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

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

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

#rockbox log for 2006-04-20

00:00:57 Quit TCK (Read error: 110 (Connection timed out))
00:01:10 Quit muesli__ (Read error: 110 (Connection timed out))
00:01:29 Join muesli__ [0] (n=muesli_t@
00:02:31peturSereR0KR: got file view set to display all files?
00:02:39SereR0KRdunno o.O
00:03:47peturgeneral settings -> File View -> Show Files
00:04:27peturand do you have a game WAD next to rockdoom.wad - like shareware doom or so?
00:04:58peturthere are links to some in the wiki
00:05:02preglowyou people need to read the wiki more
00:06:54 Quit ender` (" Does anyone here have a computer?")
00:10:14 Quit wefds (Read error: 110 (Connection timed out))
00:12:19kkurbjunamiconn, lostlogic, or preglow: could you look at this asm code and tell me if you see anything blatantly wrong, it doesn't seem to be copying to %[dest]
00:12:58amiconnSlasheri, hardeep: With the correct handling of playlists > max_files_in_dir_buffer, building an all-tracks playlist got somewhat slower (64 secs for 3307 tracks on my archos recorder), but that's still quite fast considering the huge number of tracks
00:13:00SereR0KRrockdoom rocks :>
00:13:34amiconntagcache builds that 3000+ tracks playlist faster than old tagdb built a 500 track playlist...
00:14:51 Join wefds [0] (n=muesli_t@
00:15:13 Join dj-fu_ [0] (
00:15:43 Quit dj-fu (Nick collision from services.)
00:15:47lostlogickkurbjun: lines 8 and 9 why? you already require those vars to come in on data registers
00:15:53amiconnI have to say I'm impressed...
00:15:54SereR0KRI got so much wads
00:15:59SereR0KRbut how do I use them O.o
00:16:09 Nick dj-fu_ is now known as dj-fu (
00:16:11SereR0KRI put them into addons
00:16:15 Part larcher
00:16:33SereR0KRbut it always says that this wad doesn't have the standard graphics type or something like that
00:17:24kkurbjunlostlogic, I didn't think of that, I wasn't sure that the values coming in were being loaded to registers.. I'm not very familiar with gcc inline or colfire asm...
00:17:51lostlogickkurbjun: nod nod −− if you say it's gotta be a data register, then it's a data register by symbolic name
00:18:14lostlogickkurbjun: looks like you do some stuff to D3 and then overwrite its lower byte?
00:19:23lostlogickkurbjun: ignore me, I hadn't read the whole thing and the C version yet
00:20:18 Quit muesli__ (Read error: 110 (Connection timed out))
00:21:48lostlogickkurbjun: I think d3 will have only its lower byte set to the right value, and you need to clear its other bytes before you use it as an index
00:23:07kkurbjunamiconn, thanks, I'll read that
00:23:23kkurbjunlostlogic, so I should do a clr.w on it then?
00:24:17 Join muesli__ [0] (n=muesli_t@
00:24:30 Join predius_ [0] (
00:24:37 Join thouters [0] (
00:24:44predius_does the ipod app work with amd64?
00:24:56amiconnkkurbjun: No. The problem is that the first move.b only replaces the low byte of d3, but in the next insn you use it as an index
00:25:08thoutershi, anybody know what (linux) program I can use to set id3 tags on a wavpack file?
00:25:36 Join ashridah [0] (
00:25:45amiconnSince coldfire can only use longword indices (unlike 'real' m68k), you need to make sure you get a full 32 bit index
00:25:51midkaypredius_, um, that isn't an app?
00:26:00midkayhave i missed something? :)
00:26:07predius_midkay: binary.
00:26:08amiconnExtending unsigned is costly, so I would suggest to not immediately reuse d3
00:26:10 Quit Kohlrabi ("Fast alle Menschen sind Regenw├╝rmer")
00:26:18lostlogicamiconn: probably just use another register
00:26:21lostlogicer kkurbjun
00:26:25midkaypredius_, i mean, 'that isn't an app'.
00:26:41amiconnThen you can do a clr.l before the first move.b
00:26:57predius_oh, and it was a misunderstanding on my part
00:27:03kkurbjunok, thanks, I'll try that
00:29:00amiconnkkurbjun: Some immediate optimisations:
00:29:29amiconn(1) the tst %[count] in line 20 can be left out. already sets the flags
00:29:54amiconn(2) sub.l #1,... should be replaced by subq.l #1, ...
00:30:37 Join qwm [0] (
00:31:03kkurbjunamiconn, thanks alot for the pointers, I'll add those in
00:31:38qwmwho's the tripper here?
00:32:01qwmare you daniel?
00:32:17BagderI am
00:32:36qwmso you're in the rockbox loop then.
00:32:42qwmhow is it?
00:32:55Bagderits life in the fast line
00:32:58Bagderfame and glory
00:33:01crashdhe gets all the chicks
00:33:02Bagderdrinks and parties
00:33:04qwmi can imagine..
00:33:05crashdand the beers
00:33:10 Quit wefds (Read error: 110 (Connection timed out))
00:33:15CtcpIgnored 1 channel CTCP requests in 0 seconds at the last flood
00:33:15*qwm is sooo jealous
00:33:16amiconnkkurbjun: It also seems that you forgot the equvalent of xfrac += ds_xstep; yfrac += ds_ystep;
00:33:20*Bagder giggles
00:33:29preglowi'm congratulating myself on the high score with a beer right now
00:33:33*preglow pats himself on the back
00:33:46Bagderbeerscore is good score
00:33:57qwmi totally agree.
00:34:23 Quit thouters ("leaving")
00:35:29qwmactually considering buying a player..
00:35:36 Join JBGood25 [0] (
00:35:43qwmneed a small, cheap and reliable one.. for walks and such.
00:35:58Bagderlike nano perhaps?
00:36:01qwmmidkay: and i don't want your opinion.
00:36:10qwmhow much does it cost?
00:36:14Bagderno idea
00:36:14kkurbjunamiconn, oops, gotta add that too
00:36:16midkayqwm, i told you, nano, before! :p
00:36:30qwmyou did not.
00:36:32scorcheqwm: in USD?
00:36:40qwmin SEK, preferably.
00:36:55scorchetoo bad then
00:36:55Bagder2250 SEK for 4GB
00:37:08scorche(199 USD)
00:37:22scorche249 stock price for 4 gig
00:37:44scorchebut no..he wants SEKs!
00:37:54qwmcalm down. ;)
00:38:02scorcheqwm: im bored
00:38:13scorchei could use some seks too....
00:38:15qwmso am i, and calm!
00:38:24qwmsame here!
00:38:35qwmlet's meet and solve that.
00:38:45scorchei want heterosexual seks
00:39:04qwmwho says i'm the same sex as you? ;)
00:39:17scorchegirls dont exist on the internet
00:39:24qwmah, true.
00:39:45scorcheon the internet, guys are guys, girls are guys, and little girls are FBI agents
00:40:56scorchemidkay: see what i mean whenever i talk in here?
00:41:14midkayscorche, what do you mean?!
00:41:24scorchewe discussed this last time...
00:41:27sharpeit makes no sense anymore!!
00:41:46 Part midgey34
00:41:49midkayAAAA!! blah .. blah.. louD NOISES!
00:42:03kkurbjunamiconn, for some reason that asm code seems to think that count is equal to 0 after 1 test, it's setting the first pixel, but nothing after that
00:42:19 Join qwm_ [0] (
00:42:37 Quit muesli__ (Read error: 110 (Connection timed out))
00:42:51kkurbjunhere's my latest version:
00:46:02qwm_telia is some unreliable crap.
00:47:37 Quit e0f (Remote closed the connection)
00:48:27kkurbjundid you see anything wrong with the setup?
00:49:40 Join damaki__ [0] (
00:51:02kkurbjunwhat about that post increment after assigning %[dest]: "move.b (%[colormap],%%d4.l), (%[dest])+ \n" is that proper syntax?
00:51:50SereR0KR"this wad doesn't follow the standard doom graphics"
00:51:53SereR0KRwhat can I do?
00:52:32 Quit JBGood (Read error: 110 (Connection timed out))
00:52:43kkurbjunSereR0KR, you can fix the wad that you're trying to run with deutex or some other wad editor, you have to size all the graphics to be lese than 320x200
00:53:15SereR0KRyeh I am using a mac :>
00:57:00 Quit qwm (Read error: 110 (Connection timed out))
00:59:21SereR0KRI tried to compile it and then it says: src/tools.c:61:22: malloc.h: No such file or directory
00:59:22SereR0KRmake: *** [src/tools.ot] Error 1
01:00:45 Quit Paprica (Read error: 110 (Connection timed out))
01:01:29mikolaskkurbjun, shouldn't the loop have bne in the end? beq will end at the first run as count is not zero
01:01:36mikolasline 21
01:03:31kkurbjunmikolas, I'll try that, not very familiar with the branch stuff, thanks
01:04:59kkurbjunhmm, that works now, the code is drawing all the way across the spans
01:05:06 Join Febs [0] (
01:05:38kkurbjunit's just not the right way to draw it...
01:05:46 Quit damaki_ (Read error: 110 (Connection timed out))
01:06:15 Join filoktetes [0] (
01:08:10 Quit bluebrother ("Leaving")
01:12:26mikolaskkurbjun, what exactly are you trying to do?
01:12:49qwm_looking for japanese porn can make you SOOOOO fucking frustrated. :(
01:13:23 Quit filoktetes ("Leaving")
01:13:44kkurbjunmikolas, I'm trying to write a faster version of doom's drawspan routine which draws all the floors and ceilings
01:14:13ashridahthou canst not outcode carmack!
01:14:45kkurbjunit's a fairly critical routine as it's called for every line of a floor or ceiling being drawn.. : ), I'm not trying to outcode him, just gcc
01:15:55dj-fuwrite a faster routine to decode mp3 faster on ipod nano ;]
01:16:29lostlogicashridah: but you can outcode gcc :)
01:16:43twisted`anyone here using linux and an iPod?
01:16:56*lostlogic mutters about gcc's coldfire support
01:17:37*amiconn wants m68k-elf-gcc for amd64, damn
01:17:49 Quit dj-fu ("leaving")
01:18:08 Join dj-fu [0] (
01:18:15amiconnlostlogic: That's not really hard most of the time...
01:19:49 Quit aegray_ (Read error: 113 (No route to host))
01:19:56preglownot hard at all
01:19:58preglowit's outright easy
01:21:02amiconnYeah, just remove some hilarious stuff like moving values back and forth between registers
01:21:02*preglow also wants m68k-elf-gcc for amd64
01:21:20preglowamiconn: yeah, and the unnecessary stack accesses
01:21:21*ashridah wants his amd64 machine to be finished :(
01:22:29SereR0KR"These wads can be fixed using a wad editor like Deutex to change all the graphics that are oversize or out of range to follow the doom standards." <- yeh, but HOW
01:22:41SereR0KRI compiled deutex successful for my mac but have no idea :>
01:22:58preglowtime for bed
01:23:54***Saving seen data "./dancer.seen"
01:24:00kkurbjunSereR0KR, you have to look through the graphics folder for images that are oversize
01:24:14kkurbjunafter you disassembled the wad
01:24:24SereR0KRbut how do I disassemble it?
01:24:28 Quit obo (Read error: 110 (Connection timed out))
01:25:47kkurbjunthere's an extract command
01:25:54kkurbjunread how to use deutex
01:26:17SereR0KR"-xtract [<in.wad> [<dirctivs.txt>]]
01:26:17SereR0KR Extract some/all entries from a wad."
01:26:21kkurbjunyou just have to extract and then edit the graphics with something like gimp
01:26:26kkurbjunyes, that's it
01:26:35 Quit petur ("Zzzz")
01:26:36SereR0KRwhat is <dirctivs.txt
01:26:52kkurbjunyou have to have a copy of doom or doom2 in the dir otherwise it'll complain
01:26:58kkurbjunjust ignore that for now
01:27:51SereR0KRso I just have to put doom2.wad in the same folder?
01:28:26kkurbjunas the wad you're extracting, yes
01:29:05SereR0KRand how do I create the wad when I edited the images?
01:29:53KyomiAnyone know whats up with the experimental build?
01:29:57mikolasis the correct coldfire syntax %%a0 for address registers or does it even have any?
01:30:13Kyomikkurbjun: Are you the one doing the rockdoom?
01:30:14 Quit ashridah ("uni")
01:30:15mikolasi'm redoing the assembler bit
01:30:33 Join sanit [0] (n=test@
01:32:15qwm_Evilnipple: interesting nick of yours.
01:32:17kkurbjunKyomi, yes
01:32:58kkurbjunmikolas, yes, that's the correct syntax
01:33:15kkurbjuncoldfire has 7 address registers I believe
01:33:24kkurbjuna7 is the program counter though
01:33:42Kyomikkurbjun: Is there a way to have you code a key setting for strafe left/right in the Set Keys menu?
01:33:59KyomiI'd sooo love that
01:34:03kkurbjunKyomi, how do you plan to use that?
01:34:23KyomiJust set strafe left/right to the left/right on the h320 pad
01:34:32KyomiIt'd be ALOT easier to dodge stuff
01:34:44KyomiAnd you'd only have to press 2 keys instead of 3 :)
01:34:46kkurbjunI guess you could get rid of the strafe modifyer and automap in the game..
01:34:59kkurbjunKyomi, I'll look into adding that
01:35:01KyomiCan't you just add it?
01:35:06KyomiAnd not take out anything?
01:35:17KyomiI mean back and open can be the same key
01:35:27KyomiWhy not the strafe keys? :)
01:35:35SereR0KRwhats the maximum size of pictures for doom?
01:35:37kkurbjunmikolas, are you working on the asm for doom or something else?
01:36:19Kyomikkurbjun: Heres an idea
01:36:26KyomiHave it like the run thing
01:36:28kkurbjunyou have to make sure that other graphics are not going out of bounds too though.. in that halflife wad for gp32 all the menu graphics are messed up
01:36:47kkurbjunso you have to resize the difficulty settings and all that
01:36:48Kyomiflip the hold key to make left/right strafe
01:36:55KyomiFlip again to turn it back to normal
01:37:09KyomiSince Always Run will most likely already be on
01:38:18kkurbjunKyomi, that would be alot more involved than just addind a strafe right/left in the options, I've thought about doing that, with the hold switch flipping between two sets of key layouts though, and I might add that in sometime later
01:38:39Kyomiyay :D
01:38:43amiconnkkurbjun: a7 is the stack pointer...
01:38:45*Kyomi hugs kkurbjun
01:39:08SereR0KRhow do I create a wad from these folders kkurbjun
01:39:11 Quit Evilnipple ()
01:39:26kkurbjunamiconn, ahh, yeah, that's what I meant got them mixed up
01:39:37KyomiCreate a wad? With WADed or something similar
01:39:49kkurbjunSereR0KR, you use deutex to create a wad
01:40:01KyomiI never used that
01:40:03kkurbjunyou have to pass in wadinfo.txt as an option
01:40:04KyomiHow is it?
01:40:25kkurbjunit's a commandline wad editor
01:41:11kkurbjunI'm starting to think my asm won't be any faster then gcc.. or not much
01:41:15 Quit Daishi ("Client exiting...")
01:41:28SereR0KRok really easy
01:41:31SereR0KRlet's test it ^^
01:42:27 Join Daishi [0] (
01:42:34SereR0KRshould get a bit smaller than the old wad
01:43:35 Join TCK [0] (
01:44:37SereR0KRit works xD
01:45:13kkurbjunSereR0KR, great, what wad did you fix up?
01:45:22SereR0KR Doom 64
01:45:51kkurbjunIf you posted it somewhere I'm sure a bunch of people would be happy
01:46:01midkaySereR0KR, cooool :) can you do the halflife wad? :)
01:46:10SereR0KRI will do all my wads
01:46:18midkayincl. halflife!?!?!?!?!!?!?!?!!! :)
01:46:23SereR0KRperhaps i'll create a package after that
01:46:27SereR0KRyes incl. halflife
01:46:36midkayi neeeed halflife :)
01:46:54midkaythat'd be great, if you posted them at the PluginDoom page, people would go nuts with enjoyment :)
01:46:57kkurbjunSereR0KR, the only wads I've come across that need to be fixed are the gp32 wads so far just as a note
01:47:59*Moos noticed the doom Wiki page is one of the most read
01:48:04SereR0KRI have Duke Nukem, Half Life, Heretic, Quake 2, Rest in Peace, Star Wars, Twilight Warrior and Western
01:48:22SereR0KRyou know which one works without editing it?
01:48:29 Quit Aviv` ()
01:48:43kkurbjunno, you just have to test them and see if they play or not
01:48:51SereR0KRok -.-
01:48:53SereR0KRlet's start^^
01:49:08midkaySereR0KR, i know halflife, quake2 and dukenukem don't work..
01:49:09kkurbjuneither in rockdoom or with prboom and it's software renderer
01:49:34midkaykkurbjun, how about some hardware rendering?
01:50:07SereR0KRheretic doesnt work
01:50:17amiconnlostlogic: I didn't observe the behaviour described on SoftwareCodecPlayback, but another one.
01:50:35Moos (read 13091 times)
01:50:42amiconnSkipping tracks on-buffer sometimes produces audible glitches before the skip happens
01:51:23SereR0KRI found a wad that works xD
01:51:42SereR0KRRest In Peace does
01:51:44kkurbjunmidkay, I wish
01:51:55midkaykkurbjun, i as well :)
01:52:06mikolasi made the loop use registers for everything so it should be quite a bit faster
01:52:10midkaykkurbjun, the 5G's broadcom thingy wouldn't be of any use, would it?
01:52:32mikolascheck out the code and make it work as i don't have the ability to test it right now
01:52:42SereR0KRTwilight Warrior also works :>
01:52:51midkaySereR0KR, cool :)
01:53:38 Join Rousenshi [0] (
01:53:40SereR0KRStar Wars/Quake 2/Half Life/Duke Nukem doesn't
01:53:48SereR0KRI will fix them -.-
01:53:49mikolasactually i already found a few bugs in it
01:53:52SereR0KRWestern also doesn't
01:53:55KyomiDuke Nukem doesn't work?
01:54:06 Quit webguest29 ("CGI:IRC")
01:54:10SereR0KRsomeone said that Kyomi
01:54:17SereR0KRand it's true :P
01:54:19SereR0KRI need to fix it
01:54:20KyomiI never knew you could try that on H320 doom plugin thing
01:54:20kkurbjunmikolas, I think they were already using registers
01:54:22midkayKyomi, i remember trying it and it didn't..
01:54:25 Quit sanit (Connection timed out)
01:54:42KyomiAlright.. who has a H340 in here?
01:54:56KyomiOr knows where they are available for cheap?
01:55:06KyomiAnd in good condition
01:55:11Rousenshihey, I know there are a few articles in the forum about it, but I was looking at the code and couldnt find where the scroll speed for the ipod (4gen is the only one in the code, which I guess is combatable with 5gen) wheel (on menus and such) is located, or if it can even be changed at all... anyone know?
01:55:14KyomiI soo need a h340 for video XD
01:55:26kkurbjunmikolas, the inputs are copied to registers and then you can reference them with the %[] syntax
01:55:28mikolaskkurbjun, dunno. have you checked gcc output?
01:55:33kkurbjunI believe
01:55:36SereR0KRsomeone wished to fix half life -.-
01:55:39SereR0KRI do now
01:55:45kkurbjunyes, that's what Ive seen from objdump
01:56:19mikolaswell there's always room for improvement :-)
01:56:43kkurbjun: )
01:56:46Moosthat is the Rockbox anthem
01:56:48Moos: )
01:57:29KyomiYou can play half-life on an H320?
01:57:46midkaySereR0KR, i wanted half life :)
01:57:52Rousenshiits a wad isnt it?
01:57:55midkayKyomi, a rather low-quality WAD version of it, yes..
01:58:12MoosSereR0KR: it seems you will make lot of users happy : )
01:58:35SereR0KRwhy? :P because I have the time to do something like that? ;>
01:59:15 Join aegray [0] (
01:59:27*Moos try just one time to play doom on h1xx and failed cause screen unreadable
01:59:29SereR0KRok Half Life wad is creating
01:59:35Moosand no X5 support yet
01:59:57midkaySereR0KR, awesome :)
02:00:09midkaySereR0KR, will you put them up at the wiki page then?
02:00:25SereR0KRI can
02:00:30SereR0KRneed to upload them somewhere
02:00:33Moosthat is the better place, no?
02:00:48midkaySereR0KR, wiki allows upload..
02:00:51 Join Zzompp [0] (
02:01:13MoosSereR0KR: how big is the zip?
02:01:27mikolasdoes coldfire have similar pipelining as 680x0 (x > 2)
02:01:39SereR0KRthe wads will be about 10-15mb
02:01:40mikolasthat is an endless source of optimization
02:01:55MoosSereR0KR: not too big then
02:02:05SereR0KRor 10-20 dunno
02:02:09SereR0KRthe biggest is 20mb
02:02:11midkaySereR0KR, what about the individual files and then a complete zip for people who wants them all?
02:02:21Moosgood idea
02:02:23midkayzip = optional :)
02:02:28SereR0KRi'll see -.-
02:02:31Mooshehe : )
02:03:13SereR0KRError in game :o
02:03:20MoosSereR0KR: what are you doing for make this wads working?
02:03:24kkurbjunmikolas, I'm not sure what the pipelining is like on the coldfire, I'm sure some of the instructions could be rearranged more optimally. here's the latest version of what I have:
02:03:28SereR0KRI scale the graphics
02:03:32Mooswhat was wrong?
02:03:35Moosoh ok
02:03:38kkurbjunit still doesn't work right though
02:03:52SereR0KRbut I think in this HalfLife wad I forgot something :>
02:04:10kkurbjunSereR0KR, the halflife wad has alot of messed up graphics
02:04:31kkurbjunyou have to rescale all the difficulty options alwo
02:04:34kkurbjunalso that is
02:04:56SereR0KRI do Star Wars first :>
02:05:03kkurbjunthey are too long I believe for where they are drawn on the screen
02:07:12 Join Aditya|Nap [0] (
02:07:26 Part Aditya|Nap
02:07:36SereR0KRis the maximum width 320 or 300?
02:08:26SereR0KRso the problem is the height
02:08:59kkurbjunSereR0KR, you have to make sure taht their offest drawn+width isn't greater then 320
02:09:26kkurbjunthe offset is determined in that wadinfo.txt and doom's menucode
02:09:55kkurbjungotta go
02:11:11 Join [TCK] [0] (
02:11:16mikolaskkurbjun, if you enable save-temps for doom you will see that the gcc generated code can be optimized quite a bit :(
02:11:16SereR0KRI will go to bed now
02:11:40midkaySereR0KR, noooooo... :)
02:11:53midkayhalfliiiiiiiiiiife. blah.
02:12:03SereR0KRI can't get it work NOW
02:12:16midkaysomething wring?
02:12:17 Quit Daishi (Remote closed the connection)
02:12:46 Join Daishi [0] (
02:12:55amiconnmikolas: I prefer m68k-elf-objdump -m m68k:5249 -d xyz.elf
02:13:01SereR0KRI didn't found the graphic I forgot to change yet ^^
02:13:08SereR0KRstar wars works -.-
02:13:20SereR0KRI can upload my first 2 wads
02:13:22mikolasamiconn, ok
02:13:22midkaySereR0KR, i'll try it myself.. what command do you use for deutex?
02:13:47SereR0KRdeutex -xtract wadname.wad
02:14:12SereR0KRand to create a wad: deutex -create wadinfo.txt wadname.wad
02:14:13mikolasamiconn, last time i programmed assembler on 68k it was with asm-one
02:14:27mikolasso my skills with gnu toolchain are somewhat limited :)
02:14:57midkaySereR0KR, thank you :)
02:15:02amiconnWhile I had an amiga for a long time, and still have one, I never coded in asm for it
02:15:03SereR0KRno problem
02:15:05midkayi'll upload it if i can figure it out, maybe..
02:15:14mikolasi still have an amiga 3000
02:15:17SereR0KRa tip: look for the biggest file size ;> these are the biggest pics
02:15:37midkayah, good idea :)
02:15:37 Quit Febs ("CGI:IRC (EOF)")
02:15:42SereR0KRwhere can I upload files to the wiki?
02:15:43mikolashaven't been using it for a while after my scsi drive melted
02:15:59amiconnmikolas: A 4000 here. It's my mail machine, plus occasional webserver.
02:16:06amiconnAnd it's running dnetc
02:16:12hardeepamiconn: i totally agree with you on tag cache performance. On an h300, for 5325 tracks it took 19s without dircache, 13s with dircache and 10s with dircache + tagcache in ram −− very impressive
02:16:12 Quit TCK (Connection timed out)
02:16:24midkaySereR0KR, "attach" at the bottom of the plugindoom page.
02:16:24 Part Rousenshi
02:16:27amiconnhardeep: wow
02:16:44hardeepnow everyone can stop asking for the shuffle all feature −− it's in!
02:17:09SereR0KRI need to register -.-
02:17:16midkaySereR0KR, you do indeed :)
02:17:29Mooshardeep: and thanks for it !
02:17:38hardeepthank Slasheri
02:18:13midkaySereR0KR, at the main wiki page there's a register link.
02:18:20Mooshardeep: tagcahe and your playlist things get a good mariage : )
02:18:58 Join ze [0] (
02:19:02amiconnhardeep: While it might not make that much sense, tagcache is fun to use on Ondio too
02:19:25SereR0KRI will update my wads that are working now
02:19:53amiconnThat means it doesn't do lengthy disk accesses as that would slow it down (reading the MMC is slow). Otoh, seek times are zero on a flash based player...
02:19:55hardeepamiconn: hehe, nice. You have, what, 100 songs on it? =)
02:19:57midkaySereR0KR, cool, thanks :)
02:20:02amiconnhardeep: 300
02:20:06Mooshardeep: out of curiosity, which devices you owned?
02:20:21 Join ProgramZeta [0] (
02:20:27amiconnThat's a 2GB MMC, plus the internal flash
02:20:29hardeepMoos: Archos recorder (v1), h120 and h300
02:20:38hardeepah, nice
02:22:58SereR0KRok both games are uploading
02:23:05SereR0KRI will go to bed now!
02:23:13 Quit aegray ("leaving")
02:23:25SereR0KRdoes sound on rockboy work?
02:23:27gromit`/go kw
02:24:09BHSPitLappySereR0KR: yeh, mostly
02:24:34SereR0KRand does fullscreen? ^^
02:24:49SereR0KR"no it doesn't" -.- ^^
02:25:17midkaySereR0KR, ty again, nite :)
02:25:30SereR0KRI won't go yet
02:25:34SereR0KRneed to get GameBoy roms
02:26:01SereR0KRMario/Donkey Kong roms
02:26:14Mooshardeep: I just tried to reproduce one bug I just reminded about "warn if playlist change", but it seems to be broken
02:26:22amiconnHmm, do we still need databox? Maybe it should be turned into an editor for tagnavi.config?
02:26:26Moosplayback restructuration related?
02:27:08hardeepMoos: what's the problem?
02:27:12Mooshardeep: is this warn thing working for you with recent CVS?
02:27:19hardeepMoos: it should be
02:27:30hardeeplet me try
02:27:38MoosI just test and apparently it doesnt
02:28:45hardeepworks for me
02:29:08Moosuch :(, what I did wrong??? let me find out
02:29:10hardeeper, wait, it won't work from the id3 db... is that what you're using?
02:29:29Moosdir browser
02:30:21Moosreseting settings...
02:30:35midkaySereR0KR, aha, found the problem you probably had.
02:31:33midkaywith halflife..
02:31:39midkaydid you get two errors?
02:31:53hardeepanother bug, it doesn't work from playlist viewer either −− but dirbrowser is working fine for me
02:31:55midkayit's the very easy and easy bitmaps in /graphics...
02:31:56midkaytoo wide.
02:31:58SereR0KRno midkay
02:32:04hardeepi'll definitely fix those two cases
02:32:10SereR0KRI got one error
02:32:21midkaySereR0KR, do you remember it?
02:32:33Mooshardeep: definitively not working with reset settings, but maybe because I don't have very lastest build, I'll test will bleeding edge
02:32:38SereR0KRsame as before, only a few moments after starting halflife
02:32:55midkaySereR0KR, then you must have missed the title screen bitmaps and stuff..
02:32:59midkayor something. :)
02:33:05SereR0KRI dunno which graphics
02:33:40midkayresized the bitmaps + retesting.
02:33:56SereR0KRok I got 13 roms now :>
02:34:17midkaySereR0KR, anything of interest? :)
02:34:24hardeepMoos: this code hasn't changed in a while. What happens when you start a new playlist?
02:34:48SereR0KRI will tell you the roms
02:35:20Mooshardeep: I just start one song, and then go to file browser and play other track, and no warn like before
02:35:39midkaySereR0KR, got halflife working..
02:35:49SereR0KRupload it to the wiki ^^
02:35:56midkayslightly weird though.
02:35:59Moosbuild for Avril 01 on my iriver, let's test with current build on X5
02:36:04midkaya couple textures are acting up..
02:36:18amiconnHmm, generating a playlist from tagcache is for some reason slooow on Ondio :(
02:36:19hardeepMoos: Are you inserting the first track or playing it?
02:36:21SereR0KRDonkey Kong 2,3/ Harvest Moon 1-3, International Superstar Soccer 2000, Zelda Links Awakening DX, Oracle of Ages, Oracle of Seasons/ Super Mario 1,2 / Warioland / Tetris
02:36:55amiconnThat's in fact weird, since browsing is really fast
02:36:57midkaySereR0KR, cool :)
02:37:21midkaySereR0KR, halflife is screwed :)
02:37:30SereR0KRgood night
02:37:30midkayflashing/changing textures everywhere, and error after the first level..
02:37:32Mooshardeep: no
02:37:39midkaySereR0KR, night :)
02:37:50SereR0KRwhat is the key configuration of RockBox O.o
02:37:59BHSPitLappySereR0KR: bad
02:38:00KyomiIs anyone avail to do an updated Experimental build with the latest version of the SID plugin?
02:38:01hardeepamiconn: When creating a playlist it writes to /.rockbox/.playlist_control file. Is write slow on Ondio?
02:38:07Mooshardeep: this is not the same behaviour I had som weeks/month ago
02:38:16hardeepMoos: what do you mean?
02:38:16SereR0KRyeh I've seen BHSPitLappy
02:38:27BHSPitLappyiBoy is much better regarding keys
02:38:30amiconnhardeep: Well it depends how it's handled. .playlist_control is usually small
02:38:37SereR0KRbut I can't get Rockbox AND iPL working
02:38:48BHSPitLappywhy's that
02:38:52hardeepamiconn: in the case of tagcache, every track is inserted into .playlist_control
02:38:58amiconnMMC access is quite slow (~330KByte/sec), but seeking is done in zero time
02:39:00Mooshardeep: I remenber tried this when you made it, and I remenber it warned fine, but now no warn at all
02:39:03SereR0KRbecause I have no windows PC here to use the installer for iPL :>
02:39:10amiconnGenerating a 317-track playlist takes 80 seconds
02:39:22BHSPitLappyiPL is best installed as FAR away from a Windows PC as possible, SereR0KR
02:39:23 Quit _rj (Remote closed the connection)
02:39:28Mooshardeep: how do you do for make the warn pop? insert...?
02:39:30amiconnThat's slower than the 3300-track playlist on recorder v1
02:39:39SereR0KRI dunno
02:39:48SereR0KRhow to install it manually using ubuntu :>
02:40:01Moosamicoon: haha : )
02:40:18SereR0KRI will try it when I'm in school next week :D
02:40:31amiconnhardeep: .playlist_control grows to 27KB so that's not the problem.
02:40:39SereR0KRI can't create ext2 partitions on OS X that's a problem
02:40:50hardeepamiconn: but there's a write() and a sync() for every track inserted
02:40:55amiconnBut, how is this file handled wrt opening & closing?
02:40:59hardeeper, after every track inserted i mean
02:41:14amiconnThe sync for each track insert *has* to go then
02:41:35amiconnIt's not only slow as hell, but also wearing the flash
02:41:39hardeepyeah, that's only on the archos devices... i'll look into removing that
02:41:57hardeepactually, it's pretty easy to do before we begin playing
02:42:02hardeepi'll change
02:42:26amiconnThe write doesn't hurt, since the file system code caches one sector
02:42:33 Join ^BeN^ [0] (
02:42:40amiconnfsync() or close() updates everything, including the fat
02:42:41hardeepMoos: The warning only appears on "Dynamic" playlists, ie. if tracks have been inserted
02:43:20hardeepamiconn: yeah, the sync is so we resume correctly... but it's only relevant once we've started playing
02:43:22Mooshardeep: silly me, i forgot
02:43:52amiconn(the write() used to be a problem back when porting to Ondio. Saving a .cfg took a minute because of flushing each byte write...)
02:44:21amiconnThe fun thing was that this flush was not intended, just went undetected for years
02:45:53 Join _rj [0] (
02:47:21Mooshardeep: great, I canot reproduce the bug that was before (when warn pop and push no for don't change playlist, when you stoped songs and wanted to resume, that resumed the songs you didn't want to play), hopefully I'm understandable
02:49:21Moosresume thing is the only thing made me don't use this setting a lot, cause if you want stop player and want to restart, if you don't want to resume it will warn anyway, and here never resume songs : )
02:50:34 Quit SereR0KR ("XChat Aqua")
02:51:07hardeepMoos: strange, nothing was changed to fix the problem you mention
02:51:56Moosor maybe is one shuffle bug, I'll try to find it out again, if it's still here
02:55:40Moostime to sleep here, bye all
02:55:54 Quit Moos ("Glory to Rockbox !!!")
02:56:00pixelmasince I disconnected my Ondio from my computer i get the following error message: IllInstr 370902F2...
02:56:07pixelma...what does this mean?
02:56:34*amiconn sees another can of worms open up regarding dircache as soon as we have support for usbotg, and hence multivolume and volume hotswap on H300...
02:57:15 Quit mphill (Read error: 104 (Connection reset by peer))
02:57:23pixelmarebooting showed no effect
02:57:32midkayamiconn, well, i'd suggest removing dircache, except that it's a useful, innovative, very important feature!
02:59:14 Quit ProgramZeta (Read error: 104 (Connection reset by peer))
03:02:01 Join fatjack [0] (
03:03:45KyomiIt is?
03:04:04fatjackhi... please i need help with vmware
03:04:18 Join qwm [0] (
03:05:15fatjacki'd like to get it compiled with the sid plugin
03:06:15 Quit lostnihilist (Read error: 104 (Connection reset by peer))
03:07:34fatjacki can't manage to get rockbox source in my computer, i can't even think to patch it!
03:09:19mikolascvs -z3 co rockbox-devel
03:09:35fatjacki did that already
03:09:48mikolasso you have the source then?
03:10:13fatjackif so, where should i look for it?
03:10:17 Quit ravon (Read error: 113 (No route to host))
03:10:35mikolasyou should have rockbox-devel directory in the dir you made checkout in
03:11:49fatjacki'm sorry, should the check out be entered like...
03:12:02fatjackcvs checkout
03:12:16mikolasdo you have the source tree?
03:12:37mikolasthe rockbox-devel dir with all the sources?
03:13:02 Join maeck [0] (
03:13:03fatjacki'm sorry but i don't know
03:13:15mikolasbut you are in vmware, right?
03:13:38mikolasthen, if the checkout fails it might be bcos you have a firewall blocking the traffic
03:13:59midkayor maybe he just doesn't know if it worked?
03:14:06midkayi don't think a firewall would block that..
03:14:20mikolasopen a shell and do the checkout as i said earlier
03:14:23mikolasit should work
03:14:34predius_is there something i need to do to get a theme working on an ipod 5g?
03:14:47midkaypredius_, just copy it over and run it..
03:14:59predius_midkay: nothing happens when I change it.
03:14:59 Quit qwm_ (Read error: 110 (Connection timed out))
03:15:05predius_just the fonts on some.
03:15:08midkaydo you run the .cfg file?
03:15:21sharpewell that sounds like a personal problem. :D
03:15:22maeckdid you select it in the menu?
03:15:41maeckclick menu -> browse themes
03:15:48predius_i'm there
03:15:57maeckis it listed?
03:16:08maeckdid you click it?
03:16:08midkay.. select it?
03:16:15predius_i have
03:16:23predius_still default.
03:16:23KyomiI wish paulheu would update the experimental build already >_<
03:16:41KyomiI want the v1.2 or 1.3(?) SID plugin
03:17:15predius_Version is CVS-060419
03:17:27 Join root_ [0] (
03:17:30predius_That might be the problem?
03:17:51midkaypredius_, not sure what to say, no, that's up-to-date.. make sure you're extracting correctly (full paths).. select one and try rebooting the iPod, maybe..
03:18:07predius_i did, same problem.
03:18:21midkaydid what?
03:18:24maeckso you click it, it shows loading for a while then when you click |<< the font does not change and the play screen isnt different
03:18:47midkaypredius_, is it installed/extracted correctly?
03:19:10maeckwhich wps did you download?
03:19:12midkaypredius_, well - what theme is it you're trying?
03:19:33pixelmaplease, can someone help me or at least say what's wrong with it - i mean the IllInstr 370902F2?
03:19:53pixelmaCos I cannot use the device anyhow :(
03:20:11 Nick Zzompp is now known as t (
03:20:29maeckPixelma, what is wrong? sorry I cannot see the message you sent out earlier
03:21:13pixelmasince I've disconnected the Ondio from my computer i get this error message
03:21:26pixelmaand rebooting didn't help
03:21:52midkaypredius_, that one would look about the same.
03:21:56maeckif you hook it up to the pc, can you still access the drive?
03:21:58midkayit doesn't appear to change the colors itself in its CFG file..
03:22:15midkaypredius_, try playing a song and see how the WPS looks.
03:22:50pixelmayes I can - the PC finds it whether it is the build in one or the MMC
03:23:34maeckcan you copy Rockbox back onto it? Did you try that?
03:23:34predius_not black and white
03:23:51predius_has ROCKbox in the middle, under a line.
03:23:57***Saving seen data "./dancer.seen"
03:24:03predius_which probably happens because the 5g resolution is hugher
03:24:14pixelmano - i'll give it a try
03:24:30midkaypredius_, it would not be black and white.
03:24:47predius_midkay: the theme is.
03:24:52 Quit mikolas (".")
03:25:00midkaypredius_, i told you, the cfg file does not include the color setup, that must be done manually.
03:25:13 Quit fatjack ("CGI:IRC")
03:25:24 Join fatjack [0] (
03:25:30predius_so, after unzipping the file, what exactly do I have to do?
03:26:03midkayrun it from Browse Themes, set your colors however you like (e.g. background black, foreground white), and that should be all..
03:26:41predius_thing is, the layout is wrong too.
03:27:36midkaypredius_, well, that's odd, it looks fine..
03:30:25pixelmastill the same error message
03:31:06fatjacki got to the point where i don't have the knowledge to think by myself :/
03:32:41root_i just installed rockbox on my iriver; if setup is essentially the same, maybe i can help?
03:33:17 Join aegray [0] (
03:34:42 Join RoC_MM [0] (
03:34:50fatjackwhen i do make dir what happens to he, where is it? should i check it out? after it where is it going to be?
03:36:00root_the zip should create a .rockbox/ directory (note leading dot); a rockbox.* file should be at the same level in the tree; all from the root
03:36:13 Quit p0wr-toc ("Leaving")
03:36:27root_rockbox.* = rockbox.iriver on my iriver
03:37:05root_you don't need to keep the .zip after it's been extracted
03:37:13fatjack but these files shouldn't be on my computer? where are they?
03:37:26 Quit mikearthur (Read error: 104 (Connection reset by peer))
03:37:27root_they should be extracted onto your iriver/ipod/etc.
03:37:58fatjackdirectly? form vmware to the ipod? (mine is an ipod btw)
03:38:47root_i mounted my iriver on /mnt/iriver/ ,changed dir, and extracted the zip. it should be similiar for you
03:39:43root_you could also recursively copy all the extracted files to the ipod. same thing, just an extra step
03:41:01root_any devs around? i want to pass out the chocolate for this awesome software! :)
03:41:53fatjack" i mounted my iriver on /mnt/iriver/ ,changed dir, and extracted the zip" i'm very confused, so confused that gives me to laugh... i'm sorry!!!
03:42:25root_heh, sorry. i'm on a linux system. how do you access the storage on your ipod?
03:42:40maeckpixelma: did you reboot? I asume you did. I have no Ondio so I am quite on dangerous grounds giving you advise on this.
03:43:24pixelmamaeck: sure i did
03:43:38maeckwhen do you get this error?
03:43:39fatjackdouble click on "My computer" than "G:\"
03:43:56pixelmaright at start up
03:44:29maeckis an ondio one of those devices there you have to flash the firmware? I do not know but just throwing a thought out...
03:44:53root_fatjack, then you should be able to drag (or copy/paste) the .rockbox directory to G:\ as well as the rockbox software (rockbox.ipod probably); should go into the root of the ipod storage
03:44:58 Join quobl [0] (n=quobl@tor/session/external/x-773015b7348943a6)
03:45:07pixelmaas far as i know, you don't have to - but someone did it for me
03:46:21pixelmaso my knowledge about this is also very limited
03:47:21fatjackroot_ ... that i can do, my problem is that i'd like to be able to compile since i'd like to have some patches running that seem that aren't yet available on present downloadable rockbox versions... would you like to compile it for me ? <big grin> please>?
03:47:22 Quit Jumpfroggy ("CGI:IRC (EOF)")
03:47:22maeckpixelma, I am checking the manual for you
03:49:14root_i could compile stuff for you but it'd be a gcc-based linux x86 binary, not something that would run under windows...
03:49:48maeckpixelma: did you follow the instructions in the manual? and did you use the windows installer or such?
03:49:56maeckJust want to be sure
03:50:11root_i'm not sure how to cross compile ipod binaries if that's what you need
03:50:58maeckroot_ there are nice instructions on the website about building your own cross compiler. Fairly straightforward. You might want to give that a try
03:51:42root_more time sinks. i'll be sacrificing the last 4 hours of sleep a night that i've got left... time to pin my eyelids to my eyebrows... ;-)
03:51:51pixelmamaeck: do you mean copying rockbox again or the flashing procedure?
03:51:59twisted`maeck: yo
03:52:04twisted`maeck: received my email?
03:52:06maeckhey Trwisted
03:52:12maecktwosted, not yet
03:52:18twisted`maeck: what client u use?
03:52:23maeckpixelma: I mean the steps in the manual:
03:52:54fatjacki know how to put rockbox running (like, duh fatjack!!!, you may think) i just don't know how to do everything to get it patched... i need that sid patch urgently... i'm going on hollidays with no internet connection... pleaseplease!!!!
03:52:55maeckI do not use clients, I work for them. But then again... oh well
03:53:27root_let me check real quick what it takes... brb
03:53:28maecktwisted: chatzilla
03:54:13twisted`maeck: hmm, I dunno if that uses tabcompletion :P
03:54:52biffherook, you guys are too smart. there's a sample.emacs file in the tools/ directory. woo hoo!
03:55:34maecktwisted`: nice, good tip
03:55:45twisted`maeck: weird, I send the mail like... long time ago
03:55:45maeckbeen looking for that...
03:56:09maeckwell, I have been busy, then dinner, then hang out in the bookstore, then here...
03:56:11twisted`maeck: I send it like 6 hours ago
03:56:18maeckhave not even seen my gmail account
03:56:23twisted`maeck: ahh I thought u meant u havn'[t received yet
03:56:30sharpewoo gmail!
03:56:37twisted`maeck: cause then I would have been _real_ worried
03:56:40maecktwisted`: that is what I said, work, dinner, bookstore, rockbox
03:56:49twisted`true true
03:57:04twisted`brb, takin a shit
03:57:07maecksharpe: actually I forward everything to my hotmail from gmail
03:57:11twisted`ff bruut het riool vervuilen :P
03:57:27pixelmai do not how my ondio was flashed, cos someone else did it for me - to get a newer build running i just have to copy the .rockbox folder and execute the rockbox.ucl
03:57:42pixelma*do not know
03:57:50maeckpixelma: doesnt look like you need to flash an ondio. Anybody knows?
03:58:09pixelmabut it was working before
03:58:39maeckpixelma: did you use the right firmware? the one for the ondio sp or ondio fm?
03:59:25maecksharpe: but I do not want you guys to know that I *love* hotmail.
03:59:56 Join lostnihilist [0] (
04:00:07root_fatjack, i'd have to rebuild gcc with arm cross-compiler support. i'd have to look into how to get portage to do that.
04:00:36pixelmai unstalled a build of the day before yesterday and i'm pretty sure it was the right one - because it was already runnung the whole day
04:00:59fatjackforget it root_, don't worry, thank you very very much for your time :)
04:01:37maeckpixelma: I do not want to dissapoint you, but I really do have no clue
04:01:37 Quit fatjack ("CGI:IRC (EOF)")
04:01:38 Quit mtnbkr ("Microsoft is not the answer. Microsoft is the question. The answer is "NO."")
04:01:41root_i'm checking into it atm (if i figure it out quickly i'll see what i can do)
04:01:48 Join ProgramZeta [0] (
04:02:25pixelmamaeck: tnx anyway
04:03:36maeckpixelma: did the manual say something about how to restore the original firmware? If so (and you have the original firmware...), you could try to restore it. The Ondio might even be able to boot into original firmware (like the Ipod)
04:03:42maeckyou could try that too
04:04:01maecktwisted`: I did -not- look in my email box ... yet ...
04:06:20twisted`maeck: whehehe, it's ok :P
04:08:02KyomiQuestion for whoever knows something about the SID plugin
04:08:27KyomiHow does it handle subtunes? Or can you even get to them?
04:10:35RotAtoRI doubt subtunes are handled at all
04:10:56RotAtoRin rockbox one file = one song only currently
04:13:24KyomiMaybe why the ones with subtunes dont play?
04:13:58RotAtoRno idea, i haven't tried the sid patch myself yet
04:14:17RotAtoRthough i do have the hvsc pack downloaded and ready to go ;)
04:14:37sharpeblargle... i need to find a wireless mini keyboard... :D
04:14:56 Quit ProgramZeta (Read error: 104 (Connection reset by peer))
04:15:35 Join ProgramZeta [0] (
04:15:45twisted`maeck: but, I was thinkin, maybe I'll do the icons on the same way I did that icon I send you
04:15:52twisted`maeck: that, went imo, very smoothly
04:16:03twisted`maeck: I just take a photo, and retrace it and make it vector
04:16:08twisted`cus I suck at drawing with hand
04:16:16twisted`or else u get happy grunge shit :P
04:21:01sharpei want a keyboard in the shape of a rectangle, not an amoebae...
04:21:08root_fatjack, rebuilding gcc with crossdev support is a bit more involved than i have time for atm. i will see what i can do about it tomorrow
04:24:27pixelmamaeck: i could boot the original firmware, it if it wouldn't have been flashed before - now the shortcut, which would lead me there, causes an rockbox rescue boot.. it says descrambling and wants to start - but with the same effect in the end
04:26:33pixelmaso i guess I will ask the one who flashed it for me
04:28:46maecktwisted`: check thy email.
04:29:39pixelmabut tnx for taking the time
04:30:07sharpemaeck, how goes the graphical menu ?
04:31:02maecksharpe: need to break up some code. To make it work likeand menu_init, menu_run. I will do some of that in a minute
04:31:18maeckstarting that is, I wont be done in 55, 50, 45...
04:32:03twisted`maeck: booyah
04:32:42twisted`I'm not too fond of icons like that...
04:32:50twisted`when it's... well sketchy
04:32:53maecktwisted`: I did that in the last 5 minutes... spent most time finding the color settings in inkscape
04:32:59twisted`and that... just... scared me
04:33:13twisted`it looks better real size
04:33:16twisted`gmail thumbed it wrong
04:33:37maeckit will work for the menus... for now until some talent spends some serious time on it...
04:33:44 Part t ("ERC Version 5.1.2 (CVS) $Revision: 1.815 $ (IRC client for Emacs)")
04:34:03twisted`hmm... cus I thought you wanted some _clean_ icons
04:34:29pixelmagotta go to sleep anyway
04:34:51 Quit pixelma (" bye")
04:34:55twisted`maeck: received pm?
04:35:03sharpemaeck, are the icons still built in?
04:35:14maeckyeah, but the register stuff is annoying. And I want to CODE
04:35:21maecknot hang out here
04:35:41maeckanybody got peanuts and a beer?
04:36:18maecksharpe: yes, would be great if I could load them at start. Easier for themes.
04:36:56twisted`lot :)
04:37:04maeckneed to find a way to scale the bitmaps. Need to extend the lcd_bitmap function for that.
04:37:47scottderhow bout cashews
04:38:04maeckjust skip lines to scale... none of that fancy pansy bi-linear cube scaling with interpollation and sperical abberation and such...
04:38:12maecknot nuts
04:40:05twisted`maeck: btw have u checked how hard it is to rotate it like I showed in that image?
04:40:59maecknot too tough, but costs cpu cycles... (read slows down).
04:41:29twisted`hmm, is it really that much of a difference compared to the original?
04:42:03sharpeyou don't have to worry about cpu cycles on the 5g until you have to emulate a processor at 300 000 instructions per second, with at least 25 frames per second.
04:42:31twisted`how fast is the cpu of the 5g anyway
04:42:38sharpe75mhz at full speed
04:42:42sharpetwo cores
04:42:50maeckexactly. Well, sharpe, how is the Open GL emulator going?
04:42:53twisted`two? seriously?
04:43:01sharperockbox only uses one as of yet
04:43:06sharpeopengl emulator?
04:43:29maecksharpe: I am kidding about the openGL stuff
04:43:43dj-fuis rockbox currently only using one core?
04:43:48maeckor did I put a though into your clockwork?
04:43:57dj-fuit'd be nice to have both cores running.. crossfeed + eq + realtime playback
04:44:01twisted`sharpe: dude, the 5g is like 75x faster then the c64
04:44:05maeckHas two legs, but uses only one... smart eh?
04:44:14sharpeyes, yes it is.
04:44:23sharpebut the c64 didn't have to run within rockbox.
04:44:27maeckbut the Ipod does not come with a 6503 processor
04:44:37maeck6502 that is
04:44:38sharpeand video output wasn't handled within the cpu
04:44:38twisted`true true
04:44:50twisted`lot of fiddle space :D
04:44:53sharpe6510 actually.
04:45:05 Quit root_ ("[BX] Does your mommy know you use BitchX?")
04:45:10sharpetis the 6502 with more i/o lines
04:45:37sharpemy god, why is it so hard to make a keyboard that's rectangle in shape, and is wireless?!
04:45:43maeckwell, 6502 assembly is about just as risc as the ARM...
04:46:11sharpewell, right not it supposedly runs at 104% emulation...
04:46:32sharpehowever it's only displaying 10 out of 25 frames per second
04:46:46maeckemulating a ramdac is hard
04:46:50sharpewhich i don't get why...
04:47:00midkaysharpe, there is only one solution..
04:47:03sharpeemulating fire would be the hardest thing to emulate.
04:47:17sharpemidkay, what would that be?
04:47:44maecktwisted`: are you okay with the guideline I gave you? can you run with that?
04:47:45midkaysharpe, you need to compile and execute a version of the emulator that features a putsxy(5, 52, "speed: 502fps");
04:47:54twisted`maeck: guideline?
04:47:56midkayfrom then on the performance will be magically rather faster.
04:48:00twisted`maeck: the things you wrote you mean?
04:48:06sharpemmm, inflated statistics.
04:48:07twisted`yup I'll try to fix some stuff up yep
04:48:15maecktwisted`: no the little image.
04:48:16twisted`maeck: but, in the weird style you send or you want nice 'n clean
04:48:29maecksomething quick
04:48:55twisted`hmm... then I could just as well do it on paper... but I'll check what I'll draw on the pc :)
04:53:34 Part YouCeyE ("Leaving")
04:54:27 Quit ProgramZeta (Read error: 110 (Connection timed out))
04:55:43 Quit TeaSea ("Leaving")
04:59:06 Quit Galois ("Leaving")
05:00:36 Quit Kyomi ()
05:01:34 Join Rob2222_ [0] (
05:01:46 Join qwm_ [0] (
05:03:12 Join Galois [0] (
05:04:02twisted`maeck: I saw you released new code, but, how do I compile it so I get a new .rock ?
05:04:23twisted`or is there a make file included?
05:05:06sharpeokay, i'm on my third attempt at how it's going to be rendered.
05:05:50maeckthe result on the ipod video is the same as it was before. Only, if you compile for Nano, it is scaled down to fit the wee screen.
05:06:49 Quit twisted` ("Lost terminal")
05:06:50maecktwisted`: if you want to compile, first add the zip file to you dev environment, then put the new gfxmenu.c file in the plugins directory, then build the normal way
05:07:08 Quit hardeep ("BitchX: now with Olestra!")
05:11:51 Join twisted` [0] (
05:14:33 Quit qwm (Read error: 110 (Connection timed out))
05:15:57maecktwisted`: if you want to compile, first add the zip file to you dev environment, then put the new gfxmenu.c file in the plugins directory, then build the normal way
05:16:19twisted`maeck: I got no idea howto build for rockbox
05:16:35twisted`maeck: never build rockbox myself yet tho
05:17:49maeckwell, start with a clean build then. If you follow the instructions on the website it will be fairly easy if your in a linux environment.
05:17:59maeckbuild a simulator to start with.
05:18:11twisted`oh boy...
05:18:17maeckThen you experiment with a cross compiler (also good docs on the website)
05:18:19 Join P-Luc_Auclair [0] (
05:18:31maeckthen, start building some of the patches
05:18:37twisted`why cross compiler?
05:18:57 Quit Rob2222 (Read error: 110 (Connection timed out))
05:19:03sharpeyou don't want the antichrist compiler, do you? :)
05:19:04maecksince there is no intel processor in the ipod, you have to compile for ARM processor. Different instruction set.
05:19:18GaloisI never bothered building a simulator
05:19:35twisted`hmmm, ohyeah confused with xscale
05:19:37twisted`that's intel
05:19:37maecksaves me a lot of plugging and unplugging of the ipod
05:19:42P-Luc_Auclairhey I was wondering, is it possible to write an app that plays sound over the music?
05:19:59sharpeneither do i.
05:20:12Galoisif you plan to do dev work (especially UI) then it's worth it to build the simulator
05:20:23goffahey Galois, how did you switch the play action again?
05:20:34Galoisgoffa wha?
05:20:45maeckI can understand that when you are working on sound decoding simulators are not really useful
05:20:50goffalike select select=play
05:20:55goffaor whatever
05:21:14Galoisgoffa: coming right up
05:21:47Galois (ipod only)
05:22:04goffai have x5
05:22:08goffai'll look at it though
05:22:09Galoisdude, you can figure it out
05:22:21Galoisthe IPOD and IAUDIO sections are almost the same
05:22:25 Quit aegray (Remote closed the connection)
05:22:29 Join aegray [0] (
05:23:12P-Luc_Auclairsorry I'm a big newb but does rockbox run on linux firmware?
05:23:30Galoisrockbox is not based on linux
05:23:43sharpeit can dual boot with linux...
05:23:58***Saving seen data "./dancer.seen"
05:24:17P-Luc_Auclairokay, can I write an application that would play sound at the same time music is playing?
05:24:36P-Luc_Auclairlike a timer over music
05:24:51Galoisshould be possible, that's how voice menus are supposed to work
05:25:07 Quit Daishi ("Client exiting...")
05:25:59 Join Daishi [0] (
05:26:24P-Luc_Auclairok, I was thinking writing some kind of application that would play a beep say after 1 minute, 2 minute, user set, like when I go running I could have a beep that tells me when to stop running and do a walk pause etc.
05:27:10sharpeyou may be able to get something from the metronome plugin
05:27:16 Quit dj-fu (Remote closed the connection)
05:27:26P-Luc_Auclairok, I'll look into that
05:28:33goffaok ... i don't know much about c
05:28:40goffaRCS file: /cvsroot/rockbox/apps/tree.h,v
05:28:50goffadoes that mean edit tree.h and tree.v ?
05:29:04sharpeonly tree.h
05:29:11Galoisthat file is a patch
05:29:15sharpethere is no tree.v
05:29:17Galoisman 1 patch
05:29:30goffawill the patch work for x5?
05:29:37P-Luc_Auclairrockbox is all C?
05:29:42Galoisno, you need to use your head a little here
05:29:51sharpeP-Luc_Auclair: yes.
05:29:55Galoishere's the basics of patch files. - means "remove this line"
05:29:59Galois+ means "add this line"
05:30:04goffafigured that much
05:30:26Galoisin this case you will notice I replaced TREE_RUN* with TREE_WPS* and vice-versa
05:30:27maeckP-Luc_Auclair: yeah building a plugin for that would make the most sense
05:30:33goffaso... i need to find what vars the x5 uses then?
05:30:53Galois117,12 means line 117 column 12
05:30:54 Join ravon [0] (
05:30:54maeckP-Luc_Auclair: you would first start your music, then browse the plugins and start say a chronometer or something.... that would be cool
05:31:06maeckP-Luc_Auclair: not that I workout or something...
05:31:12Galoisline 117 in the file tree.h is a 10-line block of commands
05:31:23Galoisthe top of the block says #elif (CONFIG_KEYPAD == IPOD_3G_PAD) || (CONFIG_KEYPAD == IPOD_4G_PAD)
05:31:47Galoiswhich in english basically means "if the keypad is IPOD3G or IPOD4G then do this"
05:31:50P-Luc_Auclairhehe, I don't have firebox yet but it looks nice from what I've seen on my friend's X5
05:32:02P-Luc_AuclairI thought it would be interesting for that
05:32:04Galoisthere's a separate section that belongs to the X5
05:32:11Galoisit's the one that starts with #elif CONFIG_KEYPAD == IAUDIO_X5_PAD
05:32:31Galoisyou need to go into that section and do the same thing in that section that I did inside the IPOD section
05:32:39P-Luc_AuclairI have this strange addiction to watches when I run, and I keep counting aloud so it fuck my breathing up :P
05:32:45P-Luc_AuclairI badly need a beep :P
05:32:57goffaso if i change that to 143 instead of 117 i should be ok?
05:33:05maeckP-Luc_Auclair: Firebox? Oh that is a fork from Rockbox... nowadays used for browsing the 'internets'
05:33:10Galoisit's not quite that simple
05:33:24Galoisthe easiest thing to do is to edit tree.h, not the diff file itself
05:33:28 Join speacial_ed [0] (
05:33:41P-Luc_Auclairmaeck, just a typo :P
05:33:48Galoissince the main purpose of the diff file is to tell a computer how to change tree.h, and you don't really care about that
05:33:51P-Luc_Auclairfirebox is a firewire soundcard :)
05:34:47*P-Luc_Auclair covers from rocks
05:35:05Galoisgo into the section "#elif CONFIG_KEYPAD == IAUDIO_X5_PAD" and, inside that section, change all the TREE_RUN to TREE_WPS and vice-versa
05:35:56Galoisthen rebuild rockbox. It should work.
05:36:11goffaso.. you changed 4 lines basically?
05:36:34Galoisthere are four -'s and four +'s in the diff file
05:37:00goffawell... time to hold breath, knock on wood, cross fingers while throwing salt over my shoulder
05:37:03Galoisthe diff file is for a computer. Computers don't understand english so you have to be super precise when you tell a computer what to change, hence the "117,12" and other mumbo jumbo
05:37:49Galoisthe diff(1) and patch(1) manual pages describe how the computer-style patches work
05:38:22Galoiscvs(1) also has a diff command which basically calls diff on your local files against the CVS repository
05:38:22goffado i want to do normal build?
05:39:50goffagrr.. not this again
05:39:58goffathought i compiled m68k already
05:40:15goffaoh .. i'll bet i didn't put it in my path the other day
05:51:31sharpe14.2 out of 25 frames per second, 154% realtime..
05:51:40*goffa wishes he wasn't retarded
05:52:05goffagetting this: WARNING: The compiler you must use (m68k-elf-gcc) is not in your path!
05:52:18goffai go to /usr/sh1 (which is in my path)
05:52:30goffai see: bin include info lib libexec m68k-elf man share
05:52:48goffawondering what i need to compile to get m68k-elf-gcc
05:55:27goffathe only m68k-elf-gcc that i see is for cygwin
05:56:14Galoisshould do it
05:59:02 Join amiconn_ [0] (n=jens@rockbox/developer/amiconn)
06:02:05goffagrr... error too big to paste
06:02:37goffalooks like it tries to do an rm of a lot of dirs
06:02:51goffaError 1
06:02:57goffarm /home/goffa/pac/rockbox-bleeding/build/apps/plugins/fire.elf ... ...
06:05:28goffamaybe nti- i- nte ellol
06:05:35goffadamn fingers were off one
06:05:50goffameant maybe i found the error
06:05:58goffamake[2]: *** [/home/goffa/pac/rockbox-bleeding/build/apps/plugins/calculator.o]
06:06:08goffadon't know what that means though
06:08:45 Quit RoC_MM ("Leaving")
06:09:13 Join Paul_The_Nerd [0] (
06:10:09goffagrr... i can't get this to compile
06:11:40goffacalculator.c: In function `typingProcess':
06:11:40goffacalculator.c:1125: error: insn does not satisfy its constraints:
06:11:46goffawatever that means
06:12:59 Join infamis [0] (
06:14:33 Part Paul_The_Nerd
06:17:35 Quit amiconn (Read error: 110 (Connection timed out))
06:17:36 Nick amiconn_ is now known as amiconn (n=jens@rockbox/developer/amiconn)
06:22:17infamisis this a known bug: when you turn off shuffle on a playlist that you turned shuffle on, it's still shuffled?
06:24:52 Quit speacial_ed ("Bye")
06:25:28infamisfor example: 1) shuffle is off 2) you start playing 'a_playlist.m3u' 3) you hold down the menu-button to get the quick menu and hit the 'left' button to turn shuffle on & exit out. 4) then you go back into that menu and turn shuffle off 5) you view the current playlist and it's shuffled, but in a different order (or a different start position, don't know)
06:26:21infamisI usually throw a whole directory in the running playlist, turn on shuffle & skip through tracks til I find something good, then turn off shuffle to continue that artist/album
06:26:50goffai think it reshuffles when you turn shuffle on
06:27:34infamisbut when you turn it off, the order looks either undefined or still shuffled...I just know it's not in-order as it is in the .m3u
06:28:31goffahmm... i think there was a shuffle option elsewhere too
06:29:53infamisI guess the main question is, before you shuffle, is the current-order retained anywhere so that when you turn off shuffle, that order is restored
06:32:00goffaseems to be on mine
06:32:31infamisturning shuffle on and off restores the original order for you?
06:33:00sharpeno, but it could just reload the playlist if it's saved somewhere.
06:33:06 Quit carini ("Chatzilla 0.9.72 [Firefox]")
06:33:22goffagrr.. battery died
06:34:55infamismine too...making the sim right now from a fresh cvs down
06:35:09sharpeif someone implemented it...
06:35:12goffais m68k-elf-gcc 3.3.6 (303)
06:35:17goffanew enough?
06:35:22sharpeif it isn't already.
06:35:37sharpei would assume any version is generally new enough.
06:35:49goffadid you see my troubles above?
06:36:33infamisI don't know enough about how the internals work to even mess with modifying code
06:36:33goffaCC calculator.c
06:36:33goffacalculator.c: In function `typingProcess':
06:36:33goffacalculator.c:1125: error: insn does not satisfy its constraints:
06:37:01goffainfamis: me either
06:38:15infamisgoffa: what are you building?
06:38:23goffarockbox itself
06:38:30goffafrom source
06:38:55sharpewell, this probably won't help at all, but, try doing a 'make clean' then building it...
06:38:57infamislol I mean what device & is it the sim or not
06:39:13goffanot sim
06:39:15 Join ProgramZeta [0] (
06:39:19goffajust compiling in linux
06:39:37infamislatest cvs?
06:40:03infamislemme see what I get
06:40:07sharpei'll try updating and compiling mine...
06:40:29infamisI'm under cygwin so I'm gonna win
06:41:48infamiscygwin is known to be superfast
06:41:48 Part P-Luc_Auclair ("Kopete 0.11.1 :")
06:42:07sharpegoffa, x5 right?
06:42:08 Join qwm [0] (
06:42:45infamisno error for me
06:42:53infamiswith a (n)ormal x5(12) build
06:43:04goffathats what i tried
06:43:14goffai changed 2 vars in tree.h
06:43:29sharpethe #defines?
06:44:24infamiswhat'd you change?
06:44:24goffa#define TREE_WPS (BUTTON_SELECT | BUTTON_REL)
06:44:25goffa#define TREE_WPS_PRE BUTTON_SELECT
06:44:25goffa#define TREE_MENU (BUTTON_REC | BUTTON_REL)
06:44:25DBUGEnqueued KICK goffa
06:44:25goffa#define TREE_MENU_PRE BUTTON_REC
06:44:25goffa#define TREE_RUN (BUTTON_PLAY | BUTTON_REL)
06:44:27goffa#define TREE_RUN_PRE BUTTON_PLAY
06:44:29infamisyou never mentioned that :)
06:44:43goffaswitched run and wps
06:45:26sharpetry getting version 3.4.5
06:46:02ravonHCl: Got it semi-working.
06:46:38ravonHCl: Can you please try it on your device when you get the chance? I'm not sure if the error is in dumb.c, in the dumb lib or if my iAudio is just too slow with the current implementation.
06:46:54goffasharpe: as opposed to 3.4.6?
06:47:13sharpedon't think it may matter?
06:47:17sharpeone sec...
06:47:34goffait might
06:48:24sharpei suppose 3.4.x should work.
06:48:47infamismust be your version...I changed tree.h to reflect what you posted & compiles without error
06:49:06infamisI'm running 3.4.5 under cygwin
06:49:35sharpeas am i...
06:49:51GaloisI got /usr/local/bin/m68k-elf-gcc-3.4.6 on linux, and I'm an official build server, so if it works here it ought to work there
06:51:15 Quit ProgramZeta (Read error: 104 (Connection reset by peer))
06:51:17infamissure that's the only thing you changed?
06:51:56goffais m68k-gcc-elf same ver number as your gcc compiler?
06:52:39 Quit qwm_ (Read error: 110 (Connection timed out))
06:52:58sharpemine isn't
06:53:39sharpegcc for me is 3.4.4, m68k-gcc-elf is 3.4.5
06:53:40goffabut.. if you download gcc 3.4.5 will m68k-gcc-elf be 3.4.5?
06:53:46Galoisgoffa, you should have tried building it without making any changes in tree.h first
06:53:59goffaGalois: you are right
06:53:59 Join Paul_The_Nerd [0] (
06:55:05 Quit vmx (Read error: 110 (Connection timed out))
06:55:18 Join vmx [0] (
06:57:28sharpeif 1+1=2, and 2+2=4, 1+1+2=4.
06:57:56infamisdoes 4=1+1+2?
06:58:16infamisor does 4=1+3
06:58:22sharpe3 is undefined!
06:59:09 Quit RotAtoR ("zzzzzzzz")
06:59:55infamisgoffa: I don't think there's anything wrong with the source, it's most likely the have you setup the environment? look for you calculator in here:
07:00:24goffai doubt there is a problem with source too
07:01:01sharpegoffa, do you need a completely unhelpful comment(tm) ?
07:01:32CtcpIgnored 1 channel CTCP requests in 0 seconds at the last flood
07:01:32*infamis waits...
07:01:49goffaheh.. sure
07:02:06sharpemaybe it doesn't like you.
07:02:18 Part Paul_The_Nerd
07:02:28goffai've thought about that
07:02:53goffacould be a government conspiracy too
07:03:04sharpewhee, time to take a break from coding the c64 emulator, and test gimp filters on images!
07:03:29infamisI say you just need to reinstall your build environment tools
07:03:43infamisper the wiki
07:11:45goffacrossing fingers
07:11:57 Join B4gder [0] (
07:12:44maecktest gimp filters on images? wha?
07:13:15sharpei'm bored.
07:13:55goffa< - anticipating a compile
07:14:02goffaso yeah.. i'm bored too
07:14:05BHSPitLappy<ot> anyone in here know vmware server well? </ot>
07:17:34sharpei think the silence give you my answer.
07:17:59 Quit maeck ("Chatzilla 0.9.65 [Mozilla rv:1.7.12/20060202]")
07:18:21sharpethat's kind of funny
07:18:28sharpei don't know why
07:18:57goffayeah... i was thinking he was showing his geek cred.
07:19:26sharpewhat if things worked like that in real life...
07:19:43sharpei'd segfault. :(
07:19:59infamisI'd be in jail with all my illegal operations
07:20:10goffathe way things are going to night it.. same here
07:20:38sharpei think i may get a lot of undefined function errors
07:21:20goffathe chicks would be giving me null pointer exceptions all the time
07:21:34infamisfemales would throw me 'syntax error'..."you can't say that!"
07:22:02sharpei'm a complicated man. you know, like shaft.
07:22:11infamishe's a bad motha...
07:22:19infamisSHUT yo mouf!
07:22:42 Quit lostnihilist (Read error: 110 (Connection timed out))
07:23:03infamisI'd be doing some bit-banging here & there, you know, the usual
07:23:29sharpethat's a vaguely sexual innuendo.
07:23:43sharpewe don't like vague.
07:24:01***Saving seen data "./dancer.seen"
07:25:02sharpewhy doesn't life break out into song?
07:25:29 Join dj-fu [0] (
07:26:38sharpein the play, fiddler on the roof, it breaks out into song
07:27:10sharpealong with the play, rent
07:27:58sharpeseems to be a trent
07:28:10goffaits called broadway
07:29:09 Join hardeep [0] (i=hardeeps@SDF.LONESTAR.ORG)
07:30:13*goffa wants gcc to compile properly
07:33:38*infamis wants ice cream
07:34:00*sharpe wants to sleep
07:34:06infamisthen he will look like this :)
07:34:53goffayou can come over and have some.. i have vanilla and rocky road
07:35:04goffafor now...
07:35:25goffa7-11 is probably closer though
07:35:30 Quit [TCK] ("well, if you say so.")
07:35:38infamiscan you paypal it? that'll work
07:35:42Galoisso lwn is covering the [be]jewels fiasco
07:35:46HClravon: yea, got the code somewhere?
07:36:03*goffa inserts ice cream into infamis's paypal account
07:36:05ravonHCl: Haven't changed much IIRC, but I can upload dumb.c
07:36:16hardeepinfamis: when you unshuffle a playlist, the order should match the original order except that the starting track may be different. is this what you're seeing?
07:36:30infamishmm, let me confirm
07:37:07ravonHCl: http.//
07:37:53HClk.. will look at it later, not now, thanks
07:38:13ravonHCl: I've been playing around with the GET_MORE_RESULT_SIZE (which I assume defines the buffer size) define. Having it at 1024 makes it way too slow.
07:38:26ravonI think it took 12 seconds to render a 3 second sample.
07:38:50infamishardeep: no, it doesn't match the original order; still shuffled
07:38:58goffathis is pissing me off
07:39:02goffamake[2]: m68k-elf-ranlib: Command not found
07:39:08goffawhen i try to compile gcc
07:39:12goffaer make install
07:39:25goffamakes fine though
07:39:27infamischeck your $PATH
07:39:34B4gderPATH not setup right
07:39:55Galoisyou need to build and install binutils and make sure the install directory is in your path
07:40:05B4gdernow goddamn it is SNOWING again
07:40:25sharpei think i am going to eventually re-organize the emulator source...
07:40:38sharpeB4gder, i never get snow.
07:41:00hardeepinfamis: ah, yes, i can reproduce. will investigate
07:41:15goffathanks everyone
07:41:30Galoissnow in april is just *wrong*
07:41:37*B4gder nods
07:41:41B4gderit is depressing
07:43:04hardeepinfamis: i'm assuming you have dircache enabled?
07:44:11goffawhee.. it compiled
07:44:13infamislemme check...
07:44:20goffanow time to see what the damage is
07:44:43infamishardeep: yes, it is on
07:45:33hardeepinfamis: thanks, i see the problem
07:45:38infamisjust confirmed that turning it off restores the correct order
07:45:55infamishardeep: cool
07:47:06sharpei'm pretty sure that this is around the fifth time i've tried to rewrite the code that times cpu execution...
07:47:52 Quit infamis ("ip-add change for rapidshare :)")
07:49:13sharpei think i have around 200 lines commented out
07:53:09 Join infamis [0] (
07:53:29 Join Paul_The_Nerd [0] (
07:54:01 Quit BHSPitLappy (Read error: 104 (Connection reset by peer))
07:54:56goffawell that fixed the double depress problem...
07:55:24sharpei dunno. emuators are a bitch.
07:55:32goffanow all i have to do is learn c and then i can make rockbox do exactly what i want... lol
07:55:43sharpethat's what we all thought... :D
07:55:59Paul_The_NerdI may be slightly unresponsive. I'm here in case there are people who need help badly, but I'm distracted mostly, so may not respond (at all / quickly / first)
07:56:13sharpei've been helping a bit today
07:56:36 Join mymomthelush [0] (i=user@
07:56:37sharpeor was that yesterday...
07:56:45goffawell.. bedtime for me
07:58:01hardeepinfamis: should be fixed in latest cvs
07:58:28sharpei wonder if it would be a viable solution for me to just declare an array large enough to contain the emulated ram, instead of using the plugin buffer... thoughts anyone?
07:58:50B4gdersharpe: oh yes
07:59:25sharpei think the only reason i didn't do that, is because i had some problems with it in the past...
08:00:38sharpebut i look at pacbox's source, and it does that...
08:00:39infamishardeep: i'll try it out
08:01:46sharpeer, does the arrays.
08:01:57 Quit dj-fu (Connection timed out)
08:04:26sharpeB4gder, i've a question.
08:05:01 Quit ravon ("Leaving")
08:05:58B4gderask away my son
08:06:42sharpewhen declaring the arrays for the emulated parts of memory, should i add "IBSS_ATTR" after the declaration?
08:06:47infamishardeep: thanks, it works!
08:07:21B4gdersharpe: no, that puts it in IRAM and I don't think you'll fit them in there
08:08:27sharpei feel like i should just rewrite the code to be cleaner.
08:10:00 Quit infamis ("CGI:IRC (EOF)")
08:11:40 Join infamis [0] (
08:12:46Paul_The_NerdIf you feel that way, then it's probably the truth
08:13:10sharpethen i'll take a shower, and rid of the dirty code. :D
08:13:49sharpeshould i keep the character rom in a seperate place, as it seems to be eventually overwritten...
08:14:31B4gderlikely, since the C64 can re-use the rom areas as ram on demand
08:15:09 Join ender` [0] (i=ychat@
08:15:24sharpewas it possible to change the character rom while it was running?
08:15:50B4gderwhat is the "character rom" ?
08:16:31sharpesort of like the font, i suppose...
08:16:55sharpe8 bytes per character...
08:16:56B4gderI don't remember if it was possible to modify in ru-time
08:17:02sharpei don't think it would be...
08:17:29B4gderno, I don't think it is
08:17:33sharpesince, i cant seem to find it copied anywhere in the ram
08:18:21sharpeokay, i'll skip loading it into the ram then
08:19:05 Quit mymomthelush ("Trillian (")
08:21:09amiconnhardeep: Thanks! :-) Tagcache playlist creation speed on Ondio is now as it should be.
08:21:28hardeepamiconn: cool. how fast is it?
08:21:42amiconn317 tracks: Before: 80 seconds. Now: 8 seconds
08:21:51*B4gder faints
08:22:53amiconnI'll now check whether it makes a difference on recorder too
08:23:15hardeepyeah, i'm building the tagcache right now on it to see as well
08:23:26 Join LinusN [0] (n=linus@rockbox/developer/LinusN)
08:23:37hardeepthe archos is definitely needing an offline tagcache creator =)
08:23:51amiconnnah, not really
08:24:07hardeepit's been going for about 10m for me now
08:24:46amiconnHow many tracks do you have on your archos?
08:25:09amiconnThen expect 15...20 minutes for the scan
08:25:21amiconn...and 3..4 minutes for the commit
08:25:56amiconnBtw, the not-syncing-while-adding fix should help all platforms, not just archos
08:26:08 Join ravon [0] (n=ravon@
08:26:16ravonThey have GOT to be kidding me.
08:26:18amiconnI noticed that (with the old code) the HD is seeking like mad during playlist creation
08:26:22ravonIt's god damn snowing again.
08:26:36hardeepamiconn: not really. the sync is done in a separate thread for the swcodec devices
08:26:39B4gderravon: hey, I already said that! ;-)
08:26:46hardeepit might improve some, but not as much as archos
08:26:50ravonB4gder: lol. Stockholm?
08:27:00ravonCurse the weather guy.
08:27:09amiconnhardeep: Yeah, still, this thread and the ui thread will fight for the disk
08:27:17ravonB4gder: Huddinge by any chance?
08:27:39B4gderravon: I live in Huddinge but I'm in Kista at work
08:27:40 Quit infamis ("CGI:IRC")
08:27:43 Join infamis [0] (
08:27:49ravonha, cool :D
08:28:03ravonNow, if you'd live in Flemingsberg the world would indeed be small.
08:28:21B4gderI don't => Myrangen
08:28:45ravonNever heard of :/
08:28:52amiconnhardeep: 65->25 seconds on a v1 recorder for a 3300-track playlist :)
08:29:11Slasherihardeep: btw, do you know what might cause the "error while accessing .playlist_control file" quite often when trying to resume playback? that file seems to corrupt easily somehow..
08:29:35hardeepamiconn: nice
08:29:42amiconnSlasheri: I didn't see this for years...
08:29:53B4gderme neither
08:29:54amiconnNah, not years, but at least one year
08:30:06hardeepSlasheri: I've seen reports of that but haven't reproduced myself
08:30:13amiconnBut then, I don't have dircache enabled
08:30:25B4gderme neither :-)
08:30:25hardeepnot sure what could be causing it
08:30:31Slasheriweird.. i experience that almost daily..
08:30:54Paul_The_NerdI've had it a couple times recently
08:31:12infamisI've had it a recenlty also
08:31:26hardeepis everyone whose seen it using dircache?
08:32:04Slasheriat least i have used dircache.. hmm, maybe i try disabling that for couple of days
08:32:06amiconnhardeep: Btw, do you know why the playlist sync is handled differently on hwcodec and swcodec (the playlist_cachectrl thread)?
08:32:37Slasherihardeep: hmm, or maybe your latest commit could have fixed that
08:32:43infamisdircache, yup
08:32:58hardeepamiconn: that was to avoid unnecessary disk writes
08:33:18amiconnYeah, but then why swcodec only?
08:33:22infamisSlasheri: I think so too, I've tried to reproduce the playlistcontrol file access error but haven't been able to
08:33:36infamisas of the latest cvs commits
08:33:48Slasheriinfamis: ah, that sounds good. i will do some testing with that too
08:34:09hardeepit could be done for hwcodec as well... the main problem was with gapless playback on swcodec so that's what i focused on
08:35:08amiconnaha... hmm...
08:35:47amiconnI wonder what excessive disk access the playlist code does that would influence gapless playback...
08:36:18hardeepthe playback code didn't respond too well to any disk access from playlist code
08:36:20amiconnGuess I need to check the commit msg
08:36:42hardeepin the playlist_next() call
08:36:57amiconnHmm, perhaps the problem doesn't exist anymore with the latest playback code?
08:36:59hardeepfor queued tracks, we needed to delete the track from playlist in that call
08:37:56amiconnHmm, syncing that means a spinup. Maybe the was a problem?
08:38:22amiconnOr does the cachectrl thread also spin up asap?
08:38:49hardeepit was the delay in returning because of the spinup
08:39:20hardeepthe cachectrl thread doesn't sync as frequently, and it's a separate thread so no problem
08:40:12 Quit infamis ("CGI:IRC (EOF)")
08:42:53amiconnHmm ok. From the commit msg it seems that the cachectrl thread is mainly useful in conjunction with dircache
08:43:21 Join infamis [0] (
08:44:06hardeepamiconn: that was only because i made use of the cachectrl thread
08:44:25hardeepif that thread is available on hwcodec devices then it would see the same benefit
08:44:51hardeepwe only flush to disk when it's active
08:45:17 Join lostnihilist [0] (
08:45:42amiconnPerhaps I read that wrong: "On non-dircache systems, behaviour should be the same as before (all commands except shuffle flushed immediately)."
08:45:42 Quit infamis (Client Quit)
08:46:44 Quit lostnihilist (Remote closed the connection)
08:46:51hardeepthat's because they don't have the extra thread to get the benefit
08:47:15amiconnHow does the cachectrl thread monitor disk activity?
08:47:35hardeepsame as how the battery plugin works
08:47:50hardeepit sleeps for disk sleep time and then queries ata_disk_is_active()
08:48:06amiconnsounds a bit cumbersome
08:48:13hardeephow so?
08:48:26 Join infamis [0] (
08:48:43amiconnI need to add disk activity signaling anyway for switching from config sector to config file for the main settings
08:49:00amiconnToday we have the delayed sector in the ata driver. This will go away then
08:49:11hardeepyeah, i was planning on switching to that once it became available
08:49:22hardeepthis was just a nice, quick workaround until then
08:49:44amiconnhardeep: It's cumbersome because there may be multiple threads just waiting for a spinup
08:50:32 Quit quobl (Remote closed the connection)
08:50:41hardeepwell, in this case the thread was also being used for the dircache so it's not a completely wasted thread =)
08:51:10 Join quobl [0] (n=quobl@tor/session/external/x-c1f605d71006e4f4)
08:51:27amiconnHmm, "Tagcache is not ready"?
08:52:38Slasheriamiconn: hmm?
08:53:55amiconnSomehow the tagcache files are gone after updating to current cvs...
08:54:10Slasheriamiconn: did the tagcache engine code update as well?
08:54:25 Join lostnihilist [0] (
08:54:37Slasheriif the header version number changes, old files are automatically removed by tagcache
08:54:53amiconnHmm, maybe. Not sure about the exact version I was running before
08:55:01sharpethe plugin api need a "axtoi()" function :D
08:55:54Paul_The_NerdY'know, it's really easy to accidentally overwrite a modified tagnavi.config
08:56:31LinusNwhere is tagnavi.config described?
08:57:14 Join ProgramZeta [0] (
08:57:14*Paul_The_Nerd wouldn't know.
08:57:22Paul_The_NerdAll I've done is read it, and tinker. :)
08:57:26SlasheriLinusN: there is no description for the syntax yet as it hasn't been finished yet
08:57:42Slasheribut the examples should give some clue
08:59:26amiconnSlasheri: The odd splash effect on swcodec after tagcache commit is still there, including the 'tagcache not ready' afterwards, requiring a reboot
08:59:54Slasheriamiconn: ah, trying that soon
09:00:08Slasherii think that happens only if dircache is disabled
09:00:48amiconnyeah, quite possible
09:01:37sharpeusing long variable names is satisfying.
09:02:04 Join petur [0] (
09:02:29amiconnPlaylist creation speed from tagcache is obviously only limited by the disk now.
09:02:49Slasheriamiconn: hehe, sounds good
09:03:00amiconn27 seconds for 3159 tracks on h140, 25 seconds for 3307 tracks on archos recorder
09:06:05sharpewell, rewriting the code seems to be going well.
09:07:52 Quit Kyomi|off ()
09:08:29hardeepinsight rocks for debugging simulator
09:09:12amiconnLinusN: Any news about the mysterious power drain?
09:09:50 Quit jnc ("reboot maintainence")
09:12:23 Join cismo [0] (
09:12:53LinusNamiconn: no, i haven't had time
09:13:45amiconnSomeone should check my theory about the isp1362 being a possible cause...
09:14:24peturLinusN: if I would supply you with a diff to build a test bootloader, could you test it for me (brick risk)? I want to init the isp1362 at boot time just like iriver does
09:14:59peturamiconn: guess what I'm doing several nights already (without progress :( )
09:15:13peturLinusN: ok, will prep this tonight
09:15:59 Join Hotfusion [0] (
09:16:10Hotfusionhi everyone
09:20:58amiconnBah, we should really implement rolo on ipod. Rebooting (e.g. for tagcache update) is cumbersome without it.
09:22:04HotfusionI'll idle here awhile. just want to say I totally appreciate the work you guys do and can't wait for future versions for the iAudio x5 players
09:22:39peturHotfusion: join the other 113 atm :P
09:24:02***Saving seen data "./dancer.seen"
09:25:47 Quit cismo_ (Read error: 110 (Connection timed out))
09:25:48 Quit Paul_The_Nerd ("Leaving.")
09:26:35 Join jnc [0] (n=jnc@
09:27:01sharpei was just about to put "/* settings struct */" next to the line that was, "struct settings {"
09:27:08 Join Zagor [0] (n=bjst@rockbox/developer/Zagor)
09:28:17midkayi don't get it.
09:28:38amiconnreboot, brb
09:28:41 Quit amiconn (" HydraIRC -> <- State of the art IRC")
09:30:32 Quit twisted` (Remote closed the connection)
09:33:40 Join amiconn [0] (
09:35:13sharpei put, /* because that's how we roll */ explaining why i do something in the code...
09:36:35hardeepSlasheri: Sorting appears to be incorrect when viewing Genre->Blah-><All tracks>
09:36:49peturI really hate comments like this: int get_a_number(void); /* gets a number */
09:38:41Slasherihardeep: hmm, what do you mean by incorrect? currently alphabetical sorting is used
09:38:53hardeepSlasheri: it's not alphabetical
09:38:56Slasheri(but if chunked browsing had to be enabled, then sorting will be disabled)
09:39:01hardeeponly in that view
09:39:03Slasheriok, then chunked browsing was in use
09:39:08hardeepah, okay... that must be it
09:39:50Zagorpetur: yes, the golden rule is "don't document what you do, document _why_ you do it"
09:40:57peturthe only people who read comments are those who are new to the code, so you should write them with that purpose
09:48:29 Quit infamis ("CGI:IRC (EOF)")
09:48:35sharpeanyone have another word for process, or start...?
09:51:47sharpeeh, i used imitate.
09:52:49sharpewell, took about 400 lines, rewrote them into ~220 lines.
09:53:13sharpefunny thing is, the 220 lines are more readable and documented than the 400 lines.
09:53:49peturis that RB sourcecode? amiconn will be hapy ;)
09:54:04sharpenope, c64 emulator source code. :)
09:57:18sharpeso disappointing, aren't i...
09:57:33 Quit darkless (Client Quit)
10:02:50 Join wehn [0] (
10:05:38 Join linuxstb_ [0] (
10:05:40 Quit linuxstb (Remote closed the connection)
10:06:02 Nick linuxstb_ is now known as linuxstb (
10:06:53peturthese guys are really stupid
10:07:18peturI get invited, want to fill out the form and it turns out it's U.S. only ?
10:07:36peturand it's nowhere on their site
10:08:21peturcd swapping site that's starting up
10:08:45Hotfusionnever been to the site but sounds like something I'd expect coming from a site calle
10:09:15Zagorooh, nice google logo today
10:09:28Zagortheir most artistic yet afaik
10:09:29sharpefor readability, should i just reference the ram location, or use a function to fetch the value?
10:09:33sharpeyes, yes, it is...
10:10:51sharpetis the artist's birthday :)
10:13:07LinusNshit vad rostig pň pci och linux man blivit
10:13:08Hotfusionhmm. looks like he was drunk drawing it
10:13:15LinusNoops, wrong channel
10:15:20hardeepblah, i can't reproduce the accessing playlist control error, even with dircache enabled
10:15:28hardeepplease let me know if you still see it with latest cvs
10:16:23 Join linuxstb_ [0] (
10:16:25 Quit linuxstb (Read error: 104 (Connection reset by peer))
10:16:37 Nick linuxstb_ is now known as linuxstb (
10:17:17B4gder... introducing ipod mini 1G builds
10:17:23Lynx_is there an automatic way to change the backlight timeout option when playing a game that requires looking at the screen without constant input, like sudoku? or do I have to go to the settings menu before playing every time?
10:17:40*linuxstb is regretting attempting to delete the HVSC collection from his player using Rockbox...
10:17:43*sharpe sounds a party favor.
10:18:39linuxstb10 minutes and still deleting...
10:20:13hardeepsleep time
10:20:15 Quit hardeep ("BitchX: coming soon to a theatre near you!")
10:25:16B4gderhey commit something! ;-)
10:26:26webmindcan anyone tell me how to get the cursor down at the filename in the edit screen?
10:26:30webmindon the nano
10:27:33Galoisin between when it warps from top to bottom, there is one step when it is on the filename
10:27:45webmindI'll try that
10:27:49webminddidn't notice it
10:29:58linuxstbHmm.... My recursive delete has now been running for 20 minutes and still going...
10:31:55 Join JdGordon [0] (
10:32:11 Quit JdGordon (Client Quit)
10:32:16 Join JdGordon [0] (
10:32:27 Quit Lynx_ (" HydraIRC -> <- IRC with a difference")
10:33:04 Quit quobl (Read error: 104 (Connection reset by peer))
10:33:23 Join quobl [0] (i=debian-t@tor/session/external/x-32286a632252c1af)
10:33:32 Join Rob2222 [0] (
10:34:59B4gderand we have a wider table!
10:36:29peturhehe, you found something to commit yourself ;)
10:37:39linuxstbDamn. 30 minutes and still deleting...
10:38:36B4gderpetur: yeah, had to dig up something ;-)
10:46:16 Quit dpassen1 (Read error: 110 (Connection timed out))
10:48:19 Quit Rob2222_ (Read error: 110 (Connection timed out))
10:48:52Bg3rlinuxstb how many files ?
10:49:35linuxstbBg3r: About 32000
10:49:44linuxstbAnd it's still going...
10:50:29linuxstbI'm submitting a feature request to enable a user to abort the recursive delete if they are as stupid as me...
10:54:02 Join qwm_ [0] (
10:57:10Slasherilinuxstb: hehe, just with my calculations, only the open() with so many files would take at least 32 minutes..
10:57:18Slasherithat is without dircache
10:57:29linuxstbI'm now up to 50 minutes :)
10:57:37linuxstbI think I should plug the charger in...
10:57:53B4gderits good with a thorough delete speed measure ;-)
10:57:54Slasheriyeah, opendir etc. will take some time too..
10:58:39linuxstbSo any guesses about when it might finish?
10:59:18Slasherijust looking into the delete code.. :)
10:59:44linuxstbIs the delete code in apps/ or firmware/ ?
11:00:16Slasheriyes, firmwre/common/file.c
11:00:19Slasheriit's quite simple
11:00:26Slasherii would say that something over 60 minutes
11:00:56linuxstbAh, so that will make user feedback tricky.
11:01:43 Join darkless [0] (
11:02:42 Join Evilnipple [0] (
11:03:00Slasherilinuxstb: the ui could show easily how many files it has removed so far
11:03:23Slasheriand offer the stop button to stop the operation..
11:04:06linuxstbI've just looked at the code - the recursive remove_dir() function is in apps/ - only the low-level delete is in firmware/. So yes, that's fine.
11:04:17 Quit qwm (Read error: 110 (Connection timed out))
11:04:37Slasherihmm, do you have a deep directory structure or many dirs?
11:04:44linuxstbCan you go back in time a few hours and implement it for me?
11:04:49Slasheribecause the opendir would be the slowest operation
11:05:03linuxstbIt's many dirs.
11:05:09linuxstbBut quite shallow.
11:05:42*Slasheri goes to the dusty basement to put the ancient timemachine in operation
11:08:29linuxstbNow 60 minutes...
11:10:22Bg3rhm, read(),write() etc don't check the fd value ...
11:13:05 Join PaulJ [0] (
11:14:05linuxstbHmm. Deleting the same folder from another one of my players took about 30 seconds from my PC...
11:15:14 Join muesli__ [0] (n=muesli_t@
11:16:49 Join Fots [0] (
11:16:59Fotshey guys :D
11:17:02 Join Moos [0] (
11:18:22*B4gder heys
11:18:39Fotsi just installed rockbox onto my ipod and am just playing around with it
11:18:53ShonkyA quick one: markun and I have been talking RB on a Toshiba gigabeat over on #gigabeat. It's basically himself, me and another guy and that's it. Since the Gigabeat already runs full linux how feasible would running Rockbox under it's linux OS be? I know nothing of Rockbox.
11:18:56Fotsso cool seeing an mpc file play on this thing :D although it does seem to stutter a little im afraid
11:19:15B4gderShonky: you could basically run the Rockbox simulator
11:19:33B4gderShonky: which is Rockbox using SDL for UI/sound
11:19:36 Join DrMoos [0] (
11:19:47ShonkyDoes that have a high overhead though?
11:19:57B4gderwell yes, you'd have the linux overhead
11:21:10Fotsguys, im sorry, may i ask, apart from the fonts available in the package i downloaded, is there any other fonts or ways that i can use to make it a little easier to read the text ? my eyes kinda suck :D">
11:22:26 Join Coroner [0] (
11:22:43 Join Rob2222_ [0] (
11:22:55 Quit Moos (Read error: 104 (Connection reset by peer))
11:23:02 Nick DrMoos is now known as Moos (
11:23:25 Join SereR0KR [0] (
11:23:42 Join damaki_ [0] (
11:24:06***Saving seen data "./dancer.seen"
11:24:49linuxstbShonky: There are lots of embedded devices running Linux, so it could be interesting to develop a version of Rockbox capable of running in such an environment - including the gigabeat. As B4gder said, it won't be as efficient as native Rockbox. But it should be a lot less work.
11:25:33Shonkylinuxstb and B4gder: well we have a 300MHz ARM9 core to play with :)
11:25:48linuxstbYes, but "efficiency" = "battery life".
11:26:08B4gder300mhz = eato mucho batteryo
11:26:23ShonkyI realise that but the faster I get something the better :)
11:26:30B4gderthen go! ;-)
11:26:41 Join wefds [0] (n=muesli_t@
11:26:56linuxstbShonky: Has anyone managed to run their own application on the gigabeat under Linux yet?
11:26:57ShonkyHell this thing has 16 hours life. I could live with 4 for most situations.
11:27:45Shonkymarkun has got scripts running. I can't get scripts to run on my X60 though. I'm sure I could compile and run a binary if I could get the script thingo working.
11:28:00ShonkyI think the X series is a little different to the F.
11:28:01B4gderbut I agree that getting Rockbox-on-Linux decent would enable it to easier run on lots of more players
11:28:27ShonkyB4gder: in the simulator?
11:28:36B4gderbased on the simulator at least
11:28:49B4gderas then it wouldn't simulate anymore ;-)
11:29:06Galoisthere's lots of WinCE devices too if we want to go that route
11:29:32linuxstbYou would probably have to write specific button/lcd/sound drivers for each device - depending on the APIs they provide. You also have the threading to deal with.
11:29:46B4gderwell, I won't prevent anyone from attempting rockbox on a wince device
11:30:01 Quit Coroner ("visit")
11:30:10Shonkymarkun was thinking of a standalone app with it's own drivers etc like conventional rockbox. I just was coming from the other angle thinking a linux app instead.
11:30:16B4gderlinuxstb: you mean it couldn't use threading similar to the current sim?
11:30:38linuxstbThe current threading doesn't work well AFAIK. I think that's the cause of the audio problems.
11:30:39ShonkyThey don't provide any API. All the drivers are done with kernel modules inserted at boot time to get them around the GPL.
11:31:18Bg3rhm, it seems that lcd_default_contrast() isn't used practically anywhere and lcd_remote_default_contrast() isn't used at all ...
11:31:32B4gderShonky: how do you access the LCD then? /dev/fb0 ?
11:31:43linuxstbShonky: You'll need to reverse-engineer the API then. I've looked at the gigabeat firmware, and I think Toshiba have helpfully left the symbols there.
11:31:49ShonkyFortunately the modules aren't totally stripped making looking at them easier.
11:32:15ShonkyNot sure about the LCD exactly yet. I think through a dev and ioctls
11:33:00B4gderwell if you can run their program with strace or similar that could be rather easy to track down
11:33:01linuxstbIMO, the first thing would be a proof-of-concept standalone linux app which accesses the important features of the gigabeat - filesystem, LCD, buttons and sound.
11:33:23B4gderafter all, they don'
11:33:29B4gdert use uclinux
11:33:55B4gderie = full linus = proper going-through-kernel procedures
11:34:11ShonkyYes. Unfortunately I really need an X model upgrade to get me started or at least knowledge of how it upgrades. That's why I was chasing you B4gder for the X upgrade
11:34:28ShonkyYes full linux. MMU, the works.
11:34:28B4gderdidn't find any yet?
11:34:37ShonkyNo :(
11:35:03B4gderhow eird
11:35:27ShonkyIt's a relatively new product (consider it a different product to the F) with few current *bugs* so my chances are a bit slim at the moment.
11:35:38linuxstbHurray! My delete has finally finished after almost 90 minutes :)
11:36:05peturnow it needs defragging?
11:36:18B4gderShonky: well still, I found that upgrade image easily
11:36:38ShonkyAre you sure it was for an X30 though as your directoru implies?
11:36:41 Quit Rob2222 (Read error: 110 (Connection timed out))
11:36:46linuxstbpetur: It should be fine - those files were the last thing I copied on, and the first thing deleted.
11:37:24ShonkyI haven't hunted high and low. Asked everywhere sensible. Waded through lots of Japanese web pages....
11:37:32Shonkyhaven't -> have
11:37:34 Quit muesli__ (Read error: 110 (Connection timed out))
11:37:35 Join muesli__ [0] (n=muesli_t@
11:38:07Bg3rlinuxstb what player was this on ?
11:38:23B4gderShonky: hm, perhaps I was sloppy then
11:38:32linuxstbBg3r: ipod 4G Color. Don't ask me to test other players :)
11:38:41 Quit damaki__ (Read error: 110 (Connection timed out))
11:39:08ShonkyNo worries. If you ever see one again though, hang on to it ;) I think you had an F upgrade personally
11:39:19ShonkyAnyway, I have to go home (2 hour commute)
11:40:39 Quit sandoaway (Read error: 104 (Connection reset by peer))
11:41:30HClravon: host not found
11:42:47Fotsguys, i must say, great work on the firmware, im afraid i had to roll back to the apple one due to the audio stopping ever few seconds
11:42:58Fotsi tried it with q6 mpc and −−alt-preset standard mp3
11:43:02Fots:") oh well
11:43:12crashdFots: did you have latest build?
11:43:18crashdbecause, playback has been up in the air lately
11:43:26Fotshmm, well I grabbed the daily build
11:43:28crashddon't be so quick to judge ;)
11:43:30crashdwhen from?
11:43:38Fots:") just today i believe
11:43:49Fotsi will try it again, dont worry
11:44:04 Quit JBGood25 (Read error: 110 (Connection timed out))
11:44:19Fotswhen i saw the feature set, my jaw dropped for about 20 minutes
11:44:51 Join Maxiii [0] (n=on@
11:44:53HClravon: ah, nm
11:45:28linuxstbFots: Audio stuttering is caused by the CPU failing to decode fast enough. You can help this by disabling other CPU-intensive tasks, such as the Equalizer, Crossfeed and the peakmeters in your While Playing Screen (WPS).
11:45:56 Quit Maxiii (K-lined)
11:45:57Fotsok i will try that out :D
11:46:04linuxstbTo disable the peakmeters, just choose a different theme - e.g. "boxes".
11:46:18Fotscool, i'll just reinstall the firmware and retry
11:46:20Fotsbrb :D
11:46:32Slasheriah, now tagcache commit should be a way faster with a less complex search
11:46:35Slasheritesting :)
11:46:59 Quit wefds (Read error: 110 (Connection timed out))
11:47:27 Join wefds [0] (n=muesli_t@
11:48:32Slasheripromising.. it was faster but it also corrupted the data ;)
11:50:32HClmmm, i'll need to adapt databox to work on tagcache, but first i want to take a look at dumb..
11:51:34SlasheriHCl: that sounds good
11:51:45*B4gder wants a databox with ascii search strings
11:51:48Slasherithe tagcache api should be quite simple to do that
11:52:04HClB4gder: shouldn't be too hard, all it needs is a scanner to be able to read text
11:52:26B4gderit would be cool to be able to edit them on host or provide a set of generated ones etc
11:52:27HClthe only downside is that at the moment, databox forces a search query to be correct
11:52:30HCland is idiotproof
11:52:31SlasheriHCl: you should also have a look of the tagnavi.config
11:52:37HCllater, later, first libdumb
11:52:41HClvery busy with college too
11:52:45HCli barely even get enough sleep nowadays
11:52:46HCl :/
11:53:09B4gdersleep? now that's only for really old people
11:53:22*B4gder looks at LinusN but says nothing
11:53:26Bg3rbut they often have insomnia :D
11:54:11B4gdernothing beats a fine SIGSEGV
11:58:31 Join Rob2222 [0] (
11:58:44 Join filoktetes [0] (
11:59:41 Join Paul_The_Nerd [0] (
12:00:11Fotsalrighty ...
12:00:14Fotslame 320 encode plays ok
12:00:20Fotsnow trying vbr aps encode
12:00:33Fotsi LOVE not using itunes :D
12:00:49Fotsseems to be playing find ! w00t
12:01:36Fotsthanks man :D
12:01:48Fotsi made sure the eq and crossfeed was off
12:01:51Fotsmaybe it was that :D
12:05:24crashdthat really saps it
12:05:44 Quit muesli__ (Read error: 110 (Connection timed out))
12:07:07Bg3rshouldn't things like [MIN|MAX]_CONTRAST_SETTING be in the apropriate config-xxxx.h files ?
12:07:42Bg3rinstead of settings.h
12:09:24ravonHCl: Got your charger now?
12:12:30 Quit Rob2222_ (Read error: 110 (Connection timed out))
12:13:04preglowmp3 is slooow on ipod yet :/
12:13:07preglowgod knows why
12:13:48 Join twisted` [0] (
12:13:58*twisted` smacks sharpe
12:14:12twisted`sharpe: no offence :P it was just the first nick that sprung to my mind
12:14:54Moospreglow: any luck with the musepack patch?
12:23:02 Join webguest65 [0] (
12:23:47webguest65hardeep implemented a shuffle all feature, is that a shuffle album feature ?
12:24:41preglowMoos: trying to fit it in today
12:24:44preglowi don't have much time
12:24:57Moospreglow: ok, thanks
12:27:58webguest6502.16.44 # <hardeep> now everyone can stop asking for the shuffle all feature −− it's in!
12:28:15webguest65shuffle all albums ?
12:29:23LinusNall files
12:29:50 Join _FireFly_ [0] (
12:29:54webguest65albums are folders I guess
12:30:06Paul_The_NerdAlbums are albums.
12:30:39webguest65file tree system what is a folder ?
12:30:40 Quit filoktetes ("Leaving")
12:30:56Paul_The_NerdWell, if you've put your albums in individual folders then yes, they're also folders.
12:31:01Paul_The_NerdBut really "Those folders are albums"
12:31:26Paul_The_NerdBut that would be a "shuffle folders" feature really. A shuffle albums feature would only work within TagCache.
12:32:02webguest65so hardeeps shuffle all feature is nothing new then
12:33:47HClfinally, my iriver charger
12:33:57Paul_The_NerdWouldn't know
12:34:07Paul_The_NerdI was just clarifying the difference between "Shuffle Albums" and "Shuffle Folders"
12:34:37LinusNit's not a new feature at all
12:35:07LinusNit's mainly that you can play the "All tracks" virtual folder in database mode
12:35:48webguest65ah, it's for db, thanks for clarifying Linus
12:36:08LinusNshuffle all can be done in file mode too
12:37:21webguest65yes I know, I just got excited because I tought that he finally gave in and implemented shuffle all folders in File Tree
12:37:32Bg3rdoes anyone knows min/max contrast values per target ?
12:38:26LinusNBg3r: use the source, luke
12:39:05Bg3rLinusN the source doesn't help much ...\
12:40:43Bg3respecially settings.h ...
12:41:46 Quit SereR0KR ("XChat Aqua")
12:42:16 Join ashridah [0] (
12:42:21 Part webguest65
12:43:38 Quit Paul_The_Nerd ("Leaving.")
12:47:51amiconnlinuxstb: I could have warned you that recursive delete is slow.
12:48:25amiconnIt usually takes several minutes to just delete the 'System volume information' crap added by windows from time to time
12:49:11amiconnLinusN: 'Shuffle all' cannot easily be done in file mode. You can recursively insert a folder shuffled, but not the root itself
12:49:36amiconn...and generating root.m3u is now slower than using the 'All tracks' folder in tagcache
12:49:37LinusNthen have all your music in a folder
12:49:45 Join SereR0kR [0] (
12:50:16amiconnI don't see the point of having an extra layer of folders, so I have my genre folders in the root
12:51:02LinusNi have my music in a music folder
12:51:02LinusNto separate it from audiobooks and data
12:51:44amiconnI have no audiobooks. Data is in a separate folder which is hidden on target
12:52:20linuxstbamiconn: Now you tell me that :)
12:52:36 Quit `3nergy (Read error: 104 (Connection reset by peer))
12:52:50LinusNbut if you want to have an easy way to shuffle all, having it in a subdir is a way to do it
12:53:10amiconnOr use 'All tracks' with shuffle now...
12:53:35LinusNmy music isn't that well tagged, so i don't use the database
12:54:00 Join `3nergy [0] (
12:55:28amiconnI didn't use the old tagdb often, and that probably won't change much, although tagcache is really more convenient to use
12:56:01amiconnOne thing I will use it for is finding tagging mistakes
12:58:15*amiconn spots 2 new builds in the table...
12:59:03Fotsguys, is it possible to fastfoward with mpc files on ur player ? after i ffwd, i simply get no sound and have to restart the track
12:59:05Fotsany ideas ?
12:59:18 Join SereRokR [0] (
12:59:21 Quit SereR0kR ("Leaving")
12:59:36HClSlasheri: got a document somewhere on the api to the tagcache?
13:00:33 Join Paul_The_Nerd [0] (
13:01:38Fotsbtw ffwd in mp3 works perfect
13:02:38linuxstbFots: At the moment, there is no seeking in mpc files.
13:03:34Fotsaah np :D
13:05:26Fotsflac files play AWESOME ! i cant believe how good this firmware is :D
13:06:32SlasheriHCl: check tagcache.h and tagtree.c
13:07:01preglowamiconn: so ipod mini needs two builds, then?
13:07:23amiconnNot necessarily. At the moment, we do it this way
13:07:42preglowbut it's possible to merge them?
13:08:11amiconnWell, why not?
13:08:21Slasherinow full commit takes only a few seconds on iriver.. something like 20s
13:08:23preglowdunno, i haven't looked at the code
13:08:30Slasheribut it requires also more memory
13:08:30 Join mikearthur [0] (
13:08:37preglow"oh noes"
13:08:44preglowit only happens at startup anyway, no?
13:08:48amiconnIt would increase binary size a bit, because the binary would then contain both variants for the driver routines that are different
13:09:03Slasherino, it still happens at runtime if dircache is big enough.. or tagcache is loaded in ram
13:09:08preglowamiconn: ah, right, in the manner we can merge all our ipod builds, then
13:09:18Slasheribut i lowered the max files in tagcache from 20k to 10k
13:09:21amiconnThe *_init(9 functions would check the hw rev and select the appropriate routines
13:09:25Slasherithat lowers the memory requirement
13:09:32preglowSlasheri: i'd really like to see dynamically resizable mp3 buffer in 3.0 :>
13:09:41amiconnpreglow: No, we can't merge all ipod builds
13:09:46Slasheripreglow: hehe, most probably not in 3.0 ;)
13:09:59Slasheripreglow: i would like the chunked dynamic non-contiguous mp3 buffer..
13:10:18Slasheriand it would cause only something like 2% estimated performance hit
13:10:19preglowthen the file loader would have to know about the stream format
13:10:20*amiconn prepares for heavy bashing
13:10:34preglowand i don't like that idea
13:10:45Slasheripreglow: playback engine would keep the stream contiguous for the codec
13:11:01preglowSlasheri: exactly, which requires the playback engine to know about all file formats we support
13:11:08amiconnpreglow: There are some difference we cannot account for at runtime, like different framebuffers izes + format, ...
13:11:21preglowamiconn: ah, yeah, that's right
13:11:29preglowamiconn: in that case i really think we should merge the builds in the future
13:11:54preglowless builds = good
13:11:58HClravon: dumb.c totally doesn't work here.. it won't even recognise my file as .xm and attempts to load it as .mod, and it takes ages, and after wards it crashes and freezes my player..
13:12:22amiconnIn fact I think the minis are the only case where it makes sense to merge the builds
13:12:34amiconn(and perhaps the old 1g and 2g ipods)
13:12:38 Quit TiMiD (Read error: 104 (Connection reset by peer))
13:12:46amiconn...if they will ever be supported
13:12:55Slasheripreglow: hmm, why it would need to know that? isn't it enough just to make sure the guard_buf_size amount is contiguous.. (the same way it's done currently when buffer wraps)
13:13:23HClSlasheri: any better documentation? :|
13:13:28ravonHCl: hang on. I have a small chip module you can use.
13:13:28SlasheriHCl: not yet
13:13:30linuxstbWe still need someone with a 60GB 5g to start using all 64MB of RAM.
13:13:36preglowSlasheri: sure, if you'll use a guard buffer as well
13:13:54amiconnlinuxstb: Now that will be a tough one...
13:13:58SlasheriHCl: but basically you call tagcache_search(...) and after that in a loop: tagcache_retrieve(...)
13:14:05preglowlinuxstb: had a look at the libmad thing?
13:14:09Slasheriand finally tagcache_search_finish(...)
13:14:15HClSlasheri: okay..
13:14:48HClravon: still, it should work with the file i give it though..
13:14:53 Join TiMiD [0] (n=TiMiD[
13:14:58SlasheriHCl: and be sure to check the tagnavi.config, tagcache has an integrated conditional search engine also
13:15:06 Quit Daishi (Connection timed out)
13:15:10 Quit darkless (Read error: 113 (No route to host))
13:15:24ravonHCl: Yeah, but it's tedious to test with a large file. Takes forever to load.
13:15:37HCltrue, but, it's not even loading properly at the moment..
13:16:24HClSlasheri: not sure whether i can use that or not.. searchengine currently uses (a rather inefficient) iterator over all the songs
13:16:58HCland internally it uses arrays of booleans to check which files still match the conditions
13:17:50 Join Daishi [0] (
13:17:59SlasheriHCl: for best performance, you should use the tagcache search engine always when possible
13:18:29HClSlasheri: i'd still need some sort of boolean array to keep track of the matches, which would berather hard to do...
13:18:38HClravon: uhhhh
13:18:53HClravon: hm.. seems heavily overloaded / too slow
13:19:07Fotssorry to ask another question folks, im trying to install the theme from so i download and extract the file to the appropriate folders but when i switch theme and go to display my song, no images appear, any ideas ?
13:19:32HClbut at least it doesn't crash
13:19:50HClSlasheri: any documentation on that search engine?
13:20:19SlasheriHCl: well, you should check tagtree.c for tagcache_add_filter(..) and tagcache_add_clause(..)
13:20:29Slasherino better documentation yet
13:20:56HCl :|
13:21:12Slasherifilters can do rough filtering based on other matching tags, and clauses do more complex filtering based on different tags
13:21:15 Quit Siku ()
13:21:37peturFots: did you extract them into their own directory? you must maintain the same folder struct as in the zip
13:21:52amiconnI've been playing with gcc versions and -O options for archos, in order to see whether we can cut down binary size.
13:22:10Paul_The_NerdSlasheri: Have you considered adding a "not" into Tagcache. For example, a means by which one can say "Genre does not contain the substring book"
13:22:32preglowamiconn: and...?
13:22:46Fotspetur: yep i made sure i did that, the only thing im thinking is that i have a 4g ipod photo
13:22:53Fotsbut i thought it would be able to handle it :(
13:23:16peturFots: screensizes must match
13:23:17HClSlasheri: think you could write some basic documentation on it later? those functions aren't really helping..
13:23:34HCllibdumb doesn't seem very hopeful at the moment, by the way :/
13:23:37Fotspetur: aah, but :") the pics actually dont display at all, instead the filenames are displayed
13:23:54amiconnrecorder v1, sh-elf-gcc 3.3.6, -O: ~224KB, -Os: ~215KB sh-elf-gcc 3.4.6, -Os: ~212KB
13:24:08***Saving seen data "./dancer.seen"
13:24:19preglowbut didn't Os break things?
13:24:27amiconnBut: -Os needs some workarounds for gcc bugs to make the binary run properly
13:24:29Slasheripreglow: yes, not will be added soon. And i have considered also "or" to be added later
13:24:35Slasherioops, Paul_The_Nerd
13:24:53SlasheriHCl: yes, sure
13:24:55preglowamiconn: and i thought you didn't use 3.4 on archos
13:25:04amiconn(and compiling with -Os and gcc 3.4.6 breaks in gray_core.c)
13:25:27amiconnpreglow: No I didn't, because 3.3.x makes smaller binaries than 3.4.x with just -O
13:25:27HClit sounds as if tagcache can already do a lot of the functionality that searchengine can, but without the more advanced things like (year == 1999 or year == 2000) && artist == queen
13:25:41 Quit dpro (Read error: 110 (Connection timed out))
13:25:48HCldo search results have some sort of tagcache "id" that i might be able to use?
13:26:00SlasheriHCl: basically tagcache can do that already, but or operations are not yet supported
13:26:19Slasherionly "and" is supported at the moment
13:26:20HClhow far through can it though?
13:26:41HCl((year == 1999 or year == 2000) && artist == queen) or artist == beatles
13:26:55amiconnWith -Os the relation is swapped...
13:27:04HClwell, in any case
13:27:30HClSlasheri: do you think it might be possible (or do you have something like this already) to have a song id in the tagcache ranging from 0 - amount of songs on the player
13:27:56HClthat would be returned by the search result and preferably, being able to fetch a song based on that id
13:28:15Slasheriwell, for example to search titles with year >= 1999 and artist == queen, you would do title ? year >= 1999 & artist = "queen". Tagtree parses that into structs and passes these parsed structs to the tagcache engine
13:28:19 Join pixelma [0] (n=M_@
13:28:26HClyea, i saw
13:28:29Slasherithen just retrieve the results with while (tagcache_retrieve(...))
13:28:34HClbut my guess is that it can't do bracket things?
13:28:45Slasherinot yet, but i have planned to add that feature in future
13:28:49HClnot that it really matters
13:28:57HClif the tagcache can supply me with a song id
13:29:04HClranging from 0 to the total amount of songs on the player
13:29:15Paul_The_NerdShouldn't it be "songs-1"?
13:29:18HCland can return those with a search result, and can fetch the filename of a song linked to that id
13:29:21HClyes, sorry
13:29:38Slasheriyes, you can do that already with tagcache
13:29:45HClanyways, if the tagcache can do that, it should be relatively easy to adapt the searchengine plugin, *and* have it use the native search function of the tagcache
13:29:58HClcould you write up some documentation to how the tagcache works and how i would do that?
13:30:07Slasherithe "id" is however linked to the current tag requested, as for some tags there might be more than one filename
13:30:28SlasheriHCl: yes, i will do that as soon as i have tiem :)
13:30:40HClwell, it's a choice to make, really, do we return all filenames or just the first
13:30:48lostlogiclinuxstb: you moving those labels in the codecs, or want me to?
13:30:58HClwe can always extend it, i guess? i'm mostly focused on adapting searchengine to work with tagcache
13:31:06SlasheriHCl: you can always retrieve just the first entry all entries tagcache can supply
13:31:20Slasheriof course
13:33:11amiconnHCl: What would searchengine be used for in future?
13:33:25amiconnI mean, tagcache already has quite powerful search features
13:33:42HClit kind of depends on whether tagcache will be able to support brackets and such
13:33:52amiconnImho it would make sense to adapt databox for editing tagcache search definitions
13:33:59HCland it depends on what we'll do with runtime data
13:34:57HClsearchengine is relatively independant of the system below it, i'm not sure what the case is with the tagcache search since i barely know anything about it
13:36:10HClyou'd use tagcache search normally, and searchengine for more complex queries, unless we can get the tagcache search to support everything searchengin does
13:38:00 Join darkless [0] (
13:38:06SlasheriHCl: Hmm, how complex queries searchengine can do?
13:38:13HClpretty much, everything you can dream up.
13:38:24HClas long as you close your brackets properly
13:39:15Slasherihmm, does it more than brackets which are not yet supported by the tagcache or tagtree parser?
13:40:00HClit's kind of hard for me to answer that since a) i don't know how the tagcache search works at the moment and b) i don't know the extent of the proposed brackets
13:40:06HClbut i'll try to think of an example
13:42:07HCl((artist==beatles||artist==queen)&&(year!=1993&&year!=1983)&&!title contains "strawberry fields") || (artist==enya&&(title contains time||year!= 1983))
13:42:41B4gderI always do that search too!
13:42:44HClmm, forgot a bracket
13:42:44HCl :P
13:42:51HCl(((artist==beatles||artist==queen)&&(year!=1993&&year!=1983))&&!title contains "strawberry fields") || (artist==enya&&(title contains time||year!= 1983))
13:43:11Slasherihehe, that should be possible with tagcache, except the brackets and orring at the moment
13:43:25HClthe brackets and orring is just the gist of it though
13:43:43HClotherwise you'll simply get different results because it's a different search :
13:43:44HCl :/
13:43:45Slasheriyep, but those feature would be quite easy to add to tagcache also
13:43:52Fotsi just cant get my images to show up when loading themes, most the time the filename is displayed instead of the pic, for other themes some pics are displayed and some aint, iv made to sure to extract everything in the exact dir strucutre, using a ipod4g color, any ideas ?
13:44:07HClsounds good, then all we'd have to worry about is how to add runtime data
13:44:22 Join qwm [0] (
13:44:47SlasheriHCl: there would be at least two ways to do that: save it internally in the tagcache or externally
13:45:07HClgranted that such complex searches are not usually the case, imho it should be possible since we want to be as flexible as possible
13:45:19Slasheriinternally sounds good, because sooner or later tagcache will support modifying of the tagdata on the fly
13:45:29HClSlasheri: i'm not sure how the tagcache works so i have no idea which is the better solution
13:45:43HClbut basically
13:45:47HClyou'll want runtime data to not get lost
13:45:58HClyou'll want runtime data to work on same files
13:46:08HCleven if there are multiple copies of the same file on the player
13:46:10SlasheriHCl: yep, but because of modifying of the tags, tagcache must support exporting and importing data too
13:46:15HClthey should share the runtime data
13:46:24HCland not have two different entries
13:46:26Slasheriso there would be a changelog file that could be automatically imported by tagcache
13:46:43Slasheriyes, that would be a main issue
13:46:54HClthat's one of the requirements of the runtime database..
13:47:00lostlogicCan anyone tell me why mpa.c line 173 is goto next_track; and not break; ???
13:47:33HClanother one is that whatever happens, runtime data must *not* get lost
13:47:38HClunless specifically requested by the user
13:47:47 Join RedBreva [0] (
13:47:50Slasherilostlogic: hmm, not sure about that..
13:47:56Slasheriit might have a reason or not
13:48:14lostlogicSlasheri: hehe, I'm inclined to change it to break; and see what happens.
13:48:18HClthose are the two main things of the runtime data, i think
13:48:22SlasheriHCl: yes, changelogs would make sure no data gets lost
13:48:36Paul_The_Nerd"it might have a reason or not" is a statement that always resolves to true. Not terribly helpful. ;-)
13:48:38HClmust share the same runtime data between files that are the same (preferably the same song, ignoring tag data)
13:48:43HCland data mustn't get lost
13:48:43 Join amiconn_ [0] (n=jens@rockbox/developer/amiconn)
13:48:45 Quit Seed (Nick collision from services.)
13:48:48 Join Seedy [0] (i=ben@
13:48:59SlasheriPaul_The_Nerd: hehe, true :)
13:49:00 Join wideawake182 [0] (
13:49:27 Join B4gd3r [0] (
13:50:45 Join Rondom [0] (n=Rondom@
13:51:33 Part wideawake182
13:52:56HClbut yea, if the tagcache search can support brackets and not operations and such, searchengine won't be needed
13:53:12 Join PaulJ_ [0] (
13:54:00 Quit qwm_ (Read error: 110 (Connection timed out))
13:54:15Slasheriyeah, support for that should happen in future (not hard to implement to tagcache, parser needs some more work)
13:56:08HClwe really need more documentation on how things work internally in rockbox, if certain developers disappear no one will know how to extend some things :X
13:56:33HClbut yea, writing documentation usually isn't very fun >>
13:57:37 Quit amiconn (Nick collision from services.)
13:57:37 Nick amiconn_ is now known as amiconn (n=jens@rockbox/developer/amiconn)
14:00:01linuxstblostlogic: I haven't really got any Rockbox time at the moment, so please go ahead and move them.
14:00:21lostlogiclinuxstb: kk, just did :)
14:00:47lostlogiclinuxstb: ended up adding another label to most codecs, because a codec initialization error should be an error, but any file decoding error should not.
14:01:25lostlogicSlasheri: where is mp3_play_stop defined for swcodec? I see it's called in talk.c, but I only see it defined in mp3_playback.c which can't possibly be right.
14:01:46 Join Lynx_ [0] (
14:02:07*petur is impressed by the buildspeed under vmware
14:02:21Bg3ramiconn ?
14:03:16Slasherilostlogic: hmm, probably it's not defined at all.. just a dummy function
14:04:02lostlogicSlasheri: think I'll hafta define it for voice to shuttup correctly.
14:04:02Slasheriyes, it's a dummy.. in firmware/mp3_playback.c
14:04:41 Quit SereRokR ("XChat Aqua")
14:05:17 Quit B4gder (Read error: 110 (Connection timed out))
14:06:03 Quit PaulJ (Read error: 110 (Connection timed out))
14:06:42linuxstblostlogic: OK, thanks. We need to remember to do the same with the SID codec.
14:08:49lostlogiclinuxstb: yah, I haven't actually tested my changes yet, because I have another pending change in all the codec files loosely related to my voice changes.
14:14:43linuxstbPaul_The_Nerd: Regarding this forum post (false detection of USB on a 5g): - Someone else reported a similar problem in the early days of Rockbox on the 5g. It seems that on a very small number of 5gs, the current USB detection code doesn't work, and always gives a false positive.
14:15:47Paul_The_Nerdlinuxstb: I thought a change had been committed that resolved that other person's problem. Didn't you change USB detection on 5G, or did that end up being backed out?
14:16:01linuxstbI _think_ lostlogic backed it out.
14:16:26Paul_The_NerdI was working on the assumption that the change was still in, and that maybe Loader2 was initializing more/less stuff than we do and potentially causing a problem.
14:16:36Paul_The_NerdTherefor agitating the previous issue
14:16:44linuxstblostlogic: ?
14:18:41linuxstbYes, it was backed out:
14:19:32 Quit Rondom ("pwned by Chanserv")
14:19:34linuxstbBut it seems very odd that only two people have so far reported the issue.
14:21:04linuxstbOnly thing I can suggest is to try upgrading to the latest version of the Apple firmware - making sure the reflashing happens to change the Apple bootloader code.
14:21:32lostlogicwouldn't call it a backout of a fix, so much as a fix. the USB detection that was used to prevent the false positive was to detect USB based on whether or not the battery was charging, meaning that on a full battery USB wouldn't work.
14:22:24linuxstbIt was the backout of a hack.
14:22:53linuxstbI think I agreed with you at the time we should remove it - as only one person had reported problems with the USB detection.
14:23:55 Join dj-fu [0] (
14:26:30 Join webguest78 [0] (
14:31:47 Quit Paul_The_Nerd ("Leaving.")
14:38:37 Join arkascha [0] (
14:40:14 Join Kohlrabi [0] (
14:41:51 Quit arkascha (Remote closed the connection)
14:43:02 Join warthawg [0] (
14:44:59 Quit wehn (Read error: 113 (No route to host))
14:46:24 Join stoffel [0] (
14:46:55 Quit midkay (Read error: 104 (Connection reset by peer))
14:47:05 Join midkay_ [0] (n=midkay@
14:48:58amiconnBg3r ?
14:50:41Bg3ramiconn i wanted to ask you something about the iriver remotes ...
14:51:26Bg3rhm, do you know why the first change on the remote's lcd happens after ata_init (and maybe more)...
14:51:58 Quit Fots ("It's raining outside, there's a storm front on my back, trying to keep me away from you ....")
14:52:08amiconnBg3r: What do you mean?
14:52:30Bg3ri was "tracing" the main.c:init()
14:52:48amiconnThe bootloader doesn't contain the remote lcd driver, so the init happens at main rockbox boot
14:52:50Bg3rto see why the logo isn't shown on the remote at the same time as on the main screen
14:53:04Bg3ramiconn yeah, i know
14:53:22amiconnThen the remote detection timer counts down before the remote lcd gets initialised
14:53:26Bg3rfirstly i thought that the problem was in the low default (firmware/) contrast (32)
14:53:57amiconnThis countdown is necessary in order to make sure the lcd gets initialised properly
14:54:52amiconnWhen you plug in the remote, not all pins make secure contact at the same time, so initialising too early would fail
14:55:09Bg3rmy remote is plugged in all the time
14:55:37amiconnThe init doesn't check whether the remote is already plugged at startup
14:55:59amiconnand I'm not sure whether it would be a good idea to change that
14:56:16lostlogicgrr, why is voice_remaining never reaching zero *scowl*
14:57:17Bg3ramiconn i can use sleep() after system_init() and kernel_init(), yes?
14:57:57 Quit webguest78 ("CGI:IRC (EOF)")
14:58:39amiconnIt might happen that a user plugs the remote at boot, and if we detect the remote as already plugged and init too early, he would end up with no display at all
14:58:41Bg3ramiconn it seems u're right
14:58:56 Quit B4gd3r ("time to say moo")
14:59:35Bg3rthe reason is the timeout
15:04:04amiconnAhahaha, I've just realised something:
15:04:35Bg3rbtw, shoudn't the adc_init go before the remote init ?
15:04:36amiconnPlugging the non-LCD remote also initialises the remote LCD and tries to update it
15:04:50warthawgi have 7 albums on my ipod now, almost ready for my trip to san diego
15:05:00amiconnNo harm done, but wasting CPU cycles...
15:06:45 Quit dj-fu (Read error: 110 (Connection timed out))
15:07:05 Quit XavierGr ("Trillian (")
15:08:55 Quit wefds (Read error: 110 (Connection timed out))
15:13:56Bg3rmain.c:init()->lcd_remote_init()->tick_add_task(remote_tick)->if(remote_present&&countdown==0) adc_scan(...)
15:14:21 Join qwm_ [0] (
15:15:03Bg3rand lcd_remote_init() is called before the adc_init()
15:15:31Bg3ramiconn ? also, can we lower the countdown to HZ/2 ...
15:15:41amiconnYes, in order to make it bulletproof, lcd_remote_init() should go after adc_init()
15:16:13amiconnIt's currently different because there was no need to scan adc within the tick back when lcd_remote_init() was added
15:16:28amiconnIt's not a real problem because of the timeout
15:17:01Bg3rcant it be the other way: to move adc_init before lcd_remote_init
15:17:12amiconnBack when implementing remote detection, I had to increase the timeout until I got stable init
15:17:30Bg3rah :(
15:19:02amiconnHZ/2 was not enough
15:19:26amiconnEven with HZ it's possible to break the init, if you plug in really slowly
15:20:37Bg3rokay, it seems it'll stay this way ..
15:20:47Bg3rin fact i can't see the logo on the remote at all ...
15:21:10Zagoris there any way to detect the broken init and retry?
15:21:52Bg3rbtw, what do you think about adding a lcd_min_contrast() and lcd_max_contrast() and use these in the settings_menu ?
15:22:37amiconnWhat should that be good for?
15:22:55Bg3ramiconn which ?
15:23:02Bg3rthe zagor's q or mine ?
15:23:28amiconnZagor: We can't detect a broken init as the lcd data connection is write-only
15:23:39Zagoramiconn: right
15:23:48amiconn..but I have an idea how we can have early display and relatively reliable init
15:23:57 Join SereR0KR [0] (
15:24:11Bg3rt0mas :D
15:24:12***Saving seen data "./dancer.seen"
15:24:38amiconnInstead of couting down once and then init, we can try a series of inits (e.g) every HZ4 during a (e.g.) 5-second period
15:25:12 Quit qwm (Read error: 110 (Connection timed out))
15:26:24amiconnI also have some ideas how to make the detection more reliable
15:29:39amiconnIirc I observed some signs that the iriver firmware does the same (multiple inits in succession)
15:30:20Bg3ramiconn what about mine question ?
15:30:50amiconnWhen plugging in the remote very slowly in retailos, I once managed to get a flipped display with some pixel glitches, which was then corrected (orientation + glitches) after half a second
15:31:07amiconn[15:22:37] <amiconn> What should that be good for?
15:32:10Bg3ram i right that archos recorders have 2 different lcds ? or at least different contrasts
15:33:19Bg3rare their min and max contrast settings equal ?
15:34:09amiconnAfaik min and max are equal on all our current targets, just the default is model specific
15:34:28amiconnMin and max is what can be set in the controller
15:34:49Bg3rhuh, okay
15:35:12Bg3rthen, are all screens readable on their min/max settings ?
15:35:45Bg3rshouldn't we just allow such values ?
15:35:55amiconnI'd say all of them are unreadable both at extreme min and extreme max
15:36:05Bg3ririver remote is even worse
15:36:10 Join dpassen1 [0] (
15:36:12Bg3rit's not readable under 32 and over 54
15:36:22amiconnYes, that's common
15:36:44amiconnI don't think that we should limit arbitrarily
15:38:36 Join gtkspert [0] (
15:40:31Bg3ramiconn so, leave it as is ?
15:43:25preglowby god, how i hate bots being used in support channels
15:43:37 Quit gtkspert ("Changing server")
15:44:10preglownothing says friendly like an "!rtfm"
15:48:45ravonWhat alternatives are there to libdumb?
15:48:51ravonExcept for mikmod and modplug
15:49:33ravonIt feels like it's been done thousands of times by now.
15:49:59preglowit has
15:50:08preglowwhat's wrong with dumb?
15:50:14ravonThere should be a really leightweight mod-only, semi-accurate player out there.
15:50:19ravonpreglow: Seems to be way too slow.
15:50:49preglowhas it got any floats in it?
15:50:55 Join Jungti1234 [0] (n=jungti12@
15:51:49ravonIt has.
15:51:51ravonmarco@garvey ~/dev/rockbox/rockbox-all/apps/codecs/dumb $ grep -R float * | wc -l
15:52:04 Join qwm [0] (
15:52:16amiconnravon: doubles?
15:52:52ravonamiconn: 5
15:52:59ravonMeeting.. brb
15:58:48Bg3r LOL
15:59:51 Join TeaSea [0] (
15:59:54peturhahaha... kill kill kill
16:00:13Bg3r <= but this is ... <no comment>
16:00:56 Quit darkless (Client Quit)
16:02:09 Quit qwm_ (Read error: 110 (Connection timed out))
16:02:36lostlogic For anyone interested in what I've done so far with trying to make voice more reliable on swcodec. This is for code viewing and suggestions only, there is a problem that clips never end properly and it loops infinitely requesting and advancing the buffer but not ending the clip. (if anyone sees why this is, I'd love that)
16:04:25 Join Azraak [0] (
16:04:52goffaBg3r: yeah... someone needs to sterilize the people at the riaa... that way they can't reproduce
16:05:01goffacollective iq of the world will rise
16:05:51goffaamazing how greedy they can be
16:06:31lostlogicgreed I can deal with... stupid, shortsighted greed, not so much.
16:07:25AzraakHi all. Anyone got any idea at all why my wiki account isn't working? I was able to attach one file with it, and then it just died. Change password doesn't work, reset password says it doesn't exist, try to reregister unhelpfully reports that my account already exists
16:08:49goffai believe artists should be paid... corporate america, not so much
16:09:27 Quit Kohlrabi ("Fast alle Menschen sind Regenw├╝rmer")
16:09:35HClartists only get 9% (!) of your money when you buy a cd
16:09:40HClit used to be even worse, 4% seems to be pretty fair.. and i'm not sure, but i think the money bypasses the riaa
16:10:09goffa20 of all sales goes to the artist according to them
16:10:15goffa$1/cd trade
16:10:41HClno idea, but 9% is normal nowadays, actually, 9 % is "high"
16:10:53ZagorAzraak: what is your account name?
16:10:56peturgoffa: these guys are really stupid
16:11:05goffahow so petur?
16:11:06AzraakZagor: JamesTeh
16:11:08peturgoffa: I get invited, want to fill out the form and it turns out it's U.S. only
16:11:20goffai have to agree with you then
16:11:26peturand it's nowhere on their site
16:11:52t0masevery body in the Netherlands should check out as I have my first driving lesson this afternoon ;)
16:12:33*petur orders concrete bunker
16:13:06t0masghehe, be prepared ;)
16:13:11*linuxstb was planning a trip to .nl in July but reconsiders...
16:14:21goffaThe excuse i hate is that they need all of this money for promotion, advertising, etc
16:14:29ZagorAzraak: It looks like your account got included in a spammer clear-out. I'll set you up a new password.
16:14:37goffaand all they do is promote/advertise the artists i don't want to support
16:14:52AzraakZagor: eek. Did I do something to cause that or was it just my bad luck?
16:15:29Azraakgoffa: And spend the rest of it suing people for various things :)
16:15:30ZagorAzraak: just bad luck. you registered just before a batch of spam accounts were registered.
16:15:38AzraakZagor: ah. gotcha.
16:15:40goffathat to Azraak
16:16:00goffatoo even
16:16:01Jungti1234anyone is developing video playback?
16:16:19goffaJungti1234: not at the moment, i don't believe
16:17:14Jungti1234Is here a person who live in Sweden?
16:17:18ZagorJungti1234: yes
16:18:01Jungti1234Do you know other Korean except me?
16:18:31Zagorno. I can't say I know you very well either :-)
16:18:34 Join Rusty|afk [0] (
16:19:01Rusty|afki've a question ;)
16:19:37Rusty|afkwhere do i habe to put my music in, so thar i can listen to it with rockbox? it's a very simple question, but please aswer ^^
16:19:48linuxstbIn any directory you want to.
16:19:59t0maslinuxstb: A trip to the Netherlands would be a 1 hour flight for you right?
16:20:05linuxstbt0mas: Yes, about that.
16:20:09t0masjust fly to Rotterdam, not so much security...
16:20:19t0masand take a train to Amsterdam of Utrecht, depending on where you want to go...
16:20:37Jungti1234Zagor: Any Korean says, 'Rockbox's Sweden developer says that someone is developing video playback'.
16:20:40linuxstbt0mas: Too late, flights to Amsterdam are bought.
16:20:43Rusty|afkand how can i use the file browser? i can't find it...
16:20:58t0maslinuxstb: and what are you planning to visit?
16:21:01Rusty|afki just can find the settings
16:21:02linuxstbt0mas: But how long is the train journey from Rotterdam to Amsterdam?
16:21:21t0masghehe I don't exactly know... but < 30 minutes
16:21:26ZagorJungti1234: well both me, Bagder and Linus are swedes. and as far as I know none of us are working on video playback right now.
16:21:34 Quit ashridah ("Leaving")
16:21:49t0mas(I live somewhere in the middle between the citys... so everything is close to my house :))
16:22:16t0masoh woops... according to my map it's 58 minutes from Amsterdam to Rotterdam
16:22:36Rusty|afkin the menue rockbox is disabled on my ipod =(
16:22:37ZagorJungti1234: on the other hand, the quote only says one of us _says_ someone is working on it. and that may very well be true.
16:22:41 Join Gargamale [0] (
16:22:49GargamaleAnyone alive?
16:23:06linuxstbI would hope a few billion people are, yes.
16:23:10Jungti1234Zagor: Perhaps, someone isn't Sweden developer.
16:23:11 Quit _Lucretia_ (Read error: 110 (Connection timed out))
16:23:35Rusty|afkwhy is rockbox disabled on my ipod...i can't choose "rockbox" in the menue
16:23:40GargamaleCan someone tell me if this is normal behavior. If I turn Crossfade on and play a song, then scroll through artists, my 5G Video iPod crashes
16:23:48Jungti1234right, only says one of us
16:23:52Rusty|afkthere is an "X" before that
16:24:19GargamaleCrash, meaning screen locks up, accepts no input, have to do menu and ok reset
16:24:20linuxstbThat's not the menu, that's the file browser. The rockbox file is the "rockbox.ipod" file you installed.
16:24:48linuxstbIf you create a new folder (e.g. "Music") and copy some music into it, you will see it there.
16:24:50Jungti1234And battery indication(?) has problem.
16:25:24Rusty|afki'm so stupid *lol*
16:25:37 Quit ProgramZeta ("Trillian (")
16:25:41 Quit Rusty|afk ("( :: NoNameScript 4.02 :: )")
16:26:05GargamaleNo one knows?
16:26:10GargamaleOr experiences that?
16:26:46linuxstbGargamale: It's obviously not intentional behaviour. Are you using tagcache or the file browser?
16:26:49ZagorGargamale: crashes are not normal, but bugs exist
16:27:13Jungti1234Status of battery changes continuously.
16:27:53GargamaleThis happens using both
16:28:04GargamaleShow ALl Files or ID3 Database
16:28:16Jungti1234Did no one find it?
16:28:28GargamaleAnd, if while something is playing, if I scroll too fast, I experience a crash
16:28:39GargamaleW/O Crossfade
16:28:41 Part LinusN
16:29:15GargamaleI know scrolling is coded very poorly from what you guys have said before, but that seems like a major issue if I cant scroll w/o crashing
16:29:19Jungti123476% -> 58% -> 66% -> 82% -> 69%
16:29:22 Join bluebrother [0] (
16:29:34Jungti1234It shows strange action.
16:30:12ZagorGargamale: it's proabably not the scrolling itself that crashes
16:30:58Jungti1234And korean people want:
16:33:12Jungti1234It makes according to all users want booting screen, USB connection screen, charge screen etc. and make so that may can change.
16:33:43Jungti1234Translation may be queerish.
16:34:20linuxstbGargamale: Have you tried different versions of Rockbox? Are you currently using the very latest CVS build?
16:36:40 Nick Inc is now known as INC1 (i=xsst4@arcane/developer/inc)
16:38:36Jungti1234-> Want that korean people change freely 'Booting screen, USB connection screen, charge screen'.
16:39:59Jungti1234Can't someone reply?
16:40:07linuxstbDo you mean to change the icons?
16:40:19 Nick INC1 is now known as Inc (
16:40:31Jungti1234Like WPS.
16:40:59Jungti1234They want that change it freely like WPS.
16:41:36goffaprobably not going to happen soon Jungti1234
16:42:04Jungti1234I don't want it right now.
16:42:28goffalike on iaudio x5.. one of the charging screens isn't affected by rockbox
16:42:40 Nick Inc is now known as INC1 (
16:42:44goffaalso one of the usbe connection screens isn't
16:42:51peturJungti1234: put it on the tracker as a feature request
16:42:52goffalike if the player is off when you connect it
16:42:57goffaso that won't change
16:43:49Jungti1234where do you live?
16:44:06goffamontana (usa)
16:44:22Jungti1234It's thing which Korean X5 user wants.
16:44:40goffai don't think its country specific
16:45:12goffai'm sure some non-koreans probably want that too
16:45:37goffai'm just saying.. that on the x5, when the device is turned off, those screens won't change
16:45:55goffawhen its turned on, maybe
16:46:00Jungti1234ah, :)
16:46:18 Nick INC1 is now known as Inc (
16:46:45peturthis is probably bootloader specific (like plugging the ac adpter on the h300), and will be changed once the bootloader is 100% ready
16:46:56goffapetur: exactly
16:47:49 Quit RedBreva ("CGI:IRC")
16:48:14 Join hardeep [0] (i=hardeeps@SDF.LONESTAR.ORG)
16:48:27Jungti1234petur: I can't do it now. I'm busy to prepare test(exam).
16:48:32Jungti1234bye all
16:48:38linuxstbpetur: The X5 might be different. I think it's like the ipod - there's a bootloader in flash that starts the Rockbox bootloader.
16:49:02linuxstbSo if it's the manufacturer's bootloader that detects usb/charging, then there isn't anything we can do.
16:49:03 Quit Jungti1234 ()
16:52:21*petur wonders how LinusN will manage to finish the iriver bootloader (s) and look after the power drain with only so few days left before 'mayday'
16:53:48linuxstbMaybe Day?
16:54:42lostlogicdid I mention that I'm going to be out of my home country 28th-1st? Maybe day, IMO :)
16:55:06goffawhere you headed lostlogic?
16:55:34linuxstbAre you going somewhere to concentrate on the Rockbox release without real-work getting in the way?
16:57:42 Quit JdGordon (Read error: 104 (Connection reset by peer))
16:57:48goffalinuxstb: what are you talking about... rockbox IS the real work :)
16:58:04goffawho needs a job... oh wait
16:58:51 Join dpassen2 [0] (
16:58:57linuxstbTrue, Rockbox is the real work. I should have said paid-work.
16:59:22lostlogichaha, going to Paris
16:59:37lostlogicfor a zero-coding weekend of skating and hanging out with a friend :)
16:59:41lostlogicZERO CODING
17:00:12goffabut what are us cvs addicts going to do?
17:00:19goffawe'll be wiggin out
17:00:23goffahaving the shakes, etc
17:02:20 Quit PaulJ_ (".")
17:03:24 Join Philip_0729 [0] (
17:04:02goffaactually you probably need a weekend off
17:05:25 Quit petur ("CGI:IRC (Ping timeout)")
17:05:33goffahmm... wonder if i could southern engineer this for my laptop and iaudio
17:06:03hardeepstrange, anyone else seeing a problem where pressing stop in WPS doesn't do anything?
17:06:58amiconnhardeep: H300, or other colour LCD unit?
17:07:30amiconnMaybe you have the "1st keypress enables backlight only" option enabled
17:08:24hardeepamiconn: yeah, h300
17:09:54 Join new5guser [0] (
17:09:55hardeepah, yes, it was enabled
17:10:01hardeepwill see if it reproduces now
17:10:43 Join ProgramZeta [0] (
17:12:19 Join _Lucretia_ [0] (
17:13:00Azraaklostlogic: I was the guy asking about whether or not I should file known voice bugs. I assume you'd prefer me to add them to the SoftwareCodecPlayback page and not flyspray?
17:13:10 Join Seed [0] (n=ben@
17:15:42preglowravon: small wonder it's slow, then
17:15:53preglowi thought hcl said the floats had been removed
17:17:02 Quit TeaSea ("Leaving")
17:17:34lostlogicAzraak: for now −− but really hold off until I commit the first round of voice reworking, because anything it does currently will be changed pretty much when I do that
17:17:56Azraaklostlogic: Ok. Shall do.
17:18:04lostlogicAzraak: I have 1 or two pesky issues to work out that make it too broken to commit so far, but I hope to nail those tonight.
17:18:07 Join }Y{maaalin [0] (
17:18:36Azraakheh, I get the impression the voice UI has caused peskiness in general :)
17:19:05 Join macdonalder [0] (
17:19:21Azraakhrm, I thought the "crash when playing voice clip with charger connected" bug was fixed, but it seems ot have reappeared.
17:19:33Azraakeither that, or I just haven't tested it in a long time and thought I did
17:19:43lostlogicbeats me, can't imagine what charger has to do with voice.
17:19:56Azraaknor can I; it's really strange. plug charger and hit menu and rockbox takes a dive
17:20:14Azraakso basically, you can't use the voice UI except on battery
17:20:42linuxstbWhich player do you have?
17:20:46AzraakIriver H300
17:21:17 Part macdonalder ("bye bye ;D")
17:21:35 Join macdonalder [0] (
17:24:14***Saving seen data "./dancer.seen"
17:26:26 Quit Seedy (Read error: 110 (Connection timed out))
17:27:18 Join petur [0] (
17:29:17Gargamalelinuxstb: Using latest build
17:31:21 Quit macdonalder ("bye bye ;D")
17:32:49 Quit petur ("CGI:IRC (Ping timeout)")
17:33:00Gargamaleipod 5g
17:33:27linuxstbGargamale: Have you had the same problem with older builds?
17:33:53GargamaleBut I only started using rockbox about a week and as half ago
17:34:38linuxstbCould you try some older builds? You can download them from here:
17:34:39 Join TeaSea [0] (
17:35:16linuxstbSomething else to test it to reset your settings - turn the hold switch on as Rockbox is booting.
17:36:20 Quit novimon (Remote closed the connection)
17:36:59GargamaleIll get back with you
17:38:34 Quit gantrixx ("Ex-Chat")
17:39:26 Quit Seed (Read error: 110 (Connection timed out))
17:40:05 Quit TeaSea (Remote closed the connection)
17:44:38 Join qwm_ [0] (
17:46:43 Join mikolas [0] (
17:49:57 Join TeaSea [0] (
17:50:20amiconnSeems we would gain 11KB on recorder v1 with sh-elf-gcc 3.4.6 and -Os
17:50:37amiconnSome gcc bug hunting is due...
17:50:43 Quit TeaSea (Client Quit)
17:51:19 Join TeaSea [0] (
17:52:18 Join RotAtoR [0] (n=e@rockbox/developer/rotator)
17:57:57mikolaskkurbjun, it seems that coldfire has pipelined architechture
17:58:27mikolasso you can get a performance boost by carefully rearranging the execution order
17:58:33kkurbjunso I should try and do other operations after move's to memory
17:58:59kkurbjunyeah, I was going to try and do that once I had the code working well
17:59:10mikolasand also try not to access same registers in a row
17:59:16mikolasinstead rearrange the code
17:59:41mikolasdid you get the code working yet?
18:00:23kkurbjunthe other function in doom that I was going to look into writing in asm was the drawcolumn function as it's responsible for all the walls.. not yet, there's some bug that I'm not sure what's caused by it yet
18:00:40kkurbjunis accessing the same register on 1 cycle operations bad?
18:00:52 Quit qwm (Read error: 110 (Connection timed out))
18:01:01mikolaswell, you might have an arithmetic operation pending for the register
18:02:00kkurbjunI was starting with the span code as I'm not very familiar with 68k asm, I've done some mips and the 68hc12 microcontrollers, but that's it
18:02:19kkurbjunok, I'll look out for that
18:02:22mikolasok, i can take a look at the code once you get it working
18:02:31 Quit ravon (Remote closed the connection)
18:02:34mikolasi don't have h3x0 so i cannot test it
18:02:34kkurbjunthat'd be great
18:02:42 Join ravon [0] (
18:03:13kkurbjundo you have an h120?
18:03:20mikolasactually i do
18:03:29mikolasit's my spare player
18:03:34mikolasand iaudio x5 also
18:03:49kkurbjunyou can test on that, the span drawing code is still used on that player as well
18:03:50mikolasbut i'm selling x5 atm and getting 5g ipod instead
18:03:54 Join darkless [0] (n=darkless@
18:03:59kkurbjundoom doesn't run on the x5 yet though
18:04:07 Quit }Y{maaalin ("( :: NoNameScript 4.01 :: )")
18:04:32mikolasif nobody is working on arm optimization of doom i can start looking at it
18:04:41mikolasonce i get my ipod
18:05:10kkurbjunyou should be able to see if the floors and ceilings are at least displaying correct.. yeah, the first things to look at/easiest would be the fixed point divide and multiply
18:05:22ravonDoom does run on the X5
18:05:36mikolasis it working already? did not work last time i checked
18:05:39kkurbjunreally? I wasn't aware
18:05:57ravonWorks like a charm. Slightly choppy sound every now and then, but it works :)
18:06:06kkurbjunI knew it ran fine in the sims, people kept telling me it didn't
18:06:10kkurbjunthat's great to hear
18:06:10ravonx5 as in iAudio X5, right?
18:06:21ravonYeah, then it works :)
18:06:32mikolasok, time to start vmware then :)
18:06:38kkurbjun: )
18:11:48 Quit new5guser ("CGI:IRC 0.5.7 (2005/06/19)")
18:12:13ravonsmeg. I wish libdumb was better documented regarding the internal design :/
18:13:02qwm_is there any work on support?
18:13:02kkurbjunmikolas, I think I have the span drawing working now
18:13:24kkurbjundoesn't seem much faster though
18:13:56kkurbjunI think the column drawing will have more room for improvement, but I havn't looked at it yet
18:14:49 Join novimon [0] (
18:15:33 Quit ProgramZeta (Read error: 110 (Connection timed out))
18:16:42mikolasdoom on x5 doesn't work
18:16:48TeaSeaWe know
18:16:51mikolashangs on "Starting graphics engine"
18:16:58TeaSeaIt's said it doesn't work.
18:17:09mikolaswell i was commenting to ravon :)
18:17:18TeaSeaAhh okay :D
18:17:21TeaSeaSorry :(
18:17:24ravonIt does. I played it yesterday :)
18:17:37mikolasdoes not with latest cvs :)
18:17:46ravonSometimes it hangs at "Initializing graphics engine..."
18:18:16ravonmikolas: It does. I can try it up to the latest commit.
18:19:47mikolashangs every time for me
18:20:01 Quit Hotfusion (Read error: 110 (Connection timed out))
18:21:52goffagrr... my x5 battery doesn't seem to be holding a charge
18:21:58goffahad it plugged in all night last night
18:22:15goffaonly been using it for 4 hours
18:22:49goffahoping a cable was loose or something
18:22:51GargamaleHat eit when shit doesnt work
18:22:56goffame too
18:24:16kkurbjunmikolas, could you look at this code: and see if you see anything that could be improved on?
18:24:48ravonmikolas: Sorry about the horrible picture quality:
18:25:39kkurbjunravon, are you running with any patches on your player?
18:25:49ravonkkurbjun: SID and dumb-patches
18:26:12mikolasis dumb working?
18:26:34ravonNot really. It's rendering the sound too slow.
18:26:53mikolaskkurbjun, just minor changes
18:27:17mikolasand i'm not sure if clr.w ; swap is faster than doind lsr #8 2 times
18:28:17amiconnkkurbjun: A minor one: you can move the clr.l %%d4 in front of the loop, as you only ever change the lowbyte within the loop
18:28:27mikolasamiconn, see my diff :)
18:28:44kkurbjunthey're both 1 cycle operations
18:29:09amiconnkkurbjun: In front, meaning outside, meaning 1 cycle saved per iteration
18:29:13kkurbjunamiconn, would it be more benefitial to setup the registers myself instead of having gcc do it?
18:29:34kkurbjunyep, I saw mikolas' change
18:29:52mikolaskkurbjun, more pipelining
18:30:29amiconnmikolas: I strongly doubt that moving insns around helps anything wrt the pipeline, except with memory stores (and *perhaps* memory loads, need to check that)
18:30:29 Quit dpassen2 ("CGI:IRC (EOF)")
18:30:34mikolasactually it will perform worse
18:30:43mikolasi just read through the changes :)
18:31:12mikolasamiconn, if the pipeline works the same way it did with 680x0 it should help
18:31:45mikolasbecause the core is superscalar it has multiple alus and moving register accesses around did indeed help back in the days
18:31:57amiconnColdfire isn't superscalar afaik
18:32:15amiconnColdfire is m68k based, but much simplified
18:32:16mikolasi was all the time thinking it acts like 060
18:32:25amiconnNope, very different
18:32:41 Join qwm [0] (
18:33:09amiconnThere's a 3x32 bit instruction fifo - that's about the pipeline
18:34:21amiconnIt helps interspersing non-mem-writing instructions with mem-writing ones, but even that only if the accessed memory has waitstates
18:34:56amiconn(I verified this verified by experiment)
18:35:37kkurbjunamiconn, so that means that since all the vars are in iram it won't help out by shifting the move's around?
18:36:29kkurbjunwell, the span drawing code does help out some in the timedemo..
18:36:40kkurbjunonly a little bit though
18:39:33mikolasit's version 2 core
18:39:52mikolasso it is not superscalar at all
18:42:16 Join petur [0] (n=petur@rockbox/developer/petur)
18:46:56 Quit qwm_ (Read error: 110 (Connection timed out))
18:47:06 Quit pixelma (Read error: 110 (Connection timed out))
18:50:20 Join PaulJ [0] (
18:51:38 Join Farpenoodle [0] (
18:53:28GargamaleGRR, I love Rockbox, but it needs to stop being gay
18:54:02warthawgbeing gay doesn't mean you're not a nice mp3 player
18:54:22GargamaleYa true
18:54:30GargamaleJust an annoying one
18:54:46warthawgi think that was a line from a movie, a lont time ago, "no way to treat a lady"
18:54:47GargamaleThe scrolling while playing freeze bug is annoying
18:54:52GargamaleI even loaded an old ass binary
18:55:09GargamaleYou have similar issues?
18:55:20GargamaleThats good to know
18:55:26GargamaleAt least Im not alone in my misery
18:59:49 Quit linuxstb (Read error: 110 (Connection timed out))
18:59:58 Quit hardeep ("BitchX: need we say more?")
19:00:07 Join linuxstb [0] (
19:04:07sharpei'm pissed.
19:04:17qwmwhy is that.
19:04:34kkurbjunmikolas, or amiconn: what is the proper way to address a constant address? this is what I was trying: "move.w #33,(0xf0000000) \n"
19:04:45kkurbjunwhere 0xf... is the address
19:05:02qwmsharpe: tooth cavity?
19:05:12qwmah.. :(
19:05:38sharpei figure later i'll drown it in listerine. to be a bitch.
19:06:42amiconnkkurbjun: Without the parentheses, but afaik this won't work, since you're trying to use 2 literal constants in the same instruction
19:06:54qwmhm, my last visit to the dentist gave me dental phobia.. the drill hurt so fucking badly.
19:06:54*amiconn points towards CFPRM.pdf again
19:07:03qwmdespite two shots. :(
19:07:33sharpethey kept poking at it, so like, "yes i fucking know it's there!"
19:07:44sharpeis what i wanted to say, instead, "ow."
19:08:12qwmhaha. :\
19:08:25qwmdamn sadists.
19:08:57kkurbjunamiconn, ahh, ok, I'm loading that value to an address register from the start now. Thanks
19:17:32 Join webguest47 [0] (
19:17:56 Join JBGood25 [0] (
19:17:58 Part Philip_0729
19:22:22 Quit _FireFly_ ("Leaving")
19:23:55 Quit webguest47 ("CGI:IRC")
19:24:17***Saving seen data "./dancer.seen"
19:26:36 Join obo [0] (
19:34:19 Join pixelma [0] (n=M_@
19:41:41 Join optimistic`t0ken [0] (
19:41:50optimistic`t0kenhey where can i get pacman!
19:41:54 Join cismo_ [0] (
19:42:38 Part Azraak
19:43:13sharpethe roms you mean?
19:43:37optimistic`t0kenyeah , i just installed rockbox and i need the roms
19:43:51optimistic`t0kendo i just download any pacman rom and put it in the folder
19:43:59sharpehas to be the mame pacman roms
19:44:19sharpeshould be six roms with sequencial extensions
19:44:37sharpei'm sure you can use google to find the mame roms.
19:45:08 Join larcher [0] (
19:45:10optimistic`t0kenk thanks
19:45:41sharpewelcome, if you need to know the exact filenames of the roms, they're on the "PluginPacbox" wiki page
19:46:01sharpeand those are the only ones you need in the folder.
19:48:15optimistic`t0kenoh ok
19:48:39sharpethe page also has the mp5 checksums to be sure they're correct also :)
19:48:47optimistic`t0kenso wait
19:48:51optimistic`t0kenthe emulator is already there
19:48:54optimistic`t0keni just need the roms
19:50:52 Join JBGood [0] (
19:54:23optimistic`t0kendo u think that other roms would word
19:54:30optimistic`t0kenlike old school DK
19:54:52t0maspetur: you can cancel the order for a concrete bunker
19:55:22t0maseverybody has survived
19:55:56 Part t0mas
19:56:00 Join t0mas [0] (n=tomas@rockbox/developer/t0mas)
19:56:29 Quit cismo (Read error: 110 (Connection timed out))
19:56:35 Join jbauman [0] (
19:57:40 Quit quobl (Remote closed the connection)
19:58:22 Join quobl [0] (n=quobl@tor/session/external/x-7e1fcbeea1de66b8)
19:58:27mikolashas anyone considered porting uade to rockbox? :)
19:59:45 Join Seed [0] (
20:05:45goffahe he he
20:06:11mikolasi think that would be really cool
20:06:19mikolasbut porting won't be easy
20:06:26goffayeah... wouldn't affect me at all
20:06:31goffabut support is good
20:07:57 Join cismo [0] (
20:08:15 Quit JBGood25 (Read error: 110 (Connection timed out))
20:10:19 Quit JBGood (Read error: 110 (Connection timed out))
20:10:27 Join muesli__ [0] (n=muesli_t@
20:13:45 Join tvelocity [0] (
20:13:57 Join qwm_ [0] (
20:15:53 Join JBGood [0] (
20:17:05t0mashm... has this song propagated to the rest of the EU alredy?
20:17:05*t0mas is now playing: Ali B feat. Yes-R en Partysquad - Rampeneren
20:18:41 Quit warthawg (Read error: 110 (Connection timed out))
20:20:41goffaheh.. hasn't propagated to northeast montana :)
20:20:47 Join Hotfusion [0] (
20:20:56 Quit Hotfusion (Client Quit)
20:21:04goffabut then we have 1 radio station that plays both country AND western
20:21:19goffaoh and oldies for 1 hour per day
20:21:23goffabut i never listen to that
20:21:31goffaso i have to discover my own tunes
20:21:40 Join lost|X40 [0] (
20:22:06 Join JBGood25 [0] (
20:22:27 Quit cismo_ (Read error: 110 (Connection timed out))
20:22:40 Part larcher
20:23:17t0masgoffa: the first two are Dutch "rappers" (note the ")
20:23:42t0maspartysquad is a quite popular DJ group here
20:24:02 Quit lostnihilist (Success)
20:24:22 Quit qwm (Read error: 110 (Connection timed out))
20:24:33goffayeah.. havent heard of any of them
20:24:58 Join Vladoman [0] (
20:26:53goffabut i don't listen to rap much...
20:27:04goffaand well even less dutch rap :)
20:31:23 Quit Rick (Read error: 104 (Connection reset by peer))
20:31:34muesli__sounds like gauda ;)
20:31:34 Quit jbauman (Read error: 110 (Connection timed out))
20:31:45 Quit Seed (Read error: 110 (Connection timed out))
20:32:21 Join Rick [0] (
20:35:02*goffa shouldn't have put that big pile of sedatives into his lunch
20:35:11goffadamn i'm tired all of a sudden
20:36:46lostlogicmmm, food coma
20:37:11goffayeah... i'd never make it as a truck driver
20:37:15Mikachubeen away for 6 days, did i miss anything exciting?
20:37:48goffaproject dissolved... everyone lost interest :(
20:37:57Mikachuoh no
20:37:57 Quit SereR0KR ("XChat Aqua")
20:38:01goffahe he he
20:38:06 Quit JBGood (Read error: 110 (Connection timed out))
20:39:36sharpeMikachu: the c64 emulator caused all players to be bricked.
20:39:52Mikachuit's april, but not the 1st :)
20:40:03goffait is the 20th though
20:40:07 Join Hotfusion [0] (
20:40:16sharpeit's a multiple of one, so it still counts. :)
20:41:17lostlogicMikachu: welcome back
20:41:46lostlogicBrace yourself for more playback breakage this week when I commit voice refactorage.
20:41:54 Join adiamas [0] (n=adiamas@
20:42:12goffajust fix it before you go to paris
20:42:17adiamashowdy all...
20:42:23lostlogicahhahahahaha, I'll do my best :)
20:42:40 Quit damaki_ (Read error: 104 (Connection reset by peer))
20:42:44lostlogicotherwise I'll hafta install the vmware stuff on my work laptop and work on it a little bit on the plane.
20:42:52muesli__one night IN paris? :-)
20:42:57adiamasso yeah... i just moved to cvs-060411 and my volumes whacked out...
20:43:10adiamasi have to bring volume to -26 to be not deafening
20:43:17adiamasarchos recorder
20:43:19goffayeah... those air phone connections can get expensive lostlogic
20:43:35sharpewhee, i get to update the raster count 7800 times per second, along with 12500 emulated instructions per second.
20:43:48sharpefor accurate emulation
20:44:06 Part PaulJ
20:44:10 Join damaki_ [0] (
20:44:49 Join Kyomi [0] (
20:45:43amiconnlostlogic: I've experienced another playback bug
20:46:00sharpeor wait, only 12000 processor instructions
20:46:28KyomiSooo... whats happening with the experimental build? It's like....vanished
20:46:32sharpeso around, 19800 function calls per second...
20:46:40amiconnPlayback itself was fine, but when I skipped tracks, it happened at least once that the playback code updated the track info several times at the start of a track
20:47:14amiconnThis is not noticeable unless there's a long line which will scroll
20:47:19 Join hardeep [0] (i=hardeeps@otaku.freeshell.ORG)
20:47:49amiconnThen the scrolling line does get reset to the 'start position' at the time of the trackinfo update
20:49:39 Quit lost|X40 (Connection timed out)
20:50:50Mikachuwho wants to guess how many files will conflict when i cvs update?
20:52:37 Join Kohlrabi [0] (
20:53:06lostlogicamiconn: yeah, I've noticed that there seem to be a couple of extra updates happening, the signaling that the track information changed is less than ideal (it's just a boolean that's reset on read by the WPS thread) currently.
20:53:14 Join ep0ch [0] (n=ep0ch@
20:53:21lostlogicamiconn: definitely not a priority, at least until after voice works *mutter*
20:54:03goffadamn train delay... i want my e5c's
20:54:14lostlogicthe voice codec is doing: read 8192b, read 0b, advance 0b in an infinite loop after the first clip finishes, and I can't for the life of me figure out why. I wonder if it tries to seek or something, which is ignored by the voice callbacks.
20:55:15 Join lostnihilist [0] (
20:55:22 Quit Kyomi ()
20:55:54goffaheh.. i guess 2 patties of beef isn't enough.. lets add steak
20:56:43 Quit Rick (Read error: 104 (Connection reset by peer))
20:59:08 Join Rick [0] (
21:05:03sharpei seem to not be able to type correctly these past few days
21:05:13peturlast time I was in the states I got disgusted by the food... they were even eating steaks as breakfast
21:06:27peturand in the shops I could only find loads and loads of sweets and stuff.. oh yeah, and buckets with butter which I probably wouldn't get empty in 5 years
21:06:45lostlogicwe don't all eat that way.
21:07:07peturbut we had a hard time finding 'normal food'
21:07:27sharpei've never had steak for breakfast...
21:07:36hardeepwhat city were you in?
21:07:39petur(this was in the Seattle area)
21:08:13hardeepseattle's a relatively healthy city −− surprised you couldn't find anything
21:08:22HotfusionI usualy eat steak for dinner.
21:08:27HotfusionSteak fajitas mmmm
21:08:55sharpemy code is all neat and tidy.
21:08:57sharpeWrite(0xD011, (Read(0xD011) & 0x7F) | ((vic.raster & 0x100) >> 1);
21:09:49sharpethat seems slightly redundant...
21:10:16lostlogicpetur: I eat a lot of sushi and asian food
21:10:25lostlogicwait, sushi is asian food
21:10:26sharpeoh well, i'll keep it so i know what it should actually be doing.
21:12:19amiconnlostlogic: What should that voice codec sequence you posted tell me?
21:12:37amiconnDoesn't that just mean the voice codec tries to get some data to decode?
21:12:57lostlogicamiconn: Yeah, but then it never advances the buffer that distance, so the clip never ends, and it just loops forever on those three requests
21:13:09 Join filoktetes [0] (
21:13:22amiconnAh, so advancing is a separate operation?
21:13:37*amiconn would have expected that a read implies advancing
21:13:58lostlogicamiconn: no, it actually uses the request_buffer function, so it's requesting a pointer to memory ont he file buffer
21:14:09lostlogicso it doesn't want to invalidate that buffer by advancing the pointer until it's done with it
21:15:28amiconnHmm, in fact I made a similar mistake while coding the (dead-simple compared to playback.c) playback engine for the wavplay plugin
21:16:02amiconnAdvancing a pointer too early isn't going to be nice when the refill task catches up...
21:16:42lostlogicof course for voice there isn't really a refill task, because that data is on a buffer that stays in memory
21:17:14 Join damaki [0] (
21:17:36lostlogicI'm going to hafta log the other voice codec functions to try and understand why libmad/mpa aren't advancing the buffer.
21:18:07amiconnYes. For voice, the flow of things changes a bit. Imho the voice thread should stop completely when the voice queue is empty
21:18:33amiconnPerhaps it retries because of an error?
21:18:57amiconnAlthough, the voice clips are plain mp3 frames, without any extra headers
21:19:16 Quit damaki_ (Read error: 110 (Connection timed out))
21:20:13lostlogicamiconn: with what I'm working on, when the voice queue goes empty and playback is stopped, the voice thread will go to sleep (on an indefinite queue wait)
21:20:41lostlogicamiconn: I think the problem is that mad is asking for more data than what is available and expecting a retry to give it more, but there isdn't any more or some such
21:21:01amiconnOk, that's what I thought it should be. So cvs does different?
21:21:14 Quit ep0ch (Read error: 110 (Connection timed out))
21:21:38lostlogicCVS busy loops with a sleep(hz/16) waiting for a clip to play
21:22:04peturamiconn: iriver does a special write sequence to the isp1362 to put it in suspend (lowpower) mode - this could be it?
21:22:21amiconnpetur: perhaps, that needs to be tested
21:22:45amiconnlostlogic: But why should 8K not be enough?
21:22:53amiconnAn mp3 frame is never that large
21:23:11lostlogicI think it requests 8k but not that much is left of the clip or something... need to put more logging there.
21:23:25peturamiconn: I'm prepping code and a bootloader for Linus to test
21:23:40amiconnPerhaps you should ask preglow as he's the codec expert...
21:24:00amiconnpetur: Why not test with a rockbox binary first?
21:24:07sharpewow, i should count how many memory accesses are made for each frame when i finish all the code...
21:24:19***Saving seen data "./dancer.seen"
21:24:22lostlogicpreglow: linuxstb: any thoughts what would make mad/mpa perform that sequence of request 8192b, request 0b, advance 0b indefinitely?
21:25:37peturamiconn: I seem to be unable to access the registers, and I saw that iriver does some configurations (including configuring the chip to wake up on CS). I'm guessing that my code runs too late
21:26:57amiconnI don't think that's the case
21:27:24amiconnThe isp should be accessible at all times, provided the various pins are set up correctly
21:27:49amiconnIirc the isp is connected to cs3, i.e. to the bus. Did you check the waitstates?
21:28:07peturI've set it the same as iriver
21:28:21peturincluding the silly nops between reads and writes
21:28:52peturI'll put my efforts in the wiki for review later on
21:30:05lostlogiccodec asks for 8192b, gets 48b, never advances the buffer over those 48b
21:30:18peturyou can wake up / sleep the chip by hardware pins or software. I haven't found any GPIO action related to isp1362, but I did find the wake-up-on-CS config, so that's my lead to follow now
21:31:34preglowlostlogic: hmm, no
21:31:51preglowwhy does it get so little?
21:31:52 Join qwm [0] (
21:34:37petursilly compiler
21:34:41peturand.l #0xFFDF, d6
21:34:50peturand.l #0xFFFF, d6
21:34:57 Join ep0ch [0] (n=ep0ch@
21:35:26*amiconn sighs
21:35:52amiconnI can't split myself...
21:36:16*amiconn would like to do at least 5 different things at once
21:37:56lostlogicpreglow: because that's all that's left in this 'getmore' from the talk.c side
21:40:24lostlogicwhat's weird is that a single 'getmore' from the talk.c should be a complete clip, and so should end on an mp3 frame boundary, so those 48 bytes should complete a frame for mad and it should advance past them and then ask for more again which would result in a new getmore
21:43:43kkurbjunmikolas, amiconn, lostlogic, preglow: thanks for the help on the coldfire asm. Finally committed some things that actually work
21:44:18 Quit qwm_ (Read error: 110 (Connection timed out))
21:44:59lostlogickkurbjun: great :) how's the speedup?
21:45:52kkurbjuneh, it's a littlebit, the spandrawing didn't have as much effect as I was hoping, the asm H300 lcd update I made did have an impact though
21:45:57mikolaskkurbjun, i just came up with something that you should look at
21:46:23mikolasas i recall, writing longs instead of bytes was faster back in the 68k days
21:46:34mikolaslong aligned of course
21:46:55mikolasso maybe you should collect full longs in the loop and write those to memory one at the time
21:47:07lostlogicdefinitely, using movem instead of move will be way faster
21:47:18lostlogicread 16 bytes into 4 registers
21:47:20lostlogicwrite all
21:47:46mikolassame stuff i used to optimize 68020 vector routines back in the days of _slow_ chip ram on amiga
21:47:56lostlogicsee amiconn's memmov and memcpy routines for the quickest alignment techniques
21:48:04kkurbjunhmm, so I'd have to unroll it and then have another loop for checking the leftovers
21:48:40 Quit quobl (Client Quit)
21:48:50mikolasstart by collecting one long at the time and elaborate on that
21:48:59mikolasjust to see if there's any initial speedup
21:49:45lostlogicthat's true, it should be a pretty big speedup just to get it to longs vs. bytes
21:49:58mikolascan you put the code online
21:50:02mikolasi can take a look at it
21:50:15kkurbjunmikolas, yes, it's in cvs now
21:50:22kkurbjununder r_draw.c
21:50:49kkurbjunI can't look into it right now though
21:51:14mikolasi'll check it anyway, i can give you the patch later
21:51:52 Join amiconn_ [0] (n=jens@rockbox/developer/amiconn)
21:52:07 Quit amiconn (Nick collision from services.)
21:52:07 Nick amiconn_ is now known as amiconn (n=jens@rockbox/developer/amiconn)
21:52:16kkurbjunmikolas, great, thanks for the help
21:53:18mikolasyou'll have to split the loop in three parts in order to align pointers correctly, but i think it's pretty easy
22:07:24 Quit lostnihilist (Read error: 110 (Connection timed out))
22:07:31 Join lowlight [0] (
22:08:24lowlightlostlogic: are you aware of this bug/feature of libmad?
22:08:37lowlightit won't decode the last frame
22:09:33 Quit midkay_ (Read error: 104 (Connection reset by peer))
22:09:49 Join midkay_ [0] (n=midkay@
22:11:06lostlogiclowlight: hmmmmmmmmmmmmmmmmmmmmmm
22:11:35 Join SereR0KR [0] (
22:11:45sharpeit never crossed my mind to update the single video chip registers for that register, only when it's written to.
22:12:02sharpeer, update that certain register when it's written to.
22:12:09sharpeand not all of them at the same time.
22:12:44sharpei should rewrite this code more often. :D
22:13:57lostlogichmm... so I somehow need to track how much mad has gotten from the voice_request_buffer, and when it has _received_ all bytes, offer it 8 more bytes of zeros
22:14:25 Join Seed [0] (
22:17:59 Join matsl [0] (
22:19:55sharpelostlogic, do you know ahead the number of bytes?
22:20:16lostlogicwhich number of bytes?
22:20:41sharpenumber of bytes from voice_request_buffer...
22:20:43sharpei think.
22:20:49sharpeto recieve.
22:21:52*amiconn wonders about that weird feature of libmad
22:22:00sharpei just read that, i dunno if you've done it already.
22:22:33lostlogicvoice request buffer receive what who which?
22:22:47sharpei don't know? :(
22:22:47amiconnlostlogic: You only need to pad if the voice queue is completely empty. Otherwise just advance to the next clip
22:24:22amiconnI'm not sure whether it's possible to continuously feed the codec from non-contiguous data
22:25:49amiconnBagder: The frontpage is in need of a model list update... (mini 1g)
22:25:55lostlogicamiconn: yeah, that's the problem, I can't just call the callback, because I haven't advanced the buffer past those original 48 bytes yet, so I think it's safer to feed it just 8
22:26:47sharpeshould i do clever positioning of the switch cases that handle memory writes, so i only have to have the line "ram[Addr] = Value" once, maybe twice, instead of for each special case?
22:28:40mikolasaddx.l #0, %%d1 does not work on coldfire?
22:28:52mikolasis it only register, register?
22:29:17sharpewait, i have a better idea... not that anyone was rushing to help... :)
22:29:24preglowlostlogic: you hit the buffer padding "feature" of libmad?
22:30:45lostlogicpreglow: I think so −− not sure why it refuses to consume the 48 bytes of real available data though :(
22:30:58mikolasaddx is a cool way to optimize fixed point math by having the decimal part in the upper word, but it seems you'll have to waste one register on coldfire
22:32:43preglowaddx? isn't that what you'd use to do a full 64 bit add?
22:33:06mikolasyou can only use it so that you have the 16 decimal bits in upper word
22:33:25mikolasso you do add.l %%d0, %%d1; addx.l #0, %%d1
22:33:34mikolaswithout any shifting or swapping
22:33:53mikolasgiven that the lower word won't be wrapping
22:34:06preglowwhy do you need an add with carry for that?
22:34:36 Join powr-toc [0] (
22:34:58mikolasfixed point math
22:35:24mikolasyou have 16h.16l where l is the decimal part
22:35:33preglowi know how to use fixed point math...
22:35:41preglowi'm just wondering what you're doing and why you need to use addx
22:35:45mikolasnormally you would do 16h.16l + 16h.16l
22:36:08mikolasand you will be doing a swap to get the 16h, right
22:36:38mikolasbut you can also do add.l 16l.16h, 16l.16h; addx.l #0, 16l.16h
22:36:44mikolasno swapping/shifting
22:37:21mikolasand in the best case you can interleave those to remove some unneccessary code from loops dealing with fixed point
22:37:56mikolasthat's how you do lightning fast line calculations for example
22:38:19 Join lee-qid_ [0] (
22:38:26mikolasyou need to use addx in order to add the overflow from the decimal part
22:39:32mikolasi think i will have to comment my code for doom optimization in order to understand it myself tomorrow...
22:40:41 Join mirak [0] (
22:40:42 Quit midkay_ (Read error: 104 (Connection reset by peer))
22:40:51 Join midk [0] (n=midkay@
22:42:01preglowi can't really claim i see the point in it at all
22:42:13preglowaddx with a constant is even slower than doing swapping/shifting
22:43:06mikolasbut you can interleave stuff
22:43:27mikolaslike in a loop you have two fp variable
22:43:53 Join BHSPitLappy [0] (i=Steve-O@
22:44:29mikolasyou set it up like x16l.y16h and y16l.x16h and actually interleave it so that you get result like add.l %%d0, %%d1; addx.l %%d2, %%d3 and so forth
22:44:38amiconnmikolas: You mean with that addx you can add two byte-swapped values to produce a byte-swapped result again?
22:44:54mikolasso you have the decimal part interleaved in the regs
22:45:16mikolasthat's how i did it in the amiga days to calculate stuff fast :)
22:45:17 Quit filoktetes ("Leaving")
22:45:31amiconnI wonder how that is faster than an ordinary add
22:46:02mikolasit's as fast as an ordinary add. but you can remove swapping
22:46:24amiconnI mean, why place two variable in one register that way?
22:46:37 Join midkay [0] (n=midkay@rockbox/developer/midkay)
22:46:53amiconnSure it's a neat trick by itself, but I can't think of an application where it actually pays off...
22:47:21mikolaswhen you calculate lines it pays off pretty well. or in the doom code.
22:47:24peturamiconn: I've attached my usbotg code on the wiki page if you care to do a 6th thing right now ;)
22:47:56 Quit midk (Read error: 104 (Connection reset by peer))
22:48:22mirakI was playing doom
22:48:58mirakwell just a bit. there is really no way the H300 pad can handle height directions instead of 4 ?
22:49:04mikolasin doom code you can eliminate 4 lsrs per loop iteration
22:49:38mikolasso it's about 30% of the code in the loop
22:50:00mirakdoom is barely playable, since you can turn and go forward at the same time
22:50:18mirakI was wodering if it was hardware limitation or drivers limitations
22:50:36 Quit linuxstb (Read error: 110 (Connection timed out))
22:50:40 Part optimistic`t0ken
22:50:49amiconnhardware limitation
22:51:01mirakdamn I am crying now :'(
22:51:33 Join linuxstb [0] (
22:51:55 Join quobl [0] (n=quobl@tor/session/external/x-810351f1ca7f634d)
22:51:59 Join lostnihilist [0] (
22:52:26amiconnAny devs knowing the wps parser code around?
22:52:34mirakprobably with forward on record button it could be more playable
22:54:29 Join eEye_ [0] (
22:56:05lostlogichmm... libmad/stream.c:84 seems to be the problem.
22:56:33lostlogicnext_frame is set to this frame, and then used for the next call to advance_buffer_loc() which of course advances the buffer by exactly zero bytes.
22:57:27hardeepcan you use the zero byte advance as a flag to go to next clip?
22:58:25eEye_anyone have an iaudio x5?
22:59:05amiconnCrappola. Who used a non-existing macro in debug_menu.c for checking for iriver arch?
22:59:13amiconn#ifdef IRIVER, haha
22:59:35sharpeso far my code totals up to 560 lines for the c64 emulator, minus the cpu emulator code...
22:59:53preglowamiconn: hahaha
22:59:56preglowhope it wasn't me
23:00:05*amiconn checks viewcvs
23:00:26*preglow does the blame dance
23:00:42sharpewow, deja vu...
23:00:49amiconnIt was Linus...
23:01:28 Join lost|X40 [0] (
23:01:35preglownow i've got some dirt on him!
23:02:40 Quit Hotfusion (Read error: 104 (Connection reset by peer))
23:02:44 Quit adiamas ("Chatzilla 0.9.72 [Firefox]")
23:02:53oboare there any strings that contain the model of the DAP?
23:02:54goffaeEye_: i have an x5
23:03:13eEye_how long have you had it for?
23:03:20eEye_i'm worried about the battery
23:03:29goffasince like october
23:03:31goffai have an x5l
23:03:40eEye_i have just an x5
23:03:47 Join qwm_ [0] (
23:04:00eEye_i've read on the forums that some peoples batteries last only 6 hours now
23:04:06goffawait... no... july
23:05:11goffayeah... that seems to be what mine is doing
23:05:19goffabut it seems to be only in the last week or so
23:05:21goffaso i don't know
23:05:35goffaand i'm not sure that i had it plugged in tightly last night
23:05:40goffai think i have to clean the contacts
23:07:42mikolasmy x5 only gets about 7 hours of battery
23:08:10goffabut even as advertised ... there are a lot of variables that factor in
23:08:30goffalike file types, if random is on, tag cache, dir cache, etc
23:08:51mikolasi get only 8 on the stock firmware
23:09:07mikolasso it is not rockbox specific
23:09:39goffaas a rule, the smaller the file type that you use, and no random seems to give best battery life
23:09:53goffabecause the disc access seems to suck the battery
23:10:06amiconnBg3r: around?
23:10:17goffaso the less disc access, the better from a battery standpoint on the x5
23:10:34goffabut no player seems to tell the truth on battery life
23:10:34mikolasyes, cos it only has 16 mb of memory
23:10:46 Quit lostnihilist (Read error: 110 (Connection timed out))
23:11:00mikolasso it accesses disc a lot if you have high bitrate files
23:11:07goffai would imagine apple is pretty close if you ONLY use itunes
23:11:22goffajust because they are highly compressed
23:11:33amiconngoffa: The effect of random is neglectible
23:11:40mikolasi was happy with h120
23:11:41goffanot on my x5
23:11:48mikolas15 hours of battery
23:11:51amiconn(with rockbox)
23:12:30goffaamiconn: it makes a difference on my x5
23:12:36amiconnAnyone with an H1x0 or H300 and H300 remote(s) (lcd and non-lcd) around?
23:12:49goffaunless 3-4 extra hours is negligible
23:12:59mikolasi have h120 + remote but they're in another location at the moment
23:13:04goffawell... i take that back
23:13:11goffano random, and ogg q6
23:13:20goffabecause with flac it becomes negligible again
23:13:59 Quit qwm (Read error: 110 (Connection timed out))
23:14:44amiconnmikolas: I'm looking for someone with a H1x0 or H300 and one of the H300 remotes. I have a H100 remote myself, and both a H1x0 and a H300.
23:14:57 Join webguest48 [0] (
23:15:05mikolasshould have read what you said :)
23:15:14eEye_well what i mean is the battery life degrades over time
23:15:16mikolasi have the original h100 remote
23:15:17axion_turn off voice too
23:15:21amiconnI want to improve the remote detect + init
23:15:24 Join ScootScat [0] (
23:15:34eEye_i just got my x5, so if in 6 months the battery life is going to degrade so much i might as well bring it back
23:15:50goffaeEye_: yeah... i haven't noticed unless it goes fine and there is a _sudden_ drop
23:16:06goffai was without power 2 nights ago
23:16:18goffaand it drained all the way.. and i think i had a loose connection last night
23:16:29goffaso i'm unsure
23:17:02axion_goffa: turning off that text to speech frees up a lot of cpu doesnt it?
23:17:44yobesoomwhoah.. axion!
23:17:46yobesoomyou too?
23:17:56yobesoomjeeze I'm meeting all my old IRC buddies here
23:18:27goffajust like cheers :)
23:18:39yobesoomwhere's the beer?
23:18:48axion_hi moose
23:19:07 Part lowlight
23:19:09goffapreglow: these are guys from the mpc channel from way back... when bushel was developing
23:19:37 Join dj-fu [0] (
23:19:50 Quit lost|X40 (Read error: 110 (Connection timed out))
23:19:50axion_whoa RotAtoR is here too
23:19:57goffabefore i saw him a couple of days ago, i don't think i've seen yobesoom online for like 3 years
23:20:15axion_hey RotAtoR, long time
23:20:18preglowso, what are you all doing here? :)
23:20:31goffawe just kind of bumped into each other
23:20:35goffafunny, huh?
23:20:45axion_i still talk to a lot of people
23:20:48RotAtoRmusepack supports unites the world :)
23:20:58goffahad to be it
23:20:59preglowso i unite the world, do i? :P
23:21:02 Join lostnihilist [0] (
23:21:04goffayep :)
23:21:06sharpeshould i define a long string over one line, or over multiple lines to fit the 80 character width?
23:21:24goffapreglow's fan club is growing
23:21:34preglowthis'll look nice on my resume
23:21:51preglow"Took a brief stint off for uniting the world. Succeeded in doing so."
23:21:52goffapreglow: i think moose encoded that stuff i sent you
23:21:53Mikachusharpe: why would you define a long string instead of making a variable that you reuse?
23:21:56goffawell yobesoom
23:22:07preglowgoffa: yeah, haven't had time to look more at it yet
23:22:18preglowi'm going to add seeking support before i do anything else
23:22:23goffayeah.. it plays. i'm happy already
23:22:26preglowfirst spdif recording, then seeking support
23:22:33goffasomeone had a patch on flyspray
23:22:37preglowi know
23:22:37goffadon't know if that was you
23:22:43preglowno, that was snowgoon
23:22:44sharpeMikachu: the long (in size, not the c keyword) string is a character alphabet for keyboard input.
23:22:46goffaoh ok
23:22:47preglowwho developed it
23:24:22***Saving seen data "./dancer.seen"
23:25:49 Join ashridah [0] (
23:27:38 Join Paul_The_Nerd [0] (
23:28:04 Quit matsl (Remote closed the connection)
23:28:25 Join taz5612 [0] (
23:28:57 Quit webguest48 ("CGI:IRC (EOF)")
23:29:48 Join DrMoos [0] (
23:29:49preglowamiconn: is there any reason not to use the or_l and and_l macros when flipping bits?
23:30:22amiconnFlipping bits where?
23:30:38 Join ProgramZeta [0] (
23:30:54amiconnFor gpio, these macros should always be used because they're atomic.
23:31:09preglowahh, only gpio
23:31:28preglowdoesn't gcc have enough brains to do an atomic operation anyway, btw?
23:31:33amiconn..unless you need to do a lot of flipping and want to really squeeze speed
23:31:36preglowif not, then using or_l and co. should always be faster and use less space, yes?
23:32:08amiconnIt's not necessarily faster than what gcc uses when you do |= or &=
23:32:42*yobesoom encoded a lot of sv4 and sv5 mpc back when it was still called mp+
23:32:45 Quit tvelocity ("Ex-Chat")
23:32:50amiconnor_l and and_l should be same speed, but interrupt safe because they're atomic
23:32:57preglowbut yeah
23:33:14preglowwill anyone cry if i don't support monitoring for spdif recording right away?
23:33:21amiconnAccessing gpio is really slow on coldfire
23:33:41amiconnand_l and or_l take 24 cycles when used for gpio
23:34:12 Quit Gargamale ("poop")
23:34:23yobesoomgood god
23:34:42amiconn...and it's not possible to access only parts of the port (.w or .b)
23:34:43yobesoomcan you just copy off the register from gpio to cpu and cmp?
23:34:59amiconnWhen I tried, I got lovely PDIR1FULL interrupts...
23:35:36amiconnyobesoom: and_l and or_l are not for reading, but for flipping individual port bits
23:35:45yobesoomohhh my bad
23:35:51 Quit eEye_ ("fractal2 mirc script (ver 1.0betar2) Ě")
23:36:08yobesoomI thought you were trying to check a bitmask
23:36:14amiconnReading isn't critical (apart from that just reading gpio also takes 14 cycles)
23:36:21taz5612Is there any planned support for the Ipod using firewire instead of USB? USB works just fine, however I've not found anything about getting it to work with firewire.
23:36:33lostlogicpreglow: you didn't happen to take a look at that libmad/stream.c line I pointed out, did you? Seems patently wrong to me
23:36:42preglowlostlogic: what line?
23:37:11preglowoh, it does
23:37:26preglowseems weird
23:37:27Paul_The_Nerdtaz5612: Technically USB isn't supported either. All it does is detect a USB insert and reboot into disk mode, which you can just do manually for firewire.
23:37:35preglowif you can't understand it, just mail rob leslie about it and ask
23:37:37lostlogicand that results in mpa.c line 195 advancing the buffer to back where it already was
23:37:38preglowhe seems quite helpful
23:38:39taz5612Paul_The_Nerd: I tried to force a reconnect while the firewire cable was connected and it would boot into Rockbox
23:38:46 Quit qwm_ (Connection timed out)
23:39:13 Quit lee-qid_ ("Trillian (")
23:39:40Paul_The_Nerdtaz5612: Did you boot into Disk mode, or Retail? Did you use Menu+Play after a Menu+Select to get into Disk Mode, or did you just plug in USB, then try to quickly switch or something odd like that?
23:39:42preglowanyone know of a reason why sample rate selection isn't supported for coldfire recording yet?
23:40:07preglowand doesn't the uda generate the sample clocl when recording?
23:40:20*preglow nudges petur
23:40:26taz5612Paul_The_Nerd: I used Menu+Select and Menu+Play to reboot to disk mode
23:40:37 Join TCK [0] (
23:40:39taz5612and that was before I switched to the USB cable
23:40:51Paul_The_Nerdtaz5612: Try simply booting retail OS then.
23:41:05 Quit ProgramZeta (Read error: 104 (Connection reset by peer))
23:41:12Paul_The_NerdThough I don't know what you mean by "force a reconnect" then.
23:41:18taz5612Paul_The_Nerd: Which is done with Menu+Select
23:41:58Paul_The_Nerdtaz5612: Which statement were you referring to with that? Menu+Select simply hard reboots an ipod...
23:42:15taz5612Paul_The_Nerd: Force a connect would be more accurate
23:43:01Paul_The_NerdAh, and see being accurate is important as I was imagining you try to pull out USB and plug in Firewire before it decided to reboot, because that's the only situation I could imagine a re-connect. :)
23:43:14sharpei'm starting to type hoystick instead of joystick... is that bad?
23:43:18taz5612:) many apologies
23:43:19Paul_The_NerdAnyway, so have you gone into Retail and tried it, or *just* Disk Mode?
23:43:36taz5612I haven't gotten into Retail mode.
23:43:43*taz5612 is looking up how now
23:43:47Paul_The_NerdI'd suggest that for now. There's no reason Retail shouldn't work
23:43:55Paul_The_NerdJust hold Menu immediately after turning on the iPod
23:44:08taz5612Paul_The_Nerd: ok. Thanks a lot.
23:46:25*petur returns
23:47:16 Quit Moos ("Glory to Rockbox !!!")
23:47:45twisted`sharpe: hoystick?
23:47:51twisted`sharpe: the hell is a hoystick
23:47:55preglowpetur: any idea?
23:48:31 Join San [0] (
23:48:39peturpreglow: I haven't looked at the configuring of the UDA, but I think it's just that nobody wrote code to do it...
23:49:01preglowi think i remember linus saying we can't just use any sample rate when recording either
23:49:14preglowbut i can't see why, it looks like the uda is set to drive the iis clock
23:49:24peturi think it it
23:49:35preglowi need to write code to allow for other sample rates now anyway
23:49:52 Quit San (Client Quit)
23:49:54 Join San [0] (
23:49:59 Quit San (Remote closed the connection)
23:50:10peturbut that has nothing to do with the UDA, no?
23:50:38 Nick taz5612 is now known as taz (
23:50:41 Join Zetan [0] (
23:50:49preglowbut i thought i might enable other sample rates for the uda as well, while i was at it
23:51:12petursure :D
23:51:51sharpehmm. i don't suppose that there is a format specifier for binary in rb->snprintf(), is there...
23:52:38lostlogicpreglow: I think the problem only happens with regard to our supposed handling of partially corrupted streams, there are cases where the next_frame pointer has not been advanced after the decode call, and the error condition is such that we advance to that location, (ie nowhere).
23:53:21lostlogicpreglow: by changing the conditional under the comment /* Fill the buffer */ to :stream.next_frame && stream.next_frame != stream.this_frame it works.
23:54:52lostlogicgah, nevermind, that breaks some normal files. makes they play all slow with artifacties.
23:55:10 Quit petur ("credits left: 0 - sleep a bit to get new credits")
23:55:10 Join Rondom [0] (n=Rondom@
23:55:44sharpeokay, i have all the code rewritten so far, with the exception of the emulation loop.
23:56:53 Join RedBreva_ [0] (
23:56:56lostlogichardeep: I think you were right, I hafta take a zero advance as a Q to end the current clip.

Previous day | Next day