glad that's cleared up
05:35:40LloreanOh, it's definitely not 100% clear what's going on there.
05:35:54LloreanI'd never seen it before and had to stare at it a bit to put things in order in my head
05:35:57Bossierjust so you know in all honesty I haven't coded in 10 years, coding is a young mans game
05:37:18LloreanAnyway, it works, and I would *guess* that it's like that for a slight increase in efficiency. Though it's entirely possible that it's like that because someone simply didn't think of doing it the other way, you never can tell.
05:38:38Bossierthanks again, btw, I'm tdtooke, on irc i'm always bossier and i'm too lazy to change nicks, gotta run now. on a personal note you're much nicer in person that you seem at times in the forums
05:38:57Bossierdont' take that the wrong way
05:39:06LloreanYeah, I'm actually even nicer in person-person, as the Devcon West guys can attest.
05:39:23LloreanEnforcing the rules at a level where people know you're serious never makes a person look nice, sadly.
05:40:12krazykit:-(. all the gigabeat flash stuff is windows apps.
05:40:27 Quit Bossier ("( :: NoNameScript 4.02 :: )")
05:47:24 Join ender` [0] (
05:50:59*ender` yawns
05:55:30toffe82_krazykit: what about the gigabeat flash ?
06:00:18 Quit relaxed (Remote closed the connection)
06:00:27 Join BHSPitLappy_ [0] (
06:01:12 Quit BHSPitLappy (Read error: 104 (Connection reset by peer))
06:03:19 Join jhulst [0] (
06:03:30 Join Destroyer14094 [0] (i=98a3644b@gateway/web/cgi-irc/
06:04:27 Quit tedrock (Client Quit)
06:08:08 Join tedrock [0] (
06:10:58 Quit Destroyer14094 ("CGI:IRC (EOF)")
06:17:53 Quit Nibbier (Read error: 110 (Connection timed out))
06:18:20 Join Nibbier [0] (
06:23:08 Part Llorean
06:24:09 Join MrKeuner [0] (n=kudo@unaffiliated/mrkeuner)
06:24:54MrKeunerHi, I am trying to get voice support or my rockbox. there is no .rockbox/lang folder as the manual says. there is a .rockbox/langs is that the one?
06:26:09 Join aliask [0] (
06:26:18 Join chrisjs169 [0] (n=jack@unaffiliated/chrisjs169)
06:31:16 Quit BHSPitLappy_ (Read error: 110 (Connection timed out))
06:31:48 Join piroko [0] (
06:35:33 Join webguest87 [0] (i=4a77d6fc@gateway/web/cgi-irc/
06:35:57aliaskhi webguest87
06:36:06webguest87can anybody help me with installin like supermario on the ipod with rockbox?
06:36:32aliaskYou don't need to install anything
06:36:36webguest87ive looked around and cant find a toutorial
06:36:44aliaskCopy the gameboy rom to the ipod
06:36:53aliaskAnd select it.
06:36:54aliaskThat's it.
06:36:57webguest87do you know where i can get it
06:37:09aliaskNo. Unless you own a copy of the game, it's illegal.
06:37:23webguest87well thankx
06:37:34 Quit webguest87 (Client Quit)
06:38:45pirokoA newbie to the world of emulation
06:40:32 Part piroko ("Leaving")
06:42:48 Quit Febs (Read error: 110 (Connection timed out))
06:45:19 Join DarthShrine [0] (n=Angus@pdpc/supporter/student/DarthShrine)
06:52:32 Part toffe82_
07:03:25***Saving seen data "./dancer.seen"
07:04:00 Nick swolbhsub is now known as bushblows (n=bushblow@
07:12:04 Join Ic0ne [0] (
07:13:02Ic0neHey all, i got rockbox on my F20 ( last version ) but the video/ Rock boy / and extension are very speedy, how can i slow down the speed ? :S
07:14:31 Quit MrKeuner ("Ex-Chat")
07:23:41 Join BHSPitLappy [0] (
07:24:18 Quit billytwowilly (Remote closed the connection)
07:24:35 Join billytwowilly [0] (
07:24:39 Quit BHSPitLappy (Read error: 104 (Connection reset by peer))
07:24:53 Join BHSPitLappy [0] (
07:25:38Ic0neHey all, i got rockbox on my F20 ( last version ) but the video/ Rock boy / and extension are very speedy, how can i slow down the speed ? :S
07:33:14aliaskIc0ne: Have you enabled the Limit FPS setting?
07:33:39aliask(for mpegviewer)
07:37:39Ic0nebut, for all its speed, the ''plasma'' extension his speedy so...
07:40:30aliaskIt's because the gigabeat has a much faster CPU than all the other DAPs that rockbox supports
07:41:12aliaskThere is no way to slow down plasma (that I know of), but I don't think it's very important...
07:43:22aliaskamiconn: Was that commit to stop all the frequency scaling bugs in PP5020? Or are they still there?
07:43:43Ic0neHeum...what ?
07:43:51amiconnIt makes things better, but doesn't fix hangs completely
07:44:13aliaskHow common are the hangs now?
07:44:18amiconnThere is more to come, I hope. Still not done analysing things
07:44:29aliaskWow, excellent.
07:44:32amiconnThat needs testing.
07:44:57Ic0neso...i really cant slow down anything ? :S
07:45:16amiconnI only have one PP5020 target to test with. Before my changes it froze at boot in 50% of all cases, and if not, it froze within a few minutes of browsing
07:45:17aliaskIc0ne: I just told you how to slow down video
07:46:04amiconnNow I can play music, and it even played a whole album (with lots of boosting/unboosting) without freeze, but then froze in a second test after ~15 minutes
07:47:08Ic0nealiask: yeah but, the FPS limiter is enabled..and no more difference
07:47:34 Join darkless_ [0] (
07:47:48aliaskIc0ne: Use the display FPS option, how many FPS is it showing?
07:48:24Ic0ne29 ~
07:48:51aliaskThat sounds about right. Are you sure you encoded the mpeg properly?
07:49:22Ic0neyeah, on my PC they run perfectly
07:50:16aliaskHrm, you might need to talk to someone who knows a bit about mpegplayer then...
07:50:27Ic0nelol :P
07:50:36Ic0neyeah but i cant register on the forum :S
07:50:40aliaskHow big is the mpeg?
07:51:21Ic0nehum...what do you mean ?
07:51:42aliaskHow many megabytes.
07:51:51Ic0neaww :P
07:51:53Ic0neit depend
07:52:07Ic0neaprox: 3 mb
07:52:23aliaskCan you send me one of them so I can test using it?
07:52:34Ic0neyeah, MSN ?
07:53:11aliaskemail would be faster
07:53:26aliaskaliask at gmail
07:55:17Ic0nethe video i'll send you is 6.54 mb
07:55:39aliaskThat's fine.
07:56:42 Quit darkless (Read error: 110 (Connection timed out))
07:57:40 Join lee-qid [0] (
07:59:12Ic0nedid u receive it ?
08:01:52Ic0neperfect :)
08:01:56 Nick bushblows is now known as swolbhsub (n=bushblow@unaffiliated/bushblows)
08:03:35Ic0neand ?
08:04:34aliaskYou're right, limit fps seems to do nothing (on the sim at least)
08:04:54Ic0neit do the same thing ?
08:05:05aliaskYes, but plasma works fine....
08:05:21Ic0neok, and the rockboy ?
08:05:26Ic0nemine his speedy :P
08:05:32aliaskGotta find a rom
08:05:40Ic0netake mario bros...
08:06:03Ic0nethe sound is accelerated, and the image too
08:06:38*amiconn hopes Ic0ne uses the latest official build to test this
08:06:56Ic0neyes i do
08:07:07aliaskWell, on the simulator, it's very very slow.
08:07:21Ic0nethe simulator ?
08:07:33 Join GodEater [0] (
08:07:41aliaskI don't have an MP3 player to test on, so I run rockbox under linux to test.
08:07:42 Join ansichart [0] (
08:08:01Ic0neokay okay
08:08:25Ic0nei hope someone wil help me tomorow :P
08:08:33ansichartI have a question, are we allowed to share our the music we have created on here?
08:08:54ansichartBecause I would like to post a link to some black metal vocals that i have done
08:09:07ansichartAnd get some feedback from people who know a lot about black metal
08:09:09scorcheansichart: what does that have to do with rockbox?
08:09:19 Quit Ic0ne ()
08:09:28GodEaterscorche: he used Rockbox to record them perhaps ? :)
08:09:35ansichartI was assuming that this is a channel for music
08:09:38scorcheGodEater: dont encourage them ;)
08:09:43scorcheansichart: it sint
08:09:45ansichartNo, I used Audacity
08:09:47DBUGEnqueued KICK ansichart
08:10:15ansichartDo you know of a channel that is about metal?
08:10:24 Join bdgraue [0] (
08:10:26ansichartoh well
08:10:57GodEateransichart: you're on the wrong network for that - freenode is mostly about open source software
08:11:04GodEateransichart: it's not really a social network
08:11:12ansichartI see
08:11:38ansichartthank you
08:11:47 Part ansichart ("Konversation terminated!")
08:12:27GodEaterscorche: happy now ? :)
08:13:09GodEaterI need to train myself to use those smileys more - I'm such an old stick in the mud
08:15:20 Join darkless__ [0] (
08:22:22 Join darkless [0] (
08:25:32 Quit Weiss (Read error: 104 (Connection reset by peer))
08:25:39GodEaterscorche: do we have better documentation on using your own keyboard layouts than this : ?
08:26:31CtcpIgnored 1 channel CTCP requests in 0 seconds at the last flood
08:26:31*scorche is watching the new doctor who and does not care to be disturbed =P
08:29:18 Join Weiss [0] (
08:30:30 Quit darkless ("Leaving")
08:30:46 Join pondlife [0] (
08:31:01 Quit lee-qid ("aufwiederbyebientotsayonara")
08:31:32 Quit darkless_ (Read error: 110 (Connection timed out))
08:32:09GodEaterwhat are you even doing logged into IRC then ? :)
08:33:21scorchei always am...
08:35:44 Quit darkless__ (Read error: 110 (Connection timed out))
08:38:49markunscorche: ?niaga yaw dlo eht gnipyt trats uoy did ,yeH
08:39:24pondlifeJdGordon: Would you like me to look at your buffering code?
08:39:38JdGordonif you want
08:39:43scorcheyaw rehto eht reisae ylthgils si ti
08:39:45JdGordonmine and nicos is on my svn server now
08:39:46pondlifeIf it's worth it...
08:40:18JdGordonive abbandoned mine, I tihnk nicos will work better once its finished
08:40:49markunscorche: (: thgir era uoy sseug I ,sey
08:41:02pondlifeJdGordon: I'm attempting to check it out, but I only get an empty mob directory... probably because I don't know enough svn ;p
08:41:25pondlifeDo I need a login?
08:41:27JdGordonsvn co svn:// should be enough
08:41:33scorchesdrawkcab si gnithgilhgih ruoy :nukram
08:41:33JdGordonno, anon read access
08:42:14pondlifeThat's what I was trying - it gives an error "svn: Failed to add file 'mob/aux.h': object of the same name already exists".
08:42:23pondlifeBut there's no local aux.h
08:42:24 Join webguest60 [0] (i=c023110f@gateway/web/cgi-irc/
08:42:38markun...noitelpmoc otua ot eud :ehcrocs
08:42:54pondlifeI can see the history with svn log...
08:43:34markunscorche: so, this is a bit faster
08:43:34JdGordonthats wierd :p
08:44:05GodEater?klat sdrawkcab eht htiw stahw
08:44:43scorchednatsrednu tndluow uoy :retaEdoG
08:49:12 Join miepchen^schlaf [0] (
08:49:39 Join RaRe [0] (
08:49:56 Quit DarthShrine (Read error: 104 (Connection reset by peer))
08:52:36 Quit gtkspert_ (Remote closed the connection)
08:53:04 Join gtkspert [0] (
08:53:18 Join nick89 [0] (
08:53:40 Quit nick89 (Client Quit)
08:54:03JdGordonsweet :) got nicos code working with my test app
08:56:01 Join darkless [0] (
08:56:07 Join petur [0] (n=petur@rockbox/developer/petur)
08:58:24 Join desowin [0] (
08:59:39 Join Rob2222 [0] (
09:03:09 Quit RaRe` (Read error: 110 (Connection timed out))
09:03:26***Saving seen data "./dancer.seen"
09:03:32GodEater¿ sıɥʇ ǝʞı1 buıɥʇǝɯos : ǝɥɔɹoɔs
09:04:50scorchei thought about it, but it appears as gobbley-gook on thsi client with this font...
09:05:19GodEaterthat's a shame - looks fine here :)
09:05:52JdGordonGodEater: haha awesome
09:06:45 Join RaRe` [0] (
09:06:52 Quit darkless (Read error: 104 (Connection reset by peer))
09:06:53scorche¿ sıɥʇ ǝʞı1 buıɥʇǝɯos : ǝɥɔɹoɔs
09:06:57scorchedoes it copy fine?
09:07:37scorche?ou ?noq? ?o?
09:07:47scorcheyeah...when i paste it wont copy
09:08:16peturmy irc client doesn't like it too
09:08:29 Join darkless [0] (
09:08:35*petur reminds the kids to stay on topic
09:09:00 Join Zagor [0] (n=bjorn@rockbox/developer/Zagor)
09:09:01scorchepetur: devs and people involved int he project have immunuty =P
09:12:00peturmorning Zagor, I could use your FAT32 expertise. I'm trying to fix a bug in fat.c for but somehow my solution doesn't work (see patch attached to task, yes I know it's not yet foolproof)
09:12:37 Join kaaloo [0] (
09:13:26 Join Weiss_ [0] (
09:13:36 Part kaaloo
09:13:48scorchepondlife: that was sad...
09:15:11 Quit Weiss (Read error: 104 (Connection reset by peer))
09:15:55Zagorpetur: ok, I'll take a look
09:16:31 Quit Rob222241 (Read error: 110 (Connection timed out))
09:22:01 Quit RaRe (Read error: 110 (Connection timed out))
09:24:03 Join Genre9mp3 [0] (n=yngwiejo@
09:27:30 Quit jhMikeS (Read error: 110 (Connection timed out))
09:30:52 Join HellDragon [0] (i=JD@unaffiliated/helldragon)
09:31:13 Part Ryuu
09:31:33 Join HellDragon_ [0] (
09:31:34 Nick HellDragon_ is now known as Ryuu (
09:31:35 Quit miepchen^schlaf (Read error: 110 (Connection timed out))
09:31:43 Join miepchen^schlaf [0] (
09:32:38 Join hcs [0] (n=agashlin@rockbox/contributor/hcs)
09:34:25 Quit gtkspert (Read error: 104 (Connection reset by peer))
09:36:09 Join gtkspert [0] (
09:42:11 Join LinusN [0] (i=linus@gateway/web/cgi-irc/
09:54:02 Join Wiwie [0] (
10:00:37 Join printfXh4 [0] (
10:00:52 Join XavierGr [0] (
10:01:15 Quit miepchen^schlaf (Read error: 113 (No route to host))
10:06:51*JdGordon slaps Nico_P with a large trout and the 100000 seg faults I keep getting
10:07:38GodEater100k segfaults ?
10:14:20 Join crop [0] (i=c27f0812@gateway/web/cgi-irc/
10:14:25JdGordonim special :D
10:14:35*JdGordon got some magic happening :)
10:14:39cropdionoea: ping
10:15:30 Quit Wiwie (Read error: 113 (No route to host))
10:17:58 Quit billytwowilly (Remote closed the connection)
10:18:19 Join billytwowilly [0] (
10:25:29cropAny reversi plugin experts here?
10:26:16dionoeacrop: pong
10:28:37cropdionoea: hey! First of all, thanks for the nice plugin! I like the game!
10:28:59 Join spiorf [0] (
10:29:04 Join RaRe [0] (
10:29:12dionoeawell i only coded the (stupid) AIs, the original patch was from someone else :)
10:29:30cropAlso the cursor movement is smart. he-he. I can reach every cell by just holding LEFT (or any other key) pressed
10:30:03dionoeathat was from the original patch :)
10:30:09 Join pixelma [0] (i=pixelma@rockbox/staff/pixelma)
10:30:56cropI have two questions about the code. 1) Shouldn't the RB api field be moved from the game to the strategy struct? Since the game should be API neutral and only the strategy needs the API (for rand I think)
10:31:50cropAnd 2) Shouldn't the threshold consts (for the AI) be moved from the general .h to a specific .h (if there is one)?
10:31:58dionoeacrop: well the game struct is the only one that were sure to have everywhere (like if we ever need the rb struct in reversi-game.c)
10:32:23dionoeathe CHAOS consts ?
10:33:23scorcheBagder: around?....i havent seen b4gder yet =P
10:34:44cropdionoea: couldn't it be made a globar var that's declared in strategy.h? I just think that it belongs more to strategy than to the game itself. But making it a global isn't nice either. Hmm...
10:35:17dionoeai could declare it as extern in the .c files which need it
10:35:30dionoeabut then i doubt that it really matters :)
10:35:37dionoeaas long as it works
10:35:51cropdionoea: and yes, I mean the CHAOS consts
10:36:03dionoeathose will be used in multiple strategies
10:36:19dionoeawhich is why i put them in the strategy.h header
10:36:39dionoeabut i haven't commited those yet
10:36:46dionoea(since i didn't finish porting them)
10:36:47cropdionoea: yes, that's what I meant: declare it as an extern in strategy.h which is included by all strategy-*.c
10:37:14dionoeayeah, i might do that
10:37:16cropdionoea: ah, ok then
10:37:32crop(reg CHAOS)
10:38:41cropdionoea: you've made an FS cleanup raid, no? Ranger dionoea! He-he :-)
10:39:14dionoeawell yeah, i find it sad that some of those patches don't get applied :)
10:42:57 Join obo [0] (
10:44:06 Quit RaRe` (Read error: 110 (Connection timed out))
10:44:29 Join RaRe` [0] (
10:48:19cropdionoea: another question: why are xxx_strategy (in -gui.c) declared as const? Shouldn't it be possible to change strategy in the middle of the game? I.e. if you're in trouble, let the xpert play a couple of moves and then switch back :-P
10:49:14peturre tracker clearing: this was actually something that was decided on devcon: clean out the patch tracker a bit by accepting patches, and if not, rejecting.
10:49:38 Quit RaRe (Read error: 110 (Connection timed out))
10:51:40pondlifeDoes anyone here have a recently built sim available? I'm getting a segfault when I enable/disable crossfade during playback, but I seem to get a lot of mystery sim segfaults...
10:51:49cropdionoea: is there a "supervised full automatic mode"? I.e. when two robots are playing but a move is only made if the user presses a key?
10:51:58 Join kaaloo [0] (
10:52:11 Join \sic [0] (
10:52:34peturpondlife: all my sims segfault the moment playback starts :/
10:52:44pondlifeHmm, even worse than me!
10:53:05pondlifeSomething changed for the worse in the past few weeks.
10:53:28pondlifeGDB doesn't even give a useful backtrace.
10:53:52pondlifeGuess I'll try binary-chopping into past revisions...
10:54:45peturmay be a simple issue
10:55:07JdGordonwhats the best way to diff two bin files?
10:55:46petursomething else that's worrying me: I've had stkov in power when attaching usb this weekend (not reproducable anymore, it went away), and somebody reported stkov in main in the tracker
10:56:07JdGordonon which target?
10:56:07peturJdGordon: in dos: fc /b
10:56:13dionoeacrop: the struct is defined as const, not the pointer to the current strategy struct which is defined in reversi-gui.c
10:56:13JdGordonpetur: linux...
10:56:37dionoeaand about the supervised mode, sure, everything is possible. You only need to add the code that goes with it :)
10:57:08 Part kaaloo
10:57:41cropdionoea: I mean this: const game_strategy_t *white_strategy; Doesn't it mean that white_strategy is a const pointer?
10:58:09cropOr a pointer to a const struct?
10:58:20Zagorpetur: what goes wrong with your patch? boken filesys, early return or something else?
10:58:40cropAh, probably the latter
10:59:42dionoeaelse i would never be able to change that value
10:59:48dionoea(even during init)
10:59:49amiconnpondlife: Is this stil fs #7245?
11:00:01peturZagor: it doesn't fix the issue at all. chkdsk complains that the . and/or .. entries are wrong. The .. entry should obviously be changed to point to the new parent but somehow my fix doesn't do that (ie it still complains)
11:01:04Zagorperhaps it's time to bring firmware/test/fat/ back to life
11:01:21cropdionoea: another thing (sorry for disturbing you): wouldn't it be better to initialise the game so that one party is a robot and the other is the user at the start? Or will it be a config that can be saved (not implemented now but later)?
11:02:01dionoeayeah, it would be better to initialise black as human and white as AI and save config changes :)
11:02:15dionoeapatches welcome
11:02:20pondlifeamiconn: That still happens, but this is a different segfault
11:02:28dionoea(I'm at work so i can't really code on rockbox now)
11:02:44pondlifeI shall FS it, but wanted to see if it was just me
11:03:11pondlifeQuite possibly the same cause of course
11:03:28***Saving seen data "./dancer.seen"
11:03:56*petur knows the feeling dionoea has ;)
11:05:39cropdionoea: ok, I just wanted to talk about it.
11:06:38dionoeahehe :) (but if you feel like coding those small changes you talked about, go ahead ... i already have the AIs to finish. Current AIs are a bit too simple to beat)
11:06:51 Join pontape [0] (
11:08:03dionoeaI also need to fix a bug in the simple-ai (or the ai calling code) which makes it enter an infinite loop sometimes when playing simple vs naive or simple vs simple.
11:08:23cropdionoea: and what about your commit comment about not being able to beat the simple AI?
11:08:37dionoeawell i only tried once before doing that commit :)
11:09:03dionoeain fact it's really easy to get the four corner stones against any of those AIs
11:09:15dionoeaand once you've got the corners you've basicaly won
11:09:30dionoeawhich won't be easy to do with the next AI
11:09:57cropdionoea: :-) BTW: I've read that reversi is a game that can be played well by computers. I.e. much better that chess. So if a good AI will be implemented we'll have no chance!
11:10:31cropthat -> than
11:10:35dionoeaI'm porting the AIs from hinversi. I don't know if they're considered as good ... but i hope that they are :)
11:10:42dionoeaElse i'll have to look for others
11:11:37dionoeaOn a fast computer i believe that you can test all possible moves all the way to the end ... i doubt that it's possible on a DAP though
11:11:52pondlifeamiconn: I've put a note and my config on FS #7245... maybe you could try to reproduce it in linux-x86 and see if you get anything helpful?
11:14:14 Join Entasis [0] (
11:14:28 Join RaRe [0] (
11:21:12 Join Wiwie [0] (
11:21:54 Quit Wiwie (Client Quit)
11:27:40 Quit RaRe` (Read error: 110 (Connection timed out))
11:29:06amiconnpondlife: Enabling/disabling crossfade is surely related, as it needs to restart playback (and hence a stop is involved)
11:36:01pondlifeTrue, but the routine it crashes in is different. Must be something in the s SWCODEC sim stop code corrupting memory/stack?
11:37:19pondlifeOdd that the original problem doesn't seem to happen when the WPS is visible too.
11:37:56 Quit RaRe (Read error: 110 (Connection timed out))
11:40:37 Join linuxstb [0] (n=linuxstb@rockbox/developer/linuxstb)
11:46:06 Quit printfXh4 (Read error: 113 (No route to host))
11:53:46 Join bluebrother [0] (i=cEpzE6DY@rockbox/staff/bluebrother)
11:54:47 Join Nico_P [0] (
11:59:24 Join mud-rb [0] (
12:02:12GodEaterNico_P: I managed to try out what we were discussing on Friday, and it seemed to work okay - I ended up with another branch in my "clean" git-svn repo of rockbox
12:03:01Nico_Pcan you remind me what it was ?
12:03:32GodEatergetting a branch from my public repository into your rockbox git-svn tree
12:03:49GodEaterand being able to then merge it in and commit,
12:03:54GodEateror choosing to ignore it
12:04:03GodEaterI believe that's what you were asking me to do ? :)
12:04:16Nico_Poh yes
12:04:26Nico_PI'll check that out
12:04:35GodEaterthere *may* be issues with binary files though
12:04:46GodEaterI did have a brief hiccup with one of the sokoban bmps
12:04:52GodEaterbut it went away again
12:05:02GodEaterso it might just have been a PEBKAC issue
12:07:02GodEaterI get those a lot :)
12:08:24Gabrilhi, are there any news about mpegplayer and native ipod-video-hardware support?
12:08:32markunGabril: no news
12:11:38pontapehi all, i wanna ask if rockbox runs stable on ipod 5.5g 80gig ?!
12:12:18GodEaterpontape: stable enough
12:12:22GodEaterI use it daily
12:12:42pontapeah ok fine
12:13:09GodEaterjust bear in mind the usual ipod caveats
12:13:34GodEater1) lower battery life than original firmware, and 2) the processor sucks so don't go mad with the DSP effects and expect silky smooth playback
12:14:47pontapehehe, well i am an absolutly rockbox newbie, first i have to get it installed somehow :>
12:15:02GodEaterpontape: go to our wiki and the RockboxUtility page
12:15:07pontapei want it mainly for playing my .acc's
12:15:07GodEaterthat makes it pretty easy
12:16:39pontapeok i will
12:17:18pontapeis the installtion possible without running linux ?!
12:17:29Gabrilyes, work's fine with windows
12:18:00Gabriljust read the howto - and within 10 minutes you have your rockbox up and running
12:18:39 Join inversions [0] (
12:23:50 Quit aliask ("ChatZilla [Firefox]")
12:25:29 Quit billytwowilly (Remote closed the connection)
12:25:46 Join billytwowilly [0] (
12:26:34 Quit \sic (Read error: 110 (Connection timed out))
12:26:47GodEateris rbutil ready for primetime yet do we think ?
12:29:22 Join Thundercloud [0] (
12:30:15JdGordonNico_P: hey
12:30:40JdGordongood news
12:30:45JdGordonseen the svn log?
12:30:49Nico_PI've seen your changes and I tested a bit
12:31:08Nico_Poh another change :)
12:31:10JdGordonthere is a bug in the bufread() when the data wraps
12:31:32Nico_Pthere was one in bufclose that I fixed
12:31:46JdGordonmy debuggin pretty much confirms its there and not in the actual buffering
12:31:46pondlifeYour svn? I'm still unable to get to it... :(
12:32:01Nico_Pout of my 4 files, the 2 last were wrong and now it's only the third
12:32:06Nico_Pthe third is the one that wraps
12:32:41JdGordonodd... my test files all are fine except the one that wraps
12:32:52JdGordonI was gonna fix it now but heading out for a while
12:33:01Nico_Pok, maybe I'll find it :)
12:33:05JdGordoncan you help pondlife with svn
12:33:10Nico_PI'll try
12:33:17 Join maffe [0] (
12:33:17pondlifeGo on then ;p
12:33:23 Quit reezeh ("Remember kiddies: Rub your balls so you don't get cancer!")
12:33:29Nico_Ppondlife: what problem do you get ?
12:33:39JdGordonthe 2 memcpy() calls are the problem in bufread, the first isnt reading the correct amount
12:34:04pondlifeI'm under Cygwin on Win2K. svn co just gives an error.. what's the URL again?
12:34:50JdGordonyep, /mob
12:35:20*JdGordon gone for a few hours, good luck
12:35:24pondlifeOK, it gives "svn: Failed to add file 'mob/aux.h': object of the same name already exists"
12:35:34 Join PaulJam [0] (
12:35:38pondlifeAnd I don't have a local file of name aux.h
12:35:43Nico_Ppondlife: just on a checkout ?
12:36:01Nico_Pand do you have a local .svn dir ?
12:36:14pondlifeYes, created by this command, in a /mob dir.
12:36:31Nico_Pwhat if you wipe everything clean and start again ?
12:36:43pondlifeThe same
12:36:55pondlifeIt recreates /mob/.svn/...
12:37:13pondlifeBut no other files except in the .svn dir
12:37:43pondlifesvn works fine for Rockbox, must be something I'm lacking in my setup
12:37:57 Join DerPapst [0] (
12:38:11DerPapsti'm having the same problem as pondlife.
12:38:25DerPapstalso cygwin
12:38:33Nico_Pvery strange, I don't have that problem (just tried a frexh checkout)
12:38:46DerPapstalso on cygwin or linux?
12:38:49pondlifeDid you have to do any kind of user/login config?
12:39:02Nico_Panonymous has read access
12:39:19Nico_Pyou're using svn in cygwin ?
12:39:30DerPapstwe both
12:39:31pondlife$ svn st svn:// gives "svn: warning: 'svn://' is not a working copy"
12:39:56pondlifeBut svn log works ok
12:39:58Nico_Pyes, you're supposed to run st on a working copy, not an URL
12:40:22Nico_Ptry info instead
12:40:42pondlifeYes, that seems ok
12:40:59Nico_Pare you suree verything's right with the file permissions ?
12:41:02pondlifeSo why no checkout..?
12:41:47pondlifeWell, I have my Rockbox working copy in D:\Cygwin\home\Steve\rockbox... and I just made a D:\Cygwin\home\Steve\jdgordon to work in.
12:41:53DerPapstafaik there are no file permissions in cygwin/windows
12:42:16Nico_Phmm very strange
12:42:35Nico_Phave you tried a windows svn client like tortoisesvn ?
12:43:01DerPapsti can successfully co rockbox iPL SDL and some others i forget.
12:43:14 Nick Weiss_ is now known as Weiss (
12:43:24DerPapstso i doubt it's an issue with cygwins svn...
12:43:35pondlifeDerPapst: Are you getting the same error as me? /
12:43:49DerPapstexactly the same
12:43:53GodEatersvn version ?
12:43:55 Quit Thundercloud (Remote closed the connection)
12:44:14pondlifer23084 - Apr 7 2007
12:44:48GodEater1.4.4 here, but of course I'm not on cygwin
12:45:07Nico_PI have the same as you guys
12:45:20pondlifeNico_P: But yours works, right?
12:45:39pondlifeDid you do anything before you did an SVN co ?
12:46:43pondlifeNo "unusual" filenames involved?
12:48:22Nico_Pmaybe try again. I made a change but without real conviction
12:48:44pondlifeNo difference
12:52:25pondlifeDo you have any filenames which would suffer from Windows case-insensitivity in there? e.g. Aux.h and aux.h?
12:52:40 Join Lear [0] (i=chatzill@rockbox/developer/lear)
12:52:57Nico_Pno, everything is lowercase
12:54:15Nico_PI can send you the files if you want
12:54:27Nico_PI've just fixed a bug so it seems to work nicely
12:54:29DerPapstNico_P: You aren't uing cygwin, are you?
12:54:36Nico_Pno, I'm under linux
12:55:41amiconnaux is a forbidden filename on dos/windows
12:55:58amiconnThat applies irrespecive of extension
12:56:10pondlifeMaybe rename it to LPT.h ;P
12:56:18Nico_PI'll change that then
12:56:54DerPapstwhy is that forbidden?!
12:57:06pondlifeMS-DOS 1.0 back-compatibility
12:57:23amiconnaux, con lpt<n>, com<n> are device names
12:57:34 Quit crop ("CGI:IRC (EOF)")
12:57:55pondlifeFrom the days before directories.... smells like CP/M
12:58:05GodEaterI'd forgotten that.
12:58:10amiconnThey work in all versions of msdos
12:58:50Nico_Pyou can try again now
12:58:54GodEaterwhat was AUX used for ?
12:59:02GodEaterI know COMn, LPTn, CON and NUL
12:59:18amiconnYou can e.g. send raw printer data by doing copy <file> lpt1
12:59:26idnarAUX is the same as LPT1, iirc
12:59:31idnaror maybe I mean COM1
12:59:32crwllPRN was that?
12:59:36pondlifeNico_P: Yep, that works now
12:59:48idnarah yes, AUX was COM1, PRN was LPT1
12:59:52Nico_Pwoohoo !
12:59:55Nico_Pthanks amiconn
12:59:59pondlifeYes, thanks
13:00:01GodEatersilly DOS hangovers
13:00:30pondlifeI never realised that it would still fail with a file extension.
13:00:31idnarthen you get other fun stuff like CLOCK$, and CONFIG$
13:00:58GodEaterpondline: same way you couldn't call a file "D:.txt"
13:01:03idnaroh, another fun one
13:01:09idnarif you had a CDROM driver loaded, called "CDROM"
13:01:14idnarthen you couldn't do anything with a file named cdrom.sys
13:01:31 Join gromit` [0] (
13:02:04amiconnidnar: The cdrom device name could be specified when loading the driver
13:02:19idnaramiconn: yeah
13:02:19amiconnLater installs used MSCD001
13:02:28Nico_Ppondlife, DerPapst: you just feed the app a series of files as arguments. It will buffer them then unbuffer them and hopefully the unbuffered files will be exactly identical to the originals
13:03:18 Join Febs [0] (
13:03:29***Saving seen data "./dancer.seen"
13:04:05 Join jba [0] (
13:04:44crwllI remember that ms-dos (or maybe some cd-rom related driver) had a funny bug, which allowed the user to cd/chdir to unexisting directories in an CD drive if the directory had an umlaut at the beginning and at the end of the name - "cd E:\öblahö" worked fine (if E: was the cd drive)
13:06:11DerPapstgood ol' times
13:07:04DerPapstprintf("ERROROROROR\n"); <−− wow.. that sounds drastic ;)
13:09:52GodEaterDerPapst: or my personal favourite : printf ("Falmunction.");
13:10:23peturprintf("This should never happen");
13:10:54peturit somehow always gets printed when a customer uses it
13:11:55PaulJamHi, i'm just curious: why does the revision number in the rockbox-info.txt sometimes have a "M" at the end and sometimes not?
13:13:03PaulJamreally? i'm almost sure i have seen this with official builds.
13:14:10DerPapstGodEater: hehe.. i love such cool errors. Also reading comments in the source code is sometimes very amusing :D
13:19:56 Join Rob222241 [0] (
13:20:56 Join spiorf_ [0] (
13:22:35pondlifeNico_P: I'm getting ERROROROROR too, expected?
13:22:53pondlife"./a buffering.c buffering.h" is my command line :)
13:24:01pondlifeIT seems to be attempting to open a 3rd file with a name of (null) though.
13:24:39DerPapsti didn't get that error. but i've added the .h files too ;)
13:25:04 Join miepchen^schlaf [0] (
13:25:12pondlifeShould the test in line 395 be "next_file < argc" ?
13:26:25pondlifeI'll have a proper look later when I have time...
13:26:47GodEaterDerPapst: you should track down the obscure error messages Apple once put in one of their C compilers
13:26:50GodEaterthey were very funny
13:26:56DerPapstbah.. i'm too used to makefiles ^^
13:27:38 Quit Rob2222 (Read error: 110 (Connection timed out))
13:27:54DerPapstGodEater: heh.. i've only heared micosoft had some fun comments in the win2k source code :P
13:28:32*DerPapst has written a small makefile for Nico_P's mob :-P
13:32:21 Join jba_ [0] (
13:32:46*DerPapst −−> work
13:32:51DerPapstsee you
13:32:52 Quit DerPapst ("So Long And Thanks For All The Fish!")
13:37:42GodEater <−− for those that have never seen them
13:42:40Nico_Ppondlife: still get the error ?
13:42:56Nico_Pyou need to delete the output files before running the program
13:43:54GodEater <−− also amusing, but quite handy advice
13:44:58 Quit spiorf (Read error: 110 (Connection timed out))
13:46:17 Join Wiwie [0] (
13:49:24 Join Xerion [0] (
13:50:54 Join cyberstream [0] (
13:51:05 Quit jba (Read error: 110 (Connection timed out))
13:51:45 Quit webguest60 ("CGI:IRC (Ping timeout)")
13:53:44 Join thegeek [0] (
13:53:45cyberstreamcan i speak to one of the rockbox developers please?
13:53:54GodEateryou are speaking to them
13:53:59GodEaterjust ask whatever it is you want to ask
13:54:02 Quit hcs (Read error: 110 (Connection timed out))
13:54:50cyberstreamit has to do with compatibility of apple ipod video and an charging akku pack
13:55:14GodEatera what ?
13:55:22cyberstreamneed t speak to one person in query
13:55:32GodEaterno you don't
13:56:05dionoeaon the other hand you need to ask your question
13:56:15cyberstreamthan i am sorry to
13:57:04cyberstreamno. because i have to speak directly to an developer , not in public. i have to give him an email adress of an ANSMANN (producer of batteries and akku packs) Product manager!
13:57:29GodEateryou could still ask your question and then we could direct you to the right person
13:57:58cyberstreamsorry for my bad english at first
13:58:22pondlifeNico_P: What's the output file..?
13:58:38Nico_Ppondlife: file*.mp3
13:59:10Nico_Pthese are supposed to be perfectly identical to the files you give as input
13:59:22cyberstreamThere is a product for recharging the ipod with an external battery pack. with apple software, it charges and display the recharge. on rockbox, it displays nothing
13:59:33pondlifeOK, deleting that file fixes the error...
13:59:47GodEatercyberstream: this sounds like it uses the AAP which we have no knowledge of
14:00:01GodEatercyberstream: and I doubt your product manager contact is at liberty to reveal details of it
14:00:30pondlifeNico_P: Yay! Now to try with some really big files!!
14:00:31cyberstream@godeater: maybe in that case i could help you because i have spoken to an product manager of ANSMANN and he want to get contact to you guys
14:00:45Nico_Ppondlife: it won't work well
14:01:37GodEatervolunteers to talk this chap then folks ?
14:02:20pondlifeNico_P: OK...
14:02:30pondlifeI'll keep an eye out though.
14:02:54Nico_Ppondlife: maybe you could try with a big list of small files if you have the courage
14:03:15Nico_Por actually not so small... just smaller than 32 MB should do
14:03:25crashmatrixcyberstream, how do you usually charge your iPod, with a usb connection to the computer or standalone charger?
14:04:36pondlifeNico_P: So part-files don't work yet, but otherwise should be ok?
14:04:59Nico_Ppart-files ?
14:05:25 Join Thundercloud [0] (
14:05:34pondlifeFiles bigger than the buffer - i.e. files that have to be buffered in multiple parts
14:05:58 Join kaaloo [0] (
14:06:13Nico_Pyes that's right
14:06:17 Part kaaloo
14:07:19Nico_Pit's because currently the only way to free some space in the buffer is to remove a handle
14:08:28pondlifeThat'll need work then :)
14:08:36pondlifeSorry I'm not more able to help.
14:09:52 Quit Wiwie (Read error: 113 (No route to host))
14:12:18GodEatercyberstream: I think your best bet it possibly to post this suggestion to our development mailing list - as not all of the core dev team appear to be awake in here currently
14:12:36GodEaterand then if someone is interested they can contact you directly
14:14:20 Quit Febs (Read error: 110 (Connection timed out))
14:14:22Nico_Ppondlife: what I was thinking is maybe have the internal buffer of a handle be a ring buffer for big files
14:15:05Nico_Pbecause we'll want to have other handles there too and we don't really want them to move
14:15:22pondlifeThat will work, but wouldn't you need to have a guard buffer per file... rather than just one per buffer?
14:15:47Nico_Pwell... what exactly is the guard buffer for ?
14:16:00pondlifeAh, you should look into that!!
14:16:31pondlifeI'm not totally sure, but the MAS has certain requirements - amiconn should be able to help.
14:17:38pondlifeIIRC you need to be able to guarantee to feed complete chunks (contiguous in memory)..
14:18:04pondlifeBut I might have got horribly confused (again).
14:19:08LinusNthe guard buffer is a MAD thing
14:19:18pondlifeMAD or MAS or DMA?
14:19:29LinusNMAD as in libmad
14:19:49pondlifeSo not a MASCODEC requirement at all?
14:20:05*pondlife *is* horribly confused
14:20:37LinusNunless the guard buffer is 30 bytes long
14:21:43LinusNwhich guard buffer are we talking about here?
14:21:48Nico_Pbut what is it used for exactly ?
14:22:28Nico_Pit's not a requirement for every ring buffer ?
14:23:35LinusNare you talking about the guard buffer mentioned in RockboxMemoryLayout?
14:24:17 Quit PaulJam (".")
14:25:07pondlifeThe one used in playback.c - search for GUARD_BUFSIZE...
14:25:48LinusNi see
14:25:52pondlife(But I thought it was also used for MAS somewhere, seems not)
14:25:54LinusNi have no idea what that might be
14:26:09LinusNthe MAS code doesn't use playback.c
14:26:14pondlifeI know
14:26:28pondlifeBut I thought it was referred to somewhere else too.
14:27:15LinusNi wrote the MAS playback code, and i have never heard of the need for a "guard buffer" there
14:27:16pondlifeIt's basically a buffer to ensure a codec's request can be fulfulled with a contiguous block
14:27:46pondlifeSee playback.c - codec_request_buffer_callback()#
14:28:19LinusNok, i see now
14:28:32LinusNthere is nothing of the kind in the MAS playback code
14:28:34preglowcodecs can't access a circular buffer, the guard buffer is just to make sure all data can be access linearly
14:28:43markunLinusN: can you delete VKeyboardSamples for me?
14:28:58LinusNmarkun: is it a wiki topic?
14:29:03*pondlife is sorry he mentioned MAS at all... ;)
14:29:12Nico_Ppreglow: how is it used ?
14:29:42Nico_Pdata that wraps is copied to the guard buffer to make it linear ?
14:29:45markunLinusN: I started it and later merged the info into LoadableKeyboardLayouts
14:29:54LinusNmarkun: done
14:30:34GodEatermarkun: oddly, when I search for "keyboard" in the wiki earlier, your page is what came out, and LoadableKeyboardLayouts wasn't returned at all
14:30:53LinusNNico_P: yes
14:31:03 Quit billytwowilly (Remote closed the connection)
14:31:15GodEateralthough not it is. Maybe I just missed seeing it
14:31:16Nico_Pmaybe the buffering api can avoid the use of this guard buffer
14:31:25 Join billytwowilly [0] (
14:31:31markunLinusN: thanks
14:31:41Nico_Phow does decoding work ? the codec has reads the buffer directly or is there some copying invloved ?
14:31:49 Quit cyberstream ()
14:34:02preglowNico_P: yes, that's how it's used, afaik
14:36:39Nico_Ppreglow: and the codec accesses the buffer directly ? that's why it needs the data to be linear ?
14:36:59 Quit z35 (SendQ exceeded)
14:39:40*linuxstb now has his own asf parser working (outside Rockbox)
14:41:03linuxstbNico_P: Yes, most (if not all) codecs use the "request_buffer" API to get a pointer to some compressed audio that needs to be decoded. This generally needs to be one compressed frame of audio, otherwise the codec would need to continually range-check on the size of the input buffer.
14:42:32Nico_Pwhat size is a frame ?
14:42:37Nico_Pis it always 32K ?
14:42:42GodEaterI would imagnie that depends on the codec
14:42:43linuxstb−−> <−− this big...
14:42:51Nico_Pok :)
14:42:57linuxstbIt's always less than 32KB - which is why the guard buffer is that size.
14:43:20linuxstbOr at least, if the frames are bigger than 32KB, (e.g. APE) the codec handles them in < 32KB chunks.
14:43:53Nico_Pok so for files bigger than the buffer I'll need to have a guad buffer too
14:44:03*Nico_P needs to do some thinking
14:44:12amiconnShouldn't the new api allow to select the guard buffer size from the track loader, to support non-streamable formats?
14:45:13Nico_Pamiconn: you mean specify the guard buffer size when requesting that a file be buffered ?
14:45:16linuxstbIt at least needs to handle those files - IIRC, varying the guard buffer size was just one possible solution.
14:47:19Nico_Pwhat if I define a 32 KB buffer and copy the data to it for the codec to decode ?
14:47:59amiconnMany non-streamable codecs (the mod formats) need the whole file as a single contiguous block
14:48:22Nico_Pah so these can't even wrap... ?
14:48:25linuxstbNico_P: Do you mean for every frame, or only when needed?
14:48:31 Join Arathis [0] (
14:48:39linuxstbNico_P: Yes.
14:48:50Nico_Plinuxstb: well it would only be when needed
14:48:51 Quit SirFunk (Remote closed the connection)
14:49:27Nico_Pthe thing is I might have something like an "internal" ring buffer
14:49:49Nico_Pso it won't be contiguous with the guard buffer
14:50:41linuxstbAssuming the guard buffer is 32KB, that doesn't seem a problem. But if the idea of variable sized guard buffers is implemented, then it will be.
14:54:37preglowi don't think the guard buffer will be needed for nonstreamable formats
14:54:48preglowi a file doesn't fit without wrapping, don't load it
14:55:10 Quit Arathis ("Bye, bye")
14:55:28preglowit needs to be contiguous before use anyway
14:56:20Nico_Pthat sounds logical to me
14:56:41Nico_Pand that way I only need a 32KB guard buffer, which I can use to store a whole frame if needed
14:57:07dionoeacan files always fit in the buffer ? (can't you have cases when they're bigger than the whole buffer?)
14:57:32Nico_Pdionoea: yes, some files can be bigger than the buffer (think of a long flac file)
14:57:51Nico_P(or a small buffer)
14:57:57dionoeathen you need a way to buffer those even if they don't fit (I never use flac :) )
14:58:14preglowwe already do that
14:58:31preglowmy comment was for codecs that can't be streamed and access files in their entirety
14:58:39Nico_Pyes, and with MoB we also need to guarantee that the associated metadata is also on the buffer too at all times
14:58:56 Join Arathis [0] (
14:59:20Nico_Pfor that, I was thinking of treating the data buffer for the big file's handle like a ring buffer
14:59:26 Quit Arathis (Client Quit)
15:02:05 Quit SaLoMoN (Excess Flood)
15:02:17 Join SaLoMoN [0] (
15:02:26 Join webguest60 [0] (i=c023110f@gateway/web/cgi-irc/
15:02:42 Quit SaLoMoN (Excess Flood)
15:02:47 Quit jba_ ("Leaving")
15:02:54 Join Febs [0] (n=chatzill@
15:03:34***Saving seen data "./dancer.seen"
15:04:46 Join SaLoMoN [0] (
15:04:58amiconnNico_P: Why not just copy the metadata of the current file to a static buffer when it starts playing?
15:05:27dionoeawhat if you want to play 2 different files at a time ? :) (To do some fancy audio effect)
15:05:30*dionoea goes back to work
15:05:39Nico_Pwe don't want to limit the size of the metadata. By metadata I mean anything like a bitmap, cuesheet, lyrics file, etc...
15:07:20 Join Guile`` [0] (n=Guile@
15:07:58amiconnSo you think having a piece of metadata blocking the centre of the main buffer would be a good idea?
15:08:01linuxstbThen why not allow the metadata to move?
15:10:48Nico_Pamiconn: maybe that would be better, I don't really know
15:11:09Nico_Perr that was for linuxstb
15:11:30Nico_Pamiconn: I was hoping to be able to avoid having the metadata block the center of the buffer
15:14:51 Join aliask [0] (
15:15:50 Quit spiorf_ ("Read error: 110 (Connection timed out)")
15:16:03 Join spiorf [0] (
15:30:33 Quit Thundercloud (Remote closed the connection)
15:33:05 Join lazka [0] (
15:35:20 Part LinusN
15:36:04 Join SirFunk [0] (
15:36:29 Quit Lear ("ChatZilla [Firefox]")
15:42:19 Join printfXh4 [0] (
15:45:50 Quit inversions (Read error: 113 (No route to host))
15:46:11 Join perplexity [0] (
15:51:26 Join gregj [0] (
15:52:21gregjroxbox hangs on ipod G4, if buffer is set to 2mins - on mp3 playback
15:52:26gregjI used todays build
15:52:55gregjeven worse, it always stops
15:53:08gregjfreezes, and that's it
15:53:12gregjno output on screen, nothing
15:53:24 Join relaxed [0] (n=relaxed@unaffiliated/relaxed)
15:54:26aliaskgregj: That sounds like it could be caused by amiconn's commit from earlier today.
15:56:39gregjit just hangs afeter around 2mins of playback
15:57:23 Join RaRe [0] (
15:57:23gregjcan it be caused by playback-skin ?
15:57:33GodEatergregj: very doubtful
15:57:44aliaskIf it's what I think it is, then it's more fundamental than that
15:57:59gregjI see
15:58:02gregjanything I can do ?
15:58:09GodEatergregj: do you mean the anti-skip buffer ?
15:58:09gregjlike to help you test it, or someting
15:58:10aliaskgregj: Go back to an old build for the time being
15:58:22 Quit Entasis ("Leaving")
15:58:25 Join crop [0] (i=c27f0812@gateway/web/cgi-irc/
15:58:26gregjyeah, but I changed the settings and it didn't helped
15:58:34gregjaliask: define "old build"
15:58:42gregjor even better point me to the one I should test it on
15:58:42linuxstbThe last one that worked for you.
15:58:54gregjthis is the first time I am running rockbox :)
15:59:02crashmatrixaliask, in a few hours I can test the same build for a g5.5, is there need for it?
15:59:10gregjand I see so many thing not quite working
15:59:16 Quit crop (Client Quit)
15:59:16GodEatercrashd_: no - they don't use those processors
15:59:18dionoeagregj: use a version older than this morning
15:59:19aliaskOlder than 15:16 GMT+10 today.
15:59:26crashmatrixrgr that
15:59:36linuxstbOlder builds than amiconn's commit are probably even more likely to freeze.
15:59:37GodEaterand of course I meant crashmatrix ;)
15:59:50 Join Rincewind [0] (
16:00:03GodEatergregj: did it crash before you started changing settings ?
16:00:21aliasklinuxstb: Wasn't scaling disabled in SVN?
16:00:29linuxstbNot on the G4 ipod
16:01:00linuxstbamiconn's commit didn't touch the PP5020 config-target.h files.
16:01:09gregjGodEater: I only listeined to one mp3 beforehand, and it worked fine
16:01:12gregjit seems
16:01:15aliaskYeah, I just looked through the .diff
16:01:30 Quit Gabril ()
16:01:43GodEatergregj: so if you reset back to the default rockbox settings (turn on the hold switch as soon as you see the rockbox logo when you power on) it still works ?
16:01:54linuxstbgregj: Rockbox freezing on the G4 is a long-term problem which hasn't been fixed yet. But today's commit that people are referring to is hopefully a step towards improving it.
16:02:10 Join relin [0] (
16:02:26linuxstbgregj: When did you download the Rockbox you installed? And did you download the "current build" or an "archived build"?
16:02:45amiconnaliask: Only for H10, and I left that for now b/c freezing isn't 100% fixed yet
16:03:20*gregj tries
16:03:45*amiconn hopes to decipher this ln-resembling function tonight
16:03:49gregjalso , I have a question
16:04:14gregjorginal apple's firmware, seems to be loading "as much as it can" into ram - and than turns the HD off
16:04:24relinHow can u load new music on the ipod so that u can hear it with rockboxß
16:04:28gregjwhich seems to be the most resonable way of doing it
16:04:41linuxstbgregj: Rockbox does the same thing.
16:04:55gregjso why the 'skip buffer' option than ?
16:04:58linuxstbrelin: Just copy the files to anywhere on the hard disk.
16:05:15gregjis rockbox mostly written in C ?
16:05:15relinokay ty :)
16:05:16 Join jonshannow [0] (
16:05:21linuxstbgregj: IIUC, that the "low-buffer watermark" - i.e. how empty the buffer gets before it's refilled.
16:05:24 Quit Zagor ("Client exiting")
16:05:58gregjlinuxstb: I see
16:06:07 Quit relin (Read error: 104 (Connection reset by peer))
16:07:18gregjis the photo/g4 ipod have any sort of 'video acceleration' ?
16:07:32gregjanimations seem to be much smoother on apple's firmware
16:08:18*GodEater didn't even know you could play movies on the G4 with apple's firmware.
16:08:34gregjis there any software, iTunes like - that can do rockbox stuff ? I mean - specific to rockbox
16:08:40gregjor plugin for amarok, for instance
16:08:56gregjfor the record, it seems to be working okay with yesterday's firmware
16:09:12aliaskThe new amarok (1.4.6) has rockbox'd iPod support
16:09:27linuxstbWhat does that mean?
16:09:52dionoeagood question :)
16:09:53aliaskIt means it can copy file from your computer's collection to your iPod. I don't think it updates the database
16:10:08gregjit updates the apple's database
16:10:10aliaskI think it also filters the files it copies based on Rockbox's supported codecs.
16:10:22gregjI am asking if I can manage playlists n'shaite
16:10:45gregjhey, the first reason I wanted to try rockbox is to go for oggs
16:10:56gregjI have many more files in ogg than mp3, and I don't feel like converting it
16:11:15aliaskThe best way to manage playlists (IMO) is to use the playlist catalogue feature
16:11:38gregjyeah, this is the achilles foot of rockbox
16:11:46gregjit is very hard to manage playlists, and do the stuff
16:11:50gregjit is far from userfriendly
16:11:51 Join relin [0] (
16:11:59gregjbut hats off for you guys, for enabling it in player!
16:12:37*linuxstb pings saratoga
16:12:53aliaskgregj: What exactly do you want to do with playlists
16:13:12*JdGordon back
16:13:18*amiconn doesn't really understand the need to manage playlists
16:13:25*aliask neither
16:13:49gregjlike, when I first turn on my rockbx, I would be very pleased with it automagicaly scan my ipod, and create default playlist
16:14:00linuxstbHave you tried Rockbox's database features?
16:14:04gregjtoo many ppl seem to have problems and ask about it
16:14:05gregjI did
16:14:07DBUGEnqueued KICK gregj
16:14:07gregjit works
16:14:14gregjbut, now I copied over some files
16:14:22gregjwill it automagically pick up the new files ?
16:14:30FebsIt can.
16:14:32gregjI probably have to rebuild it
16:14:34linuxstbIf you enable the right combination of options...
16:14:47gregjwhich should probably be default for first-start
16:15:10linuxstbI agree. I never use the database though.
16:15:22JdGordonNico_P: nice work :) read the logs and sounds like we have a fair bit to go still :)
16:15:25gregjwhere do I find the database options ?
16:15:30gregjoptions->database options ?
16:15:34gregjdon't exists
16:15:50gregjgot it...
16:16:15aliaskgregj: If all else fails, you could read the manual.
16:16:15Nico_PJdGordon: we have...
16:16:26linuxstbgregj: You can also do a "long-select" press on the Database option in the main menu.
16:16:40Nico_PI've been working on move_handle and adding a test suite for the linked list
16:16:48gregjis rockbox based on linux, or does it have its own multitasking os
16:16:58linuxstbIts own.
16:17:09gregjreason I ask, is I would love to have a look on sources, and try to tweak few things here and there
16:17:16gregjbut without understanding the source - this is hard
16:17:37linuxstbIf you have any specific questions, just ask here.
16:18:46gregjatm, none :)
16:18:56bluebrotherI doubt that if Rockbox would be based on linux it would be any easier understanding the sources.
16:18:59gregjother than - where to start, and if there is any sort of 'overview' for newbies
16:19:16bluebrothercheck the "for developers" section in the wiki
16:19:18linuxstbbluebrother: Unless of course you were familiar with the linux kernel...
16:19:25gregjwell, I understand a lot of linux internals, because I kinda work on simmilar stuff (virtualzation, based on linux kernel)
16:20:37bluebrotherthere is still much around it −− like all the gui stuff etc.
16:22:08gregjis it in C ?
16:22:11gregjor c++
16:22:18GodEatermostly, with a bit of ASM
16:22:28gregjI hope you guys didn't made that mistake, and go for scripting languages
16:22:29CtcpIgnored 1 channel CTCP requests in 0 seconds at the last flood
16:22:29*bluebrother rereads his first answer and realized he wanted to say "much" instead of "any" at first :
16:22:45bluebrotherport Rockbox to JavaScript! ;-)
16:22:54gregjnope, python is now popular!
16:23:04GodEaternot here it isn't
16:23:06dionoeapython on embeded devices ...
16:23:09GodEaterjust ask Zagor
16:23:12dionoealua could work :)
16:23:19gregjyou wanna do it in python, so it is sluggish like hell, and requires reboot every 30s
16:23:24gregjor java....
16:23:46bluebrotheryou'll find some people that pretty much dislike Java here. Like me ;-)
16:24:02linuxstbgregj: No, Rockbox is straightforward C, with no dynamic memory allocation.
16:24:20GodEaterpeople who don't do C are followed round by everyone else shouting "Witch! Witch! Burn them!"
16:24:24gregjmy screen is bit strangly looking
16:24:27gregjbut it last 3 songs!
16:25:00linuxstbgregj: Disabling the cpu clock scaling generally fixes the freezing problems on the G4. You'll need to build Rockbox yourself to do that though.
16:25:04gregjit crashes on mp3 called 'the god that failed' - metallica, black album
16:25:17gregjalways on the same mp3
16:25:57gregjnow database browsing crashed it
16:26:01gregjit seems to crash more and more
16:26:06gregjand now crashed on splas screen
16:26:28gregjapple firmware displays battery flat message
16:26:32gregjso that would explain a lot
16:26:39bluebrotheralways the same mp3? Maybe it's a metadata issue.
16:26:44gregjprobably HD decided to die, because there was not enough power
16:27:12gregjit dislpays metadata alright
16:30:57 Join Wiwie [0] (
16:30:57 Join saratoga [0] (i=980398fe@gateway/web/cgi-irc/
16:31:31linuxstbsaratoga: Hi
16:31:42saratogai saw you have a parser working?
16:31:47linuxstbYes ;)
16:31:55 Quit Wiwie (Client Quit)
16:32:46Nico_Pwhat parser ?
16:32:48saratogai got the code you sent me yesterday working in integer
16:32:49linuxstbI now have code which can quite easily be ported to Rockbox - both in apps/metadata/ for the metadata reading, and in the codec itself.
16:33:08linuxstbSo hopefully this evening I'll be able to implement the codec.
16:33:19linuxstbNico_P: An ASF parser
16:33:25linuxstb(for wma)
16:33:28Nico_Pnice :)
16:33:52gregjyeah, another thing mpeg playback here was really out of sync (audio against video)
16:34:09preglowlinuxstb: saratoga should be excited about that
16:34:13gregjI don't have nerves enough to lookup the implementation, but I think there is something wrong
16:34:14preglowoh, there he is!
16:34:14saratogaI am
16:34:21linuxstbgregj: Did you change any of the default settings in mpegplayer?
16:34:24gregjit should be that hard to do proper fram edropping
16:34:25saratogano more of this ffmpeg nonsense
16:34:54gregjlinuxstb: I did - I asked it to drop frames, and display fps (which is not working either, it only blinks from time to time)
16:35:13GodEatergregj: did you convert your movie for your ipod as in the wiki ?
16:35:23preglowshop time, brb
16:35:40saratogalinuxstb: could I look at what you have? I'd like to add the latest integer decoder
16:35:42linuxstbgregj: "skip frames" should have been enabled by default. As well as "limit fps"
16:36:01saratogaits somewhat different then the fp version, so i think it makes sense to just use the int version for development of the parser
16:36:28linuxstbIt shouldn't affect me at all. All I'm doing is giving the wma decoder the packets of data to decompress.
16:37:10linuxstbBut give me a moment, I'll upload it. I'll disable some of the debugging info first though.
16:37:45saratogalinuxstb: ok thanks
16:38:10aliasklinuxstb: There was another report earlier today where mpeg playback was really fast, and enabling limit FPS didn't do anything (tested in sim as well, same result)
16:38:17JdGordonis there a Bagder in the house?
16:38:35*bluebrother goes looking for coffee ...
16:38:42gregjGodEater: yup sir
16:38:57*gregj feels like pain in the back
16:39:29GodEatergregj: the expression is "a pain in the neck" ;)
16:40:02linuxstbsaratoga: It's here -
16:40:48linuxstbsaratoga: If you could replace the wma decoder with the integer one, and give it me back (feel free to add wav writing to it instead of raw output), then I'll try and make a codec out of it this evening (EU time - i.e. starting in about 3-4 hours)
16:42:34Nico_PGodEater: isn't "a pain in the backside" used too ?
16:42:35GodEaterlinuxstb has become the new codec monkey ;)
16:42:45GodEaterNico_P: not so much
16:42:51GodEaterat least - not where I'm from
16:43:36linuxstbNico_P: Yes, that phrase is used often (at least where I'm from...)
16:43:36pondlifePain in the neck or PITA
16:44:12GodEateryeah, I'd go with PITA over Pain in the backside.
16:44:20GodEaterlinuxstb: different sides of the river huh ? ;)
16:44:50linuxstbGodEater: Only just...
16:45:16Nico_PGodEater: I also prfer PITA :)
16:46:01pondlifeOf course, the A stands for different words on each side of the Atlantic...
16:46:08JdGordonlinuxstb: saratoga: untill Bagder sets us up with a svn, do you guys want to use my server for svn?
16:48:45 Join amiconn_ [0] (n=jens@rockbox/developer/amiconn)
16:49:05 Quit amiconn (Nick collision from services.)
16:49:05 Nick amiconn_ is now known as amiconn (n=jens@rockbox/developer/amiconn)
16:49:34gregjGodEater: yeah, I know - I am kind of using english for 6 years now - eversince I moved to UK
16:50:40 Join kaaloo [0] (
16:50:49 Part kaaloo
16:50:59GodEatergregj: ah - another UK resident - welcome :)
16:52:54saratogaJdGordon: I'll defer to linuxstb/preglow on that
16:53:43 Join low_light [0] (i=c730180a@gateway/web/cgi-irc/
16:55:20 Quit billytwowilly (Remote closed the connection)
16:55:25gregjGodEater: cambridge, for the record
16:55:41 Join billytwowilly [0] (
16:55:42gregjGodEater: so if someone needs to lpay a bit with my ipod - for dev purposes, just gimme a shout
16:57:07JdGordonok, who else is on OCAU....
16:57:17JdGordonsomeone made a thread about the upside/backwards text....
16:59:02 Join mickleby [0] (i=456b3a5c@gateway/web/cgi-irc/
17:01:08 Quit webguest60 ("CGI:IRC")
17:02:00micklebygreetings hallowed realm of wizened sages
17:03:12*JdGordon lookes around...
17:03:18JdGordonno wizended sages here....
17:03:35***Saving seen data "./dancer.seen"
17:03:53 Quit jonshannow ("ChatZilla [Firefox]")
17:04:03*GodEater preens his non-existent beard
17:05:48gregjGodEater: I can confirm, the rpoblem with playback was a battery actually
17:06:03GodEatergregj: good for you ;)
17:06:14gregjGodEater: so rockbox seems to crash on hardware problems caused by low power
17:06:31gregjwhich isn't good, and not very easy to track down probably either
17:07:26GodEatergregj: I think that will depend on the platform. This doesn't happen on my 5.5G ipod, Gigabeat F, or iriver H140
17:11:50 Quit thegeek (Read error: 104 (Connection reset by peer))
17:17:59gregjGodEater: sure, apple is able to do it - than it should be doable here as well :)
17:18:03preglowsaratoga: you guys make it sound like you've found some other wma codec, this is just me being wrong, of course?
17:18:07gregjGodEater: it is probably some sort of hardware event
17:18:28 Quit relin ("ChatZilla [Firefox]")
17:18:28gregjpreglow: other than ffmpeg one?
17:21:14GodEatergregj: apple have the advantage of the hardware documentation however
17:25:28gregjtrue indeed
17:25:35gregjpreglow: show as the source!
17:32:10linuxstbpreglow: saratoga is still using the ffmpeg wma decoder itself, but I've written a new ASF parser, so we don't need to use the ffmpeg one - meaning we can delete a lot more ffmpeg code.
17:32:45aliaskJdGordon: I've seen upside down text popping up on my msn contact list, so I'd say it's become a meme
17:33:07JdGordonoh dear :(
17:35:44linuxstbsaratoga: What's the state of your integer decoder? Is it generating accurate output now?
17:37:43saratogalinuxstb: yes its been accurate for a while
17:37:53saratogaworks on wma9-11 files between 32 and 320k
17:38:11saratogadoesn't work on anything less then 32k because I haven't worked out how the LSP stuff is supposed to work
17:38:34pondlifeAny idea how much optimisation is going to be needed yet? Or too early to say?
17:38:34saratogatheres also some code paths in the joint stereo stuff which I haven't tested, mostly for lack of files that use them
17:38:37linuxstbI didn't spot an answer last night - is the 12-bit accuracy a limitation of your integer decoder, or of the ffmpeg codec?
17:39:04saratogalinuxstb: I get agreement with the ffmpeg fp decoder to 12 bits
17:39:15saratogai'm not sure if thats because I'm more accurate or its more accurate
17:39:20saratogabut I get at least 12 bits
17:39:26preglowcompared with microsofts decoder?
17:39:32GodEaterJdGordon: hehe - freenode won't let me set my nick as upside down :(
17:39:37saratogaI haven't checked their decoder yet
17:39:46linuxstbWas that the old ffmpeg decoder, or the SVN version, which claims better accuracy?
17:39:48JdGordonshame :p
17:40:01linuxstb^current SVN version
17:41:06gregjcrash again, batter flat again :(
17:41:15gregjon such a nice tune - "forsaken" by dream theater
17:42:47saratogamine does a few things a bit better then the ffmpeg decoder, so maybe I have better results
17:43:13saratogadepends how much I lose in the fixed fft
17:43:25 Join My_Sic [0] (
17:46:20gregjcan rockbox record to mp3/ogg/whateverelse than plain wav - in real time ?
17:48:48saratogalinuxstb: got the parser outputing wav in fixed precision
17:48:54peturdon't know how ipod recording is... *looks at linuxstb*
17:49:25saratogathough oddly it returns -1 (though the file decodes anyway)
17:49:42 Join bluey- [0] (
17:50:26linuxstbsaratoga: Are you talking about the -1 at the end of asfplay?
17:51:00saratogalooks like a problem in wma_decode_superframe somewhere
17:51:04saratogashould I look into it?
17:51:59linuxstbNo, I think that's a -1 from my parser - because I'm not detecting the end of stream correctly yet. I think you can just ignore it.
17:52:27 Part maffe
17:52:45 Join maffe [0] (
17:53:15saratogaheres what I got
17:53:35saratogathe malloc stuff is still in mem.c, but I don't think anything still calls it
17:55:07*linuxstb deletes mem.c to find out...
17:55:57linuxstbHurray, a malloc free codec ;)
17:56:18linuxstbNothing in either the asf parser or the wma decoder.
17:56:23lazkajust wondering, why is malloc bad?
17:56:42linuxstbThere's a wiki page about that, but I forget the name
17:57:23lazkafound it, thanks
17:57:27saratogalazka: rockbox doesn't have a malloc function, so its bad if you try and call it on the target :)
17:58:16preglowsomeone finally made a page about it, eh
17:58:29pondlifeIt's been there a while.
17:58:50preglowif less than 50% of it is four letter curse words and derivatives, i'll be surprised
18:00:15saratogalinuxstb: so how do you want to proceed?
18:00:58linuxstbsaratoga: I'm working on implementing a codec based on what we've got at the moment. Hopefully it won't take too long (an hour or two).
18:00:59 Join jgarvey [0] (
18:01:04 Quit bdgraue (Read error: 104 (Connection reset by peer))
18:01:12saratogalinuxstb: thats great to hear!
18:01:36linuxstbIt should be straightforward now we've exorcised malloc from the code.
18:01:47*preglow crosses self
18:01:47pondlifeWooh! Some kind of WDM playback in two hours...
18:01:55saratogai'm getting lunch, then working on into my new apartment, but I should be on later
18:01:56*pondlife counts down
18:02:09*linuxstb starts to feel some pressure and regrets mentioning a timeframe...
18:02:32saratogafeel free to gmail me at mgiacomelli@ if i'm not on irc
18:02:47saratogai can't wait to see how slow this codec really is ...
18:03:04linuxstbWhich reminds me, I wonder how the Gigabeat S port is coming along...
18:03:13preglowthere is an effort?
18:03:15saratogaand how much of a difference changing from 64x64->64 mul to 32x32->64 really makes
18:03:27saratogapreglow: yeah, theres actually someone writing code for it now
18:03:32preglowsaratoga: who?
18:03:33 Quit saratoga ("CGI:IRC")
18:03:38preglowwhy, you!
18:03:39 Join saratoga [0] (i=980398fe@gateway/web/cgi-irc/
18:03:41linuxstbpreglow: I believe so, at least an effort to run third-party code.
18:03:51saratogapreglow: ptw419 (I think)
18:03:55pondlifelinuxstb: No pressure, but it would be nice to have a realtime port on the H300 by the morning.
18:04:08linuxstbSend me your H300 in the next 2 hours, and I'll make it so.
18:04:22saratogalast I heard they were having some trouble getting the display driver to do more then just blink the screen
18:04:23linuxstbDamn, you're near London, aren't you?
18:04:26*pondlife jumps in his car
18:04:39*linuxstb should pay more attention to rasher's map
18:04:51pondlifeSadly I have badminton to play shortly, so you can blame me for the lack of codec.
18:06:20saratogagigabeat stuff:
18:06:23XavierGrI am itching to know how slow wma will be on coldfire...
18:06:54 Join bdgraue [0] (
18:06:57preglowpft, some iram and a wee bit of asm, and it'll be cruising along nicely
18:07:18XavierGrand the moment wma will sound on an ipod hell will freeze over :)
18:08:53pondlifeBut that'll happen before Linux on a Zune... :)
18:09:27 Quit JdGordon ("Konversation terminated!")
18:10:52XavierGrhehe indeed
18:11:29 Quit petur ("work->home")
18:11:39 Quit pondlife ("disconnected has pondlife")
18:11:39linuxstbpreglow: Has saratoga told you how much RAM the wma decoder is using?
18:11:49 Quit My_Sic ("Miranda IM! Smaller, Faster, Easier.")
18:12:23*linuxstb waits for saratoga to tell preglow
18:16:36*gregj looks at dev wiki. and stuff
18:16:42gregjit seems like you guys have girls in the team
18:16:44 Nick swolbhsub is now known as bushblows (n=bushblow@
18:16:58gregj 's link to "tower" is 404
18:17:08gregjthis one
18:17:10preglowthere are some around, yes
18:18:07 Quit aliask ("ChatZilla [Firefox]")
18:19:19 Quit Genre9mp3 ("I don't suffer from Rockbox psychosis. I enjoy every minute of it.")
18:19:20 Join Thundercloud [0] (
18:22:05 Join thegeek [0] (
18:22:20 Join My_Sic [0] (
18:23:04lazkacan someone tell me where i can find the "Show ID3 Info" in the source?
18:23:33linuxstbLook in apps/lang/english.lang for the string, then search the source for the language ID
18:23:51low_lightlazka: in screens.c
18:24:29 Quit My_Sic (Read error: 104 (Connection reset by peer))
18:25:58lazkalow_light: thx
18:26:29 Join Soroosh [0] (
18:27:24low_lightha...the iphone hacking has begun...
18:29:38 Join inversions [0] (
18:30:06 Quit Thundercloud (Remote closed the connection)
18:32:19Sorooshhi. What is the safest way to upgrade rockbox?
18:34:30preglowthere are multiple ways?
18:34:43preglowjust overwriting the old rockbox folder with a new one should work wonders
18:35:09 Join anathema [0] (
18:36:03 Quit obo ("KVIrc 3.2.6 Anomalies")
18:36:36 Quit thegeek ()
18:37:32 Join kaaloo [0] (
18:38:07 Join pseudo_ [0] (
18:38:32 Part kaaloo
18:40:36Sorooshok, that's what I was thinking of doing, but wanted to be sure that works.
18:41:24 Join The-Compiler [0] (
18:45:45 Quit mickleby ("CGI:IRC (Ping timeout)")
18:46:37 Join My_Sic [0] (
18:49:37 Quit printfXh4 (Nick collision from services.)
18:49:39 Nick pseudo_ is now known as printfXh4 (
18:51:52amiconnlinuxstb (or anyone else): Is there a h10 firmware dump available somewhere?
18:53:30 Quit inversions (Read error: 110 (Connection timed out))
18:57:16 Join kaaloo [0] (
18:57:21preglowamiconn: isn't the timer interrupt reenable after clock frequency change needed anymore on pp5020?
18:57:27 Quit kaaloo (Client Quit)
18:58:05amiconnIt still isn't 100% freeze free
18:58:08preglowrelated to your change or something else?
18:58:12 Join kaaloo [0] (
18:58:26 Part kaaloo
18:59:46low_lightamiconn: the h10 bootloader is on bagder's mi4 page
18:59:51amiconnThis kludge never was very logical, and my change makes it unnecessary. However, PP5020 still freezes from time to time, otoh I'm not yet done with analysing what the OF does
19:00:03amiconnlow_light: H10 *firmware*, not bootloader
19:00:40preglowit was a kludge indeed
19:00:50preglowpure guesswork
19:01:08preglowany idea what the magic sequence does?
19:01:31amiconnI think it "kicks" the pll counters
19:01:35 Join obo [0] (
19:02:02preglowas in how?
19:02:14amiconnThe lower 16 bits of PLL_STATUS are equivalent to the lower 16 bits of PLL_CONTROL on PP5020
19:02:39amiconnThat is, they are equivalent after locking... no idea how to read the short transient status
19:02:58amiconnOn PP5022, PLL_STATUS works very different
19:03:27amiconnNot only that there is the lock indicator bit; the lower 16 bits have a different value independent of pll numerator and denoominator
19:03:37***Saving seen data "./dancer.seen"
19:09:03 Join chuckwag0n [0] (
19:09:45chuckwag0nwhat's the best way to create a playlist using rockbox?
19:10:24linuxstbamiconn: The H10 firmware should be on Bagder's mi4 page. You'll probably need his/MrH's mi4code program to decrypt it. (google for mi4code)
19:10:26low_lightamiconn: i know that I said's probably easier to re & it contains a diagnostic/test mode for the hardware
19:10:56amiconnYes, and that's why I do _not_ want the bootloader.
19:11:06 Join lost|X40 [0] (
19:11:21amiconnThe bootloader handles booting and diagnostics. Real-life clock change is more interesting...
19:13:44low_lightamiconn: whatever...thought these strings were interesting though "Clk: 24MHz" "Clk: 33MHz" "Clk: SLOW" "Clk: FAST" "Clk: PLL"...shrug
19:14:43 Join Siltaar [0] (
19:17:49 Join p00p [0] (n=hoor@
19:18:59 Quit p00p (Client Quit)
19:19:14 Quit Nico_P (Remote closed the connection)
19:21:20 Join petur [0] (n=petur@rockbox/developer/petur)
19:21:25 Quit chrisjs169 ("Konversation terminated!")
19:22:55peturmarkun: visa bill finally arrived ;)
19:22:59*amiconn is a bit puzzled about the ipod video clock setting routine
19:23:32amiconnIt looks very similar to the mini version, but handles yet another register, and seems to check for a pp5026 (??)
19:24:00peturgeneric pp code?
19:24:49preglowlow_light: i believe that stuff has already been investigated
19:25:52 Quit lostnihilist (Read error: 110 (Connection timed out))
19:27:57saratogapreglow: the version using ffmpeg was something like 500k for everything
19:28:10 Quit desowin (Read error: 110 (Connection timed out))
19:28:19gregjhmm, is there a mp3 decode aiding or decoding chip on ipods ?
19:28:26saratogathe new version will be smaller, though probably still around 400kb
19:28:43saratogathe wma decoder itself is enormous due to massive numbers of lookup tables
19:28:55saratogagregj: no
19:29:34gregjWolfson Microelectronics WM8975 CODEC
19:29:35preglowwhat are the tables used for?
19:29:47saratogagregj: that is the DAC
19:30:05saratogapreglow: the FFT and MDCT use seperate trig look up tables, one for each window size
19:30:16saratogaso thats 12 tables there
19:30:30gregjsaratoga: ok, I was hoping it at least does FFt/ MDCT and perhaps huffman
19:30:36gregjlike some chips
19:30:41preglowsounds weird, i would expect one big table to be able to be used for all of them
19:30:44gregjcos than you could use it for ogg as well
19:30:50saratogapreglow: I will eventually do that
19:30:54preglowsince the lower sized tables really should be resampled versions of the bigger ones
19:31:06saratogagregj: I don't think any modern players use a coprocessor for audio
19:31:46gregjthere are still some arm cpus sold nowdays that have mp3/dac caps
19:32:06saratogapreglow: the table system is a legacy of how ffmpeg works, not something that should be there
19:32:15saratogagregj: no I don't think there are
19:32:19preglownot a prio anyway
19:32:41saratogai imagine it will hurt when we need to move the stuff into IRAM, but yeah, i'm putting it off
19:33:03saratogai need to print out the tables, plot them in matlab, and figure out how to fold them together
19:33:04amiconnBtw, the huge tables for wma are the reason why wma cannot be implemented for the mas
19:36:21markunpetur: they really took their time
19:37:09peturseems the hotel was a bit slow at passing the bill to visa, they only did it end of the month
19:39:01 Join ackbahr [0] (
19:40:05 Quit My_Sic (Read error: 110 (Connection timed out))
19:42:02 Quit ackbahr (Client Quit)
19:42:55saratogaheh, plotted the tables ffmpeg uses, and the 256 point table is just the 512 point table with only the even points ...
19:43:08saratogaso this optimization should be pretty easy
19:45:06*linuxstb listens to a WMA file in the sim
19:45:51saratogais it my test file?
19:45:58linuxstbBut it won't compile for a target - too big for the codec buffer...
19:46:08saratogahow big is the codec buffer?
19:46:19saratogai'm surprised we went over
19:46:32saratogacan you send me the map file?
19:46:37linuxstbThe output buffer is too big - it looks like it can decode up to about 80KB of data.
19:46:50saratogawhich buffer is this?
19:46:52linuxstbIt's OK - we're not using the 512KB codec malloc buffer - I'll just use that, and it will fit.
19:47:02 Join DerPapst [0] (
19:47:13linuxstbThe buffer I'm passing to the wma_decode_superframe function.
19:47:27saratogai don't think a superframe can be that large
19:47:47gregjcan you point me to a file/web site where I can find info on how to build the thing from sources
19:48:00saratogai think superframes can be at most 2 or 3 frames, and frames are 2048 samples
19:48:14amiconnThis logarithm function looks like it is floating point, while it takes integer arguments
19:48:28DerPapstsaratoga, linuxstb: nice work :)
19:48:35amiconnAt least a code path leads to something that looks like floating point exceptions
19:48:46DerPapstgregj: there is a guide on
19:48:48saratogagregj: check the wiki, theres a useful "for developers" section
19:49:00linuxstbsaratoga: I think we should change it to decode one frame at a time, but that's for later.
19:49:05amiconnNow the question is how to test this function....
19:49:16DerPapstgregj: and afaik you've aready seen the "For Developers" headline at the index.
19:49:24saratogalinuxstb: doesn't it decode one superframe at a time now?
19:49:29DerPapstthat one for petur ;)
19:49:45amiconnWe have no SIGFPE on target ;\
19:49:48saratogaor do you mean call decode frame directly?
19:50:57linuxstbsaratoga: Yes, I mean decode the small (2048 sample) frames in each iteration of the main loop, rather than the whole superframe - which I've seen can be 10 frames (maybe more)
19:51:19saratogalinuxstb: woah, thats surprising
19:51:31saratogai thought superframes were just a frame + bit resivior
19:51:43saratogasince it doesn't use them if you disable the bit res
19:52:07 Join davina [0] (
19:54:53linuxstbHmm, I'm just getting nasty noises on my gigabeat though...
19:55:05markuntypical rockbox sound
19:55:05 Quit chuckwag0n ()
19:56:32saratogalinuxstb: the code doesn't check the return value of decode_superframe, so its possible its just erroring out
19:56:40saratogaand by possible I mean virtually certain
19:56:55peturmarkun: bag of shit?
19:57:04markunpetur: they make nasty noises, right?
19:57:11DerPapstpetur: of course :P
19:57:44 Join My_Sic [0] (
19:57:47GodEater_esp. if you hit people with them
19:58:05saratogacould add if(res<0) goto exit; to line 784 and see what happens
19:58:16saratogaline 784 of asfplay.c
19:58:25saratoga(if you didn't notice that already when you made the codec)
20:00:15linuxstbYes, I've done that now, and it stops the garbage output, but doesn't help fix the problem...
20:00:28linuxstbGive me a moment to clean the code up, and I'll post a patch.
20:02:03 Quit billytwowilly (Remote closed the connection)
20:02:20 Join billytwowilly [0] (
20:03:10 Join Lear [0] (i=chatzill@rockbox/developer/lear)
20:03:20 Join Nico_P [0] (
20:04:34 Join chrisjs169 [0] (
20:09:07linuxstbIt's OK in the sim, but still not working on target (at least, not the gigabeat)
20:09:28linuxstbApply it to a clean source tree, and run configure
20:10:17XavierGrlinuxstb: you can listen to wma in the sim now? (without any hiccups or something like that)
20:10:18 Quit pontape ()
20:11:01linuxstbXavierGr: Yes.
20:11:10XavierGrnice going :D
20:11:39amiconnlinuxstb: x86 sim?
20:11:49linuxstbXavierGr: It's my asf parser and codec, using saratoga's integer and malloc-free wma decoder.
20:11:51linuxstbamiconn: Yes
20:12:24XavierGrwell bravo to both of you, it is getting very interesting
20:12:41 Join Genre9mp3 [0] (n=yngwiejo@rockbox/contributor/Genre9mp3)
20:12:51XavierGrI would try to apply the patch too but now I have to go
20:12:59*linuxstb doesn't want to know how much the asf parser will add to the core binary size...
20:14:12 Quit The-Compiler (Connection timed out)
20:14:21 Join The-Compiler [0] (
20:14:28 Join desowin [0] (
20:15:26 Quit bluey- ("Leaving")
20:17:01saratogalinuxstb: any idea where it dies?
20:17:37saratogaI'm wondering if its the same place in decode_block where it died with the ffmpeg parser
20:18:10 Join AdmiralJ [0] (
20:18:13linuxstbsaratoga: No idea...
20:18:26saratogawell, i know the first place to check at least
20:18:33linuxstbAlthough it doesn't crash, so you can enable logf
20:19:20saratogaI don't have my gigabeat handy today (I'm at school)
20:19:26saratogai will check tonight
20:20:02saratogabut I bet if you uncomment line 2455 of wmadeci.c you'll see where it gives up
20:21:22 Join ompaul [0] (n=ompaul@freenode/staff/gnewsense.ompaul)
20:22:33linuxstb"rb->splash" doesn't exist in codecs, but I'll enable LOGF and see what v is equal to.
20:22:52saratogamostly just need to check if it returns -4
20:23:00saratogathats what it always did before witht he old parser
20:23:12saratogaits entirely possible that my "static" codec doesn't work correctly
20:23:24linuxstbI can't see how the parser would affect things.
20:23:44saratogabefore I thought it was a problem in the ffmpeg parser
20:23:47saratogai'm guessing its not now
20:24:02 Quit printfXh4 (No route to host)
20:26:25 Join digin4 [0] (n=nbo@
20:33:29lazkadoesn't work on my sansa..
20:35:28*amiconn hates this usb reboot bug :(
20:35:48*preglow too
20:38:18amiconnThis supposed logarithm function is in fact a simple integer division
20:38:29amiconn(or not so simple −−- )
20:40:13 Join linuxstb_ [0] (n=linuxstb@rockbox/developer/linuxstb)
20:41:11 Join Thundercloud [0] (
20:41:38 Quit linuxstb (Nick collision from services.)
20:41:40 Nick linuxstb_ is now known as linuxstb (n=linuxstb@rockbox/developer/linuxstb)
20:44:33gregjhow do yo uguys debug ipod rockbox over serial console ?
20:44:50gregjI would belive, it should be doable over FW or USB , but I have no idea
20:44:53gregjhence the question
20:44:59saratogagregj: I don't believe anyone knows how
20:45:09linuxstbIf we had firewire, USB or serial port drivers...
20:45:14markungregj: tomasz wrote a driver for that while working on the iriver ifp
20:45:50markungregj: firmware/usb_serial.c
20:46:00 Nick bushblows is now known as swolbhsub (n=bushblow@unaffiliated/bushblows)
20:46:03linuxstbHmm, my asf parser doesn't seem to work at all on Coldfire...
20:46:14saratogalinuxstb: endianness?
20:46:18linuxstbI expect so
20:46:28linuxstbThe patch I posted didn't even compile...
20:46:54gregjI am asking, because it would be very usefull - for crash detection, and shaite - to be able to at least get a core or backtrace off that device
20:47:11*DerPapst glances at his clock ;)
20:47:19saratogagregj: USB support would be useful for more then just that!
20:48:47gregjsaratoga: true
20:48:55gregjanother of top of my head question
20:49:10gregjwhy thread function doesn't take void*, as it does oeverywhere else
20:49:22gregjit is usefull, so you don' tneed global variables than
20:50:44gregjI can create patch against it, and add that - but obviously can't modify plugins and shaite to use it :)
20:51:16 Quit low_light ("CGI:IRC")
20:52:02gregjI just looked on "todo" list for ipods, and seems like handling crashes is on it at high priority
20:52:16gregjI guess the rockboxes OS is unable to generate backtraces
20:52:27gregjshame :'
20:55:18preglowamiconn: what logarithm function?
20:55:20*gregj is asking too many questions, has few in queue waiting :)
20:55:33amiconnpreglow: The one that isn't....
20:56:13gregjI see it already has cpu_boost()
20:56:22gregjcool, wonder if that works on ipods
20:56:30 Join KlrSpz [0] (n=KlrSpz@
20:56:45amiconnpreglow: There is a function that is used in calculating some register values, and I needed to know what the function does
20:57:29KlrSpzany huge strides worth looking at for the ipod port? i haven't updated rockbox in about 10 months
20:58:04GodEater_KlrSpz: a few new codecs
20:58:14KlrSpzi remember some patches/hacks to run some alternate WPS screens, any worthwhile changes for ahat
20:58:16KlrSpzok, loading
20:58:29KlrSpzwell actually i'm on that page
20:58:50KlrSpzmight just have to try it out when i get home
20:59:08KlrSpzso much cool stuff oing on these days.. been messig with compiz-fusion a lot lately
21:01:04 Join EnterUserName [0] (
21:01:09EnterUserNameis nano 2nd generation impossible
21:01:15EnterUserNameis that why it hasn't been supported yet :)
21:01:43GodEater_EnterUserName: currently yes
21:02:08EnterUserNamedoes that mean it is possible but still not been successful :P
21:02:18GodEater_no - it means we have no idea where to start
21:02:38EnterUserNameare there 2 gig second generations?
21:02:43EnterUserNameor are they old generations still?
21:02:54GodEater_no idea
21:03:04saratogacheck wikipedia
21:03:14GodEater_or apple's website
21:03:24EnterUserNamehow do you tekll if its 2nd generation?
21:03:33markunthey look different
21:03:33GodEater_it's coloured metal
21:03:37GodEater_not black or white plastic
21:03:38***Saving seen data "./dancer.seen"
21:03:57markunthe 2nd gen looks more like the mini
21:04:07GodEater_only smaller obviously ;)
21:04:27EnterUserNamehows the video process
21:04:34EnterUserNameis it still lagging when you watch it or has it improved
21:04:41EnterUserName(i have my first generation right now withr ockbox)
21:04:42GodEater_on what ?
21:04:47markunon the nano it's good I think
21:04:49EnterUserNamefirst generation..
21:04:50EnterUserNameoh ok
21:05:01EnterUserNameI havent' updated in a while :)
21:05:22linuxstbsaratoga: I've enabled logf, and am getting "v was 6", "v was 0", "v was 6", "v was 0", "v was 0", "v was 0"
21:05:22markun32fps at full resolution
21:05:26GodEater_beat me to it markun :)
21:05:51markunGodEater_: you could have added the #Performance to the URL :)
21:06:11GodEater_I leave that as an exercise for the reader :)
21:06:51EnterUserNamelol im not sure what that means markun
21:07:08GodEater_32 frames per second
21:07:12GodEater_that's very watchable
21:07:45GodEater_cinema = 24 frames per second
21:09:13linuxstbsaratoga: So yes, wma_decode_block is failing with error -4...
21:09:21 Quit My_Sic (Read error: 110 (Connection timed out))
21:09:24*GodEater_ goes for food
21:09:52KlrSpzwoah wait, you can watch videos on your nano with rockbox now?
21:10:09saratogalinuxstb: yeah thats my old bug then
21:10:10KlrSpzi thought that was considered "impossible" last year
21:11:17saratogaits really strange, the init code seems to work normally, but decode_block somehow gets the wrong values when it reads in the stream
21:11:27*amiconn wonders what can be different between x86 sim and arm target
21:11:45amiconnBoth are little endian 32 bit, and arm doesn't crash so it's not an alignment issue
21:12:13amiconnsaratoga: Is there some struct mapped to the block?
21:12:21amiconnPerhaps it's a struct padding issue...
21:12:38linuxstbchar is another issue (signed/unsigned), but I can't see any problems like that
21:12:54saratogaamiconn: i haven't looked at that part of the code carefully, but I believe it just has a pointer to the buffer and then reads in bytes and words as it goes
21:13:19saratogaits a mess of #defines that obscure how it actually works, so i could be mistaken
21:13:45 Quit mud-rb (Remote closed the connection)
21:14:05 Quit desowin ("use linux")
21:14:09linuxstbstruct padding could be the issue though...
21:15:54gregjwhat's the 'ADC driver' thing ?
21:16:28peturmostly reading buttons & co
21:16:37saratogagregj: I believe its for teh wolfson codec's ADC
21:16:37gregjonly SPI thing I know is 'simple programming interface', thing used to program and debugg embeded cpus
21:16:40gregjbut that's probably not it
21:17:39gregjbut it works, otherwise I wouldn't be able to hear anything - right ?
21:17:42gregjok, this one I know too
21:17:44gregjthanks :)
21:18:07gregjmost embeded audio things use it
21:18:13saratogawell ADCs are for recording, but I believe it is working now
21:18:18gregjlike hardware mp3 decoders, voip dsps, and stuff
21:18:35gregjif i start recording, playback doesn't work anymore
21:18:42gregjI have to reset the boc
21:18:59saratogaamiconn: does this look ok: #define unaligned32(a) (*(uint32_t*)(a))
21:19:30linuxstbsaratoga: You've copied a load of ffmpeg stuff into the wmadeci.c file?
21:19:59amiconnNot at all...
21:20:34amiconnIf a were actually unaligned, it would cause an exception on arm
21:21:06*gregj is trying to find something to hack for himself, like within hour or two
21:21:40peturgregj: start with setting up the crosscompiler, that will take some time ;)
21:21:49gregjI got it compiled and on the box
21:21:55gregjI have 4 way pentium D here
21:22:02gregjit wasn't that long ;)
21:22:15gregjI already compiled rockbox, and put that on my device
21:22:27 Quit The-Compiler (Read error: 104 (Connection reset by peer))
21:23:56saratogalinuxstb: actually, someone else did that and I just never changed it back
21:24:12linuxstbWell, I never did.... IMO it needs fixing.
21:24:33saratogai'd rather leave it alone for now
21:24:42linuxstbe.g. there is the wmadecodecontext struct shared between wma.c and wmadeci.c with different definitions...
21:24:45saratogai can fix it once we have something working
21:24:57linuxstbI think you need to fix it in order to have something working...
21:25:19 Part chrisjs169 ("Konversation terminated!")
21:25:25saratogawhat wma.c ?
21:25:27EnterUserNamewhere's there the good custom firmwares
21:25:29EnterUserNamefor nano ipod
21:25:34linuxstbThe codec.
21:25:38EnterUserNamewhich one should i do for videos any suggesstions?
21:25:43gregjthanks to you guys I can listein to another 500 oggs I have :) makes my day
21:26:01linuxstbThe codec declares the wmadecodecontext struct (and allocates the memory for it), and passes a pointer to it to the wmadeci.c functions.
21:26:06EnterUserNameU know the ppl who compiled it from source and added patches
21:26:10saratogalinuxstb: I don't have a wma.c except for the one you added as part of the codec interface
21:26:30linuxstbYes, that's the one I mean.
21:27:11saratogaoh wow, just noticed wmadec.h
21:27:12EnterUserNameanyone? can someone help me find a link :)
21:27:27saratogaEnterUserName: they're called unsupported builds for a reason
21:27:31EnterUserNamethank u
21:27:32EnterUserNamefound it
21:28:06saratogathis is a problem
21:29:20EnterUserNameI think i have an older senabs build..
21:29:28EnterUserNameit seemed to work good
21:31:09EnterUserNamedoes the latest firmware on rockbox's site accept album art?
21:31:26lazkaEnterUserName: no
21:32:16EnterUserNameanyone using senabs build?
21:32:27 Join kaaloo [0] (
21:32:33 Nick swolbhsub is now known as bushblows (n=bushblow@
21:32:44 Part kaaloo
21:33:15EnterUserNameanyone compiling from source using gentoo?
21:34:01*gregj renews the question, anything simple that I can do for starters - usefull ?
21:34:34peturgregj: try making a plugin
21:34:39gregjI can alreayd see - for my personal benefit - I can make the metronome plugin bit more future complette
21:35:00peturso go get started
21:35:13peturpost your code to the tracker
21:35:41amiconnNow if I knew what 0x70000034 _actually_ does :\
21:36:08peturgregj: you know that we have a real name policy? We only accept code if you give your real name to be put in the credits file
21:36:45gregjeasy easy
21:36:47gregjI am not a bot
21:36:57gregjare you a man:<yes> <no>
21:37:02*petur is just mentioning up front
21:37:23gregjwhere do you want me to fly to, and who to kiss/shag/buy a beer
21:37:25peturwe had some anonimity freaks before
21:37:37gregjbut first - I want to try to do something that works
21:38:04gregjI already see that you guys could use some architectural help, but most of the ideas I had - I can see already written down on 'todo' pages
21:38:09*amiconn should have checked boost ratio before the change
21:38:40gregjmy guess is that for instance - cpu saving code - is quite half finished
21:39:15gregjquestion number two, is there a way for kernel to know how much cpu is used currently? (like on 'top') - this would be usefull for freq scaling
21:39:42amiconnWe *do* freq scaling
21:39:49gregjI know
21:40:01gregjbut it seems to have problems, is it feature complete ?
21:40:54gregjalso - someone mentioned, that it is causing problems in ipod - so...
21:41:09 Quit KlrSpz ()
21:41:27amiconnI am working at that very problem
21:41:39gregjalright, got you - I am trying to make a plugin
21:41:40amiconnCPU scaling on PP5020 causes occasional freezes
21:41:56amiconnNow if we had a datasheet... but we don't
21:42:12amiconnPortalplayer is the kind of "keep it secret" company, unfortunately
21:42:38scorchewell, nvidia now
21:42:49peturnot any better ;)
21:43:19gregjok, so question regarding plugins
21:43:33 Quit Ryuu (Read error: 104 (Connection reset by peer))
21:43:33 Quit HellDragon (Read error: 104 (Connection reset by peer))
21:43:35gregjI copied over hello world, to some other name, added it to SOURCES
21:43:41gregjwhat's next
21:43:47 Join HellDragon [0] (i=JD@unaffiliated/helldragon)
21:44:15gregjyeah, I did that
21:44:27gregjbut can't find my file in rocks/
21:44:34gregjwhich means - eee
21:44:34 Nick HellDragon is now known as Ryuu (i=JD@unaffiliated/helldragon)
21:44:53scorchewas it compiled?
21:44:54 Join c250Owner [0] (i=44ce4ef1@gateway/web/cgi-irc/
21:44:59gregjfsck knows
21:45:21scorchewell, there is a reason you see what goes on in the compiling process....scroll up
21:45:24gregjdoesn't seem to
21:45:28c250OwnerIs anything being done to put Rockbox on the Sansa c200 series?
21:45:39gregjanywhere else I need to add it ?
21:45:45scorchec250Owner: not atm
21:45:47 Join HellDragon [0] (i=JD@unaffiliated/helldragon)
21:45:57*amiconn wonders whether handling 0x70000034 does the trick, and what other impacts it has (battery runtime etc)
21:46:04c250OwnerAlright, thanks
21:46:07 Quit c250Owner (Client Quit)
21:46:16amiconnAt least I have it running, and it *seems* to work like the OF would do it
21:46:42gregjamiconn: about that PP, so you're not able to register as developer there - and get some sheets, or even sign a document - that will not allow you to release docs, but source - yeah
21:46:42amiconnInterestingly enough, boost ratio on mini g1 and mini g2 is different when playing the same album (!)
21:46:45peturhow do you tell?
21:47:05gregjright, so how can I make sure my plugin do compile ?
21:47:08amiconngregj: You need to sign an NDA (and probably pay $$$)
21:47:33amiconnBut even without the $$$, docs under NDA are mostly useless in the opensource world
21:47:45gregjusualy not
21:48:13amiconnlinuxstb: Playing an ~128kbps AAC album, I get ~60% boost on mini g2, and ~55% boost on mini g1.
21:48:21gregjsee, I work for opensource company (xensource), we have access to docs and stuff from intel and amd, under nda
21:48:30 Part Soroosh
21:48:35gregjbut we develop software that's mostly opensource (around 50%)
21:48:45gregjplease tell me what's with plugins
21:49:51 Join jhMikeS [0] (n=jethead7@rockbox/developer/jhMikeS)
21:49:53amiconnlinuxstb: No crash so far...
21:50:31amiconn(I know that doesn't mean much at this point... :-| )
21:51:01EnterUserNameHEy i know its not supported but does anyone use EVilg's build?
21:52:20EnterUserNameCan someone tell me if evilg's work with nano?
21:52:25peturgregj: make sure you didn't add your file to some ifdeffed section of SOURCES?
21:52:27scorcheEnterUserName: have i not told you that they are unsupported for a reason?
21:52:33EnterUserNameok nevermind
21:52:46saratogahow do I revert my source to svn?
21:52:47gregjpetur: yeah, I figured - waiting for make to finish
21:52:48scorchewe have the unsupported builds area in the forums for a reason as well..
21:53:53EnterUserNamescorche: yup i figure i'd post a question
21:56:11 Join SeanInglis [0] (i=566fa167@gateway/web/cgi-irc/
21:56:41amiconnIf that fixes the freezes, I have a strong suspicion what actually caused them...
21:58:11 Quit digin4 (Read error: 110 (Connection timed out))
21:58:21preglowamiconn: how long does it usually take for a crash to appear?
21:58:32amiconnANything from minutes to hours
21:58:33preglowamiconn: what suspicion? :>
21:58:33 Join digin4 [0] (n=nbo@
21:59:25amiconnIf you touch the pll too early after switching to the fixed 24MHz source, things might not be fully decoupled yet
21:59:28EnterUserNameHEy guys i cant find this in the wiki whats the Ipod Nano - SIm?
21:59:46amiconnI now added handling for the 0x70000034 register, and that goes in between...
21:59:50saratogaEnterUserName: theres a simulator for all rockbox targets
21:59:54saratogacheck the wiki
21:59:56SeanInglisDipping into IRC during the day, lots of discussion on frequency scaling. Any good links explaining what this is WRT Rockbox? (allowing the CPU to idle when there's no demand, as opposed to furiously doing nothing?)
21:59:59petursaratoga: I'm not an svn expert but I think: svn revert . -R
22:00:32saratogapetur: was missing the .
22:00:34preglowamiconn: what handling is there? not a wait-loop, i assume? that would have been a dead giveaway
22:01:06saratogaSeanInglis: theres a wiki page for it somewhere, but yeah, thats the gist of it
22:01:15*linuxstb listens to a wma file on his gigabeat...
22:01:18scorche"Frequency scaling is, in computer architecture, the technique of ramping a processor's frequency so as to achieve performance gains."
22:01:21amiconnregister 0x70000034 needs a frequency-dependent value to be set (in both bit 0..3 and 8..1)
22:01:36amiconnThis was the calculation I was analysing
22:01:37preglowlinuxstb: wooot
22:01:39saratogahmm I did an SVN revert and now linuxstb's patch doesn't compile
22:01:45linuxstbSounds like a bag of shit...
22:01:48EnterUserNameahh thanx saratoga i get it
22:01:51preglowwell, of course it does
22:01:53preglowit's still rockbox
22:02:12saratogaheh looks like linuxstb fixed the problem while i was battleing SVB
22:02:16amiconnErm, that would be bits 8..11
22:02:36gregjit still doesn't make that plugin
22:02:46linuxstbsaratoga: I just removed a lot of the code that didn't belong in wmadeci.c, and restored the original ffmpeg files. It still needs some more cleaning though.
22:02:58linuxstb(my cleaned versions of the ffmpeg files)
22:03:15saratogalinuxstb: can you send me what you have?
22:03:16*amiconn wants a turbo for the G5.5's emergency disk mode :/
22:04:15gregjbtw, how many exactly implementations of mdct, huffman, and friends do yo uguys have there ?
22:04:18gregjI already can see about 4
22:04:23*gregj ducks
22:04:26preglowgregj: well, each codec that needs one has its own
22:04:44preglowwouldn't make any sense to put them in the core anyway
22:04:51markungregj: if it's possible to share some code, maybe you can work on that?
22:05:19SeanInglisAh, so effective frequency scaling equates to better battery life, broadly speaking. In the changes you're currently wrestling with, are you at full pelt by default, and tweaking downards when appropriate then? (promise I'll head to the wiki in a tick :-))
22:05:34markungregj: would be interesting to compare the speed of the fft based implementations with the normal imdct ones
22:05:56scorcheSeanInglis: we boost when something that needs it comes up, and turn it back down when it doesnt need it anymore
22:06:17gregjmarkun: yeah, having like one MM library would be indeed cool
22:06:33scorcheon targets that work fine with freqscal that is..
22:06:41linuxstbsaratoga: Running test_codec on the gigabeat gives 272% realtime.
22:06:59markungregj: but maybe they all use different fixed point precisions
22:07:10gregjno idea
22:07:16gregjit is probably too big for me for starters
22:07:20preglowlinuxstb: bitrate?
22:07:27gregjlike i said, I already see some room for improvements in metronome plugin
22:07:27saratogagregj: the wma codec uses an MDCT based on the FFT in 17.15 fixed precision
22:07:42linuxstbpreglow: 80kbps
22:07:43*gregj is musician, 2nd skin after programming
22:07:49preglowwell, that's not too bad
22:07:56saratogai believe MAD uses a similar approach except with 2.30 fixed
22:08:05preglowsaratoga: mad uses no fft for the imdct
22:08:07gregjyou can start by reusing some other code in wma implementation than :)
22:08:12amiconnlinuxstb: Could you please run a test on your color?">
22:08:30preglowsaratoga: also, it uses s3.28 precision :)
22:08:34linuxstbSure. Give me a few minutes to run a test_codec with the wma codec first, then I'll try your patch.
22:08:59gregjamiconn: what it does ?
22:09:06gregjI can give it a go on mine ipod photo
22:09:06SeanInglisscorche: ...and hopefully those that don't *currently* seem "fine" by the looks of it. Bon courage anyway.
22:09:14amiconnIt should avoid the freezes on pp5020 devices handling clock setup more correctly (aka more like the OF does)
22:10:42saratogais the pure MDCT approach preferable to the FFT based one?
22:11:03SeanInglisWorth reading up on by the sound of it. Thanks folks.
22:11:04 Part SeanInglis
22:11:09saratogaalso, I really don't understand why svn revert breaks my source
22:11:17preglowsaratoga: depends how you look at it, from an accuracy viewpoint, it might be
22:11:22linuxstbamiconn: Is it correct that that patch contains fewer files than your first one?
22:11:23saratogais there someone to force it to revert completely?
22:11:28preglowsaratoga: but with an efficient fft the combo will be faster
22:11:33amiconnlinuxstb: Yes. It's against svn...
22:11:35linuxstbsaratoga: rm -fr always works...
22:12:03gregjamiconn: you sure this patch doesn't make it explode in smoke ?
22:12:12gregjamiconn: build finishes, just will give it a shot :)
22:12:21preglowsaratoga: also, a fast direct approach needs factorisation i think is dependent on size, so you don't get a function that can work for multiple sizes
22:12:33preglowdisregard broken english
22:12:38saratogapreglow: that makes sense
22:12:52saratogaand also not very interesting to me then, since I need 6 sizes
22:14:25 Join kaaloo [0] (
22:14:26EnterUserNamehow long does it take for the forums to send you out an email for registering/
22:14:32 Part kaaloo
22:14:46*amiconn wonders whether he should try to max out the cpu
22:14:56amiconnI.e. set 80MHz for CPUFREQ_MAX
22:15:40gregjamiconn: so far it works okay
22:15:56gregjamiconn: if you need any specific tests, let me know, I want to help :)
22:16:28linuxstbsaratoga: It feels about 50% realtime on my ipod - I'm still waiting for the test to finish...
22:16:44amiconnThe main part here is testing for extended times on many devices
22:16:56gregjhow do I do that
22:17:00 Join kaaloo [0] (
22:17:04 Part kaaloo
22:17:24amiconnThere should be no freeze "out of nowhere" anymore
22:17:33preglowlinuxstb: i'd say that is pretty decent
22:17:39saratogalinuxstb: thats pretty awesome considering we're doing 64 bit multiplies for everything
22:17:48amiconnJust let it play music, preferably such formats where the cpu needs to boost/unboost often
22:18:02gregjamiconn: just froze
22:18:08gregjamiconn: 2:50 into the song
22:18:13amiconnhmm :/
22:18:22gregjno info on the screen whatsoevre
22:18:28gregjbut I am compiling in "normal" mode
22:18:39linuxstbpreglow, saratoga Better than that - 71% realtime (same 80kbps file)
22:19:06preglowsaratoga: why do you use 64x64 multiplies? do you actually need the coef precision?
22:19:07linuxstbIt's still broken on Coldfire though - I blame my asf parser...
22:19:23saratogapreglow: no, but theres no way to do a 32x32->64 one in c
22:19:27saratogaat least that I can think of
22:19:34preglowsaratoga: true, true
22:19:41gregjamiconn: funny it just stops
22:19:47saratogaa few lines of asm should make for a nice speedup
22:20:00gregjamiconn: make you are turning on 'sleep' mode, or 'step by step' mode, or something
22:20:13preglowsaratoga: though i have observed gcc being clever enough to use a 32x32 asm instruction
22:20:32preglowsaratoga: basically, if it is apparant that the multiplicands are 32 bit, the up-casting to 64 won't matter
22:20:46gregjamiconn: it was working very well on yesterday's snapshot, it is only today's that I have problems (with)
22:20:48 Quit saratoga ("CGI:IRC (EOF)")
22:20:52 Join saratoga2 [0] (i=980398fe@gateway/web/cgi-irc/
22:20:57preglowsaratoga2: didya get that?
22:21:35saratoga2preglow: didn't i show you the asm for that function ages ago?
22:21:51amiconnlinuxstb: Could you send me a .bin of the color's OF? I'd like to compare routines...
22:21:54saratoga2it was definately 64x64, and quite dumb too
22:21:55preglowhrm, i think you did, but i can't remember what i said about it
22:22:13 Join rhet [0] (
22:22:38saratoga2preglow: I will get it in a couple seconds when the logs load
22:22:50gregj udelay(500); /* wait for relock */
22:23:02gregjsounds like candidate for unwelcomed states
22:23:12preglowi think retailos uses that wait too
22:23:18 Join kaaloo [0] (
22:23:20amiconnIt does
22:23:23 Part kaaloo
22:23:24*gregj extends it to 1500
22:23:34amiconnThe OF uses 500
22:23:40EnterUserNameHi i'm having problems with the forums where do i go for help for that
22:23:49scorcheEnterUserName: what problems?
22:23:53gregjudelays are nice, when you are able to check if hardware catched up
22:23:57amiconnOn the video it's even reduced to 255 (but never used as there is no video with PP5020)
22:24:36 Join stripwax [0] (
22:24:40EnterUserNameI haevn't gotten an email validation yet and i want to make sure it has the correct emaila ddress
22:25:23gregjshaite, it crashed on usb handler - damn
22:25:28saratoga2 davegermiquet at gmail ?
22:26:00gregjnope, that was only me
22:26:03EnterUserNamethats the right one
22:26:22EnterUserNamethank u
22:26:27EnterUserNamei'll wait for activation.
22:26:52gregjlets give it a go with extendend usleeps
22:27:46saratoga2ok going to start testing this
22:27:56saratoga2preglow: will you be on later tonight if I have questions?
22:28:08preglowhow late?
22:28:12rheti have a question..i have a mac formatted ipod..if i format it to fat32 will i be able to synch it with itunes on a mac after installing rockbox?
22:28:18preglowi'll be on four hours from now, but no later, i'd guess
22:28:21linuxstbrhet: Yes.
22:28:43rhetalright cool. thanks linuxstb
22:28:57saratoga2mainly i'm wondering about the MAD asm code for doing fixed multiplies
22:29:06EnterUserNamesaratoga2: when i click on send a reactivation it says 'An error has occurred" then says a new activation email has succesfully been sent is that normal?
22:29:33gregjamiconn: works like a charm.. so far
22:29:36saratoga2EnterUserName: I have no idea, I don't actually know much about the forums, I just clicked on your user profile and read the email address
22:29:40gregjamiconn: twice as longer than previous one
22:29:44EnterUserNameoh ok
22:29:51preglowsaratoga2: and i'm fairly intimate with that code, yes
22:29:59gregjamiconn: so my guess is - there is something about those usleeps that apple does afterwards, they must check some flags, and stuff
22:30:09saratoga2preglow: i wanted to try that with my stuff
22:30:23saratoga2seems like all i need to do is change the constants for the decimal point
22:30:23gregjamiconn: usleep is better than for() - in terms of bat usage (probably0
22:30:34saratoga2any reason you can think of thats not going to work?
22:30:47amiconngregj: Checked the source? There is no usleep()
22:30:58gregjwell, I do work with hardware - on os level everyday, but those are 'big boyz' - amd/intel
22:31:02gregjamiconn: so what it does ?
22:31:11amiconnAnd I am looking at the clock setting routine in a disassebler
22:31:15preglowsaratoga2: do you use lots of different fixed point formats?
22:31:19gregjvoid usleep(int){}; ?
22:31:25saratoga2preglow: just two
22:31:32preglowwhich are?
22:31:38saratoga2and only 1 in the performance critical parts
22:31:46saratoga217.15 I believe
22:31:51 Quit SirFunk (Remote closed the connection)
22:31:52saratoga2let me double check, its been a while
22:31:57preglowhmm, 17.15 is not good
22:32:05preglowthat is, for coldfire, heh
22:32:10preglowfor arm it doesn't matter too much
22:32:11amiconngregj: rockbox has udelay() and sleep(). The former only exists on portalplayer targets, and the latter is useless here, as it works in multiples of timer ticks
22:32:13*linuxstb wants to commit this wma codec, but not with it broken on coldfire...
22:32:18gregjamiconn: edelay
22:32:19gregjamiconn: sorry
22:32:33gregjamiconn: I changed all udelay(500) to udelay(2000)
22:32:42gregjmagic, works so far for ~10 mins
22:32:48saratoga2preglow: sorry, its actually 16.16
22:32:49preglowlinuxstb: 'fraid i don't have the time to be involved with any actual testing/coding right now :/
22:33:10gregjamiconn: that's 7 minutes longer than previously (on average it made up to 3 mins)
22:33:13saratoga2i have no problem with leaving this as a patch anyway
22:33:28saratoga2we're just going to get about a billion sansa users complaining if we commit it before its real time anyway
22:33:32preglowthe sooner it's in svn, the better
22:33:39saratoga2and half of them will decide to format their recovery partition to see if it helps
22:33:43preglowwe'll just not enable it
22:33:58linuxstbpreglow: Don't worry, I wasn't asking - I'm sure it's my asf parser that's broken on coldfire, I'm about to investigate now.
22:34:01saratoga2ok that'd be fine
22:34:12preglowlinuxstb: sure, know that, it's just that i'd love to help but can't
22:34:12gregjamiconn: if it is going to last longer, I will start to really laugh very loud :)
22:34:19saratoga2i was planning to just post a patched build on anythingbutipod once we get it real time and ask for broken samples anyway
22:34:30gregjamiconn: I know this kind of issue, with udelays, and stuff - from my own expierence..
22:34:31 Quit Siltaar ("Aurevoir -")
22:34:37saratoga2i'm sure theres tons of stuff that should be fixed before we let normal users near this thing
22:34:50linuxstbLots of Rockbox is like that...
22:34:52preglowwell, we have been known to let users try out unstable stuff...
22:35:15linuxstbNo-one reads the changelogs anyway...
22:35:16saratoga2preglow: why is 17.15 a problem on coldfire?
22:35:24gregjudelay(5000) ?
22:35:33gregjif that will last even longer, trace would be clear
22:35:42*gregj gievs it ago
22:35:42amiconnNo it wouldn't
22:35:51amiconnOF uses 500 and doesn't crash
22:35:53preglowsaratoga2: because the 32x32->64 mul instruction we have there only preserves the top 32 bits
22:36:03preglowsaratoga2: top 40, actually, but only the top 32 are accessible really fast
22:36:10*linuxstb should try getting the sim working on his (PPC) ibook
22:36:13preglowsaratoga2: but anyway, don't care about that for now
22:36:21gregjamiconn: what's OF ?
22:36:21saratoga2can do
22:36:23gregjopenfirmware ?
22:36:42saratoga2i need to switch buildings, bye
22:36:46 Quit saratoga2 ("CGI:IRC")
22:36:51 Quit Genre9mp3 ("I don't suffer from Rockbox psychosis. I enjoy every minute of it.")
22:37:16gregjamiconn: than obviously it is something else, some tiny little detail
22:37:51preglowamiconn: i think retailos switches freqs far, far less frequent than we do anyway
22:38:00gregjamiconn: but anyway, if it wasn't clear - yesterday's version was working okay
22:38:13gregjamiconn: so that means - all those changes are not brining in any stability
22:38:19gregjat least here
22:40:05 Quit Nico_P (Remote closed the connection)
22:44:33amiconnlinuxstb: Haha, know what? Even the ipod color OF has the pp5022 check....
22:45:11Bagdermost likely due to it being in the SDK they use I'd say...
22:45:11gregj6min- no cash
22:45:12 Join saratoga [0] (i=9803c6dd@gateway/web/cgi-irc/
22:45:20saratogaok great its working on my gigabeat
22:45:34amiconnBut it also has that magic 0x600060A0 handling as the video, but unlike the minis
22:45:59amiconnBagder: Yeah, but the color is from before-pp5022 times...
22:46:25Bagderbut is the firmware you check really from pre-pp5022?
22:46:38BagderI mean, pre even in PortalPlayer's labs
22:46:41amiconnGood quiestion...
22:46:50Bagdersince they most likely do the SDK
22:46:55linuxstbThe firmware is from August 2006 I think.
22:47:04Bagderor did rather
22:47:48 Join kaaloo [0] (
22:47:54 Part kaaloo
22:48:03*amiconn slaps forehead
22:48:12amiconnJust noticed that my magic value is wrong...
22:48:15gregjlight bulbs flying around ?
22:48:16markunsaratoga: nice!
22:48:38saratogamarkun: thank linuxstb, he got it working
22:48:40scorcheBagder: there are a few decisions to be made with the new theme repository that seem to only be able to be answered by a to help?
22:48:44*amiconn blames beer and late evening y'day
22:48:55amiconnCouldn't distinguish a B from an 8
22:49:01*petur discovers the beer highlight works
22:49:04markunlinuxstb: nice :)
22:49:16saratoganow to see if we can get it real time on the PP chips
22:49:21*gregj mumbles something about hex numbers better to be displayed in lowercase
22:49:36amiconnTell that the disassembler...
22:49:38*preglow raises a bottle of aass bock and exclaims "cheers!"
22:50:06*gregj doesn't wanna know what 'ass bock' is
22:50:29preglowsaratoga: it already is faster than a lot of other codecs were at first
22:50:29amiconngregj: Yould try something: change the udelay back to 500, and then change the 0xd198 in line 185 to read 0xd19b
22:50:30Bagderscorche: sure, but isn't that a topic better suited for a mailing list discussion or similar?
22:50:38gregjamiconn: ok, give us the patch to test than :)
22:50:47gregjamiconn: okie dokie
22:50:50gregjstand by
22:50:59gregjit works for 10 minutes again
22:51:02scorcheBagder: well, it deals with the site itself, so mainly just swedes
22:52:24 Quit miepchen^schlaf (Read error: 110 (Connection timed out))
22:52:35saratogapreglow: I put some efffort into speeding i up as I went along, but mostly its because the codec is based heavily on the vorbis decoder, and so is pretty good to begin with
22:52:46 Join miepchen^schlaf [0] (
22:52:53*petur wonders if the swedes are occupied by a dead camel on the road
22:52:54Bagderscorche: are we talking about hosting stuff on our servers now?
22:52:59saratogapreglow: can you look at mad_f_mul(x, y) in fixed.h and tell me what it does?
22:53:00Bagderoh yes
22:53:20preglowsaratoga: sure
22:53:24saratogaverses MAD_F_MLX ?
22:53:33scorcheBagder: well, as you likely read, we decided to move away from the wiki based system, and move to a type style
22:53:45saratogabasically, which one is the fixed multiply
22:53:57Bagderscorche: yes, that's the only sensible thing
22:54:07saratogaguessing mad_f_mul but I'm not sure why the others are there
22:54:12 Join sean__ [0] (
22:54:41Bagderscorche: but still, since Redbreva has a system setup for it, it also makes sense to use his...
22:55:08gregjamiconn: ok, trying it now
22:55:18preglowsaratoga: both are, mad_f_mul does a 32x32 multiply with a 64 bit result in two registers, then adjusts for fixed point format and stores the upper part of the result in a 32 bit register
22:55:37sean__how can i have ipodlinux and rockbox live harmoniously? (5.5)
22:55:43preglowsaratoga: MAD_F_MLX just does a 32x32 multiply and leaves the 64 bit answer unadjusted in two 32 bit variables
22:56:14saratogapreglow: so I want to use mad_f_mul then
22:56:37 Join DoctorPeper [0] (i=4a841a2a@gateway/web/cgi-irc/
22:56:46saratogai don't understand the line " "M" (MAD_F_SCALEBITS), "M" (32 - MAD_F_SCALEBITS) \"
22:57:06preglowsaratoga: indeedey, but it depends on your fixed point format, then, this one assumes MAD_F_SCALEBITS is defined to be the number of fractional bits in your fixed point format
22:57:22sean__how can i add rockbox to the iplboot thing?
22:57:27saratogathats saying to do shift right by adding zero?
22:57:35preglowsean__: by going to #ipodlinux and asking
22:57:43sean__fair enough
22:57:58saratogasorry I mean "do a shift, right? by adding 0"
22:58:06 Part sean__ ("Ex-Chat")
22:58:12saratogawhy does it take two constants though?
22:58:17gregjamiconn: bang, crashed after 3mins
22:58:36gregjI will change usleeps to 255,and see, I am really sticking to that theory
22:58:44preglowbecause there are two shifts, namely (hightpart << x) and (lowpart >> (32 - x))
22:58:45scorcheBagder: indeed it does, but the one thing i came up against is if we make his site official, the copyrighted material themes will disappear, and i think that there are a lot of people who would have quite a negative effect with that...i was curious if you were at all opposed to having point to the official section of with no links to the unofficial section (which would be redbreva's and we would ahve no access o
22:58:48scorchethemes we would be
22:59:10saratogapreglow: oh so thats doing two shifts at once
22:59:17DoctorPeperI was browsing the Flyspray thingamajig and i found a page on a rockbox startup sound. this sounds like a cool thing to do. what would you need to do to get this working
22:59:18saratogai didn't realize that was possible in arm
22:59:22Bagderscorche: nope, I would be happy do arrange that
22:59:38preglowsaratoga: well, it's not, really, there are two instructions doing the shifting, one mov for the first, and an adc for the second
22:59:48gregjDoctorPeper: the worse idea ever, I don't want it to do any sounds on startup..
22:59:50preglowsaratoga: arm supports doing shifts in parallell for a lot of instructions
22:59:50Bagderand that was what I wanted Redbreva to do from the start...
23:00:00gregjmaybe apart from 'welcome master, awaiting orders'
23:00:28scorcheBagder: and have split up into an official version and redbreva's personal theme collection (with anything he wants on there) on a different subdomain and not linked on our official section?
23:00:30saratogapreglow: so doing " "M" (16), "M" (16) \" would mean I'm shifting both registers by 16 then?
23:00:38preglowsaratoga: indeed
23:00:50preglowsaratoga: which is what you'd want for your .16 format
23:01:10Bagderscorche: I really don't care what happens to the existing one, as long as we can get an "official" one with fine themes on it
23:01:22*amiconn adds a debug readout for 0x600060a0
23:01:23preglowsaratoga: does the code do any multiply-accumulate operations? i'd guess not
23:01:35gregjamiconn: want me to test something mroe ?
23:01:40saratogapreglow: theres more then a few places it could do them
23:01:42DoctorPeperFunny... But in the article. It talks about being able to disable it and to change the sound. I still suport it. Whos with me?!?!?!
23:01:42scorcheBagder: then i will let you know when i need something from you =)
23:01:47saratogawill try that next
23:01:55Bagderscorche: I'll stand by! ;-)
23:02:02preglowsaratoga: cool, since you'd benefit from using another set of macros for that
23:02:07*scorche is set up with an FTP account to and is working on the official version
23:02:16preglowsaratoga: or preferably, of course, asm. i'll have a look at that when i don't have deadlines looming over my head
23:03:42***Saving seen data "./dancer.seen"
23:04:10gregjamiconn: it seems that this change has no impact whatsoever (line 185)
23:04:29gregjamiconn: I have G4/photo/color or whatever ti is called this week
23:04:37 Quit ompaul ("Leaving")
23:04:51saratogapreglow: thats fine, I'm ok with asm, just not arm or gcc's, so this will be a good excuse to learn about it
23:05:06DoctorPeperDoes anyone like the idea of a start-up sound
23:05:31preglowsaratoga: you'll find it all a quick study, i expect, arm asm is pretty straight forward and gcc's syntax is very nice when you get to know it
23:05:47preglowfeel free to ask whenever while i'm still here
23:05:53gregjamiconn: I changed udelay to 255, crash after 2mins
23:05:57amiconnpreglow: Btw, where is this pll denominator value '8' from? ipl? arbitrary?
23:06:03*gregj see the pattern emerging
23:06:06preglowsaratoga: btw, just found out i'll probably have internet access for the next two weeks too, but it'll be more sporadic
23:06:46saratogaif I don't see you online, i'll try email
23:07:05preglowi'll be around until thursday now anyway
23:07:39DoctorPeperStartup sound? Anybody?!
23:07:54preglowamiconn: hmm, you mean the ones in the pll_control = line?
23:08:31DoctorPeperscorche: do you want a startup sound?
23:08:35preglowamiconn: i'm quite certain i got that from discussion with the ipl people
23:09:01preglowamiconn: but not verified through re or anything like that
23:09:04DoctorPeperLook at the startup sound flyspray
23:09:42 Quit maffe (Remote closed the connection)
23:10:31preglowamiconn: what parameter does the retailos function take? mhz? some arbitrary integer?
23:10:45saratogapreglow: 311% realtime :)
23:10:46amiconnpreglow: It takes 3 parameters....
23:11:02amiconnFirst 2 are numerator and denominator for the pll
23:11:21amiconnThe last one is a post-divider that only seems to work on pp5022 (??)
23:11:25preglowsaratoga: holy fecking shit, does that output white noise or what?
23:11:37saratogapreglow: this is on the gigabeat
23:11:43preglowi forgot that
23:11:48saratogayeah it plays fine though
23:11:53preglowstill good
23:12:07saratogamaybe 400% means it'll work on the PP players?
23:12:09preglowanyway, did you mention what the other fixed point format you use is?
23:12:19preglowsaratoga: possibly, but we want wider margins than that anyway
23:12:37saratogasorry, 2.30 I guess since its signed
23:12:51saratogano, 1.31
23:13:13saratogaactually, I change the precision all the time as I go, but theres only two types of multiplies
23:13:17preglows0.31 or 1.31? i like to include information on signedness
23:13:26 Quit DoctorPeper ("CGI:IRC (EOF)")
23:13:43saratogarange is +1 to -1
23:14:01preglowextremely cool that's ideally suited to coldfire
23:14:05saratogaits the multiply used for anything to do with trig functions
23:14:17saratoganow to make it fast
23:14:20preglowbut we'll see about that later anyway, for some reason i always keep coldfire in mind :)
23:14:52 Quit miepchen^schlaf (Connection timed out)
23:14:54amiconnWell, at least coldfire is documented
23:15:19preglowyeah, but it's also the least common denominator when you think about fixed point maths
23:15:35preglowthe arm part is well documented anyway
23:15:53amiconnOnly the core
23:15:53preglowi enjoy coding for coldfire the most anyway
23:16:02 Join miepchen^schlaf [0] (
23:16:07preglowmy favourite target is still the h1x0
23:18:12saratogapreglow: the asm version of the 1.31 multiply is no faster
23:18:19saratogamaybe gcc is smart enough to figure that one out
23:18:27preglowperhaps, it depends on the code
23:18:30saratogaor rather 0.31
23:18:49preglowwell, the sign bit does behave as an int bit, so 1.31 is the most descriptive
23:18:55 Quit rhet ("Leaving")
23:19:05saratogaok, wasn't sure what the convention was
23:19:06preglowit's just me being weird and using 's' as a prefix for sign, heh
23:19:15preglowthe convention is Q1.31 or something
23:19:21saratogai tend to think of the whole number as 16 bits anyway since its twos comp
23:19:22preglowbut that still doesn't tell me if there's a sign bit
23:19:52saratogawell i'm off to the gym
23:19:56preglowhave fun
23:20:00saratogai consider this a good afternoons work
23:20:05saratogawill do the really fun stuff tonight
23:20:05preglowand you should, heh
23:21:03 Quit saratoga ("CGI:IRC (EOF)")
23:21:05preglowprobably won't be around thanks to bedtime, but feel free to pelt me with whatever questions you might have in gtalk/gmail/whatever
23:21:13preglowthat was for the logs, apparently
23:27:19 Join webguest52 [0] (i=44610111@gateway/web/cgi-irc/
23:27:52webguest52cn anyone help me real quick
23:28:05*amiconn is running a test with pll denominator == 4 and CPUFREQ_MAX == 78 MHz
23:28:39preglowand first impressions are...?
23:28:46*linuxstb has fixed wma for coldfire
23:28:58digin4how do i enable shuffle in nano?
23:29:02amiconnBoost ratio is a bit lower, and it works
23:29:13amiconnBut it worked for a whole album y'day as well
23:29:18webguest52how do i put .flac files on my ipod without using itunes
23:29:21preglowlinuxstb: speed test, ples
23:29:28amiconn(with old pll setup)
23:29:31webguest52i'm dumb
23:29:37linuxstbpreglow: Give me a week...
23:29:48amiconnI'll let it play overnight if it doesn't crash before...
23:29:50scorchewebguest52: just a plain copy will work..
23:29:55linuxstb(it's running, or rather walking, now)
23:30:08webguest52a plain copy?
23:30:12webguest52just drag it?
23:30:20preglowlinuxstb: so not crawling? good news!11
23:30:30webguest52it shows up in files but not in database
23:31:12linuxstbHave you updated your database?
23:31:27webguest52how? heh
23:31:38linuxstbDunno, but the manual does.
23:32:03bluebrotherenable auto-update of the database or manually update the database from the settings menu.
23:32:20bluebrotherthe manual has the exact path through the menu to that entry ...
23:32:35amiconnpreglow: I wonder whether I should even go for 32/80 MHz instead of 30/78 (with denominator == 3, which is the default setup on mini g1 from the apple loader)
23:32:41webguest52ah thank you
23:32:49amiconnBut the 32 will cost battery...
23:33:00webguest52should i load to ram?
23:33:36bluebrotherdepends on your usage. If you use it a lot you most likely would want to load it to ram.
23:33:49digin4how do i enable shuffle in nano?
23:33:51preglowamiconn: deed
23:34:10webguest52hmm i turned on auto update for the database but the songs are still in files and not in my database
23:34:32bluebrotherwell, you need to enable dircache and load the database to ram for that to work
23:34:47bluebrotherand of course you need to reboot (as a splash should have told you)
23:34:59bluebrotherbut check the manual, it has all of this information ...
23:35:08webguest52ahyeah i jus told me to reboo
23:35:13webguest52thanks for your help
23:35:16linuxstbpreglow: My estimate is about 5%-10%, I'll know the result in about 10-15 minutes...
23:35:40amiconnpreglow: Of course we could also go 24/80, but that might be too sluggish especially on the G5....
23:35:50gregjamiconn: well, svn up -r13741 firmware/target/arm/system-pp502x.c
23:35:54gregjamiconn: and the box is stable again :)
23:36:10gregjamiconn: why are yo upushing for those changes anyway ?
23:36:22amiconnBecause you're just lucky
23:37:05amiconnBefore my changes, the mini g1 I have here for testing already froze at boot with 50% probability
23:37:22 Quit Febs (Read error: 110 (Connection timed out))
23:37:30amiconnIf it didn't, it was merely possible to browse a minute or 2. Playing music? No way...
23:38:27 Quit webguest52 ("CGI:IRC (EOF)")
23:38:31gregjbut like I said - on my box it only gets worse
23:39:05gregjso either this just makes another bug worse, or makes the whole thing worse and you are fixing wrong bug
23:39:09gregjdunno to be honest
23:40:12pixelmaIIRC linuxstb reported that it's always been this way - some builds were more stable than others (without code changes)...
23:40:37scorcheseemingly random
23:41:03gregjwell, if I use the one I mentioned - it is rock solid stable
23:41:10gregjplayed on it for a half of the day
23:41:19gregjapart from rpoblems with batery, no other issue
23:41:49amiconnlinuxstb, gregj: My latest patch">
23:41:50gregjif it is a luck, I wish I had that much luck with lotery tickets
23:42:35gregjamiconn: so yo ualso changed: pll_control = 0x8a020000 | 8 | (postmult << 8);
23:42:43gregjamiconn: besides last change I tested
23:42:49amiconnNot only
23:43:01amiconnThere are related changes in .h files
23:43:26amiconnIf you just change the 8 to 4, you will overclock the cpu by a factor of 2
23:44:36amiconnlinuxstb, gregj: Also, if you try this patch, check the debug menu (i/o ports) for the value of CLOCK_0xA0
23:44:45amiconnI expect it to be 0, but you never know...
23:45:23*amiconn wonders where all the H10 owners are gone
23:46:07*scorche wondered if they existed at all
23:46:20amiconnWell, there must have been some
23:46:30 Join My_Sic [0] (
23:46:49 Join Entasis [0] (
23:46:53*petur knows one personally.... can always contact him if required
23:47:07Bagderspeaking of that, it's been a while since barry was around
23:47:26gregjamiconn: so you want me to configure that build in debug mode ?
23:47:32gregjamiconn: I use 'normal' usualy
23:47:40amiconnJust normal build
23:47:49amiconnDebug builds don't work for ipod anyway there is no way of communication, and hence no gdb stub
23:48:34 Join entheh [0] (
23:48:42gregjamiconn: system->debug
23:48:46gregjamiconn: and than what
23:48:54amiconnview i/o ports
23:48:57gregjview io
23:49:17gregjclock a0 is:
23:49:25gregjall the time
23:49:40amiconnNow that makes sense... no voodoo involved
23:49:51amiconnOn alll my 3 PP targets it's 0
23:49:57 Join Desiato [0] (i=4bc162a8@gateway/web/cgi-irc/
23:49:58amiconn...all the time
23:49:59gregjsee, I am helpfull - apart from the fact that I am pain in the arse, and whiner
23:50:11gregjdon't ask
23:50:14Bagderoh, the zune-linux site was compromized and is currently gone
23:50:15gregjand - frozen
23:50:20bluebrotherBagder: was just wondering about the devcon t-shirts −− are there leftovers? ;-)
23:50:22linuxstbpreglow: Not as bad as I estimated - 18% realtime on Coldfire (80kbps wma)
23:50:28amiconnAnd the OF clock setting routine sets the top 2 bits to 0 while fiddling with the clock
23:50:32scorcheand that you seem to have limited your lines to 2 words per...
23:50:35gregjamiconn: this was the fastest crash ever
23:50:40gregjamiconn: not even a minute
23:50:44Bagdergregj: hey, that site could cheer you up like few others ;-)
23:50:48*amiconn should add that as well
23:50:52 Quit Jon-Kha (Remote closed the connection)
23:50:52linuxstbgregj: Which ipod do you have?
23:51:02Bagderbluebrother: yes, but I've not yet started to deal with them
23:51:13amiconnlinuxstb: Could you also check CLOCK_0xA0?
23:51:20amiconnMaybe the apple loaders are different...
23:51:22linuxstbWith your #3 patch?
23:51:24 Join rotator [0] (n=e@rockbox/developer/rotator)
23:51:35amiconnThat one adds the display of CLOCK_0xA=
23:51:36gregjrunning again
23:51:41gregjresurected after pluging in usb
23:51:50scorchegregj: please try to place your thoughts on one line
23:51:50bluebrothernice ... /me hopes to notice that quick enough to get one ;-)
23:51:53DBUGEnqueued KICK scorche
23:52:13amiconnBroken thoughts?
23:52:19gregjsort of resurected
23:52:20amiconnHappens to me as well
23:52:34gregjscorche: sure
23:52:44scorcheyeah, but it gets annoying with it being done all the time...and it fills the logs up
23:52:44gregjscorche: I usualy want to add something later
23:54:16CtcpIgnored 1 channel CTCP requests in 0 seconds at the last flood
23:54:16*bluebrother just stumbles upon the greyscale Rockboxed stuff
23:54:27bluebrotheralready more than 3 month ago :o
23:54:28peturnah, he just wants to make it in the irc stats for the shortest lines :p
23:54:40gregjamiconn: now my battery is gong, gotta recharge :)
23:55:27scorchepetur: or most lines =P
23:55:51gregjthere should be an option to just draw power from lead, but without usb disc kicking in
23:56:15linuxstbamiconn: CLOCK_0xA0 is C0000000
23:56:24*gregj smiles
23:56:24amiconnVery nteresting....
23:56:41peturgregj: there is iirc, read the manual
23:56:50pixelmagregj: hold "menu" when plugging the usb cable - make sure you hold it for a while
23:57:08peturoh look, our walking manual
23:58:05bluebrotherwalking manual?

Previous day | Next day