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 2005-04-22

00:06:54amiconneek - cvs binutils don't build
00:07:10amiconnmake[3]: Entering directory `/home/Administrator/build/binutils/bfd/po'
00:07:10amiconnfile=../../../../binutils-20050421/bfd/po/`echo fr | sed 's,.*/,,'`.gmo \
00:07:10amiconn && rm -f $file && PATH=../src:$PATH msgfmt -o $file ../../../../binutils-20050421/bfd/po/fr.po
00:07:10DBUGEnqueued KICK amiconn
00:07:10amiconnmsgfmt: not found
00:07:26preglowi've got it
00:07:29rasherdamn french
00:07:49amiconnwth is msgfmt ??
00:08:16rashermessage format?
00:09:33preglow msgfmt - compile message catalog to binary format
00:10:01amiconnmsgfmt: command not found
00:10:20amiconnFound out it's part of the 'gettext' package
00:11:22amiconnHmm, strange. gettext is installed according to cygwin-setup
00:15:22amiconnAh, I needed gettext-devel
00:20:52 Quit ashridah ("Leaving")
00:21:39***Saving seen data "./dancer.seen"
00:38:47 Quit Sucka ("a bird in the bush is worth two in your house")
00:42:53 Join TCK [0] (
01:07:34 Join blackhole1 [0] (~blackhole@
01:08:32blackhole1do anyone of u hav IDA pro edition
01:09:35blackhole1Please reply
01:09:46preglowthis isn't a warez channel, you know
01:10:34blackhole1preglow: sorry but if some of u can help i will be thankful
01:10:57MoosCamaropreglow: you don't sleep?
01:11:05preglowMoosCamaro: apparently, no
01:11:09 Quit courtc (Read error: 110 (Connection timed out))
01:11:17MoosCamaronever you sleep
01:11:25 Join courtc [0] (
01:11:25preglowoh, i sleep plenty, just not right now
01:11:39MoosCamaroit's good for rockbx project
01:11:48MoosCamarobut for your mind?
01:12:09preglowi'm not doing anything rockbox related anyway
01:12:24MoosCamaroa ok
01:14:17preglowno, school work
01:14:27preglowcombined with relaxation ;)
01:15:26MoosCamaroyou work in univesity?
01:18:27blackhole1which one.,..
01:18:40preglownorwegian one, ntnu
01:19:42preglownorwegian acronym, stands for norwegian university of science and technology
01:19:54preglowwhat is it?
01:19:59preglow404, btw
01:20:05blackhole1sorrypasted by mistake
01:20:28blackhole1preglow u r prof of which science
01:20:48MoosCamarostudent, no?
01:21:09blackhole1student ((:
01:21:49preglownono, student
01:21:57preglowand electronics
01:22:08MoosCamarovery good student
01:22:22preglowdepends on how you look at it ;)
01:22:23blackhole1preglow: so u must have good knowlege of asm and disassembly
01:22:33preglowthat i have
01:22:37amiconnWow, I actually successfully built sh-elf-gcc 4.0.0 on cygwin :-) :-)
01:22:44preglowamiconn: tried making any code?
01:22:53amiconnI'll do that immed
01:23:00preglowamiconn: i'm pretty anxious to see how it performs
01:23:18blackhole1preglow: did u try making some kind of robot
01:23:26preglowi actually think i'll just go ahead and build a cross compiler for 68k
01:23:27amiconnBtw, building gcc itself took ~40 minutes
01:23:37rasherget a real OS :)
01:23:51preglowblackhole1: well, no, not really
01:24:23blackhole1amiconn: u doing some kind of LFS
01:24:44amiconnNo, building the sh1 cross compiler on cygwin
01:25:23preglowdamn, one would have thought they actually tried propagating these new releases to the mirrors fast
01:25:26preglowbut no, not so
01:25:50blackhole1amiconn: why are u buliding them...
01:26:00 Quit matsl (Remote closed the connection)
01:26:06amiconnpreglow: Building rockbox failed :((
01:26:10preglowon what?=
01:26:23blackhole1amiconn: can u explain me what rockbox is...
01:26:35blackhole1and why r u building sh1
01:26:40amiconnFirst, a ton of new warnings like "drivers/lcd-recorder.c:404: warning: pointer targets in passing argument 1 of 'strlen' differ in signedness"
01:27:01amiconnThen "drivers/power.c:78: error: impossible constraint in 'asm'"
01:27:10preglowblackhole1: what're you doing here if you don't know what rockbox is?
01:27:24amiconn...together with a truckload of asm constraint warnings
01:27:29preglowamiconn: then we have a fun porting project on our hands :PPPP
01:27:33blackhole1trying to find what rockbox is
01:27:45preglowblackhole1: rockbox is a opensoruce firmware for a bunch of mp3 players
01:28:11blackhole1preglow: how can i contribute to it
01:28:15preglowby programmin :D
01:28:22blackhole1in which lang
01:28:30blackhole1i m good in C n c++
01:28:32preglowc and asm
01:28:42preglowmostly c, depends on what you do
01:29:00blackhole1what can be the easiest part to start from
01:29:17rasheronly hardware work and really time-intensive stuff needs asm (like codecs)
01:29:36rasherthe easiest part is by far the plugins
01:29:50rasherbut you don't get your hands dirty much there
01:29:58rasherbut I guess it's still a good place to start
01:30:07rasherwhy what?
01:30:37blackhole1tell me should i read all documentation to understand the proj
01:30:54blackhole1means i will need a liitle bit of help as how can i start
01:31:56blackhole1rasher: ..?
01:32:43blackhole1rasher: hello
01:32:46blackhole1preglow: hello
01:33:14preglowtake it easy, we don't always respons in ten seconds
01:34:14amiconnpreglow: It seems they changed the integer constant constratint format. Again!! Grr
01:34:15blackhole1see i m a computer engginer in india i m looking to contribute to some open source project if anyone of u can help on how can i start i will b thankful
01:35:08preglowamiconn: how nice of them
01:35:14preglowi'll try out gcc4 for m68k now
01:36:55 Join ehntoo [0] (
01:36:57rasherblackhole1: I guess the best way to start is to play around with Rockbox.. and then find something you want to add/change or even remove
01:37:15rasherjust figuring out how to do that should give you an idea of the project
01:38:46rasherdo you have a supported player?
01:39:08blackhole1what u mean by that
01:39:16blackhole1i use xmms in linux
01:39:43rasherHm.. :-/ do you know what rockbox does?
01:40:14blackhole1i just joined this room as some of my frd told me this is growing proj and u can contribute to it
01:40:27rasherI suggest you read up about it :)
01:40:45blackhole1documentation is quite long
01:40:48rasheryou can, but it almost requires you to have a supported digital music player
01:41:17ehntooor a play that's in the prcess of porting
01:41:23ehntoothen you can REALLY contribute
01:41:40ehntoo<3 my iRiver.
01:42:18preglowbinutils fails on _MAKE INSTALL_ of all things
01:42:30blackhole1what are Jukebox 6000
01:42:38blackhole1is they some kind of mp3 players
01:42:53blackhole1so rockbox works on them only
01:43:08MoosCamarogood night(or other) all
01:43:09preglowit works on a number of other archos players as well
01:43:21MoosCamaropower and courage
01:43:21preglowand we're porting it to iriver's h1x0 line of players right now
01:43:24blackhole1what u mean by archos players
01:43:28ehntooand the iRiver h1xx and Archos gmini are in progress
01:43:37ehntoogoogle, ph00l.
01:43:53blackhole1is archos some kind of company players..
01:44:00ehntoojust google it
01:44:07blackhole1yeh i got same
01:44:25blackhole1so to test rockbox u need some this kind of player
01:44:37 Part MoosCamaro
01:45:20ehntoowhoa. we've got a wikipedia entry
01:47:24blackhole1ehntoo: i got what exactly is rockbox
01:47:46ehntoorockbox, in a sentence, is this:
01:48:07ehntooRockbox is a third party firmware aimed at increasing and improving the capabilities of your existing mp3 player
01:48:23amiconnpreglow: Seems that the constraint format is still the same... so I'm puzzled why gcc suddenly complains
01:49:01blackhole1ehntoo: so i can build rockbox for my mp3 player
01:49:42ehntooit depends on your mp3 player
01:49:43preglowdepends on what your mp3 player is ;)
01:50:25blackhole1which kind of mp3 player can be made supported by using rockbox or bulding it
01:51:39amiconnpreglow: btw, gcc is *right* about the signedness warnings about pointer targets in assigments
01:52:03amiconn...and we've got at ton of 'em, even in the small part of rockbox it compiled before the error
01:52:54rasherwell, as long as they're legitimate warnings
01:52:58rasherand it still builds
01:53:00preglowblackhole1:, they're all listed there
01:53:09preglowamiconn: what signedness?
01:53:18preglowamiconn: i don't get the warning
01:53:31amiconnNice example: Look at firmware/common/memcmp.c, especially lines 98 and 99
01:53:33CtcpIgnored 1 channel CTCP requests in 0 seconds at the last flood
01:53:33*preglow tells gcc make to hurry already
01:53:54ehntooI hate waiting for compiles to finish.
01:54:03amiconns1 and s2 are pointers to unsigned char, yet a1 and a2 are casted to char* ....
01:54:39amiconngcc says this:
01:54:40amiconncommon/memcmp.c: In function 'memcmp':
01:54:40amiconncommon/memcmp.c:98: warning: pointer targets in assignment differ in signedness
01:54:40amiconncommon/memcmp.c:99: warning: pointer targets in assignment differ in signedness
01:54:41preglowlike that
01:54:54preglowwell, that should be fixed
01:54:58preglowit's a valid warning
01:55:28preglowghah, assembler error
01:55:32amiconnI got like 20 of them, although the compile already stopped at power.c
01:56:16amiconnDo you know what switches I need to tell gcc to keep intermediates?
01:56:32amiconns/intermediates/intermediate files/
01:56:46amiconnI'm especially interested in the .s
01:56:47preglowwell, this is a bitch
01:56:52preglowi too need newer binutils
01:56:55preglowand i can't install them
01:57:13preglowthat'll just produce an asm file, though
01:57:20preglowit won't keep them, it'll just never assemble it
01:57:39amiconnHmm, this likely won't work
01:57:49preglowit'll work up to the first linker step ;)
01:58:01amiconnI vaguely remember there is a switch...
01:58:46preglowwhy, thanks
01:58:48preglowi'll remember that
01:59:07ehntoono you won't... =P
01:59:46preglowwhy the bloody hell does make install need to compile stuff?
02:00:59rasherthat's curious
02:02:48amiconnMrf... -save-temps doesn't work
02:03:19 Quit blackhole1 ()
02:05:13preglowi can't make gcc4
02:05:34rasherThis ahx seems intriguing
02:05:39rashershame that the code is c++
02:06:13rasher"an average AHX tune has a length of about 200 bytes - 5 kbytes (unpacked)."
02:06:19rashersounds like fun
02:07:09preglowc++ should be ok with rockbox
02:07:14preglowas long as a c interface is exported
02:07:22rasherno idea about that
02:07:29preglowwell, you'll just make it
02:08:33rasherplus the available code appears to be an xmms plugin >_<
02:09:19preglowseems like a cute format
02:09:50rasherthe xmms-ahx tarball contains an xmms and an ahx dir
02:09:56rasherthe ahx dir has one (1) cpp file
02:10:07rashercalled, amusingly, Windows.cpp
02:10:34rasherported over from a winamp plugin it seems
02:10:54preglowbut yeah
02:10:59preglowthat should be very rockbox friendly
02:11:06preglowno ram use, just synthesis
02:11:06rasherah there.. xmms/AHX.cpp I think
02:11:24rasherI wonder if that's it
02:11:36ehntoosounds like it'd be it.
02:11:42preglowi'll try compiling this shit on this machine, then
02:12:00rasher"The core being only C++ without system calls, it needs only some glue to adapt to specific players like xmms."
02:13:49rasherhaha.. from the ahx page:
02:13:53rasher26-Nov-98: AHX v2.3d-SP3 released (complete release)
02:13:56rasherlatest news item
02:14:37preglowwhy, oh why, can't people keep their cvs trees working
02:15:14ehntoo"Finally the 68000-Version of the Editor is out" Hmmm... that'd help as far as processor porting.
02:15:33rasherexcept that bit doesn't appear to be out under PL
02:15:38preglowwe're not porting the tracker itself :PP
02:15:45ehntoogood point.
02:15:53ehntooI should prolly start reading before I copy and paste
02:15:58 Quit cYmen ("zZz")
02:16:05preglowthen again
02:16:13preglowhaving a tracker on h1x0 would be _very_ cool
02:16:14rasherthat'd be a first )
02:16:17preglowvery little sound dj-ish
02:16:18ehntooit would be.
02:16:37ehntoothat would be a sweet thing to have.
02:16:38preglowoh my god
02:16:41preglowrockbox and little sound dj
02:16:45preglowrockboy <-
02:18:06rasherwhat the...
02:18:15rasherThis plugin is under GPL. For a long time I couldn't
02:18:16rasherdistribute the core AHX source, but now the authors
02:18:16rasheragreed to open the source, with the following license:
02:18:16DBUGEnqueued KICK rasher
02:18:16rasher> do whatever you like with the source codes, BUT:
02:18:16rasher> - DON'T sell them
02:18:18rasher> - DON'T claim they're yours.
02:18:20rasher> - DON'T use commercially.
02:18:23rasher> - bla..bla..
02:19:14rasherThat's... rich
02:20:24preglowof, how blissfully fast this box compiles
02:21:43***Saving seen data "./dancer.seen"
02:22:14rasherI guess ahx support is unpossible then
02:23:34CtcpIgnored 1 channel CTCP requests in 0 seconds at the last flood
02:23:34*rasher sighs
02:23:43amiconnHmpf. I managed to make gcc -S produce an output file
02:23:46rasherI wonder if mailing them would do any good
02:23:56preglowthere, older binutils in
02:24:02*rasher nominates amiconn to do so
02:24:09amiconnHowever, this doesn't contain the opcode that it complains about!
02:27:21amiconnIt seems gcc4 actually checks asm statements itself before passing them to gas
02:27:36preglow../../gcc/libgcc2.c:1256: internal compiler error: in do_SUBST, at combine.c:462
02:27:52amiconnHowever, we use some instructions that gcc doesn't use itself...
02:28:00preglowamiconn: i find that hard to believe
02:28:16amiconnpower.c, line 78:
02:28:22amiconnor_b(0x20, &PBIORL); /* Set charging control bit to output */
02:28:42amiconnsystem.h, line 82ff:
02:28:43amiconn#define or_b(mask, address) \
02:28:43amiconn asm \
02:28:43amiconn ("or.b\t%0,@(r0,gbr)" \
02:28:43amiconn : \
02:28:43***Alert Mode level 1
02:28:43amiconn : /* %0 */ I_CONSTRAINT((char)(mask)), \
02:28:44***Alert Mode level 2
02:28:44amiconn /* %1 */ "z"(address-GBR))
02:29:17amiconnThere is no or.b instruction in the intermediate power.s!
02:30:10amiconngcc says:
02:30:11amiconnpower.c: In function 'power_init':
02:30:11amiconnpower.c:78: warning: asm operand 0 probably doesn't match constraints
02:30:11***Alert Mode level 3
02:30:11amiconnpower.c:78: error: impossible constraint in 'asm'
02:30:38amiconnBtw, that behaviour is very different from gcc 3.x
02:30:56preglowvery possible
02:30:57preglowbut yes
02:31:05amiconnWith gcc 3.x, I got the actual warnings from gas, referring to some cryptic intermediate .s files
02:31:09preglowgcc 4.0.0 is very much a primary and secondary platform release
02:32:18preglowi find it very, very hard to believe that gcc knows anything at all about the contents of asm statements
02:32:23preglowapart from the constraints, that is
02:32:32preglowwhich reflect directly gcc internals
02:33:01rasher>< .. so.. ahx source is NOT gpl.. and the xmms-ahx is basically claiming to be GPL but isn't
02:33:19rasherAnywhere we go, license problems jump at us and bite us in the face
02:33:29preglowwell, let's see if gcc4 works better on a 32 bit machine
02:35:56ehntooin the words of old what's his name from the first Macintosh development team
02:36:05ehntoo"There would be legal problems, but we could just ignore them."
02:37:49rasherThat'd be fairly stupid
02:38:02rashersince we rely on people to respect our license terms
02:38:03 Quit micoo (Read error: 110 (Connection timed out))
02:39:19preglowthey're corporate, we're not
02:39:47ehntooit was a joke, you see.
02:39:50preglowthey more or less have to think that way to survive
02:40:03amiconnpreglow: Eeek - seems I found the problem - and it's not gcc's fault
02:40:03preglowlots of open source people ignore licensing issues
02:40:09preglowamiconn: spill it
02:40:12***Alert Mode OFF
02:40:18amiconn#if (__GNUC__ >= 3) && (__GNUC_MINOR__ > 3)
02:40:27preglowin what source?
02:40:28amiconn.. is FALSE for gcc 4.0
02:40:47preglowhow can that be false?
02:40:53preglowof course
02:41:13amiconnI need to add || __GNUC__ >= 4
02:41:27*ehntoo wishes I knew C in order to assist in development
02:41:48preglowi've got gcc4
02:42:00preglowehntoo: never too late to learn
02:42:29amiconnpreglow: You first (unsuccesful) try was on your amd64 machine?
02:43:22preglowamiconn: my first was with their obviously broken binutils cvs, my second on amd64
02:43:29preglowamiconn: and that bombed out with an internal error
02:43:49preglowamiconn: in gcc 4, during bootstrapping, it looks like
02:44:19preglow#if (__GNUC__ >= 3) && (__GNUC_MINOR__ > 3) || (__GNUC__ >= 4)
02:44:21preglowthat'll do?
02:44:36amiconnBleh. Now the build bails out by massive errors in system.c :(
02:44:50amiconnsystem.c:569: error: 'GII' aliased to undefined symbol 'UIE4'
02:45:00amiconn...and a whole ton of them
02:45:27*preglow sees all the errors
02:45:29*preglow vomits
02:45:47preglowrecorder/peakmeter.c:59: error: static declaration of ‘peak_meter_range_min’ follows non-static declaration
02:45:50preglowthis is where i die
02:45:59preglowand god DAMN, the sheer number of warnings is staggering
02:46:02amiconnI actually read about more strict checking on symbol aliasing, but didn't think we use that
02:46:42rasherswitching the builds over to gcc4 should produce an amusing daily build matrix
02:46:51preglowamiconn: there is an incredible amount of signedness warnings
02:46:58amiconnpreglow: yup
02:48:40preglowbut yeah, what kills me is that a static variable is declared extern in the header
02:48:45preglowwhich is quite obviously not entirely correct
02:49:32amiconnYeah, fun. It's puzzling though that gcc 3.x doesn't complain
02:50:19amiconnThe aliasing errors are sh specific (of course)
02:50:34preglowwhy define everything as 'extern' in a header?P
02:50:37preglowit's default anyway
02:54:46preglowmy synth_full opt exhausts the register store
02:55:30preglowand o
02:55:37preglowand i'm not about to rewrite that in pure asm now
02:55:48amiconnGah! Seems I can't solve the 'alias to undefined symbol' problem
02:55:59preglowi didn't even understand it
02:56:03amiconnThe actual symbols are defined in asm...
02:56:27 Quit Aison ("( :: NoNameScript 3.72 :: )")
02:57:02amiconnC doesn't know they exist. Declaring them is obvoiusly not sufficient
02:57:36preglowc shouldn't care, it's a linker job
02:57:55preglowbut i don't exactly get what's happening there anyway
02:58:19amiconnYes... but read , section "New Languages and Language specific improvements"
02:58:36amiconnThis part is the initialisation of the exception vector table
02:59:30preglowahh, it's that kind of an alias
03:02:17preglowoh well, looks like an asm rewrite of synth_full is in order tomorrow, then
03:02:25preglowthink i'll call it a night now
03:02:35rashersounds like a plan
03:02:52preglowi actually think it's more correct to call it a morning
03:03:22 Join amiconn_ [0] (
03:03:54 Quit amiconn (Nick collision from services.)
03:03:55 Nick amiconn_ is now known as amiconn (
03:04:06amiconnSeems this exception vector thing is done differently for coldfire
03:06:23preglowwell, apparently
03:06:25preglowit built fine here
03:06:35amiconn...better, afaiu
03:06:38preglowit got as far as to libmad before it stumbled
03:09:08preglowbut yeah, bedtime
03:09:13preglowsee you later
03:09:22 Quit preglow ("ni!")
03:47:28 Join amiconn_ [0] (
03:50:02 Join DMJC [0] (
04:05:44 Join QT_ [0] (as@area51.users.madwifi)
04:06:03 Quit amiconn (Read error: 110 (Connection timed out))
04:06:03 Nick amiconn_ is now known as amiconn (
04:11:02 Quit ehntoo ("Leaving")
04:13:13 Join Byron [0] (byron@
04:13:27Byrondoes the Archos pda thingy have an ftp client anywhere?
04:17:19 Quit QT (Connection timed out)
04:21:45***Saving seen data "./dancer.seen"
04:24:10 Quit StrathAFK (Read error: 110 (Connection timed out))
04:31:05 Part Byron
04:51:47 Quit DMJC (Read error: 110 (Connection timed out))
05:04:38 Join Strath [0] (
05:34:30 Quit tvelocity[away] (Read error: 110 (Connection timed out))
05:35:38 Join tvelocity[away] [0] (
06:07:56 Join webguest61 [0] (
06:08:43 Quit webguest61 (Client Quit)
06:21:48***Saving seen data "./dancer.seen"
06:25:51 Join odd [0] (
06:26:07oddanyone around?
06:27:07*odd is reading about the conflict between the rockbox-iriver team and the iriverlounge folks
06:27:12oddso ridiculous
06:27:24oddi don't understand why they're so protective of their info
06:27:34oddit's like they don't even understand the concept of free software
06:27:41oddfree as in freedom, that is ;)
06:27:57Strathlink? :)
06:31:55Strathidiots :)
06:33:23Strath(statements about bush era crud...)
06:33:32oddyeah that was just absurd, heh
06:33:47oddi'm about as anti-bush as anyone can be, and i still could only shake my head and laugh at that
06:34:34oddeventually they simply say that the wiki is taking away hits from their board... but if the only reason people were visiting their board was to get hardware info about the iriver, then perhaps they have some bigger problems to deal with
06:35:17oddi've never seen such disgusting conduct from hackers/modders/OSS-developers
06:39:33Strathi host the site containing a majority of the hardware info relating to the gmini. It was nice that someone asked first before copying information over to the rockbox wiki. So i can understand why they would be a bit irritated if they happen to find thier research 'stolen'
06:41:35oddi can surely understand too... but it was a simple mistake, and the information was public anyway... i mean seriously, forgive and forget - it's all for the greater good
06:42:06Strathright.. :)
06:53:02 Join LinusN [0] (
07:15:17Strathgah... annoying egotism to the effect of: "we knew it first, credit us even if we forced you to duplicate the effort"
07:17:21LinusNi agree with them about how the "theft" was done
07:17:58LinusNthe first wiki pages contained very little credit to konrad and his team
07:18:30oddyeah, and that was wrong
07:18:38LinusNthe sad part is that they continued accusing us of theft even when i published my own research
07:18:44oddbut they should've just said "hey guys, can you add a little credit line and maybe a hyperlink back tot he source?"
07:19:25Strathya, i understand that and was glad that i had been asked before gmini information was duplicated in the rockbox wiki
07:19:52LinusNthat's how we normally do it
07:20:29LinusNin the iriver case, we were approached by the misticriver guys who asked what they could do to make rockbox happen on the iriver
07:20:49LinusNwe replied that they should gather as much info as possible in the wiki
07:20:53Strath(a third group?)
07:20:58LinusNand it went on from there
07:26:00Strathit's just my first time reading that exchange, i agree with odd that he just doesn't(didn't) seem to understand open collaboration, "free as in freedom" :) though the burned ego surely had a large part to do with that...
07:31:16LinusNwow! i just realized that it has been exactly 1000 days since we started the rockbox irc channel
07:32:18LinusNor rather, 1000 days since we started logging
07:32:22Strathwoohoo! :)
07:33:04LinusNnah, the stats are lying
07:33:25LinusNit's been a lot longer
07:33:30Strathhow long had the channel been active before loggin started?
07:35:32LinusNdon't remember
07:36:07Strathawh.... i only have 2196 lines :)
07:37:27Straththe "Big numbers" section is fun :)
07:37:29LinusNlooks like we logged it from the beginning, or a few days after
07:38:49LinusNit's funny to read those really old logs
07:39:09LinusNhistory kind of repeats itself
07:39:41LinusNi do the low-level hacking, and people waiting for it so they can start hacking the high-level stuff
07:41:12Strathwhich would still really could think about there RIGHT? :)
07:48:19 Join ashridah [0] (
07:48:48dwihnoFinally friday!
08:21:50***No seen item changed, no save performed.
08:31:41 Join Harpy [0] (
08:34:19 Quit tvelocity[away] ("Leaving")
08:45:02 Quit edx (Read error: 110 (Connection timed out))
08:50:38BagderLinusN: we've logged more than 1000 days actually as the irc stat is the dancer logs only
08:51:06LinusNsaw that
08:51:18LinusN<LinusN> nah, the stats are lying
08:51:19LinusN<LinusN> it's been a lot longer
08:53:09Bagderah, right
08:53:35Bagder1039 days in fact
09:00:58 Join einhirn [0] (
09:07:35 Nick Lynx_awy is now known as Lynx_ (HydraIRC@
09:11:46ashridahokay. i thought gcc 4 was kinda a ways off yet.
09:19:57 Nick QT_ is now known as QT (as@area51.users.madwifi)
09:43:46 Quit Stryke` ("Friends don't let friends listen to Anti-Flag")
09:48:43 Join Chamois [0] (
09:57:35 Quit ashridah ("Leaving")
09:58:45 Join webguest87 [0] (
10:06:11 Join mdeboer [0] (
10:21:51***Saving seen data "./dancer.seen"
10:24:44 Join edx [0] (
10:27:17 Join Patr3ck [0] (
10:32:04 Join micoo [0] (
10:35:54 Join bobTHC [0] (
10:59:01 Join ashridah [0] (
11:10:15 Join kurzhaarrocker [0] (
11:11:28*kurzhaarrocker searches schematics of a toaster
11:11:44kurzhaarrockerI want to port rockbox to a toaster and burn mp3s on toast
11:12:19LinusNyou could toast your mp3 player
11:12:33Bagderthat hurt
11:12:56LinusNbad jokes'r'us
11:13:28kurzhaarrockertoasting recorders is easy: just reverse the polarity of the power in.
11:16:46dwihnoBeen there, seen that :(
11:19:12LinusNrepaired that :-)
11:21:51rasherwow, that iriverlounge thread seems.. childish
11:22:35kurzhaarrockernah, they're not childish. They don't make jokes about toasters.
11:27:57 Join ferenczy [0] (
11:28:01rasherThis is true
11:28:21BagderI think they just weren't up to the open source way of doing things
11:28:47amiconnhi all
11:29:56dwihnoLinusN: You realize you pushed my iriver purchase further into the future by fixing it? :)
11:31:36oddare there any reports of a successful installation of one of those new 60GB 1.8" HDD's in an iriver H-140 yet?
11:32:40LinusNnot that i know of
11:33:07Bagderodd: you can be the first!
11:33:26amiconnLinusN: Did you read about my attempts to compile rockbox with gcc 4.0.0?
11:33:28oddthe drive is too expensive =\
11:33:30oddthough i am tempted
11:33:35oddand i've got a bit of disposable cash
11:33:37LinusNamiconn: yes
11:34:02amiconnImho the restriction of the alias symbol definitions is silly
11:34:23amiconnI'm tempted to file a bug report for gcc
11:34:57oddyou should if it's an issue for you
11:35:27dwihnoLinusN: but with mp3 playback in the works, I need to get a hold of a ihp-140 asap :)
11:36:03LinusNamiconn: i didn't read the irc log thoroughly
11:41:48Bagderso what is the status for rockbox and gcc4?
11:41:52HClodd: no, but someone had a successful installation of a normal 3.5" 60gb hdd by adapting a cf connector for the iriver
11:41:57LinusNdoesn't compile
11:44:16 Join sox [0] (
11:45:54amiconnBagder: (1) A ton of new warnings, regarding signedness of pointer target types (2) SH!: Errors out at system.c because of the symbol aliases to symbols defined within the inline asm
11:47:23amiconn(2) Is the silly thing in gcc4 I was talking about
11:51:08amiconnLinusN: Btw, your building instructions are good :) There are only 2 additional remarks for building gcc4
11:51:34HClwhat about building for iriver?
11:51:39 Join Aison [0] (
11:51:51amiconn(1) You need the gettext-devel package (at least on cygwin) (2) binutils 2.15 is not sufficient
11:52:02HClso it compiles? o.o
11:52:14HCl2.15 wasn't sufficient already
11:52:18Bagderbinutils 2.15 never was sufficient on iriver
11:52:32*Bagder is an echo ;-)
11:52:37LinusN2.16 is released Real Soon Now
11:52:49Bagderits about time
11:52:53Bagderbeen a year since 2.15
11:53:30amiconnHCl: I'm talking about SH target here
11:53:39Bagderaha, on sh as well
11:53:41HClamiconn: yea, thats why i asked
11:53:56HClhas anyone tried it for iriver ? o.o
11:55:21amiconnHCl: preglow did
11:55:58amiconnIt goes beyond system.c (the exception handling is different for coldfire). It errored in libmad instead
11:56:40HCli see
11:56:52*HCl yawns and guesses he should try
11:58:58amiconn[02:55:45] <preglow> gHAH
11:58:58amiconn[02:56:00] <preglow> my synth_full opt exhausts the register store
11:58:58amiconn[02:56:44] <preglow> and o
11:58:58DBUGEnqueued KICK amiconn
11:58:58amiconn[02:56:51] <preglow> and i'm not about to rewrite that in pure asm now
12:00:56CtcpIgnored 1 channel CTCP requests in 0 seconds at the last flood
12:00:56*LinusN doesn't feel gcc4 is that important
12:01:55HCli'm mostly just curious to whether it'll be any faster.
12:02:44HClif you look at the speed difference in code between 2.9.x and 3.x when 3 was just out
12:03:31LinusNwe should of course test with gcc4, but i don't feel it's that important to port to it right now
12:03:57BagderI agree
12:04:21Bagderbut we will get questions and reports from others who try
12:04:53LinusNunfortunately, yes
12:05:03*LinusN edits the wiki
12:16:20*LinusN goes to lunch
12:16:21 Join MoosCamaro [0] (
12:16:31MoosCamarohey all
12:21:54***No seen item changed, no save performed.
12:23:57 Join idlewild [0] (
12:27:01 Join preglow [0] (
12:27:27preglowwhy, hello
12:27:56Strathbecause you're here
12:28:06MoosCamarohi preglow
12:34:55 Join idlewild_ [0] (~idlewild@
12:36:02 Part idlewild
12:41:06HClhey preglow
12:41:29HClpreglow: did you try gcc4 with iriver?
12:43:37preglowone of my opts broke it
12:43:54preglowit uses too many registers
12:44:19HClthink you can send me a copy of gcc4 ?
12:44:24HClcompiled, ofcourse o.o
12:44:58 Nick Strath is now known as StrathAFK (
12:49:02preglowwhy? it's a breeze to compile
12:49:39preglowi can send you my new cross compiler tree, but that's got binutils in it as well
12:49:53HCli don't really have much cpu
12:50:28HClbut yea, please o.o
12:51:13idlewild_does anyone use os x? I'm having trouble compiling the cvs binutils
12:54:20idlewild_or anyone who's familiar with gnu compiling errors...
12:54:34HClwhats the error?
12:54:35Bagderwhat does it say?
12:55:03idlewild_In file included from ../src/binutils/size.c:33:
12:55:04idlewild_../src/include/libiberty.h:94: warning: function declaration isn't a prototype
12:55:04idlewild_make[3]: *** [size.o] Error 1
12:55:25idlewild_make[2]: *** [all-recursive] Error 1
12:55:26idlewild_make[1]: *** [all-recursive-am] Error 2
12:55:26idlewild_make: *** [all-binutils] Error 2
12:55:38HClidlewild_: thats not the error.
12:55:48HClwarnings aren't errors, search for where it says error
12:55:50HClpreglow: thanks
12:56:07idlewild_yeah, confused me, cause I couldn't see an error
12:56:08HCl1mb/s :)
12:56:54HClwell, i suggest you try to find the error..
12:57:38idlewild_is there a reason why it wouldn't print out the error?
12:57:53*LinusN can't understand why it's so important to use gcc4
12:58:09HClLinusN: its not important, i'm just curious to the speed difference.
12:59:42preglowLinusN: you honestly don't think new people are going to be using gcc4 after a while?
13:00:09preglowresponding with "you need to downgrade your compiler" isn't a good start
13:00:34LinusNno, the response is "go ahead and fix the rockbox code if you want to run gcc4"
13:00:42preglowahh, but we'll be doing that first :P
13:01:10preglowthere aren't any fatal errors that i could find
13:01:15LinusNi have better things to do than tend to version-junkies
13:01:19preglowapart from the ones i'm responsible for, heh
13:01:54HCl :P
13:02:01HCli'm just mostly hoping it'll speed-improve
13:02:02preglowanywho, if gcc4 can make the codecs faster for me, i won't complain
13:02:09HCl2.9 -> 3 was about a 10% boost
13:02:31HClthat was a lot. if you looked at it system wide
13:03:26HClwow. talk about warnings, heh.
13:03:45 Quit idlewild_ ()
13:04:38preglowHCl: as i said, you can't compile it yet
13:04:55preglowthe signedness warnings don't stop you, but synth_full not having enough registers does, heh
13:06:34HClpreglow: i'm getting an additional error
13:06:56HClrecorder/peakmeter.c:59: error: static declaration of 'peak_meter_range_min' follows non-static declaration
13:06:59HClrecorder/peakmeter.h:79: error: previous declaration of 'peak_meter_range_min' was here
13:08:56preglowHCl: i fixed that, just remove the static decls for those variables
13:09:05preglowthey're wrong anyway, i wonder why gcc 3.3 didn't complain
13:13:52*HCl skips libmad, tries to continue
13:14:55HClwhats musepack for a codec anyways?
13:15:35preglowdoes it destroy anything?
13:16:40HClonly thing that broke is libmad so far
13:16:55*HCl is gonna try it
13:20:18*HCl prepares for it to crash horribly and set fire to his iriver
13:20:56HClIllInstr at 00000000
13:26:10kurzhaarrocker<- never really understood static; sorry for that peak_meter_range stuff
13:26:23preglowkurzhaarrocker: static means that it's internal to the file
13:26:40LinusN*if* the symbol is global
13:26:47preglowkurzhaarrocker: it's mostly ok, just two variables labeled static that are labeled extern in the header
13:26:53preglowkurzhaarrocker: and yeah, what linus says ;)
13:26:55kurzhaarrockerIsn't there something else? Concerning memory?
13:27:42LinusNkurzhaarrocker: the "static" keyword is mostly used to keep the value of a local variable static, i.e it remains between calls
13:28:20LinusNthis has the effect that the variable ends up in the .data or .bss segment instead of on the stack
13:28:28kurzhaarrockerah. That's what was dangling in my mind. static function local variables keep their value between calls, right?
13:28:48preglowthey're more or less global variables in disguise
13:29:16LinusNthe name of the static variable has to be "hidden", so it still behaves like a global variable
13:29:44LinusNthis has the side effect that the "static" keyword can be used to "hide" global symbols
13:30:21kurzhaarrockercool. Next lesson would be when and why to use const...
13:30:33 Quit Hadaka (Read error: 60 (Operation timed out))
13:30:48LinusNwhen you want to tell the compiler that the value of the variable isn't to be changed
13:31:04LinusNit serves two purposes:
13:31:13preglowi really don't ever use const unless i use c++, for some reason
13:31:22LinusN1) the compiler warns you when your code writes to it
13:31:33LinusN2) the compiler can optimize better
13:31:59kurzhaarrockerAh and 2) is the reason why the rockbox code uses it wherever possible?
13:32:38LinusN3) it makes sure that the data is placed in the .text segment instead of .data
13:33:07LinusN(when the const data is initialized in the declaration)
13:33:18LinusNthat saves memory
13:33:58LinusNif the data isn't to be changed, it is unnecessary to have it in RAM
13:34:14LinusNit can be in ROM, saving RAM memory
13:34:26kurzhaarrockerso the .text segment is placed in rom?
13:34:44LinusNif you run rombox, that is
13:34:56kurzhaarrockerAre there other segments? Like .code=
13:35:02LinusN.text is code
13:35:05LinusNand constant data
13:35:12kurzhaarrockerok is data
13:35:31LinusN.bss is non-initialized data
13:35:55kurzhaarrockerhm. Bull Shit Stuff? :)
13:36:05HCl :P
13:36:23LinusNi don't know what it stands for, actually
13:36:45LinusNBlock Started by Symbol
13:37:14LinusNnow i learned something
13:37:45kurzhaarrockerI hope we don't have to write a test next lesson ...
13:39:22LinusNfyi, the .text segment has a copy of the .data segment, which is copied to RAM at boot
13:39:53LinusNso all .data takes the double amount of RAM when running rockbox in ram
13:40:19kurzhaarrockerah! So maybe a clever implementation to save rom could keep the .data segment compressed and decompress it at startup?
13:41:00LinusNbut making sure that constant data is "const":ed save a lot of ram
13:41:20LinusNthat's what the const'ing is about in rockbox
13:42:17kurzhaarrockerSo if we were neglegting the 'hiding effect' of static: const somehow implies static behaviour.
13:42:46LinusNsort of, yes
13:43:20LinusNbut you can still have const local variabled that aren't static
13:43:46LinusNjust to hint the compiler that the data isn't to be changed
13:50:12 Join idlewild [0] (~idlewild@
13:55:09 Join Nibbler [0] (
13:58:40 Quit kurzhaarrocker (Read error: 145 (Connection timed out))
14:19:13preglowHCl: i take it you've fixed the issue already? :>
14:20:54*rasher flibbles
14:21:55***Saving seen data "./dancer.seen"
14:26:22 Quit lostlogic ("Going to the moon")
14:28:41 Join Patr3ck_ [0] (
14:33:18 Join oxygen77 [0] (
14:35:15 Join oxygen77_ [0] (
14:36:17 Quit oxygen77_ (Client Quit)
14:36:53 Quit oxygen77 (Client Quit)
14:37:18preglowi want more registers, mom
14:37:46 Quit Patr3ck (Read error: 110 (Connection timed out))
14:41:01HClpreglow: which issue? :x
14:41:36preglowwhere did that occur, btw?
14:41:44preglowtry reflashing with a gcc4 bootloader :P
14:43:18HCl :P
14:43:20HClno thanks :P
14:43:27HClit occured when i tried to load a gcc4 built rockbox
14:43:44rasherdid you make clean?
14:43:45HClnot that that file will be of much use ;P
14:43:48HClrasher: ofcourse.
14:43:59rasherwell you never know :)
14:44:04HCltrue :)
14:44:07HClbut i'm fairly awake today :P
14:48:13preglowbah, i just can't be bothered to rewrite synth_full entirely in asm
14:52:01HCldon't you have the C version?
14:52:04HClor is it already C?
14:52:21HCli say add a define that checks for gcc4 to make it back in C for now
14:54:26preglowif it doesn't even make a valid image for gcc4 yet, i don't think there's a hurry
14:55:44HCland the only one able to fix it into making a valid image is LinusN , i think, and i don't think he'll make it any priority :P
14:56:12LinusNexactly :-)
14:56:17HCl :P
14:58:00preglowi think i should be able to find out why it doesn't work, but i'd rather use my rockbox time for codec opts
14:58:46HCli'm not very interested in putting effort to get gcc4 to work
14:59:00HCleventually we might want to, but its not in a hurry.
14:59:18preglowyou too would rather spend time doing codec opts? that's wonderful news :P
14:59:24HCl :P
14:59:27HClno, other stuff :P
14:59:35HCli'm wanting to look at the runtime db
14:59:44HCli need to read my digest on the suggestions and add them on the wiki
15:01:35*LinusN wants a MELBUS protocol specification
15:02:08preglowwhat is melbus?
15:02:13preglowgoogle wasn't very helpful
15:03:13LinusNmelbus is the cd changer protocol in many volvo cars, it's a misubishi thing
15:03:37LinusNand i want to connect my mp3 player to my v70 volvo
15:03:52LinusNlooks like i have to do it the hard way
15:04:02HCli see.
15:04:10HCli wonder whether that could connect to my dads volvo
15:04:13HClnot that it would interest me must.
15:04:19HClmaybe it would interest my dad
15:04:21LinusNthere is a guy who reverse engineered it, but all he did was to start building and selling his own kits
15:04:32HClgee. how nice of him.
15:04:33rasher o.O
15:04:48LinusNand he is only in "beta testing", selling the units for $230!!!
15:05:54LinusNso i'll need to find a melbus cd changer and connect the logic analyzer...
15:06:08LinusNas if i had time for that... :-(
15:06:15HCl :/
15:06:38LinusNi guess i'll have to live with the fm modulator for the time being
15:06:48*HCl prods amiconn.
15:06:57preglowhow do the fm modulators sound+
15:07:09HClrequest for comments:
15:07:17HClmostly note the last change
15:07:37HClwondering whether there are any suggestions left, otherwise i'm interested in starting to implement it for archos
15:07:51preglowaren't there any car stereos with line in?
15:07:59HClpreglow: you can buy a fake tape.
15:08:03preglowi'm quite convinced i've seen some with line in jacks in the front
15:08:07HClmy old, old, old....
15:08:08HClcar radio
15:08:11HClwith a tape player.
15:08:15 Join lolo-laptop [0] (~lostlogic@
15:08:16HCli just stick a fake tape in it
15:08:17rasherMy parents last one had a line-in
15:08:23LinusNpreglow: there are lots of car stereos with line in
15:08:24HClwith a wire on it and a stereojack plug
15:08:30HCland i just plug that into my iriver
15:08:35rasherthere are some with usb-in as well :D
15:08:42preglowLinusN: then there's the solution! :)
15:08:46HCldidn't someone get an usb in the other day
15:08:54HCland then it wouldn't recognise stuff above 1gb or so..
15:09:02LinusNpreglow: :-)
15:09:06rasheryeah, I read that.. that was silly
15:09:25*HCl considers the fact that his fake tape thing cost 15 bucks and gives perfect quality
15:09:30preglowanywho, i think i'll buy a car first
15:09:48LinusNHCl: my v70 doesn't have a tape player
15:09:58HClLinusN: yea, i know how that goes.. neither does my dad's volvo..
15:10:03HClall cd fancy dolby stuff.
15:10:09HCli'm happy to have an old radio player
15:10:19HClthat accepts a tape without any actual cogs in it
15:10:30HClmy fake tape gets refused by his jaguar's tape player too.
15:10:47zebut then you get the noise of the things spinning
15:10:52HClyes, exactly.
15:10:54HCli took the cogs out of mine.
15:10:57HClfirst thing i did.
15:11:00HClworks like a charm
15:11:07HClbut his jaguar's tape player
15:11:11HClrefuses tapes without cogs
15:11:15LinusNheh, i'm not too keen on ripping the audio system apart either...the car will be delivered new this tuesday
15:11:25preglowthen i can understand that
15:11:35 Join t0mas [0] (
15:11:40LinusNso i won't open it up until maybe thursday :-)
15:11:41preglowi think i'd be far too busy stroking and polishing it
15:12:07LinusNstill, i'm going on a long trip, so i really want the mp3
15:12:11HClLinusN: will yours be with one of those fancy gps systems?
15:12:16HClmy dad has that, its awesome :x
15:12:26LinusNHCl: no, it's pretty vanilla
15:12:36HClyou can even buy a tv module for it to get tv on the gps screen o.o
15:12:40HClwith a remote control too o.o.
15:12:52HClyes :P
15:12:55preglowi really don't fancy too much electronics in cars
15:12:55HClit has an infrared port
15:12:57rasherjust how far are you going to sit away?!
15:12:59HClon top of its gps screen
15:12:59LinusNcheaper to buy the car with no gps and install a 3rd party gps
15:13:03HClrasher: maybe a meter :P
15:13:17HClLinusN: prolly, but its still cool to have the gps controls on the wheel
15:13:28HClyup, my dads volvo is slick :X
15:13:34HCl<.< >.>
15:13:36HClrockbox, yes.
15:13:42LinusNi have a remote control for the radio in my current car
15:13:49LinusNguess how much i used it?
15:13:50HClanyone looked at the altered runtimedb proposal?
15:13:54HCl :P
15:13:58LinusNat all
15:14:12HClprobably just as much as the dashboard control panel in the back of my dads jaguar :P
15:14:14LinusNi looked at it, looks ok to me
15:14:21LinusN(the runtimedb)
15:14:28HClanyone else?
15:15:38t0masin wiki?
15:15:45HClhmm... i have a bit of an open issue..
15:15:49*t0mas is reading
15:15:50HClthat needs to be fixed
15:15:51HClhold on
15:15:52HCllet me add that.
15:16:35 Quit idlewild (Read error: 110 (Connection timed out))
15:17:38t0maswell... it looks nice now...
15:17:47HClread the open issues section
15:17:56HCli need to have that defined before i can start implementing
15:18:00t0masthe fixed size is useble... as it makes reading a lot easyer
15:18:46t0masit is possible that there are 2 different mp3 files sharing the same CRC
15:19:10HClalso, the same hash will have to be in the tagdatabase..
15:19:17rasherdepends how you hash
15:19:25HClrasher: no, even then.
15:19:27HCli know that on my player
15:19:29HCli have a song
15:19:34HClthat exists in multiple directories
15:19:40rasheroh, yes
15:19:42HClwhich is an exact copy.
15:20:17t0mashm... that should share the runtime info...
15:20:19HClamiconns idea that they would share the runtime info was nice. but not possible
15:20:27HClour runtime db format simply does not support it.
15:20:29t0masas it's useless to have two playcounts for the same song
15:20:33HClit insists on seperate runtime infos.
15:20:45t0masmaybe use the CRC as an index?
15:20:56HClwe could add an extra field to the runtime info
15:21:06t0mas(hadn't Bagder a good idea on this? mailinglist?)
15:21:07HClthat says to look elsewhere
15:21:15HCldid we? *looks*
15:23:13HClno suggestions on how to cope with file duplicates yet.
15:23:28*HCl futilely tries to pause the tv
15:23:40HCli always try to do that..
15:24:14HClanyways, what we could do, is add a seperate field
15:24:24t0masghehe buy a HD recorder...
15:24:29HClthat pretty much says "no, the runtime info for this file isn't stored here, its there! *points*"
15:24:30t0masor a tv tuner card for your pc :P
15:24:39HCli have a tv tuner card.
15:24:48t0maswhy don't you timeshift it with that?
15:25:03*t0mas normally does that when eating... :)
15:25:03HClmy cpu isn't fast enough to capture its data realtime.
15:25:20t0mashm... my dad's P3 1.1 can do it...
15:25:23HClanyways, doesn't matter.
15:25:28 Quit webguest87 ("CGI:IRC (EOF)")
15:28:26*HCl points at
15:31:03HClany comments/agreeing/whatever?
15:31:15HCli either need "no it should be that way" or "looks fine"
15:31:16HCl :x
15:32:03 Join Shagnar [0] (
15:33:17HClhey shag
15:33:23HCli updated my grayrockbox today
15:33:27HClif you're interested
15:36:08*HCl prods rasher
15:36:12*HCl prods amiconn
15:36:57rasherbetter add that field in both dbs then
15:37:06rasheror.. well
15:37:22HClno, the tagdb has entries for every file
15:37:30HCland doesn't write to them.
15:37:31rasherwould save a lookup when going from tagdb->runtimedb
15:37:56HClit would have as much lookups o.o.
15:38:05rasherno, wait
15:38:13*HCl sighs. remembers his original design and says nothing :/
15:38:27rasherif you have a tagdb entry, and want to find the runtimedb entry
15:38:39rasherit'd be tagdb->runtimedbentry->runtimedbentry2
15:38:39HClohyea, i hadn't even considered that yet.
15:38:44HClmy proposal would create problems.
15:38:51ShagnarHCl yeah thx :]
15:38:51HClif you want to go from runtimedb entry
15:38:54HClto tagdb.
15:39:30HClit would link to the first file, not the file with the same hash thats elsewhere on disk
15:39:47rasherthat's irrelevant
15:39:52HClis it?
15:39:54rasherthe runtimedb is "songs" not "files"
15:40:13HCli guess.
15:40:15rasherso runtimedb->tagdb is not guaranteed to give a specific file
15:40:20rasherwhy would it matter?
15:40:33HClthe only time i can think of when you want to go from runtimedb->tagdb
15:40:41HClis when you want all songs with a rating higher than ...
15:40:50HCland i guess the specific file doesn't matter, if they share the same hash.
15:40:59rasherthat's my thinking
15:41:05HCli dunno.
15:41:19HClneed to try to think of a situation in which it would create problems.
15:41:23HClif i fail, it might be ok.
15:42:06HCli'm almost thinking of just storing the offset of the runtime db in the tagdb.
15:42:26HClsince with my proposal, you'd have to check the field for elsewhere anywho
15:42:41rasheryou might as well
15:42:52HClbut i guess it would break the reverse lookup..
15:42:55rashersince it'd just be songno*constant + constant(header)
15:42:58HClokay, that was a bad idea..
15:43:13HCli'll just do my proposal then and have it stored in the runtime db..
15:43:22HClit doesn't make a difference in checks
15:43:27HClbut it allows to preserve the static size
15:43:34HCland easy between tagdb<->runtimedb lookups
15:43:41rasherthis seems nice
15:44:00HClso with the "its not here, its there!" field?
15:44:17HCli'll update the proposal
15:44:47rasherI guess we'll have a limit of songcount on the size of that field
15:44:57rasherset it high
15:45:04HCli'm just gonna make it an offset field.
15:45:12HClpointing to the used runtime db entry
15:45:15HClfor the file
15:45:16rasheryeah, nevermind :)
15:48:25*rasher watches Bagder appear and change the whole thing
15:48:43HCl :P
15:48:55HClrequest for comments
15:51:04BagderI want the hash made on data that's available in the songdb
15:51:17HClhmm. why?
15:51:19 Join kurzhaarrocker [0] (
15:51:35Bagderto reduce complexity and gain lots of speed
15:51:42Bagderhow else could you verify the hash?
15:51:44HClsounds sane enough.
15:51:46kurzhaarrockerHCl: what would you think about a counter that counts how many times a song has been skipped?
15:51:46 Quit sox ("Snak 4.13 IRC For Mac -")
15:51:54HClcan we trust the songdb data to be unique enough?
15:52:11preglownot really
15:52:12HClkurzhaarrocker: grin. we can easily add that later.
15:52:14rasherwhy can't the songdb just *include* the hash?
15:52:17preglowbut most of the time
15:52:20HClrasher: good point.
15:52:22Bagderif we add path to the mix, then we can
15:52:33HClBagder: then we lose renaming
15:52:34Bagderinclude the hash?
15:52:36HCli suggest what rasher suggested
15:52:40HClmake the hash on a songdb update
15:52:41HClon the pc
15:52:47HCland put it in the tagdb
15:52:52rasheryes.. the tagdb includes a hash.. calculated however we decide
15:53:02HCl*goes to change*
15:53:15Bagderit'll take forever to resync
15:53:24 Part t0mas
15:53:28Bagderimagining scanning 12000 mpeg files
15:53:31Bagderto get the hash
15:53:41rasherit's faster than you'd think
15:53:42HCloh, the songdb creation.
15:54:01Bagderwell, it would make it totally impossible on target
15:54:03rasherI did it in *php*, and it took something like 30 seconds for 6000 files
15:54:30HClBagder: well, including the filename might make it unique, but destroy the renaming ability
15:54:31rasherDidn't we trash that idea?
15:54:34kurzhaarrockerbtw: Is there a special reason why the context menu of the db browser doesn't allow any playlist modifications like "insert" or "cue"?
15:54:59BagderHCl: yes
15:55:02 Join t0mas [0] (
15:55:09Bagderit is all a matter of what we what to sacrifise
15:55:10t0masooops :)
15:55:27 Quit ashridah ("Leaving")
15:55:34BagderI'm not vetoing anything, just bringing my opinion
15:55:52HClpersonally, i think rehashing it on a file rename should be okay, the only big problems we'd get is when you rename externally.
15:56:07HClwhich is probably not what we want, losing all runtime data when you rename files while mounted on usb
15:56:25rasherwhen would we need to hash a file on the target?
15:56:43Bagderif we don't want to resync on target we never want that
15:56:51HClrasher: only when you don't want to update the runtime db on the target / want to build the songdb off a mirror of the files on the target.
15:56:54Bagderbut then we lose runtime info until resynced
15:57:00kurzhaarrockerrasher: e.g. when recording
15:57:33rasherI'd say that's an acceptable sacrifice
15:57:35HCli dunno
15:57:41BagderI could live with that too
15:57:47HCli have no idea how harsh hashing all files is
15:57:51HClon target
15:58:03rasheruntil you resync, you lose runtime info on new files
15:58:15rasherthat doesn't sound too bad to me?
15:58:17HClrasher: resync using the pc, correct?
15:58:22HClor just resync in general
15:58:26Bagderand we _could_ make some weird hack to attempt to fix that too
15:58:43t0masisn't this done in the media thing for ipod linux?
15:58:44rasherBagder: dumping info to a third file?
15:58:45HClso hashes are still stored in the runtime db, based on the file
15:58:51Bagderlike "no in the db, I store the info here and we can try to merge it in later when we resync"
15:58:59t0mas(was it called podzilla?)
15:59:01HClyea, flatfile
15:59:12Bagderbut let's do that part last ;-)
15:59:15HClt0mas: yea, podzilla, dunno what thats done.
15:59:16kurzhaarrockerBagder: good idea.
15:59:19rashersimply write out "played this file" to a file
15:59:21 Part LinusN
15:59:26kurzhaarrocker(both, the idea and to do it last)
15:59:28*t0mas checks how they did it :)
15:59:28rasherand have the syncscript read that file and merge
15:59:31HClokay, let me get everything straight
15:59:39HClwe build hashes based on the files
15:59:47HCland we store them in the runtime db
15:59:57HCland for duplicate hashes the runtime db will have re-pointers to a single entry
16:00:25HClfor files that get added, you need the pc to update the tagdb compared to the runtimedb
16:00:25rasherand all other fields zeroed out
16:00:26Bagderhow would the runtimdb know the hashes are duplicates?
16:00:32HClrasher: yea, thats in the wiki
16:00:54HClBagder: it won't. but it can check whether the hash in the entry pointed to is the same as the hash in the entry it came from
16:01:08HClonly the pc tool who'd have all entries in ram would know about duplicate hashes
16:01:38Bagderso why use the hash at all when referring to songdb entries?
16:01:48Bagderwhy not just a seek offset?
16:01:59HClyou lost me.. what.
16:02:14HClthe hash is there to enable building of the songdb without the runtime db
16:02:18HClthere was someone who insisted on that
16:02:19kurzhaarrockerThat would require that the runtime info is kept when you rebuild the song db, Bagder
16:02:20HCli don't remember who.
16:02:32Bagderthe hash in the runtimedb is only needed for resyncing
16:02:49Bagderok, then I'm with you
16:03:24Bagderthe wiki is not clear at how it refers to songdb entries
16:03:34HClmmm, it has a bit on it
16:03:35HClhold on
16:04:02HClright under the how its supposed to work
16:04:09HClThe runtimedatabase has a static size, and a record for every song in the tagdatabase. you can get the rundb record of a song by calculating its song record number in the tagdb (its the 768th song in the tagdb, for example), and then take the rundb record with the same record number (768th rundb record)
16:04:14HClIn the same way, you can get a tagdb record of a song by taking its rundb record number
16:04:23Bagderyes, but shouldn't that then be mentioned in the table?
16:04:43HClwe don't need to store it anywhere with the current approach, since you can just calculate it..
16:05:11Bagderwhen you load the runtimedb it might contain thousands of entries
16:05:24Bagderyou want to calculate all offsets then?
16:05:33HClwhen you have a runtime db entry
16:05:37HClyou know its fileoffset
16:05:39HClat least, i assume you do.
16:05:43Bagderwhy? how?
16:05:53HClcause you know where you read it from o.o.
16:06:00HCldon't we?
16:06:19kurzhaarrockerBagder: the other way round. You browse the tag db and access the runtime db with hash keys generated from the tag db entries
16:06:28BagderI'm stupid
16:06:28HClkurzhaarrocker: what?
16:06:31ShagnarHCl thanks for updating grayscale :) works pretty fine
16:06:42HClkurzhaarrocker: no, thats not how it works at all at the moment o.o
16:06:48kurzhaarrocker<- me stops confusing everybody and pretends to be far away
16:06:51BagderI didn't quite believe you would actually store all songs everytime in the runtimedb
16:06:54HClissok :x
16:06:58Bagderseems unnecessary to me
16:07:04HClBagder: we can only update the ones that are changed
16:07:08HClwe just seek around it o.o
16:07:24Bagderif you add a seek offset to each entry, you can add only those that there is info for
16:07:25HClbut yes, all songs are stored in the runtimedb
16:07:29HClallllll of these problems
16:07:30HClwere avoided
16:07:33HClwith my original design
16:07:37rasherquiet now!
16:07:38HClbut that was rejected by everyone
16:07:43HCl :/
16:07:46Bagderyour original design was... crap
16:07:51HClthanks :P
16:07:57HClit didn't have any of these problems though :P
16:08:01Bagderit had others
16:08:09HCllike? :p
16:08:12HCli don't really care anyways
16:08:12Bagderlike 12000 songs
16:08:14Bagderon archos
16:08:18Bagderdeal with it
16:08:23HCli'm just saying
16:08:26HClthis is the new designy
16:08:28HClyou wanted it
16:08:29HCldon't complain :P
16:08:32BagderI don't
16:08:35BagderI present answers
16:08:35HClokay, good :P
16:09:01HCleverybody okay with the current design? o.o
16:09:19rasherI think so
16:09:27preglowi don't care much about databases :V
16:09:29Bagderexcept for that I don't quite agree it is the best format, I could shut up ;-)
16:09:30rasheranyone here :)
16:09:44HClBagder: you're free to propose a better one, heh.
16:10:05Bagderwell, for one I want a seek offset and NOT a 1 <=> 1 between the databases
16:11:02rasherwell it isn't 1<=>1 surely
16:11:11Bagderentry wise it is
16:11:26HClthen we'd almost have the same as my original format, heh, aside from no seperate filename lib.
16:11:37BagderHCl: yes, and no keeping everything in ram at once
16:11:38kurzhaarrockerit must be. We want each song in both db's. Don't we?
16:11:47Bagderkurzhaarrocker: not necessarily
16:11:58Bagderif we don't have info about a song, why have it in the runtimedb?
16:12:11kurzhaarrockerhow can we store song specific data then?
16:12:17Bagderwhen we get data, we store it
16:12:20HClBagder: i'm confused, why would we keep everything in ram at once any less than my old design?
16:12:34HClaside from mapping tagdb and runtimedb to a file
16:12:40HClcause we have the exact same issue at the moment
16:12:43HClif not worse
16:12:43BagderHCl: this format does not require keeping it in ram, afaics
16:12:59HClif we want to go from filename to tag/runtime db
16:13:05HClwe have to search the tagdb for the filename
16:13:07HCltake the offset
16:13:16HClthen search for that offset in the songs bit of the tagdb
16:13:23HClthen take the offset there for the runtime db
16:13:28HCland finally access the runtime db
16:14:00HClwith the current design we'd need two searches to go from filename->runtimedb/tagdb
16:14:19HClbut ok
16:14:25HCli'm not gonna whine over my old design anymore..
16:14:27 Part kurzhaarrocker
16:14:46HClbut this is how we would have to do things at the moment..
16:14:53Bagderthe 1:1 entry would prevent the second search
16:14:58HClno it wouldn't.
16:15:05HClwe don't link filenames to tagdb entries at the moment.
16:15:05Bagderwhy not?
16:15:10HClwe link tagdb entries to filenames
16:15:13HCland not the other way.
16:15:23Bagdertagdb to songdb entries, isn't it?
16:15:33Bagderruntimedb to songdb
16:16:07HClthe two searches are only for finding the tagdb entry
16:16:12HCleven if we don't have a runtime db at all
16:16:16HClat the moment, if we want tagdb info
16:16:18HClwe have to do two searches
16:16:24HClto find the correct tagdb entry
16:16:25Bagderwe could easily have the songdb have all file names in a way that allows us to do a binary search
16:16:38HClwe already have that.
16:16:43HClregardless of the way of search
16:16:49HClwe'd still have to pass through all songdb
16:16:53HClto compare the offset of our filename
16:16:55HClto the entries
16:16:59HClto find the correct entry
16:17:43HCltell me how we would find a tagdb entry at the moment
16:17:46HClif we have a filename
16:17:49*HCl flops on the couch and sighs
16:17:55Bagderthe search for a file name in the songdb should make us find the songdb entry
16:18:00Bagderon one search
16:18:01HClit should.
16:18:04HClbut it doesn't.
16:18:06HClwith our current format.
16:18:07Bagderthen we fix it
16:18:21Bagderwe fix the format and code
16:18:28HClyou can't link filenames to tagdb entries unless you have a static sized filename field
16:18:36HClwhich *was* in my original design x.x;;;
16:18:42Bagderand that WAS CRAP
16:18:52HClthen make a better one which doesn't have these problems.
16:18:59Bagderwe can have a fixed size
16:19:03Bagderbut the size should be set in the header
16:19:11HClsure, i agree with that.
16:19:22Bagderand we already do that for the songdb
16:20:15BagderI guess a 1:1 mapping is better for the search problem
16:20:35HClyou mean a 1:1 mapping between the filenames and the songdb?
16:20:58HCl(all tagdb internals, by the way)
16:21:18BagderI meant between songdb entries and runtimedb entries
16:21:26HClthats not gonna solve the search problem.
16:21:38Bagderyes, if each file name has an associated songdb entry number
16:21:49*preglow sits down and gets his barbecue equipment
16:21:56***Saving seen data "./dancer.seen"
16:22:00HClthen we can just as well make each filename and have the runtime entry number as well next to the filename
16:22:09HClso we don't have to look up the songdb entry in order to get the runtime db entry
16:22:19HCl(which waaaasssss my original design) >.>
16:22:21Bagderyou'd need the file search
16:22:27*HCl shakes his head and flops on the couch.
16:22:36Bagderand if the songdb entry and runtimedb is 1:1 then it is the same number
16:22:44HClyea, it is
16:23:45HClif we put the filename bit in a seperate file
16:23:50Bagderwhy a separate?
16:23:52HClit would allow runtime data for files not in the tagdb
16:24:09Bagderwe could allow entries without tag info in the songdb
16:24:22Bagdera separate would not help that
16:24:31Bagdersince the file with file names would need to be sorted
16:24:42Bagderto allow binary searches
16:24:49HClyes, has to be
16:24:55Bagderso adding files is painful
16:25:01Bagderbut still doable of course
16:25:50Bagderso let's have a 1:1 mapping and let's add an "entry number" associated with each file name in the songdb
16:26:09HClwhy would we insist on 1:1 mapping though?
16:26:26HClif we don't, it would allow things like no runtime info for a file, keeping the runtime db small
16:26:35HClsame thing for no tagdb info for a file
16:26:39HClkeeping the tagdb small till its updated
16:26:47HClwe can just have two entries
16:26:49HClnext to the filename
16:27:02Bagderthen how would you find the runtimedb given the file name?
16:27:10Bagderruntimedb entry
16:28:01HClthe runtimedb would be next to the tagdb number
16:28:09HCllike, instead of having a filename -> record
16:28:18HClwe'd have filename -> tagdbrecord, rundbrecord
16:28:26Bagderbut then you'd need to write to the songdb when you add a runtimedb entry
16:28:38HClno, you can simply have an empty tagdb record
16:28:39HClfor a file
16:28:50HCland make it non empty when it gets a tagdb entry
16:29:03Bagderyes, that implies writing to it
16:29:12 Quit pabs (Read error: 54 (Connection reset by peer))
16:29:33HCldon't tagdb entries only get written when songdb runs?
16:29:44 Join pabs [0] (
16:29:56HCli propose an additional hash field next to the filename
16:29:59HClto support renames.
16:30:55BagderI thought we talked about hasing the music data for that
16:31:04HClthats what i mean
16:31:09HCla hash next to the filename
16:31:12HClbased on the music data
16:34:06BagderI don't mind exactly where the hash would be stored
16:34:17HClyea, its not really vital
16:34:34 Quit pabs (Read error: 54 (Connection reset by peer))
16:34:49HCljust seems a sane place to add it in the table that links filename to metainfo, to me anyways
16:35:38 Join pabs [0] (
16:38:22*HCl stares at bagder
16:38:56Shagnarwill it be possible to move files later?
16:39:04HClShagnar: the hash should make it possible..
16:39:11HClsince its a matter of taking the filename table
16:39:15HClcalculating the hashes
16:39:18HClsearching for the hash
16:39:22HCland updating the filename in it
16:39:23HClif it has changed
16:39:34HCland resort.
16:40:55Shagnarthats nice. because, with an usb-otg-copybox you'll be able to move files and copy them on a player, so (e.g. in the holidays) you can fill a little flash player with music
16:41:15preglowi move files all the time
16:41:43Bagder the only problem still bouncing in my head is to map quickly from a songdb entry to the runtimedb entry
16:41:56HClBagder: it would go through the songdb and the runtimedb
16:41:58HClthe songdb
16:42:00HClwould have an offset
16:42:04HClto the filename table
16:42:12HCland then you can read the filename table entry the songdb pointed to
16:42:15HClto get the runtimedb offset.
16:42:32preglowsounds like there'll be a lot of nice cross references
16:42:36Bagderbut that would mean that you update the songdb run-time
16:42:37HClwithout needing to search for anything
16:43:04HClwell, the filenames is a table of the songdb anyways
16:43:13HClyou'd assume that if you update it, you have to update the songdb too.
16:43:44HClyou can put the runtime db offset in the songdb, but then you lose the ability to have runtimedb entries without songdb entries
16:44:20BagderI don't follow you now
16:45:01HClokay.. let me take it slowly..
16:45:11HClwith some examples
16:45:35HCl"\iriver.mp3" <hash> 334(songdb entry) 555(rundb entry)
16:45:45HClsongdb entry 334:
16:45:52Bagderhold it
16:45:57HClsorry :/
16:46:02Bagderisn't this file name table part of the songdb?
16:46:22Bagderok, so how does it know where the rundb entry is?
16:46:48HClit was written there when it made an rundb entry for it
16:46:57HCl(which is why i propose the filename table in its own seperate file)
16:47:09Bagdera separate file doesn't change much imo
16:47:35HClwe can just make it 1:1, and write empty tagdb entries for files with no tag
16:47:53HCli don't really mind, the only downfall would be that the runtime db and the tagdb would be bigger
16:47:58HCland that duplicate files with the same hash
16:48:05HClwould need that "elsewhere" field in the runtime db
16:48:14HClwhere with an entry in the filetable that can be avoided
16:48:22HClby simply pointing to the same entry in the runtime db
16:48:23 Quit DangerousDan (Read error: 104 (Connection reset by peer))
16:50:36Bagderthe thing is that I don't really like 1:1 and I don't really like updating the songdb entry run-time... :-) But I don't have any third suggestion right now.
16:51:33HClwell, we need a pointer from the songdb entry to the runtime db entry one way or the other
16:51:41BagderI agree
16:51:43HClwe can just choose between 1:1 or a seperate filetable
16:52:08HCli'm not really aware of any alternative
16:52:12Bagderseparate or not makes no difference to me
16:52:31HClkay, i'm slightly pro seperate, mostly because it would deal with duplicate hashes nicely
16:53:20Bagderwhy does it makes any difference if the file table is part of the songdb or not?
16:53:38*Bagder types like crap
16:53:39HClwell, it doesn't matter too much, aside from that if you want to add a filename.
16:53:50HCl(assuming the filename table is at the beginning)
16:53:57HClyou have to shift the entire rest of the database
16:54:26HClputting the filetable at the end would solve that
16:54:28Bagderbut it would be painful to have yet another file to deal with
16:54:29HClbut then you'd get a problem
16:54:35HClwhen you want to add a songdb entry
16:54:44HCland still have to shift half the database
16:55:12Bagdera file name doesn't imply any songdb entry actually
16:55:18HClwell, using offsets shouldn't make it have much of an performance impact
17:00:19Shagnarhow can i get winamp to play flac ?
17:00:53preglowby installing a plugin?
17:01:09dwihnoThe flac binary distribution contains a in_flac dll
17:01:15dwihnoWorks like a charm
17:01:39HClBagder: i guess that if you put the filename table at the end of the songdb, and assume that we won't be adding tagdb records in runtime, it'll do
17:01:40dwihnoThere U go!
17:02:27HCladding a filename to a sorted list shouldn't be too much of a problem anyways, its just a matter of doing a binary search and inserting it in there.
17:03:25Bagderthat file name list isn't even sorted atm
17:03:35HClyea, it'll need to be
17:03:53HClat the moment we're not doing any filename->tagdb links, but we're definately going to have to need that ability
17:04:41Bagdermaking it sorted should be one of our smallest problems ;-)
17:05:19BagderI looked at the code, and it could be turned into a sorted list instantly
17:08:09HClwill we put the filename data in a seperate file or at the end of the songdb?
17:08:49HClwe could always seperate it later if we bump into a need of adding tagdb entries runtime
17:09:20Bagderthere's also the option to add new ones in separate file
17:09:28HCli forgot that.
17:09:54HClso just at the end of the songdb then?
17:10:07HClwe'll want to be able to add runtimedb entries realtime
17:10:34HCl(which is just adding to the end of the rundb, binary searching the filename table and inserting the new filename with a link to the runtimedb entry)
17:13:55BagderI'm off for real life
17:14:40HClokies, bye.
17:14:50preglowthis is real life :V
17:15:04Bagdermy wife thinks differently ;-)
17:15:17preglowskewed point of view
17:18:18HCl :p
17:18:38preglownow look at that, the clouds vanished
17:18:46*preglow goes outside
17:19:53 Join webguest13 [0] (
17:20:13 Quit webguest13 (Client Quit)
17:21:45 Quit pabs (Read error: 104 (Connection reset by peer))
17:22:33 Join pabs [0] (
17:24:17 Join xen` [0] (
17:45:35 Join TCK- [0] (
17:49:05 Join asdsd_ [0] (
17:54:07 Quit TCK (Read error: 60 (Operation timed out))
17:55:47 Join Psy^Dead [0] (
17:55:53Psy^Deadhooray for linus!
17:56:02Psy^Dead(and the others)
17:56:30 Quit pabs (Read error: 104 (Connection reset by peer))
17:57:18 Join pabs [0] (
18:00:27 Quit Patr3ck_ ()
18:08:59 Quit bobTHC ("Smoke Weed Every Day !")
18:21:57***Saving seen data "./dancer.seen"
18:24:51 Quit elinenbe (Read error: 104 (Connection reset by peer))
18:25:32 Join DangerousDan [0] (
18:33:27 Quit pabs (Read error: 104 (Connection reset by peer))
18:40:29 Join idlewild [0] (
18:46:38 Join pabs [0] (
18:48:18 Quit Chamois (" HydraIRC -> <- State of the art IRC")
18:49:02 Quit TCK- (Read error: 60 (Operation timed out))
18:55:53 Nick Lynx_ is now known as Lynx_awy (HydraIRC@
18:58:08 Quit asdsd_ (Read error: 110 (Connection timed out))
19:00:26 Quit einhirn ("Miranda IM! Smaller, Faster, Easier.")
19:09:30 Join Hornet [0] (
19:13:55 Join Sucka [0] (
19:56:30 Join TCK [0] (
20:14:10 Nick StrathAFK is now known as Strath (
20:14:52 Quit xen` (Read error: 110 (Connection timed out))
20:21:14HClheh. pathetic.
20:21:35HClthe payed computer staff of our university sent an email whether any students would be able to help them secure mysql server 2000
20:22:00***Saving seen data "./dancer.seen"
20:22:08Psy^Deadgive them the link to mininova
20:22:22*HCl watches catwoman..
20:22:30HCl(yes, that supposedly crappy movie of a while ago)
20:22:42t0massql server 2000 as in Microsoft sql? :P
20:22:48HCli think so.
20:23:22t0masdamn... that's just a wizzard...
20:23:44Psy^Deadtell them to get the free one
20:23:48Psy^Deadmysql or something
20:24:29Psy^Deadi must type "n" a lot, its gone all shiney from rubbing the rough surface off
20:24:38Psy^Deadits worse than e or wasd
20:25:54rasherMy n key doesn't have an n
20:26:04HClmy n key has a b
20:26:13rasherthat's surprising
20:26:21HClis it? :p
20:26:31HClwell, it depends on what you call your n key
20:26:36HClthe key that says n on it gives b
20:26:38rashertyping dvorak?
20:26:47HCland the key that gives n says l on it
20:26:54Psy^Deadmy n says n
20:27:02preglowi tried dvorak once, then realized i'd have to learn typing all again, then stopped
20:27:07t0masweird conversation...
20:27:09rasherHaha, me too
20:27:10HClits easy.
20:27:17HCland types so much better than qwerty
20:27:24preglowyes, after you spen three months in getting used to it
20:27:24t0masto get back on the rockbox topic :P
20:27:31t0masdoes anybody know a free translation tool?
20:27:33preglowi can't stand not being able to use a keyboard any more
20:27:39t0masor open wordlists?
20:27:51t0masso I can write a dictionary lookup thing?
20:28:12 Join Hornet_ [0] (
20:28:42Psy^ or are crap
20:28:48Psy^Deadbut the best you'd get
20:28:56rasherI think maybe has some.. but I'm not sure
20:29:05rasheristr some format with mappings between languages
20:29:08Psy^Deadtranslators suck
20:29:11t0masPsy^Dead: I want it GPL-ed ofcourse...
20:29:23t0masand I don't want it to translate... just a word lookup tool
20:29:28rasherwell, the file lists wouldn't *have* to be gpl I guess
20:29:29t0maslike some nokia phones have
20:29:29 Nick Psy^Dead is now known as nobby (
20:29:44t0masrasher: as in let people download em themself?
20:29:48nobbyno idea then, sorry
20:29:57rasherno as in, as long as they're redistributable
20:30:05t0masthat's true...
20:30:14rasherit'd just be a data-source for the plugin.. I don't think that has to be gpl
20:30:32rasherthink being the operative word here
20:32:04nobbylicensing is really important to you guys, eh?
20:32:25nobbyhow come?
20:32:25rasheras it should be
20:32:41*HCl doesn't care about licensing much.
20:32:46*nobby either
20:33:54preglowi can't possibly understand how one can deal with opensource software and not care about licensing
20:34:07HClwhy not? :)
20:34:10preglowit's half of what it is
20:34:23HCli just don't think its such a big deal
20:34:30preglowit's not just a bunch of shit dropped in the public domain because you can download it freely
20:34:32HCllike the public domain code of bluechip the other day
20:34:40HClyou guys went way too harsh on him
20:34:46HCli'm aware of that.
20:34:58HCli'm also aware of that anyone violating gpl is just dansk and other weird things I can't even read :P
20:35:11HClhas no problem as long as the original authors of the sourcecode don't have a problem with it
20:35:26rashert0mas: yeah, I saw it now.. I thought it was more general, sorry :)
20:35:31preglowthey're probably just ok because it's major bother taking any action
20:35:34preglowthat doesn't mean they like it
20:35:34HClpreglow: bluechip wanted his code in pd, and the whole rockbox crew made a huge deal out of it
20:35:40HClwhen it was all bullshit
20:35:48HClbecause the code can be both pd and gpl
20:35:57preglowno, i don't think so
20:35:59HClany modifications on the gpled rockbox version would have to stay gpl.
20:36:02t0masyes it can..
20:36:07HClwhere anything based on the original pd code can be anything.
20:36:10rasherit *could*, but then you'd have to pd it all
20:36:11t0masyou can adopt code to gpl
20:36:14HClrasher: no.
20:36:23HClthere would be two copies of the same code.
20:36:24HClone under gpl
20:36:27HClone under pd
20:36:28t0masthings changed to rockbox's version would be forced to be GPL
20:36:33HClany updates on the gpled version
20:36:35preglowdon't you feel it's just better keeping everything gpl?
20:36:35t0mashis version can still be pd separate from rockbox
20:36:36HClwould have to stay gpl
20:36:38preglowit's much simpler
20:36:40HClwhere anyone can take the pd version
20:36:43HCland take that.
20:36:45 Quit Hornet (Read error: 60 (Operation timed out))
20:36:47rasherHe can not pd "his own version"
20:36:52rasherbecause it is already gpl
20:36:53HClyes he can.
20:36:56rasherhe cannot sublicense
20:37:04t0mashe wrote it
20:37:05t0masso he can
20:37:10HCli'm not even going to go into this, seriously.
20:37:22HCli think you all need to relax XD
20:37:25HCland with that i go afk
20:37:31HClcause i seriously think that that licensing stuff
20:37:32HClis all crap.
20:37:34HClbye :)
20:37:54rasheroh, maybe I'm misunderstanding "his version"
20:37:55preglowyeah, you're allowed to think so, but you're not allowed to just ignore it and tell people it's ok
20:38:20nobbywhy does the gpl not allow non-gpl stuff in it anyway?
20:38:24 Join Stryke` [0] (~Chairman8@resnet-241-86.resnet.UMBC.EDU)
20:38:34rasherit's not really like that
20:38:41preglownobby: because then it wouldn't be gpl, would it?
20:38:45rasherit just doesn't allow any *additional* restrictions to be put on it
20:38:55rasheryou can have BSD licensed code just fine
20:39:04HClyou can take pd code and adopt it into a gpl project.
20:39:09HClit has no effect on the original licensing.
20:39:13preglownobby: the whole meaning of something being gpl, is the ensure that certain freedoms are present, if the gpl would allow all stuff of shits to be bundled under it, then it would no longer be able to guarantee that
20:39:19HClit only has effect on any code based on the GPL version of it.
20:39:25HClnot on code based on the PD version of it.
20:39:33rasherWhat did bluechip write?
20:39:34HClif you can't see that, i'm not gonna go into any discussion o.o.
20:39:44HClsome games, some audio filter stuff.
20:39:46HClarchos stuff.
20:40:11preglowof course, you can do whatever you like with pd stuff
20:40:26t0masso it's no problem copying his code and putting it under GPL for rockbox
20:40:39preglowshouldn't be, no
20:40:42t0mashe can still give his copy away as PD
20:40:54t0masso there is no problem
20:40:56HClexactly, and that was all i was saying.
20:41:02t0mas(and I really can't type with dvorak)
20:41:02rasherproblem would be if his code is derived from gpl code
20:41:05t0masjust tried
20:41:18t0masrasher: that would be his problem
20:41:19t0masnot ours
20:41:22HClyou can't just switch to dvorak and expect to be able to type in it.
20:41:27preglowif his code was derived from gpl, he had no right to put it in the public domain
20:41:28rasherIt most certainly would be our problem
20:41:29t0maswell... I am now
20:41:48rasherbecause he's releasing code into the pd which he has no right to do so
20:41:53t0masI did it before... as my basis school (nederlands) had dvorak keyboards
20:41:58HClrasher: ah, didn't know that
20:42:03HClt0mas: wow o.o.
20:42:14HClgood basic school.
20:42:20rasherwell.. *if* it's gpl derived
20:42:21 Join Hornet__ [0] (
20:42:25webmindt0mas, what school was that ??
20:42:26rasherdon't know about that
20:42:32t0masbut I'm rather slow now... as I have to corret a lot
20:42:51t0maswebmind: Wethouder luidens school in Gouda
20:43:14t0masyeah... I hated it then...
20:43:28webmindwhy ?
20:43:29t0maswe had to follow typing lessons... and I had qwerty at home...
20:43:34webmindaah k
20:43:51rashertyping lessons are about the most dull thing EVER
20:44:03t0masyeah, but in dvorak it's a challenge...
20:44:04*webmind never did one
20:44:16webmindstille type with 5 fingers
20:44:24t0masand as I see now... I can still type like this..
20:44:29t0mas(on a qwerty keyboard now)
20:44:31rasherwell, if you're not used to typing anyway, it hardly makes much difference
20:44:38rasherwhether qwerty or dvorak
20:44:48t0maswell... nobody can use my pc now :P
20:45:08t0maseven if my password is set to "password" as I have a qwerty keyboard set to dvorak now :)
20:45:18webmindgood to know :)
20:45:31nobbyi have no password
20:45:39nobbyyet XP insists on asking for it
20:45:45HClt0mas: i could use it :P
20:45:54HCli always type dvorak blind on qwerty keyboards
20:46:02t0masHCl: I am doing now...
20:46:21t0masbut I need the schema with it...
20:47:31t0masbut still... I'm far better at qwerty :)
20:47:40t0mas(nice to be able to switch with a hotkey in windows :P)
20:48:43rasheryou can do that with gnome :)
20:49:06rasherand kde as well I'd wager
20:49:54HClt0mas: whats that hotkey?
20:50:10HCli need to know cause i'm always pressing it by accident and then i suddenly type qwerty which i hate
20:50:27t0masghehe... I'll set my laptop to dvorak too :)
20:50:46 Quit Hornet_ (Read error: 60 (Operation timed out))
20:50:55t0masHCl: you can set em yourseld
20:51:30t0masget the language bar on your taskbar
20:51:39t0masgo to "toetsinstellingen"
20:51:47HClgot it
20:53:54 Quit lolo-laptop ("Client exiting")
20:54:12*rasher updates ircstats
20:55:03 Join [IDC]Dragon [0] (
20:56:54*amiconn appears
20:57:04*[IDC]Dragon too
20:57:16HCli need to talk to Bagder..
20:57:34[IDC]Dragonwe're not holding you...
20:57:34HClcause i'm thoroughly confused about the tagdb/rundb now :/
20:58:04[IDC]DragonI'm not really up to date, but saw a hash mentioning
20:58:16HCllittle improvement for renames
20:58:32[IDC]Dragonwe have a CRC routine, could make that public
20:58:50amiconnIt's CRC16 only...
20:59:14amiconn[IDC]Dragon: Did you read about my fat.c fixes / improvements?
20:59:15[IDC]Dragoncould be CRC32 as well
20:59:24*amiconn updates the public copy
20:59:42[IDC]Dragonamiconn: not in detail, I only saw you asking Zagor for testing
21:00:06amiconnI'm undecided whether to commit this stuff
21:00:17[IDC]Dragonwhat's wrong?
21:00:20 Join matsl [0] (
21:00:50HClis the fix for archos or iriver?
21:01:09HCldoes it need more testing or what?
21:01:11HClwhy undecided?
21:01:34amiconnYes, more testing would be desirable
21:02:03amiconn[IDC]Dragon: The current add_dir_entry() implementation has bugs (guess who found them!)
21:02:15rasherI know! I know!
21:02:35amiconnI fixed that, plus some more fixes
21:02:38rasherbest tester ever
21:03:31amiconn(1) DOS shortname generation was slightly off (2) observe dir size limit (not very likely to hit that though)
21:03:40amiconnI also have an improvement
21:04:18amiconnWindows NT derivates can store the case of the name part and the extension part _in a shortname_
21:05:00amiconnThis is what I wondered about very often. WinXP displays 'ajbrec.ajz', but rockbox displayed 'AJBREC.AJZ'
21:05:15[IDC]DragonI saw you mentining that, yes, very nice
21:05:25amiconnIt's an undocumented feature, and my code now supports it for reading
21:05:44[IDC]Dragonwhere did you find it?
21:05:51rasherYeah, I have a few of those fileas as well
21:05:54HClif we understand how it works, wouldn't we be able to write it too?
21:05:59amiconnI googled a lot for FAT related things
21:06:25amiconnOne site mentioned that the NTRES field is used to store that info, but did not mention the details
21:06:44[IDC]DragonHCl: we always create longnames, iirc
21:06:53HClfair enough.
21:07:05amiconnThe rest was a simple deal. Created some files with different cases and shortnames on a FAT volume, then used a disk editor to look at the NTRES fields
21:07:25[IDC]Dragonyou had to reverse engineer it?
21:07:40amiconn[IDC]Dragon: (longnames) Yes, we do, except for the dotdirs, where this is prohibited
21:08:48[IDC]Dragontrevmar is back, you saw that?
21:09:08[IDC]Dragonwith his strange issues
21:09:08amiconnI think I should mail him
21:09:59[IDC]DragonI may have an hour or 2 for rockboxing now, if I don't fall asleep
21:10:09[IDC]Dragonany urgent stuff?
21:10:35amiconnI have lots of stuff I wnat to start working on :-\
21:10:38[IDC]Dragonmy midterm goal is the barebones rockbox
21:10:59[IDC]Dragonfor image 1, to gain rombox space
21:11:21amiconnI tried compiling rockbox with gcc4. Perhaps gcc4 optimises so much more that rombox will magically fit again for some models
21:11:36*[IDC]Dragon starts with a cvs update
21:11:40amiconnThe latest additions pushed the v2 over the edge
21:11:59HCli suggest not trying gcc4 for now...
21:12:04HCli tried it for iriver and it crashed horribly
21:12:14amiconnIt crashed?
21:12:24HClillinstr at 00000000
21:12:36HClthere's a compiled rockbox with gcc4 on my ftp
21:12:41HClbut i don't suggest anyone to try it
21:12:43HClsince all it'll do is crash
21:12:45amiconnThat sounds 'ill'
21:13:09amiconnPerhaps sh does better. Did you compare code size?
21:13:15HCli didn't.
21:13:18[IDC]Dragonmay be different for th SH
21:13:21HCllet me compare code size
21:13:48[IDC]Dragonthe t-shirts are removed from cvs...
21:14:01amiconnYeah, saw that. I wonder why...
21:14:16[IDC]Dragonspace? out of date?
21:14:25*amiconn wants one :)
21:14:33rasherBagder removed them to reduce the size I think
21:15:13[IDC]Dragonamiconn: how about my MMC USB mode patch?
21:16:07[IDC]Dragonwell, you 2GB are waiting, too
21:16:14amiconnyup :-/
21:16:49HClits 1300bytes bigger
21:16:56HClwith gcc3
21:16:58HCli mean
21:17:19amiconnIf someone wants to try my latest fat.patch:">
21:18:04preglowHCl: i kind of figured that
21:18:14amiconnpreglow: Why?
21:18:29preglowbecause that's the way gcc always has gone on default optimizing settings
21:18:48preglowcode has most of the time gotten bigger and faster
21:19:06[IDC]Dragonamiconn: I tried to load your midi file, is it still there? URL?
21:19:41t0masarg... I keep on switching T and N on dvorak">
21:20:14rashert0mas: you mean N and T?
21:20:47t0masand swinchitg :P
21:21:48t0mas*switching back*
21:22:03[IDC]Dragonamiconn: got it, thanks (but I don't hear anything...)
21:23:08amiconnIt's playing fine here, both on the PC and the Amiga
21:23:21[IDC]Dragonah, disabled midi soure in the mixer
21:23:43[IDC]DragonI rarely play midi
21:24:26[IDC]Dragonthe s/w synth on my onboard sound is really bad
21:25:29amiconnWinXP? Then switch to the "Microsoft GS Wavetable SW Synth"
21:25:45[IDC]Dragonswitch where?
21:26:23[IDC]Dragonah, it's already set
21:26:44amiconnI think that one doesn't sound too bad
21:26:47[IDC]Dragonso M$ is to blame
21:30:07rasher[IDC]Dragon: excellent commit :)
21:30:27[IDC]Dragonthanks, it's a start ;-)
21:35:53 Quit Shagnar ("( :: NoNameScript 3.81 :: )")
21:40:05 Join Bippy [0] (
21:40:10*amiconn did a similarly trivial commit
21:40:35 Join zezayer [0] (
21:40:48[IDC]Dragonhey, join the frenzy!
21:42:33[IDC]Dragonthe main problem with my "bootbox" project is to setup a makefile
21:42:35amiconnI'm tempted to simply commit the fat patch.
21:42:52amiconnI'll do another little test before
21:45:13amiconn[IDC]Dragon: Btw, speaking about bootbox and knowing the need for proper charging: For me, it seems the tweaked charging algorithm performs worse than the old.
21:45:14 Quit Bippy ("CGI:IRC (EOF)")
21:45:40[IDC]Dragonoh, really?
21:46:03[IDC]Dragonit worked for me, but I'm not using the box much atm.
21:46:17amiconnWell, it immedialety starts charging, unlike the old algorithm, but it seems that it stops charging prematurely
21:46:42amiconnI get fairly bad runtimes, even with my new 2500 mAh cells
21:46:47[IDC]Dragonhow prematurely?
21:47:49t0maswhat's bootbox? (nothing about it in wiki?)
21:49:27[IDC]Dragonnot everthing is in wiki...
21:49:57t0masthat's why I asked what it is :)
21:50:00[IDC]Dragonwe're short of space for rombox
21:50:23[IDC]Dragon(Archos only issue)
21:51:06[IDC]Dragonbecause that image is uncompressed, and there's also the compressed Archos image in front of it
21:51:07t0masah, I read that somewhere...
21:51:51[IDC]Dragonso bootbox is a stipped-down rockbox, meant to replace the Archos image
21:52:07t0masah ok
21:52:17[IDC]Dragoncapable of just charging, USB mode, booting
21:52:27t0masis it possible to flash archos without a booting firmware?
21:53:02t0masdo you need a working firmware to flash something in (like on iriver), or is it possible from the pc?
21:53:45[IDC]Dragonyou can do it from PC (serial), but need a small hardware mod to enter that mode
21:53:54t0masah ok...
21:53:57[IDC]Dragonelse, from rockbox
21:54:08t0masso it is a little dangerous to remove the archos firmware...
21:54:25[IDC]Dragonnot for me, I have that mod
21:54:29t0masah ok
21:54:31amiconn[IDC]Dragon: It charges around 4..5 hours at maximum (from empty cells). I'd expect more like 15 hours for a full charge
21:54:37t0masfor me neither... as I don't have an archos...
21:55:07[IDC]Dragonamiconn: can you report that to Jerry?
21:55:49amiconnPerhaps I should do that log thingy first that he built in for debugging, and send it along
21:58:06[IDC]Dragonsounds fair
21:58:12amiconnRegarding bootbox: Funny enough, I have exactly those archos devices that are capable of running rombox. All those devices I don't have also can't run rombox
21:58:45[IDC]Dragonso you have little motivation for this
21:59:23[IDC]Dragonif bootbox works, I want to change the flash zoo to rombox only
21:59:53 Quit zezayer ("Chatzilla 0.9.67 [Firefox 1.0.1/20050311]")
22:00:15[IDC]Dragonand we can freely distribute the starter image, since it contains 100% our code
22:00:38[IDC]Dragonmy webspace expires this month
22:00:44amiconnI know
22:01:00amiconnMy old page does expire as well
22:01:07amiconn...but I have a new one
22:01:28[IDC]DragonI get a new 10 MB, but don't know yet where and how
22:02:45[IDC]Dragonand am this −−><−− close to switch to 1&1
22:02:49*amiconn has 200 MB and an own domain :)
22:03:19[IDC]Dragondyndns? ;-)
22:03:24amiconnMy content is in urgent need of a mega-update though
22:03:27*rasher has "unlimited" space and own domain
22:03:43[IDC]Dragonunlimited sounds impossible
22:04:03amiconn[IDC]Dragon: That's only for occasional use, as it's not always up.
22:04:13[IDC]Dragonthen shell out some ftp accounts! ;-)
22:04:15rasherwell, not unlimited, but limited by the size of the harddisk in the server :)
22:04:16amiconnIt's the fastest way to publish small files though
22:05:56*rasher tries installing phpsysinfo on a windows xp box
22:06:04rasheruh, nt2000
22:08:12*amiconn just noticed that he didn't update his site for over 2 years
22:11:00 Quit nobby ()
22:22:03***Saving seen data "./dancer.seen"
22:32:38HClwhat does ymmv mean..
22:33:34rasheryour milage may vary
22:33:49HClaha o.o.
22:33:54*HCl blinks.
22:51:34*preglow doesn't have a site
22:51:53preglownot anything with notable content, at least
22:52:37rashermine's pretty dull as well
22:53:17preglowi've got a random index.html experiment i did in mason, and that's pretty much that
22:54:04HCli have a site.
22:54:07HCl :3
22:59:43*HCl needs to talk to bagder
22:59:58Bagdertalk my son
23:00:02HCl :P
23:00:10HClyes um, its hard to phrase the question.
23:00:21HClbut i was sort of wondering where we stand about the database about now o.o.
23:02:09BagderI was thinking if we perhaps should go back and consider one single db
23:02:32HClthen it'd be completely impossible to do a db upgrade without having the old one
23:02:43HCli hope you understand what consequences that would have o.o
23:02:46HCli don't quite remember who
23:02:53HClbut *someone* had an extreme objective against that.
23:02:59BagderI did
23:03:01rasherthat was Bagder
23:03:07HClyou changed your mind? o.o.
23:03:07Bagderbut the thing is
23:03:15Bagderall this new stuff is equally bad
23:03:19Bagderin my mind
23:03:33Bagderso whereever I turn, I don't see a design I like
23:03:46HCli guess, i'm always open for other suggestions
23:03:56Bagderso, I'm continuing thinking
23:03:57HClthough i will point out problems we would have with certain designs..
23:04:23Bagderand I'm all for starting with the existing suggestion
23:04:37Bagderwe can always fix and improve things later
23:04:41HClso what we discussed this afternoon
23:04:49HClwith the filenames -> tagdb/rundb id
23:04:50HCl ?
23:05:14HCli outlined it on the wiki a bit, but i'm scared to mention it because it closely resembles my first design :(
23:05:20*HCl goes to his mewing cat :/
23:05:55rashercats :(
23:05:58*rasher sneezes
23:07:28HClmy soft and fluffy alarm clock :3
23:08:38HClbut yea, we can always improve later
23:11:05HCloh wait, misread :x
23:11:10*HCl waits for reply :x
23:13:25 Join asdsd_ [0] (
23:14:32 Quit asdsd_ (Client Quit)
23:20:44 Quit [IDC]Dragon ()
23:27:58 Join amiconn_ [0] (
23:28:10 Quit amiconn (Nick collision from services.)
23:28:10 Nick amiconn_ is now known as amiconn (
23:36:53*HCl prods Bagder one last time, otherwise he'll just assume that we're going with the current design for now, till something better comes along
23:37:14Bagdersure go ahead, I'm a bit busy
23:57:29 Join ehntoo [0] (

Previous day | Next day