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 2005-04-25

00:00:02preglowi can't leave stuff unfinished
00:00:03preglowit haunts me
00:01:35HClamiconn: take a break?
00:01:45 Quit noclue2 (Read error: 60 (Operation timed out))
00:02:23 Join noclue2 [0] (
00:05:52preglowbut no, the odds of me doing the gnuboy asm core are slim anyway, once i'm done with codecs, i'd probably like to move on to dsp stuff
00:05:56preglowlike eqs and stuff
00:06:02*HCl nods
00:06:36preglowmy dreams of transforming my h120 to a portable vocoder must also see fulfilment :P
00:08:22 Quit keith ("Leaving")
00:08:29preglownot really, i don't think it should be too hard
00:08:37HClwhats that?
00:08:56CoCoLUSit makes you speak like darth vader :P
00:08:59preglowthose boxes that make you sound like a friggin robot ;)
00:09:08HClah. right.
00:09:13HClto use on phones
00:09:16HClwhen blackmailing people
00:09:22preglownah, not really, just because it sounds cool, heh
00:09:22HClright preglow ? XD
00:09:31preglowit's better for scaring people
00:10:08 Quit _n2_ (Read error: 60 (Operation timed out))
00:15:15preglowrasher: awake?
00:16:55 Quit t0mas ("goodnight")
00:20:36HClpreglow: replied
00:22:08*HCl wonders why he typed just o.o.
00:23:12***Saving seen data "./dancer.seen"
00:24:46HClmaybe i should've signed my email with my name XD
00:24:49HClah well.
00:26:03preglowi think he gets it
00:26:39CoCoLUSi order my collection like \artist - album\, and no artist\album, if someone plans on making stats ;)
00:26:46HCleither the mailinglist is slow or mutt is slow with checking email..
00:27:25HClCoCoLUS: yea, eventually we should make it some form of regex, really *forces a smile*
00:27:54HCli used to have them ordered artist - album myself
00:28:05HCltill it became much more practical to put them artist/album
00:28:45preglowHCl: it is slow
00:28:52preglowsending mail via smtp does take a while
00:28:55preglowit's a very slow protocol
00:29:38HClbugs landing on my hand in the dark scaring me
00:29:46HClwhats that thing doing here anyways
00:30:16preglowthink i'll indulge in one hour of rockbox now
00:31:50HCl? XD
00:31:55HCli think you forgot an h
00:31:57preglowenthusiastic 'hai'
00:32:00 Join _n2_ [0] (
00:32:05HClotherwise you just yelled out "lovvveee!
00:32:27preglowwas thinking of edo out of cowboy bebop
00:32:31preglowshe always does that
00:32:33preglowand in rocks
00:32:40preglowin, it, bleh
00:32:46 Nick _n2_ is now known as ehntoo (
00:33:25 Join gromit` [0] (~gromit`
00:33:58HCl :P
00:35:07 Join _n2_ [0] (
00:35:52HClCoCoLUS: still, wouldn't you think "artist - album" is better than "<no album tag>" ?
00:36:17HClthe problem with <no album tag> is that every song without a tag gets stuck into one album named "<no album tag>"
00:36:21preglowi order stuff like genre/artist/album
00:36:32preglowunless i've just got one album, then it's genre/artist - album
00:36:35HClwhen you can easily split them into directories
00:36:51 Quit ehntoo (Nick collision from services.)
00:36:55 Nick _n2_ is now known as ehntoo (
00:37:11HClpreglow: same, pretty much. without the genre bit
00:37:47preglowthat's how i do it on the mp3 player, at least, to minimize scrolling
00:37:52preglowi've got no genre on my 'puter
00:38:24 Quit asdsd_ (Read error: 104 (Connection reset by peer))
00:44:50 Join asdsd_ [0] (
00:48:49 Quit asdsd_ (Client Quit)
00:48:50 Quit noclue2 (Read error: 110 (Connection timed out))
00:53:57 Join asdsd_ [0] (
01:18:03*HCl guesses he'll go sleep
01:19:55*preglow still has ten minutes left of rockbox!
01:26:55HClgo go go go go!
01:29:11preglowthink i'll finish it tomorrow
01:29:20preglownow, if it'd only work...
01:29:24preglowthat'd be so very great
01:30:11MoosCamarowe pray for you:)
01:36:20 Quit TCK (Read error: 110 (Connection timed out))
01:36:21 Quit MoosCamaro (Read error: 104 (Connection reset by peer))
01:36:39 Join MoosCamaro [0] (
01:38:22*amiconn has a suspicion
01:38:25 Quit cYmen ("zZz")
01:42:50MoosCamaroit's time to bed
01:43:04MoosCamaro1:43AM here in Paris
01:43:24 Join noclue2 [0] (
01:43:31MoosCamarogood night all
01:44:22 Part MoosCamaro
01:44:51amiconngrr, %$@, you won't belive that...
01:45:45amiconnyeehah! touchdown! :-)
01:46:46 Join _ferenczy [0] (
01:46:53amiconnThe Ondio reads the 2 GB card :)
01:48:34amiconnStill Really Hacky Code (tm), but at least I'll sleep well...
01:50:30HCl :)
01:50:43*HCl goes to sleep
01:51:20amiconnNow for the reason: Would you expect that that default read block size is different from the hardware block size?
01:51:54amiconnhardware block size is 1024, but the default read block size is 512...
01:53:22preglowi'm very happy i don't have to deal with stuff like that
01:53:25HClsounds a bit like the database code, having to change code all around the file to alter sizes.
01:54:14amiconnpreglow: Imho it's similarly lowlevel as your emac (i)(m)dct stuff, and I am happy I don't have to deal with such stuff
01:54:34amiconnI'm good at bit manipulation, but bad at that kind of mathematics
01:54:39preglowwell, yes, really, it all boils down to what floats your boat, i guess
01:54:43*HCl is glad that the database is fairly easy :P
01:55:52preglowwhat i'm doing now is more or less direct translation of arm code
01:58:13amiconnSounds like tedious work
01:58:43preglowsure is
01:58:53amiconnI remember reverse engineering the Ondio MMC part. Reading several pages of disassemblies was no fun
01:59:00preglowbut i don't think i can get this routine as tightly down as the guy who wrote the arm code already has got it
01:59:32preglowheh, at least this code is commented
01:59:34amiconn..even though SH1 asm is fairly simple
01:59:41preglowhow many registers has it got, again?
01:59:53amiconnIt's a straight & simply RISC machine
02:00:07 Quit ehntoo (Read error: 110 (Connection timed out))
02:00:08amiconn16 registers
02:00:50amiconnplus a special MAC register, which is 40 bits
02:01:05amiconn(not used in any code I've seen so far)
02:01:08preglowwould be fun to try using it for something ;)
02:01:22preglowcoldfire macs are also 40 bit wide
02:01:30amiconnyup... I think it might be possible to speed up jpeg loading
02:01:31preglowwell, 48, actually, but you seldom use more than 32 anyway
02:01:57amiconnThe bad thing is that the hardware multiplier is only 16x16->32 bit
02:02:07preglowdoesn't matter much for jpeg
02:02:12preglowi think the coefs are 16 bits anyway
02:02:21amiconnSo the MAC instruction does 16x16+42->42
02:02:32preglowmore then ample for lots of stuff
02:02:44amiconnyou can choose to "clamp" the result to 32 bits
02:02:59preglowthe block dct in jpeg should be pretty mac-ifiable
02:03:14amiconnIf I only knew how....
02:03:55amiconnPlus, I have the impression that the huffman decode needs more time than the dct
02:04:42 Quit _ferenczy ()
02:05:22preglowthe dct you use is one of those decomposed types
02:05:28preglowwith tons of butterflies instead of accumulates
02:05:33 Part Yono
02:06:12preglowlibmad also has one of those, i decided to ignore it on the grounds that a cpu with a fast mac instruction probably does it faster using that
02:07:05amiconnYes, probably. I googled around a bit for info about jpeg & dct
02:09:23amiconnStill, I don't think I can write that kind of stuff, let alone write it in asm
02:10:54preglowi'm pretty sure you can, having been told the basics
02:11:16preglowgetting the data out of the accumulator, is that a slow business?=
02:11:51amiconnIf you're only interested in the lower 32 bits, it's a simple mov-like instruction (1 clk)
02:12:23preglowthey'll do nicely
02:12:30preglowall jpeg coefs fit in 16 bits
02:13:00amiconnThe SH1 MAC is probably much less flexible than the coldfire EMAC. There is no fractional mode, and the accumulate operation is always add
02:13:40amiconn..and no parallel move etc
02:13:53preglowahh, it's always add
02:13:59preglowthat's a bummer
02:14:10amiconnIt is?
02:14:40preglowyeah, like i said, there are a lot of butterflies (c = a + b; d = a - b)
02:15:09preglownot having to negate b all the time would be nice
02:15:16preglowthen again
02:15:21preglowyou wouldn't use mac for that
02:15:53preglowwell, like it is, i don't think there'd be much benefit
02:16:08preglowthe sh1 programmers manual is really confusing :D
02:17:04 Join _n2_ [0] (
02:17:15amiconnpreglow: In what way?
02:17:34amiconnI think it is fairly good (but then I'm already used to it)
02:18:29preglownever mind me, i found the interesting parts
02:18:35preglowwow, sh-2 has a 64 bit accumulator
02:19:17preglowoh well, the emac isn't that bad, since it's capable of saving the high bits of a 32x32->64 bit multiplication
02:19:39preglowyou seldom need so much precision that the low bits are important
02:20:44 Join ehntoo [0] (
02:23:13***Saving seen data "./dancer.seen"
02:24:26 Quit noclue2 (Read error: 60 (Operation timed out))
02:26:21 Quit ehntoo (Nick collision from services.)
02:26:43 Join ehntoo [0] (
02:27:44 Quit _n2_ (Read error: 60 (Operation timed out))
02:33:03preglowhas the sh1 got stuff like relative addressing?
02:33:08preglowseems like arm doesn't have anything like that
02:34:10amiconnDepends on what you mean with relative addressing
02:34:55amiconn(1) (r<m>+r<n>)
02:35:27preglowlike <baseregister + offset>
02:35:44amiconn(2)( r<n>+offset*width), where offset is 0..15 and width is 1/2/4 depending on the access type
02:35:44preglowi thought most things have that, but looks like you have to calculate it by hand on arm
02:36:23preglowit also has <baseregister + indexregister*x + offset> which is a pretty cisc like construct, i guess
02:36:26preglowi use it very seldom
02:36:37amiconnPlus, there is a "general base register" than can be used for (r<n> + GBR)
02:37:01amiconnWe need to keep GBR at zero in rockbox though
02:37:46amiconn...because some operations are _only_ possible with that addressing mode, which we do use
02:38:36amiconnThe worst thing with the register + offset addressing is that the offset is always positive
02:39:06preglowwhat about auto-increment/decrement?
02:39:40preglowthat's a feature i've grown to like, i think 68k is the first arch i've coded for that has it
02:39:46preglowno, avr has it as well
02:39:51amiconnThat's implemented, but unfortunately the directions are also hardwired. Auto-increment is only possible with loads, auto-decrement only with stores
02:40:12preglowsounds like risc indeed
02:40:22amiconnThat's why my asm-optimised memset() works backward...
02:41:11preglowwhy, we need a bunch of those in asm for m68k as well, i guess
02:41:32preglowusing movem.l for memory access is a must
02:41:46amiconnI know
02:42:14preglowat least i think most memcpy/set/etc deal with enough memory that the register saving overhead is going to be worth it
02:42:30amiconnI also utilise an interesting effect in memcpy()
02:43:14amiconnThe archos devices have fast page dram, so it's beneficial to stay within the page as long as possible
02:43:53amiconnmemcpy() alternates between source & destination page (if src & dst differ more than 1024 bytes)
02:44:19amiconnSo I optimised it to do read-read-write-write instead of read-write-read-write
02:44:39 Join ashridah [0] (
02:44:51preglowhow? where do you save the extra data in between?
02:45:49amiconnThe loops (for both long & word alignment) copy 2 longs per iteration
02:46:05amiconnThe byte aligned loop only does one long per iteration though
02:46:35amiconnThe source is always long aligned (after copying the head bytes)
02:46:39preglowbecause you have to use an entire register for a byte?
02:47:36amiconnmemcpy() is somewhat tricky because you have to deal with 2 alignments, one of them you can't adjust
02:47:53amiconn..and the SH can't do unaligned accesses
02:48:25amiconnlong->long alignment is easy. Simpy read 2 long and write them to the destination
02:49:36amiconnlong->word is somewhat trickier. (1) Read 2 longs (2) write high word of first long (3) compose low word of first long and high word of second long, and write that as a long. The SH1 has a special instruction for this extraction
02:49:45amiconn(4) write low word of second long
02:50:16preglowweird instruction
02:50:20preglowbut useful, apparently
02:50:21amiconnlong->byte is most tricky. (1) Read a long (2) write top byte (3) write 2 middle bytes as word (4) write low byte
02:50:56amiconnIf I would extend this to do 2 longs at a time, there would be 2 problems
02:51:25amiconn(1) I would have to combine bytes from both words in a way for which there is no special instruction
02:52:14amiconn(2) The case would split in 2 if I would go to the max (byte/long/word/byte and byte/word/long/byte)
02:52:43amiconnBtw, the weird instruction is called xtrct
02:53:16preglowguess they anticipated it, can't see much other use for it
02:53:23preglowsome fixed point applications, perhaps
02:53:27preglowit's a cheap two register shift
02:54:02amiconnUseful for 32 bit fixed point with 16 bit fractional part
02:54:18amiconnExactly the kind of shift necessary after a multiplication
02:55:32preglowso it has a 32x32->64 bit mul?
02:55:48amiconnUnfortunately not
02:56:33amiconnI also tried extending this memcpy() concept to do 4 longs per iteration, but that wasn't worth it. Only a few percent speedup for large blocks, at the expense of slower operation for short blocks
02:56:35preglowthen how does a fixed point result ever end up in two registers?
02:57:02amiconnThe only case where this may happen is the MAC instruction
02:57:23amiconn(but you first need to fetch the 2 registers from the mac registers)
03:06:04preglowbut i think i'll call it a night
03:06:07preglowsee you later
03:06:29 Quit preglow ("-")
04:05:51 Join QT_ [0] (as@area51.users.madwifi)
04:17:31 Quit QT (Read error: 110 (Connection timed out))
04:20:18 Quit XShocK (" HydraIRC -> <- IRC for those that like to be different")
04:23:15***Saving seen data "./dancer.seen"
04:31:18 Quit Stryke` (Read error: 54 (Connection reset by peer))
05:20:36 Quit Rick (Read error: 104 (Connection reset by peer))
05:21:10 Join Rick [0] (
05:37:51 Quit silencer_ (Read error: 110 (Connection timed out))
05:45:01 Join StrathAFK [0] (
06:03:33 Quit Strath (Read error: 110 (Connection timed out))
06:23:19***Saving seen data "./dancer.seen"
06:46:04 Join LinusN [0] (
06:56:50 Quit asdsd_ (Read error: 104 (Connection reset by peer))
07:48:23 Quit hile ("jos taas menis")
08:10:37amiconnGood morning
08:23:23***Saving seen data "./dancer.seen"
08:34:09ashridahew. dude. we don't need to see that end of you
08:44:22 Quit edx (Read error: 110 (Connection timed out))
08:45:16 Join Lost-ash [0] (
08:49:33 Join Zagor [0] (
08:53:27 Join Harpy [0] (
09:07:38 Quit ashridah (Read error: 113 (No route to host))
09:25:37 Quit Ka_ (Read error: 104 (Connection reset by peer))
09:26:43 Join Ka_ [0] (
09:29:38 Join bobTHC [0] (
09:30:58bobTHChi folks !
10:04:42 Nick Lost-ash is now known as ashridah (
10:11:46 Join cYmen [0] (
10:16:20 Join edx [0] (
10:23:27***Saving seen data "./dancer.seen"
10:25:09 Join B4gder [0] (
10:26:32B4gdergood day sirs
10:26:51LinusNhelo B4gder
10:28:35 Join webguest90 [0] (
10:28:43*ashridah bounces. sort of.
10:34:27amiconnLinusN: I'm curious who will be #100
10:35:39LinusNah, webguest
10:35:40amiconnYou added contributor #99
10:35:57amiconnI wonder about your last commit though
10:36:10amiconnNo mpeg_get_last_header() for the 3507?
10:36:41LinusNguess i'm wrong there
10:37:08amiconnPlus, you forgot to bump the api version
10:37:55amiconn#ifdef SIMULATOR
10:37:58amiconn#elif CONFIG_HWCODEC != MASNONE
10:45:01LinusNwhy should i bump the api version?
10:47:35amiconnErm, because the struct changes because the function is no longer included on all platforms?
10:47:46Lynx_awyyesterday i declined being added to the contributer list...had i know i would be no. 100 ;)
10:50:22LinusNamiconn: sure, but i don't think it's necessary to bump it more that once a day
10:51:21amiconnDoh, didn't think about that. I agree.
10:52:40 Join preglow [0] (
10:52:54LinusNi'll change the #ifdefs according to your sugestion
10:54:03amiconnThe #ifdefs were from mpeg.c, No mpeg_get_last_header()
10:55:19amiconnAnd it seems I can't read correctly :( In fact you corrected the inclusion of this function in the right direction, just not completely...
10:56:05amiconnAnyway, if it compiled correctly before, the function wasn't used in plugins for non-3587 platforms
10:56:23LinusNalmost true
10:56:38LinusNsplitedit.c was compiled for iriver
10:56:50LinusNi'm fixing this as well
10:57:35amiconnI think that including functions in the api for a platform where they're unused is just wasting resources, but then that's my opinion...
10:58:08LinusNi think they should be included if the function is supported
10:58:51amiconnFor what reason?
10:59:59LinusNso we don't have to revise the api when we start using the function
11:01:07LinusNto simplify for (3rd party) plugin writers
11:01:16amiconnHmm. There are many functions which might be useful in plugins, but are not part of the api. What about these? ;-)
11:01:37B4gderI agree with Linus
11:01:45LinusNsurely there are lots of functions not used by any plugin
11:02:23LinusNbut the already exported functions should exist on all platforms that support them
11:02:59amiconnHmm, okay. That policy should be documented somewhere...
11:03:45LinusNlots of things should be documented somewhere
11:04:22LinusNi don't find anything particularly odd with that policy
11:04:51LinusNwe export functions on a "need-to-use" basis, and then we export it for all platforms that support it
11:05:59amiconnOkies. Need to obey that when I start working on the graphics api
11:06:13dwihnoWhat is a reasonable price for a IHP-140?
11:06:23amiconndwihno: New?
11:06:28dwihnoamiconn: yup
11:06:38amiconnI'd say ~350 EUR
11:07:09amiconn(That's for germany, with 16% VAT)
11:07:17LinusNamiconn is an experienced H140 buyer :-)
11:07:48dwihnoHow many have you gotten this far, amiconn?
11:07:51LinusNstill not H140 owner though :-) :-)
11:08:04dwihnois not? :)
11:09:34amiconnmeh :(
11:10:54*preglow strokes his h120
11:12:11amiconnBagder: 'feature requests', 'bug reports' and 'patches' are still broken
11:12:12dwihnoIs a 40 gbyte model overkill?
11:12:24dwihnoI might as well get a 20 gbyte model and replace the disk whenever it wears out
11:12:33B4gderok, I better investigate
11:13:36B4gderdwihno: in 1.8" land, that is not as easy
11:13:45B4gderas in 2.5" land
11:14:06B4gderthe 40gb disk is thicker
11:14:11B4gderat least the current one
11:17:02preglowyou should buy from your current needs
11:17:14preglowyou can't count on getting anything bigger than a 30 gig as a replacement for a while
11:17:59ashridahi've managed to fill up my iriver on many an occasion. usually using it to store large chunks of non-music data, mind youi
11:18:47amiconnLinusN: The new bmp reader adds 2 warnings to the gmini build because it doesn't obey the long policy
11:19:05amiconnLine 74: int readlong(void* value)
11:19:10amiconnshould of course be
11:19:13amiconnlong readlong(void* value)
11:21:17preglowwhat does the long policy dictate? use longs for 32 bit, int for stuff that doesn't need to be 32 bit?
11:21:42B4gdersince the calmrisc has 16bits ints
11:21:42LinusN"should be documented somewhere" :-)
11:21:58LinusNCONTRIBUTING, maybe?
11:22:00preglowi hate 16 bit archs
11:22:05preglowthey need to go away
11:22:25B4gdertell that to jyp :-)
11:22:48preglowi can tolerate even 8 bit archs, but not in friggin' mp3 players
11:24:17B4gderI wonder why tomas rewrote so much of the bmp code
11:24:59B4gderbut I'll leave it
11:27:29B4gderbut there is room for quite serious speed improvemts in it
11:28:58LinusNi thought it was best to put it in cvs and improve it later
11:29:29B4gderthere aren't that many users of the code yet anyway ;-)
11:30:58amiconnpreglow: Calmrisc is 16 bit, but on the other hand it's harvard. I thought you like that ;)
11:34:40preglowno, not when i thought about it
11:37:10preglowit's good for stuff you design yourself, heh
11:39:43amiconnLinusN: Strange... the warning is still there
11:40:14LinusNmaybe it should be unsigned?
11:41:00preglowi guess so
11:41:31LinusNi think all calculations are made on the default type
11:41:38amiconnMaybe an explicit cast is necessary before the shifts
11:41:41LinusNand then casted to the return type
11:41:49LinusNamiconn: exactly
11:47:16 Join muesli- [0] (
11:47:44muesli-morning mates
11:52:32muesli-thc? well ;)
11:53:14bobTHCas usual, stoned ;)
11:53:31muesli-too early for that stuff...its just 12 ;)
11:53:46bobTHCnot really in fact, but why not a 12 ? ;)
11:53:50preglowthe earlier the better
12:01:45amiconnLinusN: I'll try to apply the long policy to the mmc driver while adding variable blocksize support
12:02:08 Join ripnetuk [0] (
12:02:15amiconnWhile it's not really used for a 16 bit arch yet, maybe it will at some time...
12:02:17LinusNamiconn: don't bother unless mmc is used on the gmini
12:02:50amiconnBtw, reading my 2 GB card already works :) I listened to some music on that card last night
12:03:17amiconnNext thing is to implement writing, and correct error handling
12:05:27amiconnHackish Code Inside (r)
12:07:43preglowbest kind of code
12:20:32rasherI see I broke the *blushes*
12:21:01preglowrasher: yo, did you notice anything out of the ordinary at all in your breezy update? mine stops completely in a /lib64 collide :/
12:21:50rasherHm.. nothing like that happened to me
12:22:00rasherbut then, I haven't upgraded for a while
12:23:31***No seen item changed, no save performed.
12:25:47preglowmight be a amd64 specific thing
12:29:07rashersounds reasonable
12:30:04preglowno milk!
12:30:35rasheryour cornflakes will be very dull now :(
12:30:49preglowmy using acronyms is an indicator of my seriousnes
12:31:17preglowoh well, shop's just 50 metres away, off i go
12:38:24 Join kurzhaarrocker [0] (
12:39:49kurzhaarrockerIs there a reason why the playlist context menu is not available when you browse the id3 db?
12:40:34*HCl reads up
12:46:57 Quit muesli- (Read error: 60 (Operation timed out))
12:48:18amiconnhi kurzhaarrocker
12:50:21HClcompiler creation class is fun..
12:50:45ashridahwish my course still included a subject on that :(
12:51:02ashridahclosest we do is grammars :/
12:51:24*HCl is planning on saving all sheets of all courses of his university soon since he's most likely gonna move down to college next year
12:51:37HClbit of a "grab what you can" idea :p
12:52:32HClgot AI class in 40 min
12:53:55ashridahah, the beauty of mandatory partner-based projects
12:54:03HCl :P
12:54:06HCli have that.
12:54:08HClwith markun
12:54:20HClwe're doing both compiler creation and AI together
12:54:44*ashridah watches the frequency of "i'm looking for XYZ" type newsgroup postings because people a) don't read their email b) don't go to lectures and c) don't bother to find their partner before their partner has to tear apart the uni looking for them
12:55:07ashridahyeah, well, it's okay if you do it with a mate. too bad if you get assigned a complete stranger who never turns up
12:55:16ashridahfortunately, i'm not in that position :)
12:55:22HCli know that idea.
12:55:34HCli once had Operating Systems with a person who was very very slow.
12:55:46HCli pretty much ended up doing the practical bit on my own with him watching
12:55:51HClwe still finished 3 weeks early :P
12:55:59HClahead of everyone else :P
12:56:27ashridahunfortunately for me, i didn't realise my partner wasn't really motivated. (of course, since i have real-world experience, and next to no social life, it's hard to find ANYONE as motivated as i am)
12:57:10ashridahbut yeah, keeping clear of him next time :)
12:58:03ashridahi'm definently looking forward to my operating systems subject next semester.
12:58:16rasherpreglow: still chugging along after a dist-upgrade just now
12:58:27HCl :P
12:58:42HClmy operating systems course was boringish :/
12:58:53HClit was mostly about creating a (unix) shell
12:59:16HClmultithreading, communication between processes, deadlock avoidance..
13:02:02ashridahtrue, but i enjoy working with complicated crap like that. leaves me lots of room for practical experience
13:03:47LinusNthe sad part is that this slow person will get the same degree as you, while you have twice the knowledge and experience
13:04:29ashridahLinusN: heh. one thiing i like about my software engineering subject is we submit a CVS tree. it's fairly easy to see who's committed the most then :)
13:05:18ashridahbut yeah, it's often hard to find another partner who's very motivated, particularly at the undergraduate level :(
13:05:26HClLinusN: well.. most courses have both a practical grade and a test
13:05:42HClbut yes, for the courses with only practical grades, the person is literally hitch hiking on your grade
13:06:04HClashridah: heh, i remember this person who refused to work together with anyone cause of that exact reason :P
13:06:22HClthen he was forced together with me, and he phoned me the other day whether we could work together on compiler building xD
13:06:36HClbut markun is nicer to work with :P
13:07:40ashridahheh. reminds me of what i read about john carmack. he got put into a group with some stupid bully, who tried to force him to do all the work for a project. so he did nothing, and torched his own mark for the assignment, just to kick the bully in the nads (figuratively)
13:08:06HCli would do that as well.
13:08:20HCli'm the kind of person you can't force to do anything without me agreeing
13:08:29HClunless someone else will get hurt if i don't o.o
13:09:01Lynx_awyin biology you have to work together on practical stuff, so your partner can even mess up the whole experiment easily, and be lazy in additon ;)
13:09:50HClthat sucks even more ;
13:09:51HCl :P
13:10:44Lynx_awyyes, especially if you get to stay hours longer than everyone else, together with that partner ;)
13:10:48 Nick Lynx_awy is now known as Lynx_ (HydraIRC@
13:11:10ashridahin conclusion, partner-based projects suck balls.
13:11:21HClonly if you don't have a nice partner.
13:11:30ashridahno, in general.
13:11:36HClmk o.o.
13:11:40ashridahbecause it also impacts unfairly on teams who partners split work evenly
13:11:46*HCl thinks of all the partner based projects he's had.
13:11:54 Join MoosCamaro [0] (
13:11:55Lynx_HCl: or at least a good looking female ;) but i guess your chances in computer science are not so great, eh?
13:11:58HClwell, its mainly to teach you to work together with others.
13:11:59ashridahit's an inaccurate method of gathering a view of someone's abilities
13:12:04HClLynx_: nah, and seriously.
13:12:14HClmost females i've met that *are* on our university
13:12:20HClare either sluts, bitches, or both o.o.
13:12:20ashridahHCl: i'd personally submit that that's not a skill that's learned at the ages we're at.
13:12:36MoosCamaroHi all
13:12:47HClthough there are a handful of nice ones *shrugs*
13:13:08Lynx_HCl: hmm, maybe thats just in your country ;)
13:13:18HClLynx_: i hope so :P
13:13:24HCli don't mind much
13:13:27HCli've grown used to it o.o.
13:13:39HCllike i said, not all are bad
13:13:46HCljust the ones i mostly come across
13:14:58Lynx_HCl: you should go into bioinformatics, then you can teach nice female students not to edit sequences in ms word
13:15:33HCli dunno what i'm gonna do yet.
13:15:50HClgonna take a career test this summer, if nothing conclusive gets out of it, i'll most likely go to college compsci
13:16:14HClshould be easy as 3.14192..
13:16:42*HCl is fairly good at compsci, just not the math thats needed for some parts
13:16:47 Join elinenbe [0] (~elinenbe_@
13:17:51preglowimdct_l still doesn't work...
13:18:59LinusNpreglow: congrats :-)
13:19:52ashridahuh. 3.14192?
13:20:02ashridahyou're missing a digit in there :)
13:20:34HCli lost my 5
13:20:38HCl :p
13:21:16preglowthis just makes me want to implode
13:21:23HCl :/
13:21:27HClcomputers suck, yupyup
13:21:28preglowi don't want to spend any more time on it, and i hate having unfinished stuff lying around :/
13:21:45*HCl knows the feeling.
13:22:06preglowbut oh well, i'll probably have a look at it again in a few days
13:25:02*amiconn prods kurzhaarrocker
13:25:22 Join Sucka [0] (
13:25:49preglowwhat should be faster, doing six clr.l on consecutive memory locations, or clearing six registers and doing movem ?
13:26:56preglowi've gone for the first one
13:35:58LinusNi think the second one is the fastest
13:36:49LinusNthen the cpu can burst both the instruction fetches and the data writes
13:36:51 Quit Sucka ("a bird in the bush is worth two in your house")
13:37:07preglowthat's true
13:37:19preglowof course, it can't burst on the first one
13:37:28preglowit doesn't matter that i'm writing to consecutive memory
13:37:32*LinusN mumbles..."damn coldfire, damn coldfire"
13:38:11ashridahnow, the big question. would you have started hacking on the irivers if you knew what you know now about the annoyances of the platform? :)
13:38:17LinusNpreglow: you're right, the cpu won't burst the writes
13:38:23preglowashridah: sure
13:38:25LinusNashridah: of course
13:38:36preglowashridah: nothing's perfect, and it's interesting nonetheless
13:38:47LinusNpreglow: or maybe it will....
13:38:56LinusNit should
13:38:57amiconnLinusN: Why can't it burst-write?
13:38:58preglowLinusN: well, that depends on the caching, doesn't it?
13:39:15LinusNamiconn: wrong thinking on my part
13:39:48LinusNpreglow: yes, the instruction fetch will burst, since it will always fill a cache line
13:40:40preglowLinusN: the burst might be interrupted, though, if the sequence is broken over the sixteen byte boundary
13:41:41LinusNif i'm not mistaken, the cache line fills are always complete
13:42:29LinusNso it will wrap
13:43:09LinusNso you might end up with two bursts to get the instructions
13:43:38preglowone burst to get some of the clr.l's, one burst the save zeros, then one burst to get more instructions, then the rest of the clr.l burst, no?
13:43:58preglowif there's not room for all of them in one cache line
13:44:54 Join Sucka [0] (
13:44:55LinusNsince the pipeline is so limited, it will not burst the clr.l writes
13:45:08 Quit Sucka (Client Quit)
13:45:13LinusNsince it interleaves fetches and data writes
13:45:25LinusNthat's why option #2 is faster
13:45:48LinusNwhere it does the data writes in bursts
13:47:03amiconnIt may still be slower overall, as you'd have to sacrifice 6 registers, which may need to be reloaded later.
13:47:10preglowamiconn: no, not in this case
13:47:18preglowamiconn: if that was the case, i'd of course not even have considered the second case
13:50:20preglowi really hate that bloody accumulator move stall they list in the errata
13:50:25preglowi bump into it all the bloody time
13:53:18preglowtwo cycles wasted here and there won't kill me, but i still dislike it, goddamn it ;)
13:53:41 Quit lostlogic ("Going to the moon")
14:07:48 Join t0mas [0] (
14:10:33preglowand behold!
14:11:46 Part LinusN
14:13:17elinenbepreglow: what are you working on?
14:13:39preglowgood old libmad optimization
14:14:22*B4gder is birdwatching
14:14:34B4gdera penguin
14:16:38*t0mas is looking at the credits list on his iriver :)
14:16:51B4gderit is quite big nowadays
14:17:04t0masyes, Linus added me ;)
14:17:21*B4gder is a commit mail reader
14:17:30t0masyeah, me too :)
14:17:36t0masnice to know what's going on
14:20:43t0mashm... rasher made a iriver.iriver file for rolo-ing right?
14:23:35***Saving seen data "./dancer.seen"
14:27:51 Join lolo-laptop [0] (~lostlogic@
14:27:54B4gderI don't think he got it working
14:47:46 Join Lost-ash [0] (
14:47:47 Quit ashridah (Read error: 54 (Connection reset by peer))
14:54:06*amiconn prods kurzhaarrocker again
14:56:25kurzhaarrockeramiconn: now I'm here again
14:58:58amiconnDid you think about my trigger mode suggestion?
14:59:43*kurzhaarrocker tries to remeber
15:00:29amiconnLet it retrigger, but don't start a new file on retrigger (only pause/unpause)
15:00:45amiconnDictaphone mode
15:00:52kurzhaarrockerAh. I in fact forgot that *blush*
15:01:34kurzhaarrockerDon't expect that to happen this week :)
15:06:18 Quit StrathAFK (Read error: 110 (Connection timed out))
15:07:18 Quit Ka_ (Read error: 110 (Connection timed out))
15:09:05 Join muesli- [0] (
15:10:12 Quit muesli- (Client Quit)
15:10:55 Join Ka_ [0] (
15:16:26 Quit kurzhaarrocker (Remote closed the connection)
15:22:29rashert0mas: didn't get it working.. tried to do it from a plugin
15:22:37rasherwhich wasn't possible, it seems
15:24:13 Nick Lost-ash is now known as ashridah (
15:35:56 Join webguest22 [0] (
15:36:58 Quit webguest22 (Client Quit)
15:40:56*HCl returns
15:40:56 Quit webguest90 ("CGI:IRC (EOF)")
15:42:45HClt0mas: ripnetuk made one that worked
15:42:51HClwe need to prod him about it
15:42:54HClripnetuk: *prod*
15:43:34rasherit didn't work when I tried it
15:43:58rasherin fact, his code was *identical* (except comments) to mine
15:44:09rasherI suspect some things changed since then
15:44:12rasherbut yes
15:44:17rasherripnetuk: *prod*
15:44:36*HCl prods rasher to talk with zagor about fixing the to support multiple artist albums
15:44:48HClalso see my email on how it works at the moment with my suggestion on what to do to get it working properly
15:45:20rasherwill do, not sure when though, at a lecture right now, and won't have time later today I think
15:45:37*HCl just returned, not sure what to do yet.
15:46:47preglowsupport your local preglow with funds to buy a new camera, perhaps?
15:48:08HClmeh, i have my money saved up for other stuff.
15:48:50HCli can choose between fixing my pda, look at my assignments for tomorrow, or make a start with copying course material to my computer
15:48:58HClor i could be lazy and play a computergame
15:49:21t0masor you can think of something to write for rockbox?
15:49:38preglowi have the money, i just need a ton of sudden income i can rid myself of equally suddenly
15:49:42HCli'd work on the database, but first i want it working properly
15:49:50HClwhich depends on zagor/rasher and maybe Bagder
15:50:55HCli should probably work on my pda..
15:51:06rasherwhat's up with it?
15:53:04HClno wireless internet
15:54:24B4gderdon't hold your breath for me working on that
15:54:35HCli'm not,
15:54:44HClbut you always have a strong opinion on how it should work :p
15:55:12B4gderI do, but I've also said that you should proceed with what you've suggested
15:55:33HClwell, i can't work on the database while the current design is broken
15:55:49HCland i don't agree with albums with multiple artists not being possible either
15:56:02HClthe current gets fairly close
15:56:09HClaside from that dbtree.c needs added filtering
15:56:12B4gderit is possible only given a few assumptions
15:56:17HClwhich it simply doesn't support yet
15:56:40HClfor multi artist albums, i'd like every artist to have a link to the album (well, i described this on the mailing list...)
15:56:55B4gderI know you did and I don't object
15:56:59rasherthat sounds reasonable
15:57:03HCland if you're coming from a browse that started from an artist, filter it with the artist
15:57:11HClokay, so its okay for me to try to fix it in the way i proposed?
15:57:22rashersounds good to me
15:57:33 Join Mirfle [0] (
15:57:44rasherI'm probably not an authorative source though :)
15:57:46HClokay, then i'll just work on my pda first, then on the database
15:57:55B4gderall fixing of broken things is goodness in my view
16:03:54 Nick QT_ is now known as QT (as@area51.users.madwifi)
16:20:03 Quit ashridah ("Leaving")
16:23:39***Saving seen data "./dancer.seen"
16:24:55*HCl slaps windows
16:25:06HCli tried to safely unmount my compactflash card.
16:25:13HClinstead, it completely detached my cardreader
16:28:11amiconn(darn return key)
16:29:06amiconnYou can't safely unmount removable media on their own, you can only safely unmount removable devices
16:29:34amiconn(and given the settings are 'optimized for fast removal', it's unnecessary)
16:30:37*HCl goes to take a look at the database
16:31:33HCli'm not gonna bugfix v1 anymore, i'll just make sure v2 works properly, then if everybody agrees, update the cvs with that.
16:31:51*HCl is tired of having to make the same changes in two different perl scripts
16:39:33 Quit B4gder ("CGI:IRC")
17:00:22t0masthe feature request list on the rockbox site is a little empty?
17:01:19 Quit Mirfle ("Chatzilla 0.9.68a [Firefox 1.0.3/20050414]")
17:01:26preglowthere are some problems
17:01:52*HCl prods rasher
17:06:25 Quit ripnetuk ("Leaving")
17:09:38*HCl bites perl
17:10:38 Quit micoo (Read error: 54 (Connection reset by peer))
17:12:22*preglow strokes perl
17:13:07HClwanna fix this script for me? heh
17:13:32preglowit's a bit too big for me to just drop in and out of
17:14:19HClcan someone explain the difference between $id3{'BLAH') and $$id3{'BLAH'} ?
17:16:39rasherYeah, can someone?
17:17:35Zagor$$id3 is for dereferencing a reference
17:17:43Zagorlike *id3 in c
17:20:41preglowwhy is it done like that, can't all data be represented as a string?
17:21:55 Join Sucka [0] (
17:22:07HClwow, Zagor is actually awake o.o.
17:22:26*HCl sucks at perl, but is trying to adapt it for albums with multiple artists anyways..
17:23:55preglowperl is a lovely language
17:28:46HCli need better testcases.
17:29:33 Join silencer [0] (
17:36:12*HCl has a dir with random songs, a dir with no album tags... *gets a properly tagged album*
17:36:21HClwhat else could it encounter?
17:36:28rashermultiartist cd?
17:36:37HClthats the directory with random songs.
17:36:52rashershouldn't they have same album tag?
17:36:58HClthey can, its not needed
17:37:22HClat the moment
17:37:31rasherwhy would they belong to the same album if they don't have the same album tag?
17:37:39HClcause they're in the same directory
17:37:59HCli'm gonna make it have two modes..
17:38:08rasherI sure wouldn't want that
17:38:24HClyea, its just a flag you can enable at the moment.
17:38:52HClnot 100% sure how to do albums with multiple artists yet
17:39:02HClif you can't assume that such an album would be in a single directory
17:39:10HClwithout extra files
17:39:29rasherI'd say you can assume that if songs in one dir have the same album, that it's a multiartist album
17:41:06HClwhat if they have different albums then?
17:41:18*HCl stares at his perl script
17:41:28HClits doing exactly the opposite of what its supposed to do. thats odd.
17:42:08webmindget a pastebot and show the url :)
17:42:14webmindalthough it seems off topic
17:42:36t0masnope, it's a rockbox script
17:43:19rasherHCl: then they're not part of the same album, surely?
17:44:40HClrasher: sure, i have plenty of albums that are like that o.o
17:44:53rasherwhy are they not tagged with the same album-tag?
17:45:00HClbecause they're compilations
17:45:00rasherI don't understand that.. :-/
17:45:15HCland its nice to know what album the song originally came from.
17:45:18rasherI'd tag those with the name of the compilatioin..
17:45:20rasherI see.
17:45:38*HCl works on it..
17:50:31preglowhere's me thinking he's getting a parameter from the stack:
17:50:34preglowmove.l (21*4 + 12), %d0
17:50:38preglowcharming, no?
17:51:33HClrasher: can you toss me a multiartist album?
17:52:22HClomg lag
17:54:42 Quit Sucka ("a bird in the bush is worth two in your house")
17:54:44*HCl prods rasher
17:54:50HClis the matrix ost one ?
17:55:44*HCl downloads it
17:57:13MoosCamaroyea all the ost films are multi artists
17:57:36 Quit Hadaka (Read error: 60 (Operation timed out))
18:01:26MoosCamaroalmost Dj's compilations, kinds retrospectives, best of...
18:01:36MoosCamaroi've got lot of them
18:02:13HClalmost works.
18:02:31HClwithout dirmode, it messes up the artist link in the album when an album has no album tags
18:03:41rasherHCl: got it sorted?
18:04:53HClrasher: almost...
18:05:03HClmy testcase of an album with all songs without album tags
18:05:12HClis messing up without −−dirisalbum
18:05:21HClwhich is odd.
18:05:25HCldon't know why yet.
18:06:06HClno, change that, its going completely wrong :P
18:06:10HClwithout −−dirisalbum
18:07:27HClits going wrong with dirisalbum too x.x
18:12:29HCli hate this kind of stuff x.x
18:12:40HCl(side note: i might very well not manage.)
18:15:57webmindHCl, what's the script suposed to do ?
18:16:32HClbuild a tagdatabase.
18:16:49HCllots of nasty cross references in the form of fileoffsets.
18:17:10HCland i didn't even write it originally.
18:21:41HClthe script doesn't seem to do it wrong though, odd.
18:23:42***Saving seen data "./dancer.seen"
18:26:43HCloddly enough, the artistcount seems to be off by one..
18:32:52HClfixed :)
18:35:16HClworks great :)
18:46:39 Join muesli- [0] (
18:53:03HClor not.
18:53:10HCl*wonders whats wrong now and glares*
18:53:26HClcomputers suck.
18:53:47preglowindeed, they do
18:55:13HClwell, at least it worked on my testcase :/
18:55:17HClit doesn't on my player though
18:55:42HClmaybe i should just enable the new −−dirisalbum option..
18:59:40*HCl throws the out the window and goes for dinner >/
19:00:45preglowgood god, how i hate gas
19:00:49preglowi thought ; meant comment
19:01:08preglowit seems that gas just ignores that character at the start of the line, at the end of a line, though, then it's bad
19:01:58 Quit muesli- (Read error: 60 (Operation timed out))
19:02:30preglowvim highlights as if it's a comment
19:14:50 Join Naked [0] (
19:15:03 Nick Naked is now known as Hadaka (
19:15:58 Join matsl [0] (
19:21:33 Quit bobTHC ("Smoke Weed Every Day !")
19:22:14HCli really need to properly tag my mp3 database
19:22:23*HCl understands why preglow stripped them, heh.
19:23:43HClwho's around with an iriver
19:23:47RickI am.
19:23:48HClexcept preglow
19:23:54HClgot some time to help?
19:23:57HClneed some testers..
19:24:03Ricktesting what?
19:24:18HClthe new tag database generator,
19:24:32Rickwindows compatible?
19:24:37HClpretty much.
19:24:43Ricklemme do a cvs update
19:24:49HClits not in cvs.
19:24:57Rickwell, then ship it over
19:25:03HClfrom there
19:25:26HClthen follow
19:25:36HClbut with my
19:26:12HClyou can alter makedb.bat, or run it by hand
19:26:14HClhold on, phone
19:27:17RickI have activeperl installed
19:28:26 Join TheVoid [0] (
19:28:31TheVoidhelp me!!! :(
19:28:40Rickso what I am looking to do HCl?
19:28:42TheVoidlol okay... heres what happened
19:28:47RickNumber of artists : 417
19:28:47RickNumber of albums : 32
19:28:47RickNumber of songs / files : 671
19:28:48HClhold on.
19:29:09HClnot you, rick x.x
19:29:18*TheVoid waits
19:29:37HCl<- phone
19:30:21HClRick: put the perl script in .rockbox
19:30:28HCland run it with −−path /
19:30:29RickI did
19:30:30RickI ran it
19:30:36HClthen install that
19:30:37RickI pasted the output
19:30:39Rickalready did
19:30:41HCland reboot your player
19:30:53HClthen go to general options -> file view -> show files -> id3 database
19:31:00HClsee if it works properly..
19:31:11HClit'll most likely misbehave if you have improperly tagged files
19:31:22Rickimproperly tagged how?
19:31:33HClmissing album tags in songs in albums
19:31:42HCland/or missing artist tags
19:31:55Rickmy backlight is turned off
19:32:03 Join stevenm [0] (
19:32:05HClTheVoid: whats the problem?
19:32:12stevenmHello people
19:32:34stevenmIt's another physics lecture, and that means rockbox development
19:32:55RickHCl: looks fine
19:32:58Rickbut I have a bunch of
19:32:59Rick<no album tag>
19:33:16TheVoidoh hi
19:33:20TheVoidheres the problem
19:33:23HClRick: can you try to find any bugs (aside from the "all songs" option when browsing artists)
19:33:44TheVoidi have an error message ATA -11, when i turn on my jukebox
19:33:46HClRick: you can optionally try the −−dirisalbumname option to see whether that helps
19:33:56HCland if that doesn't help, −−dirisalbum
19:34:04Rickthat's not why
19:34:15HClit would fix the no album tags things.
19:34:16RickI have a bunch of dirs that are mp3s that were streamripped
19:34:22Rickwell, true
19:34:46HClon my player, almost everything malfunctions except when i use the dirisalbum option
19:35:03HClcause most of my tags are incomplete.
19:35:10stevenmMaybe you can have an option, "Set all the files in this directory to album X" and it sets the id3 tag ?
19:35:15HClTheVoid: no idea, you should ask linusn, i guess.
19:35:30Ricksearching works
19:35:32HClstevenm: meh.. there are plenty of tagging tools around
19:35:38stevenmah ok
19:35:39TheVoidum... who's that? lol
19:35:47HClTheVoid: our hardware person, pretty much
19:35:48preglowTheVoid: that means your disk is dead
19:35:54Rickseems to be working fine for me, HCl
19:36:02HClRick: ok, please try to find any bugs :)
19:36:09HClaside from the all songs for an artist
19:36:14HCl(that one should be broken, i think o.o)
19:36:16Rickwhat all songs for artist?
19:36:18stevenmTheVoid, linusn is rockbox god
19:36:24TheVoidpreglow: i read somehting about them getting that error message when the batteries were very very low
19:36:35TheVoidstevenm: how do i get ahold of him?
19:36:38preglowTheVoid: then try swapping batteries
19:36:39Ricknow is ee
19:36:55*TheVoid is currently charging the batteries
19:37:08HClRick: is it broken?
19:37:12Rickwhat happens if there are different albums by different artists?
19:37:20Rickdo they list all tracks from the seperate albums?
19:37:22HClthen every artist has that album listed.
19:37:35HClits not fixed yet.
19:37:49Rickseems to be working fine except the all songs then
19:37:50TheVoidthanks... i'll let the batteries charge for a couple min
19:37:53Rickand the ton of <no album tag>
19:37:55Rickin album list
19:38:15HClyea, i added a seperate option to default to dirname when there's no album tag
19:38:18HClto fix that
19:38:26HClokay, thats good
19:38:31HCli'll need some other people to test it too.
19:38:40Ricksearching works
19:38:43HCli'm hoping that the only thing broken is the all songs thing.
19:38:48HClwhich i'm aware of and shouldn't be hard to fix
19:39:38Rickyou can't seem to search for 'no tag' though
19:39:58HClwell, i altered it so that all "<no title tag>" things
19:40:01Rickand, instead of taking you to an empty list
19:40:05HClwere replaced with the basename of the mp3 filename
19:40:12Rickit should take you back to the search query
19:40:16 Join Sucka [0] (
19:40:17Rick(when no results found)
19:40:24HClsuch details come later...
19:40:31HClfirst i want it working properly.
19:40:37HClthen i want to put it in cvs
19:40:43HCland then i'll start on the runtimedb part
19:40:45Rickgood work :)
19:40:52HClthanks :)
19:40:59HClany other people with irivers around that can help test?
19:41:26HClt0mas: can you test it please
19:41:33t0mas(intrested in reading the code too)
19:41:55t0masyes... d/l url?
19:42:23HClthe code of isn't particularly interesting
19:42:31HCland the code in rockbox itself hasn't changed much at all.
19:43:04RickHCl: oh, and a suggested thing: automatically strip drive name on windows
19:43:12Rickunless −−strip can do that
19:43:15HClRick: hmm?
19:43:16HCldrive name?
19:43:26HClyou can just do −−path /
19:43:29RickI know
19:43:30HCland it won't take the drivename with it
19:43:34HCland −−strip will do it too.
19:43:38HCl −−strip G:
19:43:47HClat least, its supposed to.
19:43:49RickNice, you guessed the drive I had mine on
19:44:00HCli'm psychic like that :)
19:44:03HCl :p
19:44:49t0masHCl: so I just need your perlscript?
19:44:54t0masand make it run on my mp3 folder?
19:44:58HClt0mas: and the
19:45:14HClthat describes what to do
19:45:14t0masah, you did change something? :)
19:45:18HClonly, you use my perl script
19:45:22HClyea, its version 2 of the database
19:45:29HClas described on HClProposalThingy
19:46:21HClfor now, i want it to work exactly like v1, when thats stable, i'll go on enabling the extra features it gives
19:46:50t0masThis is perl, v5.8.2 built for cygwin-thread-multi-64int
19:46:52t0masthat's ok?
19:46:57HClshould be.
19:48:55TheVoidanyone know how many volts the batteries have to be for minimal function?
19:49:26t0masperl −−path /cygdrive/j −−strip /cygdrive/j
19:50:05t0masbetter idea...
19:50:07t0masperl −−path /cygdrive/j −−strip /cygdrive/j
19:50:13HCl :P
19:50:20t0masthat's gonna take hours...
19:50:26t0masdoes it have to be so slow? :P
19:50:30t0masno... usb2
19:50:34t0masbut player almost full
19:50:45Rickshouldn't take too long
19:50:47HClit scanned my player in about 5 min
19:50:54t0mas5 min == long :)
19:51:11t0masarg... it doesn't check for .mp3 extension?
19:51:16t0masoh wait... never mind
19:51:17HClit does.
19:51:21t0masit just has to check all folders
19:51:37t0masmy /data is a little big ;)
19:52:05t0masso it scanned some "data"
19:53:19 Join Aison [0] (
19:53:37t0masin 3:17 min.
19:53:42HClkay, load the
19:53:55t0masMax album length : 56 (Life Is What Happens While Your Busy Making Other Plans)
19:53:57HCland go to general settings->file view->show files->id3 database
19:54:05t0masalready loaded
19:54:11t0masI don't have to copy the database?
19:54:22HClyou have to make sure your database is in .rockbox
19:54:25HCland you have to reboot the player
19:54:44*t0mas thanks Linus again silently for rolo
19:55:14t0masempty HCl
19:55:24HClhuh? o.o.
19:55:42Rickwhat's the size of your db?
19:55:45t0masand I know why :)
19:56:01t0mas[19:54:20] <t0mas> I don't have to copy the database? <−− Yes Tomas, you should
19:56:11HClyou have to reboot it when you copy a new database as well.
19:56:39*HCl prods rasher
19:56:48HClrasher: yo, hows the hash stuff going? :3
19:56:56HCl(remember kids, drugs are bad, mkay)
19:57:06preglowi started wondering for a teeny while there
19:57:11 Quit TheVoid ()
19:57:18HCl :P
19:57:23preglow"yo, how's that shipment coming"
19:57:39*t0mas swallows a rasist remark :P
20:00:18t0masdatabase works HCl
20:00:46t0masand a suggestion: make the database a part of the menu... so I still have my filebrowser
20:01:33HClsave that suggestion up
20:01:35HClfor later
20:01:42HClcan you test and see whether there are any flaws in it?
20:01:46HClaside from all songs for artist
20:01:53t0maswell.. I checked my artist list
20:01:56t0masand part of the song list
20:01:59t0masappears ok
20:02:09t0massearched some things... found the right one's
20:02:16HClif you have the time, please look around for stuff see if you see anything odd
20:02:19t0masanymore tests?
20:02:33HClwell, i mostly need people to look for bugs
20:02:35t0mashmz... will be a lot easyer with sound support :)
20:02:38HClit'd be nice if you could confirm
20:02:42HClthat <all songs>
20:02:44HClfor an artist
20:02:45HClis broken
20:03:44t0masI'll test
20:04:20t0mashalf broken
20:04:38HClsame here, pretty much
20:04:40HClokay, good
20:04:49HClas long as i know whats broken and how that part is supposed to work
20:05:16*HCl might even be able to fix it quickly, goes to look.
20:10:04HCllets see if the new fixed it..
20:10:37*HCl makes mental note: fix multiartist albums browsing before committing.
20:12:06HClfixed :)
20:12:35HClt0mas / Rick: redownload
20:12:42HClsee if you can find any bugs left
20:13:28t0masHCl: works :)
20:13:34t0masI'll be back in two hours...
20:13:46t0masI'll test when I'm bored ;)
20:13:53HCl :P
20:14:08t0maswhat's the english word voor file?
20:14:14t0mas(the dutch file)
20:16:55HCltraffic jam.
20:20:12HClwe have database v2 in cvs :3
20:20:25HCl*goes to monitor the daily builds and prays nothing breaks*
20:23:43***Saving seen data "./dancer.seen"
20:24:39HClnext stage is filtering for artist only when going artist -> multi-artist-album
20:38:12*HCl goes to shower and wait for people to yell at him about broken things
20:56:37HClno angry mob yet?
20:56:39HCl :P
20:57:15HClno emails about players catching fire either
21:09:21*HCl prods Rick
21:09:24*HCl prods rasher.
21:09:27*HCl prods everyone o.o.
21:09:35*HCl prods everyone who used the old tagdatabase.
21:10:33 Quit Sucka ("a bird in the bush is worth two in your house")
21:11:38*t0mas prods HCl stop prodding :P
21:11:47*HCl ies.
21:12:31*HCl prods everyone but t0mas who is capable of testing current cvs regarding the tagdb
21:12:52*t0mas prods people hilighting him :P
21:13:04HCl :P
21:15:25HClah well.
21:15:38MoosCamaroHCl i come back in ~1 hour, if you need i'll test for you
21:15:45*HCl guesses he'll go nap while people form an angry mob with torches and pitchforks
21:15:50HClMoosCamaro: please do
21:19:24 Join XShocK [0] (
21:19:36HClhey XShocK
21:19:49HClplease bugtest the current cvs regarding the tagdatabase
21:19:55*HCl goes to nap
21:26:07 Nick Lynx_ is now known as Lynx_awy (HydraIRC@
21:32:52 Join webguest89 [0] (
21:38:30 Quit webguest89 ("CGI:IRC")
21:43:13 Join F1^Aison [0] (
21:44:59XShocKHCl: hi
21:45:26XShocKok. how do i test it?
21:48:31 Quit Aison (Operation timed out)
21:48:32 Quit MoosCamaro (Read error: 54 (Connection reset by peer))
21:48:36 Join MoosCamaro [0] (
22:01:42 Join Biptoria_the_2nd [0] (
22:03:22XShocKgrrr.... what is that?
22:03:22XShocKCan't create /cygdrive/d/Projects/rockbox/cvs/rockbox/target/sysfont.c
22:04:23Biptoria_the_2ndOwt new to test yet ?
22:05:48Biptoria_the_2ndMy good people of rock box
22:06:53HClXShocK: follow the procedures on the tagdatabase wiki topic
22:07:02HClwith current cvs
22:16:19 Quit Biptoria_the_2nd ("CGI:IRC (EOF)")
22:20:35 Quit stevenm (Read error: 110 (Connection timed out))
22:23:46***Saving seen data "./dancer.seen"
22:26:01 Quit matsl (Remote closed the connection)
22:26:35 Join Querty [0] (
22:28:21QuertyHi HCL, just followed the wiki instructions for the songdb, but not having any luck
22:28:31HClmmm? whats it doing or not doing?
22:29:09QuertyI've created the songdb with the from the wiki (that's a cvs checkout, isn't it)
22:29:16HClyea, it is.
22:29:16QuertyIt seems to be filled ok
22:29:25QuertyNumber of artists : 323
22:29:25QuertyNumber of albums : 284
22:29:25QuertyNumber of songs / files : 1030
22:29:37Quertyupdated rockbox to latest cvs
22:29:55Quertyselected under General -> Show Files -> ID3
22:30:12Quertyno dice, not displaying nada ;-)
22:30:23HCldid you put the database in your .rockbox?
22:30:37QuertyOuch, that's gotta hurt... No I didn't
22:30:49Quertyfigured would put it where it was needed ;-)
22:30:50 Join andY`fRa [0] (
22:31:01HClno :P its described on the wiki.
22:31:01MoosCamarore HCl
22:32:14Quertyw00t so cool! It works!
22:32:28HClplease search for bugs in it.
22:32:41HClbugs not related to your own id3 tags not being correct
22:33:04Quertyhaha, lol
22:33:14Quertymy id3 tags suck :-/
22:33:26HClwhich is why i added the −−dirisalbum stuff.
22:33:43HClyou can specify that if all your albums are in seperate directories, and your id3 tags aren't all that
22:34:05Quertyoh man, once LinusN finishes codec playback, that's going to be so nice.
22:34:18Quertysounds useful.
22:34:46QuertyI'm creating a "neat" directory with properly tagged files.
22:35:00*HCl plans to retag most of his files properly soon
22:35:12*t0mas has almost all properly tagged files :D
22:35:20HClt0mas: found any bugs yet?
22:35:31t0masused it in eh \file\
22:36:02Querty Search is awesome, I can never find the files I want in the "cruft" directories
22:36:18t0masah... got it
22:36:21t0mastraffic jam
22:36:27HClt0mas: i told you earlier XD
22:36:28t0masbut that sounds to stopped...
22:36:35t0masyeah, I scrolled up ;)
22:36:40QuertyAnd iriver's tagdb sucked soooooo bad
22:36:47t0masyeah, never used it
22:37:02HCli'm wanting to modify the way we browse artists/albums a bit
22:37:20HClso that right = back one letter in the alphabet, and left = up one letter in the alphabet
22:37:24amiconnHCl: I have a little feature request for the script...
22:37:29HClamiconn: mm?
22:37:48amiconnTo complement −−strip, I could use an −−add option
22:38:04HClfeel free to add..
22:38:06HClits in cvs now
22:38:07HCl :)
22:38:08amiconn..which, after −−strip'ping, simply adds a string to the left of the path
22:38:39amiconnThis is useful for the Ondio, in case I want a database of the MMC
22:38:55HClit shouldn't be too hard...
22:38:57amiconnThe MMC is presented under /<MMC1> on the target
22:39:03HCli see
22:39:19amiconn...which is a prohibited name for a mount point
22:39:29amiconn...because of the <>
22:40:11amiconnThat would really make sense for me, having a 2 GB card, and the internal flash being 128 MB only :)
22:40:23QuertyDoes it handle .ogg files yet?
22:40:35amiconnUnfortunately it isn't possible to merge 2 databases...
22:40:54HClits easy to add, hold on..
22:41:11MoosCamaroHCl: re, you need tester again?
22:41:31HClanyone who can test the tagdatabase, please do so
22:41:37HClyou can find instructions on the tagdatabase wiki page
22:42:00MoosCamaroi know instructions
22:42:22MoosCamarolet me update cvs and db
22:42:30andY`fRahi, i got a ihp140 and wonder where to get a bootloader with the firmware in it, i can only find the bootloader, but not with the firmware in it, and as it looks like, you need linux to do so, and i don't have linux, it would be awesome if somebody could help me ;)
22:42:34amiconnHCl: Ogg support should be selectable via an option
22:42:51HClamiconn: done, committed.
22:43:05amiconnI do have some Oggs on my recorder, but since it doesn't play them, I dont want them in the db
22:43:20amiconnThat was fast...
22:43:21HClwe only support id3 mp3s at the moment.
22:43:50amiconnAh yes, we need an additional tag reader lib
22:44:28andY`fRamy problem is, howto descramble, merge, and scramble the bootloader and the firmware
22:44:48Quertynow if we could only get perl to run on rockbox ;-)
22:44:50HClyou can use the one on my ftp. on your own risk.
22:45:15HCli won't guarantee it won't set fire to your player
22:45:31andY`fRait has worked for other players?
22:45:33t0masandY`fRa: do you know how to calculate the md5 sum of the file?
22:45:47andY`fRanope, no idea
22:45:54HClits the one i have on my h140, and if you md5sum it, it should be one of the md5sums on the wiki
22:45:55t0masok, get some tool to do that for you
22:46:00t0masand calculate it for that file
22:46:08t0masand check the wiki :)
22:46:13*preglow prods his firmware patcher
22:46:21andY`fRaok nice :)
22:46:25andY`fRathank you
22:46:26HCloh, hows the firmware patcher?
22:46:30HClshould we release it yet?
22:46:33preglowfinished, more or less
22:46:36HClwe should prolly wait for linus to say its okay o.o
22:46:46preglowi'm going to put it in cvs when i have time
22:46:51preglowbut i need a makefile first
22:47:16preglowmore or less need bagder or something to tell me what the commands i need are called
22:47:46Bagderpreglow: if you add the file to cvs, I could work on that in my end
22:47:46t0mas"bagder or something"
22:48:01preglowBagder: it's several files
22:48:07Bagderadd them all
22:48:09preglowBagder: should probably be stuffed in a dir of its own
22:48:12preglowok, i'll see about it now
22:48:19Bagderyes, make a subdir in tools, imo
22:48:37preglowany preferred dirname?
22:48:47HClfirmwarepatcher ?
22:48:58preglowsounds good enough
22:49:01Bagderuhmmmmm, fwpatcher perhaps
22:49:15Bagderit doesn't really matter to me
22:49:18preglowagain, sounds good
22:50:22preglowperhaps add win32 to the name?
22:50:24 Nick F1^Aison is now known as Aison` (
22:50:34HCl ?
22:50:46Bagdernah, skip the win32
22:50:54Bagderperhaps we'll make one for other oses another day
22:50:56preglowthen i'll just sanity check the code a bit
22:51:01andY`fRak md5 is correct, now i'll get my ihp and flash it :) *nervous*
22:51:04preglowthere is no md5 code in rockbox, no?
22:51:06HCli wrote a bash script one day to do it in linux o.o
22:51:08preglowandY`fRa: good luck
22:51:11Bagderpreglow: nopes
22:51:17HClbut i think i deleted it...
22:51:20preglowBagder: ok, then i bundle along some gpl md5 code
22:51:29Bagdersounds fine
22:51:51Bagderpreglow: if you can then just write down how you build everything, I'll write up a makefile for it
22:52:08Bagderand make the daily build magic etc
22:53:28preglowiriver.c and mkboot.c should be integrated better, but it works as it is
22:54:13MoosCamaropreglow: congrates
22:55:20t0maspreglow: I already did
22:55:23t0masthe md5 thing
22:56:10t0masoh wait... you ment on pc?
22:56:14preglowi meant in cvs
22:56:35preglownevermind, i'll upload it as it is
22:56:42preglowi can't test it now, since i'm not in windows
22:56:48preglowso i wont change anything
22:57:03t0mashm... I can test it tomorrow
22:57:14t0maswhen I see the cvs commit mail
22:57:18preglowanyone got a quick sed command to remove cr on lineshifts?
22:57:35preglowi gotta learn myself how to use sed one day
22:57:36Bagdertr -d '\015' < infile > outfile
22:57:55Bagdernot sed, but anyway :-)
23:01:00rashersed doesn't deal with mulitilines well
23:01:58preglowi ended up doing 'sed -i -e 's/\r//' *'
23:02:03preglowworks in place and seems to work ok
23:02:49rasherthat's me being an idiot, please ignore
23:03:53MoosCamaroHCl: my big db works fine
23:06:43preglowwhen i add to cvs, to i need to add the dir first, then the files?
23:06:52preglowexplains it
23:07:30XShocKHCl: are you here?
23:09:00preglowthere, it's in
23:09:16preglowwindres resource.rc resource.o
23:09:16preglowgcc main.c -o main.exe iriver.c md5.c resource.o -DUNICODE -D_UNICODE -I. -Os -s -fomit-frame-pointer -lmingw32 -mwindows
23:09:21preglowthat is what i build it with
23:09:50preglowthe two unicode defs decide whether the exe will be built with unicode file name support, and requires nt or higher for the resulting exe
23:11:13preglowand no, iriver.c and iriver.h are not straight copies from the ones in the tools dir, i was hoping someone would fix them so the same source code can be used by both my patcher and scramble/descramble :>
23:12:23amiconnHCl: I just tried the new db on my 2 GB card in the Ondio, with the new −−add option. Works nicely :)
23:12:43HClXShocK: yea?
23:12:45HClamiconn: good
23:13:01XShocKi need an update version of rockbox?
23:13:07HClyea, latest cvs
23:13:11Bagderpreglow: it needs a bootloader.bin in the dir too, right?
23:13:16preglowBagder: indeed
23:13:27amiconnCurrently I can only test on the Ondio, my other 2 units need recharging...
23:13:56 Join Camilo [0] (
23:13:59XShocKCan't create /cygdrive/d/Projects/rockbox/cvs/rockbox/target/sysfont.c
23:14:03XShocKhow do i solve it?
23:14:18HCleh.. i have no idea...
23:14:28HCli haven't touched that
23:14:50XShocKi can't build rockbox?
23:15:02HClmake clean? :x
23:15:09preglowyou've quite obviously contaminated the cvs tree with bugs
23:15:09Camilois it a file permissions thing?
23:15:13XShocKtried that..
23:15:32amiconnXShocK: Sounds like a file permission thing
23:15:41Bagderpreglow: you didn't aff favicon.ico binary, did you?
23:15:56andY`fRacan i choose the firmware i want to start when i power up the iriver ?
23:16:11Bagderi586-mingw32msvc-windres: favicon.ico: read of 1384 returned 1378
23:16:50XShocKhmm.. strange..i have sufficient free space.. working in cygwin, but never had any problem
23:17:08BagderandY`fRa: when rockbox is flashed, it loads rockbox by default, or the original if REC is pressed at boot time
23:17:10 Quit lolo-laptop ("Client exiting")
23:17:16andY`fRaah thx
23:17:31preglowBagder: it says i did
23:18:22preglowi'm stupid
23:18:27preglowi ran sed on the icon as well :PPP
23:18:38preglowgimme a sec, i'll update it
23:18:56preglownot used to having binaries lying around in the source code dir
23:18:59XShocKit happened exactly after CONVBDF.
23:19:06XShocKi tried it on another harddisk
23:20:03HClwhats that for anyways?
23:20:14HClyou have an iriver right?
23:20:23preglowBagder: i hope cvs understands that file is a binary, btw?
23:20:26HClyou can grab from
23:20:37preglowBagder: because i didn't think of that either, i'm used to svn understanding that
23:20:45XShocKi have iriver.
23:20:57XShocKbuilding under cygwin
23:20:59Bagderconvbdf makes the system font
23:21:09Bagderit is part of the regular build
23:21:26Bagderpreglow: worked fine now
23:22:31 Quit Harpy (Read error: 60 (Operation timed out))
23:22:58XShocKok.. downloaded from ftp..
23:23:06HClno no, http o.o.
23:23:20XShocKyes.. http :)
23:24:58XShocKhmm. i have an exmpty space
23:25:09HClyou need to put the database in .rockbox
23:25:10HCland reboot
23:25:23XShocKaah.. i put into root.. sorry
23:25:30andY`fRai didn't find any info about rockboy, but if i understood it correct, i only need to download a *.gb file and then select it in the playlist, right ?
23:25:51HClpretty much, yes
23:25:54HClthere's a wikipage on it
23:26:11XShocK:) works
23:26:16preglowBagder: does the exe build fine as well?
23:26:19HClplease find any bugs :)
23:27:40XShocKfound.. :) there are some strange folders... "H" "J" .... M,M,N,P,P
23:27:45Bagderpreglow: sure does
23:27:50Bagdermakefile coming right up
23:27:56HClXShocK: sure they're not caused by bad id3 tags..?
23:28:15*t0mas is going to bed :)
23:28:16HCli have a few of those too, but i figured they're caused by my japanese mp3 songs
23:28:17 Quit t0mas ("goodnight")
23:28:21XShocKhhhm... it might... :)
23:29:04andY`fRawell i found the wikipage, but not really anything usefull, i downloaded the latest rockbox for the iriver and in viewers there is rockboy, but i can't select *.gb files, they are not displayed
23:29:05MoosCamaroXShock: me too, it's little bug
23:29:10XShocKBy the way folders in Russian language shows very nicely.
23:29:29HClandY`fRa: you need to change file view from music to supported
23:29:39andY`fRak thx
23:29:58 Join amiconn_ [0] (
23:30:11Bagderpreglow: can you try running that and see if it pops up as it should?
23:30:18HClmmmm, i need someone with perfect id3tags to try that..
23:30:33 Join ashridah [0] (
23:30:40HCli don't think i'll recognise it as a bug yet o.o
23:31:33XShocKok. i will remove some stuff, but original Iriver does not show anything like that..
23:31:57MoosCamaroXshock: in 1.60 fw
23:32:16preglowBagder: sorry, linux
23:32:17MoosCamarothere was a same bug
23:32:24Bagderah, ok
23:32:48preglow64 bit linux as well, so can't use wine, heh
23:32:48XShocKhmm.. i didn't see it. but i never tried 1.60 either. :)
23:32:48Bagderany windows user feel like trying?
23:33:03Bagderpreglow: I added my makefile anyway, in case you wanna try
23:34:35preglowthe checksums are not updated at all, btw
23:35:58 Join Sucka [0] (
23:38:34Bagderhow's that supposed to be done then?
23:39:26HClby hand, i guess?
23:39:44preglowwell, i can be done manually, i guess we won't update the bootloader too often
23:40:17preglowmd5 sums should be verified to be correct anyway
23:42:25Bagderso do I need to do anything when I build fwpatcher?
23:43:18preglownot if checksums.h is loaded with good md5sums, no
23:46:17 Quit amiconn (Read error: 110 (Connection timed out))
23:46:17 Nick amiconn_ is now known as amiconn (
23:47:13preglowno windows users want to try a cute little exe?
23:48:08ashridahnot if it's that one i've seen that changes nearly every image in windows to be goatse
23:48:24preglowit's my firmware patcher
23:48:27preglowjust to see if it works
23:49:50ehntooshall I try it?
23:49:53ehntooI think i shall.
23:50:00preglowplease do so
23:50:10preglowjust lemme know if it pops up and looks like it's working
23:50:49 Join lostlogic [0] (
23:50:56preglowor more importantly, let bagder know :)
23:51:45ehntoowell, it pops up
23:51:50ehntooit opens the firmware file
23:52:02ehntoowant me to grab an original 1.63 firmware and check md5's?
23:52:14preglowi don't think the md5 sums are correct anymore
23:52:28Bagderno, I put a fresh cvs bootloader.bin there
23:52:32 Join Yossariannz [0] (
23:52:46preglowyeah, and the sums in checksums.h are for the old one
23:52:49preglowunless you've replaced them
23:52:53Bagderbut it could be interesting to see that the check detects it
23:53:01preglowsure, go ahead
23:53:57XShocKi found a bug.. :)
23:54:01ehntooI just realized something
23:54:05ehntooI don't have an md5 checker on here...
23:54:10preglowehntoo: it does it for you
23:54:11HClXShocK: mmm?
23:54:26ehntoo*sigh* all the pains of using windows
23:54:28preglowehntoo: all you need for verified patching is that program
23:55:02ehntoonever expected that
23:55:27ehntoodoesn't patch it
23:55:36preglowwhat does it say?
23:56:04XShocKthat strange folders i told you about. when i go into it, the cursor (that small triangle on the left of the name of selected item) moved to the left, but the scrolling line was not cleared, so there is a mess there
23:56:08ehntoo"checksum doesn't match known good firmware. Download another firmware image and try again."
23:56:10preglowexcellent, all is good
23:56:15preglowehntoo: thanks for testing it out
23:56:31HClXShocK: not entirely sure what you mean...
23:56:44*HCl goes to check on his odd folders
23:56:55XShocKcan i take a sreenshot in the rockbox?
23:57:03HClno idea
23:57:06ehntoothat would be a cool ability.
23:57:37BagderXShocK: yes you can
23:57:38 Quit Yossariannz ("CGI:IRC (EOF)")
23:57:45HCli have exactly 3 odd folders
23:57:46Bagderenable screenshot mode in the debug menu
23:57:52HCland they all seem to be messed tags of 3 files
23:57:54Bagderthen take shots by inserting the usb cable
23:58:02HClbut my database was built with −−dirisalbum
23:58:58HCli need to build a context menu option that displays the taginfo of the tag selected

Previous day | Next day