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 2005-04-30

00:03:12amiconnAh, working now :)
00:03:18HCl :P
00:03:23HClamiconn: read my question :x
00:03:37amiconnThat only needs some testing on the other units now.
00:03:52amiconnI think I can give your database query thing a shot
00:04:04HClyea, i mostly meant my new tool to create search files :/
00:05:46amiconnrebuilding from current cvs...
00:05:52HClits not in cvs yet..
00:06:16HClits pretty much a new version of tokentool, only much, much smarter.
00:06:34HClfirst version was raw, this one is rather smart and will not allow you to create an invalid search query
00:06:37amiconnI'll still need current searchengine.rock, correct?
00:06:57HClsearchengine is pretty much done, doesn't need anything aside from memory optimalizations
00:07:21HClthis version of tokentool is the first step to a frontend for the searchengine
00:07:22 Quit muesli_ (Read error: 60 (Operation timed out))
00:08:27amiconnHow do I build the tool? There is no Makefile...
00:08:34HClsorry, i was lazy
00:08:43HCljust gcc -o tokentool `find -name '*.c'`
00:09:50HCland yes, i know that the numbers should be at the front of the line xD *fixes*
00:09:55amiconn==> core dumped
00:09:56HClnot working
00:09:59HClwoah o.o;
00:10:00HClwhat happened?
00:10:05HClcan you do a gdb trace?
00:10:19amiconnI simply pressed return on the queries
00:10:55HClumyea, :P you shouldn't. you need to pick one of the choices
00:10:59HClits still a development tool
00:11:01HClnot end user
00:11:11HClits pretty much to demonstrate the new editor core
00:12:12amiconnHmz. I don't understand the choices :/
00:12:18HClhold on
00:12:24HCli changed it slightly :X
00:12:36amiconn"numberproperty" exists twice?
00:12:59HClno, only once o.o
00:13:02HClreget it from my http?
00:13:18HCli changed the code to make what you can pick more readable
00:14:03HClyou can choose one of the numbers it lists, "<number>" is a direct number
00:14:05HCllike 1980
00:14:10HClwhere numberproperty is something like year
00:14:11HClor rating
00:15:25amiconn"numberproperty" is still displayed twice
00:15:37amiconn$ ./tokentool.exe
00:15:37amiconnOutput filename? test1.rsp
00:15:37amiconn(1) "not"
00:15:37DBUGEnqueued KICK amiconn
00:15:37amiconn(2) "("
00:15:37amiconn(3) "<number>"
00:15:37***Alert Mode level 1
00:15:37amiconn(4) "<numberproperty>"
00:15:39amiconn(5) "<string>"
00:15:41amiconn(6) "<numberproperty>"
00:15:43amiconn( ? )
00:15:45amiconnToken ?
00:15:54HClyou're so right
00:15:56HCllet me fix that
00:16:25Epoch`_you guys are some pretty focused blokes. =]
00:16:29HClthere, get it now :P
00:16:34HClno, i'm just tired
00:16:36HCland i have been all day
00:16:37HCl :P
00:16:46HClamiconn: reget it from http :)
00:16:56HClthe 6 should've read string property, 'twas a bug
00:16:58preglowi'm going to be shit tired tomorrow
00:17:13preglowconsidering the fact that the area outside is filled with people playing shitty music
00:21:22CtcpIgnored 1 channel CTCP requests in 0 seconds at the last flood
00:21:22*amiconn tries to understand the search expression language
00:21:45HCli wrote a description on the pluginsearchengine page in ebnf, but thats not very readable
00:22:00HClpretty much, you can compare two numbers with eachother, and you can compare two strings with eachother
00:22:14HCland you can use brackets and "and"s and "or"s in whatever way you like
00:22:19HCllike it were a calculator
00:23:03HClwhere a number can be a number or something like year or rating
00:23:11HCland a string can be a string or something like artistname
00:23:42HClyear >= 1980 and (artist contains "nightwish" or artist contains "air")
00:23:44HClstuff like that.
00:23:51Epoch`_Between version 2.2 and 2.4 did you guys decrease the startup load times for playing?
00:24:10oddHCl: that is profoundly awesome
00:24:18HClthanks :p
00:25:31oddcan those attributes be selected from a list from the tagdb, or must they be entered manually?
00:25:38***Alert Mode OFF
00:25:40***Saving seen data "./dancer.seen"
00:25:41amiconnToken ? 5
00:25:42amiconnString? angel
00:25:42amiconn(1) "tokentypeerror"
00:25:42amiconn(2) "tokentypeerror"
00:25:42***Alert Mode level 1
00:25:42amiconn( B@ ? )
00:25:43***Alert Mode level 2
00:25:43amiconnToken ?
00:25:52amiconnWth does that mean?
00:25:54HClthats not right...
00:26:00HClit should've printed angel
00:26:03HClwhere B@ was..
00:26:22HCltokentype errors o.o;
00:26:24oddbats in the code
00:26:36odd...ok i really need sleep
00:26:48*odd has been working furiously for like 20 hours straight
00:26:50HCli'll look at that.
00:26:53HClsorry amiconn :X
00:26:59HClit seems i have a bug, heh.
00:27:08amiconnHCl: I couldn't figure how to specify "contains"
00:28:03amiconnSimilar thing happens when I start with the stringproperty
00:28:17stripwaxHi guys. Anyone know if Linus' latest iriver bootloader fixes the reported incompatibilities with the latest iriver firmwares?
00:28:23amiconnStringproperty itself works, but then again -> "tokentypeerror"
00:28:27stripwax(like 9 days ago in CVS or something)
00:28:36HClamiconn: yea, i got a severe bug, apparently.
00:28:41HClstripwax: no.
00:29:24stripwaxNo, it doesn't, or no, no-one knows?
00:29:37HClit doesn't.
00:29:49*HCl ughs and wonders what kind of bug he's having now..
00:31:09HClyea, i kind of forgot to add contains and equals
00:31:10*HCl fixes.
00:31:29preglowstripwax: don't think anyone knows where that bug comes from
00:31:37preglowstripwax: that is, i think bagder mentioned linus had some ideas
00:32:37*amiconn realises that the simple search already present in the db browser will probably always be sufficient for his search needs
00:33:31preglowdon't think i'll ever need a search
00:34:29amiconnYeah, not really...
00:34:37HCli know what happens.
00:34:59amiconnThe db browser itself is useful for one thing though - detect improper/ inconsistent tagging
00:35:03stripwaxpreglow/HCI ok thanks..
00:35:44***Alert Mode OFF
00:37:08amiconnHCl: Does the searchengine read the whole database linearly when executing a query?
00:38:16HClamiconn: pretty much.
00:38:24HClamiconn: try now :x
00:40:09HClamiconn: it does some selective querys when "and"s are involved, but otherwise, there's not much to optimized
00:41:30amiconnI put the generated .rsp on my Ondio, and ran searchengine
00:41:44amiconn=> CPUAdrErr at 091F906A
00:41:55HClit might've ran out of memory
00:42:08amiconnNo, that looks like an unaligned access to me
00:42:26amiconnI'll check the .map and the disasm
00:42:31HClthat was something i had to fix, hold on
00:42:34HCli think i might know..
00:43:47amiconnMost likely it's in loadentry()
00:43:58HClor that
00:44:01HClor the token
00:44:05HClthe token isn't aligned either
00:44:06HCllet me fix..
00:44:15amiconnThe token is a byte?
00:44:57HClfixed that
00:44:58HCllet me commit
00:45:04HCldon't forget to make the same change in edittoken
00:45:07amiconnThen it doesn't need a special alignment
00:45:09HClor your tool won't create correct tokens
00:45:20HClno, the byte is making the long that follows after a while
00:45:32amiconnAh, yes
00:45:37 Join xen` [0] (
00:46:22HClcommitted a fix
00:46:32HClcheckout, and change the length of the string in struct token to 255
00:46:38HClit should make it aligned again
00:52:25amiconnHmm, same result :(
00:54:36HClthen i dunno :X
00:57:35HClmaybe try with a smaller id3 database? i dunno how many songs you got on there and how much ram the thing has and what searchquery you submitted
00:57:53amiconnThe db is rather small; 41 KB
00:58:04HClshould fit then
00:58:14amiconnSearchquery is really simple: <title> contains "angel"
00:58:25HClokay, that shouldn't cause much trouble
00:58:30amiconn(to compare with the built-in search
00:58:45amiconnShould match 3 titles
00:59:00amiconnCurrently looking at the disasm
01:00:02 Part stripwax
01:01:34amiconnHCl: The error definitely happens in dbinterface.c, line 77
01:01:39*HCl checks
01:01:50amiconnIt seems that entryarray itself isn't properly aligned
01:01:57HClhow do i do that?
01:01:59amiconnWhere is that done?
01:02:03HClin the malloc.
01:02:14HCl entryarray=(struct entry *)my_malloc(sizeof(struct entry)*rb->tagdbheader->filecount);
01:02:20HClthen my_malloc isn't aligning properly
01:02:25amiconnAh, yes
01:03:29amiconnCheck the malloc in xxx2wav.c
01:03:40amiconnThis one does proper alignment
01:03:49HClbit busy at the moment
01:03:52HClwill do when i got time
01:04:00HClfeel free to fix it if you can and have time
01:04:27 Join TCK [0] (
01:11:27amiconnHCl: Is there a reason why your malloc() reserves 4 extra bytes per call?
01:12:02preglowunfinished alignment? :P
01:12:24amiconnDefinitely not for alignment; it's a literal "+ 4"
01:15:15HClamiconn: no XD
01:15:20HClamiconn: checkout
01:15:23HCli tried to fix
01:15:34amiconnI have a fix myself...
01:15:46HClcan you see if mine is correct? :X
01:16:19amiconnHmz. Now it doesn't crash, but it tells me "unknown stringid intvalue"
01:16:59HClsounds like endianness messed up..
01:18:08preglowok, ok, NOW i understand nothing, it's not windowing that's slowing things down
01:18:20preglowmy new imdct_l is plain and simple just a ton slower than the first one i wrote
01:18:28preglowhow this can be, i will never understand
01:18:32amiconnThere's a slight problem with your malloc() fix. If you bump the buffer pointer, buffer_free should be decreased by the same amount
01:18:41HCldidn't i do that? o.o
01:18:51HClat the very start.
01:19:23amiconnHmm. How do I stop the searchengine? It seems to hang...
01:20:47 Quit TCK- (Read error: 110 (Connection timed out))
01:21:24HClit might be very slow, it doesn't have an abort
01:21:33*preglow tears his hair out
01:24:00HClamiconn: did it crash?
01:25:00amiconnIt displays some lines of text, then the splash "unknown stringid value". Then it sits there forever
01:25:12HClit might crash o.o;
01:25:13HCli dunno.
01:25:38HCli dunno what caused that
01:25:43HClcan you send your generated search file to me?
01:25:47HCli'll analyze it tomorrow
01:26:23amiconnBtw, it should tell the wrong value - handy for debugging
01:26:36HCli'll fix that
01:26:41amiconnunknown stringid intvalue %d
01:26:50HCli had the odd impression that no one would ever pass an unknown stringid
01:27:12amiconnYour tokentool did...
01:27:31HClits most likely an endianness problem
01:27:39HClmaybe your gcc didn't define little endian?
01:27:50HCl(that the host is little endian)
01:27:54amiconnHumm? Where?
01:28:11HClwell, with my gcc its defined and included by default..
01:28:18HClcan you like
01:28:24HCltry doing #define LITTLE_ENDIAN
01:28:29HClin databox.h
01:28:35amiconnThen it's definitely that
01:28:36HClsave your old search file
01:28:40HCland diff them
01:28:43amiconncygwin never defines that
01:28:44HClsee if that makes a difference
01:28:45preglowyou actually ended up calling it databox?
01:28:50HClpreglow: yup :P
01:28:51preglow\o/ for me
01:29:12amiconnHCl: That's the reason for my odd cygwin workaround in dbtree
01:29:34preglowi'd kill for a debugable target right now
01:29:34HClamiconn: *nods*
01:29:57*amiconn hides from preglow
01:30:05HClamiconn: let me slap that into my tool and fix my tgz on my site
01:30:38HClactually, i don't have it anymore since it got removed from cvs..
01:30:46HClsince we had a configure script that checks it now
01:31:15HCli hope you managed to reset your player
01:31:22amiconnHCl: I see RED
01:31:29HCl*goes to fix*
01:31:44preglownice red line
01:32:03amiconnHCl: Resetting the Ondio is simple. If you hold the OnOff button for ~10 secs, it powers down by "hardware force"
01:32:42preglowwhat the hell
01:32:51preglowmy h120 has locked up on the Resume? screen
01:33:08amiconnThe hw poweroff timeout is even lower on the recorder and the player - ~2 secs
01:33:13*preglow turns off hold and tries again...
01:33:32amiconnIt's so short that we can't do safe shutdown if the user tries to powerdown by holding OFF...
01:34:14HCli might've added a warning
01:34:18HClbut i'll sacrifice that over fixing red
01:34:50HClpreglow: :P
01:35:53preglowfucking drunk idiots playing football around/on/at the parked cars
01:36:02preglowhad i owned one of them, i would most probably have killed them
01:36:33preglowthe spanish exchange students around here try their very best to act like assholes
01:36:51preglowthey succeed admirably
01:38:25*HCl should go to sleep
01:38:30HCland more, test code tomorrow..
01:38:40HCli haven't even tested my searchengine optimalizations yet
01:38:47HCli don't even know if they work ;x
01:38:58amiconnThere'll be yellow builds now, probably
01:39:03HClsorry :/
01:39:11HCli couldn't find a solution fast
01:39:15HClif you know one, please fix
01:39:28amiconn(lvalue cast)
01:39:44HCloh wait.. i know one, i think :X
01:40:14preglowamiconn: ondio uses sh1 as well?
01:40:29amiconnThere's a new H-140 on ebay germany, 8 hours to go...
01:40:33amiconnpreglow: Yeps
01:40:53amiconnThe hw platform is very similar to the hd based jukeboxes
01:41:08HClamiconn: ohh, bid?
01:41:19HClat what price is it?
01:41:51amiconnpreglow: It is really very similar; one of our plugin devs did a funny experiment. He tried to load an Ondio firmware on his jukebox FM recorder
01:42:19amiconnGuess what? It booted a little, showing to boot screen flipped, then hung
01:42:29HClits fairly cheap o.o
01:42:35HClamiconn: lol.
01:42:42amiconnThen [IDC]Dragon and me decided to port rockbox to the Ondio
01:43:03amiconnHe aquired an Ondio FM, and I got an SP
01:44:02HClamiconn: that h140 offered looks good o.o
01:44:45*HCl goes to bed..
01:44:52HClafter everything is green again
01:47:22 Part MoosCamaro
01:47:37*HCl waits :/
01:47:55preglowi wont
01:48:19 Quit preglow ("jeg kaller dem en floskel!")
01:49:21amiconnHCl: I recompiled tokentool with LITTLE_ENDIAN, and updated searchengine.rock
01:49:27HCldid it work?
01:49:41 Quit ncrawla (Remote closed the connection)
01:49:43amiconnNow it goes a little farther, and does no longer hang
01:49:53amiconnThere's still an error though
01:49:57HCl :/
01:50:01HClwhats the error?
01:50:21amiconn'4' found where CONTAINS/EQUALS expected
01:50:28amiconnfollowed by
01:50:31amiconnHits: 0
01:50:36HCla 4 is..
01:51:14HClyou didn't change the struct token
01:51:15HClin the tool
01:51:33amiconnI did... but then re-extracted the tgz
01:51:42HCltgz doesn't contain the fix
01:53:26amiconn−−> 'unknown stringid intvalue' :(
01:53:33 Join DMJC-L [0] (
01:54:09HClodd :/
01:54:16HClhold on
01:54:23HCli'll add debug code to the searchengine
01:54:28amiconnHmm, seems like this has to do with struct packing.... or endianess, again
01:54:28HCland make it not crash as well.
01:54:40Rickdid anyone else get spammed
01:54:46Rickabout some survey
01:55:38amiconnHCl: Changing the string to 255 chars decreased the file size by _12_ bytes
01:56:02amiconn..though it contains 3 tokens
01:58:25HClwell, it should contain 4..
01:58:30HClEOF is a token too
01:58:45HCli committed a patch in cvs that makes error reporting better
01:58:50HCland also prevents it from crashing on error
01:58:59HClit should prevent your ondio from crashing
01:59:50amiconnThe int value for the first token should be 4 iiuc, but the field contains 0x04 0x00 0x00 0x00
02:00:00HClthats confusing..
02:00:04amiconnThat's definitely not 4 when read with a big endian machine...
02:00:24HClit should be safe to change it to a long, and change it back to 256..
02:00:30HCloh wait..
02:00:33HClno, thats not safe..
02:00:43amiconnI'd keep the 255
02:00:43HClsince it doesn't do endianconversion for it at the moment
02:02:01amiconnIt would... but it can't
02:02:15amiconnYou said I should #define LITTLE_ENDIAN in databox.h
02:02:28amiconn..but databox.h doesn't get included in tokentool.c
02:02:37HClyou can hit me if you want
02:02:40HClit should be included
02:04:03amiconnFile looks much better that way
02:04:11HCl :)
02:04:17HClrecompile the .rock too
02:04:21*amiconn thanks his hex editor ;)
02:04:21HClit should be safer now
02:04:57amiconnHits: 3 :)
02:05:03amiconn..even with the old .rock
02:05:50HCli only added error reporting :)
02:06:00HCland better error handling
02:06:12amiconnsearch.m3u is generated correctly
02:06:17HCl :)
02:06:44amiconnNighwish - Wish I Had an Angel
02:06:56amiconnWithin Temptation - Angels
02:10:40HClhm, i have a slight bug..
02:10:50amiconnI have an idea how the search token file could become much smaller
02:12:20amiconnThe string is the largest part of the token by far. What if you replace it py a pointer, and append the strings after the eof token, null terminated, and let the pointers point to where the string starts?
02:13:23HClyou could, but.. they're fairly small, and it would increase code complexity..
02:13:41amiconnI wouldn't expect a query to become larger than a few kb, so it should be possible to load them at once
02:14:46amiconn..and it only requires to walk all tokens once, adding the start address to the loaded pointers to get real char pointers
02:15:09*HCl smacks himself upside the head for a dumb bug.
02:15:34amiconnIf your code reads every token once, and exactly once, the extra walk isn't even necessary
02:15:47HClyea, it reads them once.
02:15:52HClonly the editor has to know about all of them
02:17:18t0masand wonders why HCl is still awake
02:17:55HClgood point.
02:17:59HCli have to go to sleep
02:18:09HClamiconn: i'm prolly not awake enough to understand what you're saying for 100%
02:18:17t0masyou slept this afternoon right?
02:18:22HClplease remember and explain again tomorrow
02:18:25HCla little, yea
02:18:40t0mas*yaaaaaaaawns again*
02:18:41HClamiconn: also check the dbinterface.c under the loading of artists and albums...
02:18:47t0masgood night
02:18:48amiconnProbably you're right... I should go to sleep too
02:18:51HClbig, big optimization possible there
02:18:55HClhehe :)
02:24:50 Quit Aison ("( :: NoNameScript 3.72 :: )")
02:25:41***Saving seen data "./dancer.seen"
02:31:52 Quit TCK ("Can you say "roflpwnt fgt" in IRC?")
02:32:51 Quit cYmen_ ("zZz")
02:49:48 Join ismi [0] (
02:53:30 Quit ismi (Client Quit)
03:29:06 Join TCK [0] (
03:37:39 Quit ferenczy (Read error: 60 (Operation timed out))
03:38:41 Quit xen` ()
04:03:59 Quit TCK (
04:03:59 Quit Ka_ (
04:03:59 Quit crash_ (
04:03:59 Quit Hadaka (
04:04:50NJoinTCK [0] (
04:04:50NJoinKa_ [0] (
04:04:50NJoincrash_ [0] (
04:04:50NJoinHadaka [0] (
04:05:24 Join QT_ [0] (as@area51.users.madwifi)
04:08:35 Quit QT (Read error: 60 (Operation timed out))
04:25:43***Saving seen data "./dancer.seen"
04:42:17 Join asdsd_ [0] (
04:43:10DBUGEnqueued KICK asdsd_
04:43:10asdsd_,00hey guys
04:43:25asdsd_,00have u all seen the rockbox mp3 player
04:48:19 Join ashridah [0] (
04:49:30CtcpIgnored 1 channel CTCP requests in 0 seconds at the last flood
04:49:30*odd /clears
04:49:53***Alert Mode level 1
04:49:53asdsd_,00yo odd
04:53:03 Part asdsd_
04:59:54***Alert Mode OFF
05:11:00 Part odd
05:11:24 Join odd [0] (
05:47:13 Quit ashridah ("Leaving")
06:25:45***Saving seen data "./dancer.seen"
08:07:24 Join Harpy [0] (
08:19:27 Part DMJC ("Leaving")
08:25:49***Saving seen data "./dancer.seen"
08:35:37 Quit edx (Read error: 145 (Connection timed out))
09:26:10amiconnGood morning
09:28:32 Join ashridah [0] (
09:45:18 Join [IDC]Dragon [0] (
09:45:45*[IDC]Dragon spots Bagder + amiconn
09:47:27[IDC]Dragonbut they don't spot me
09:49:23webminddoes anyone know the print's design?
09:49:29webmindget the feeling it's multilayer
09:49:45webmindand wonder what parts are used
09:49:49webmindfor what
09:50:00webmindso I can see what could be cut away
09:52:08amiconnHi Jrg :)
09:52:23[IDC]Dragonhi Berlin
09:52:32 Join preglow [0] (
09:52:32amiconnMy 2 GB MMC is working :)
09:52:47[IDC]DragonI'm trying to understand how the .lds file is built
09:53:04[IDC]Dragoncan't see who's generating it
09:53:24amiconn[IDC]Dragon: I had a look at your usb-before-mmc patch. It worked for me without problems, but then this was with the new mmc driver
09:53:43[IDC]Dragonit mostly worked for me, too
09:53:43amiconnNevertheless I fixed the issues you marked in main.c, and also some others
09:54:07[IDC]Dragonwhich + how?
09:54:21amiconnE.g: The card can only be considered non-touched if this was a coldstart
09:54:36[IDC]Dragonah, yes
09:55:05amiconnI'll make a new patch; could you test that? Maybe your cards are different...
09:55:29[IDC]Dragonfile it into the patch tracker
09:55:32amiconnAnother issue was that you relied on card->initialized
09:55:50amiconnI added separate "mmc touched" tracking to the driver
09:56:17[IDC]Dragonwe talked about it, and I didn't do so for a reason
09:56:37[IDC]Dragonwhich I don't remember now, could have been laziness ;-)
10:03:32amiconnPatch uploaded. I did not yet test it on the jukeboxes
10:03:56[IDC]Dragonthanks for looking at it
10:04:02amiconnBtw, which .lds file are you looking for?
10:04:28[IDC]DragonI tried a bit on "bootbox"
10:04:48amiconnSo you mean the main .lds, right?
10:05:08[IDC]Dragonby allowing the bootloader option in "configure" for the SH model, too
10:05:44[IDC]Dragonnow I can create a project which should become our bootbox later
10:06:29[IDC]Dragonbut the file being generated is all wrong
10:07:41[IDC]Dragonin the firmware dir, we have 3 .lds files, which are all meant to be preprocessed
10:07:57[IDC]Dragonbut I don't know from where this is done
10:08:09amiconnIt's done from apps/Makefile
10:09:02amiconn..building and from and rom.llds, respectively is processed from bootloader/Makefile
10:09:59[IDC]Dragonor, in my case
10:10:23[IDC]Dragonah, I see it
10:11:23 Join DMJC [0] (
10:11:37[IDC]Dragonso I either upgrade boot.s for the SH, or modify the makefile generator to use here
10:12:07amiconnI think you'll need to update
10:12:21[IDC]Dragonyes, I think so, too
10:12:36amiconnIf bootbox should run from rom, it will be similar to
10:12:55[IDC]Dragonno, it should run from dram
10:13:03[IDC]Dragonso that I can compress it
10:13:31amiconnHmm. Do you think it will become that large?
10:13:53[IDC]Dragonit's just unnecessary to have it plain
10:14:21[IDC]Dragonplus, the first image moves with the bootloader
10:14:40[IDC]Dragonit's not sector aligned, starts right after the bootloader
10:14:49amiconn..and the second one moves both with bootloader and first image
10:14:59amiconn...still we link to a fixed address
10:15:05[IDC]Dragonbut is aligned
10:15:26[IDC]Dragonand I hope the size will not move into a new page
10:16:01amiconnIf you get this to work, we need to add a big fat warning
10:16:30amiconnFlashing an older rombox.ucl to the new image or vice versa will lead to non-working rombox
10:16:45[IDC]Dragonwhy? should be safe to use with F1-start
10:17:22amiconnYes, but people might not expect this there should be a warning
10:18:03[IDC]Dragonfrom the flash plugin or so
10:18:08amiconnToo bad the .ucl files don't contain a field that allows to check the linkage address
10:19:04[IDC]Dragonwe can check the very first bytes of the payload, those are literal, before the compression kicks in
10:19:29[IDC]Dragonand that's the reset vector
10:19:58[IDC]Dragonmy rockboxing timeslot is running out, cu later
10:20:10 Quit [IDC]Dragon ()
10:24:07 Join edx [0] (
10:25:52***Saving seen data "./dancer.seen"
10:42:33 Join amiconn_ [0] (
10:51:20 Quit amiconn (Nick collision from services.)
10:51:20 Nick amiconn_ is now known as amiconn (
11:08:04*HCl wakes up a bit
11:08:16HClyay for good night's sleep
11:14:17amiconnNo parsing nightmares this time? ;-)
11:14:57amiconnI'll try searchengine with the much larger db on my recorder in a sec
11:23:04amiconnWorks correctly, but the search took ~15 sec compared to ~3 sec the internal db search takes
11:23:14amiconn(same simple query as last night)
11:27:32HClamiconn: lol, no
11:27:48HCli'm working on the editor
11:27:52HClhead is clear again :P
11:28:05amiconn*good :)
11:32:25amiconnHmm, searchengine doesn't get cleaned on 'make clean'. Me fix.
11:32:52HClthe 15 sec longer is most likely because the searchengine works on files listed in the filename structures
11:33:06HClwhile the internal db search works on the song structures
11:33:42HClthe searchengine has to do this because of the rundb info it'll have.
11:34:11HCli'm hoping it should be near 3 seconds if you search for a file by filename with the searchengine
11:34:23HCleither way, i never aimed for speed
11:35:32amiconnAnyway, it should be possible to speed things up a bit
11:36:13HClwhats the proper way of getting the length of a file in normal unix FILE * ?
11:36:27HClapparently, fseek(fp,0,SEEK_END); crashes when length is 0
11:37:00ashridahit does? strange.
11:37:23amiconnI have no idea.
11:37:29HClyea, this is production stage linux too.
11:37:41ashridahhrm. the file descriptor itself should have the length too iirc, but i have no idea if it's a standard field
11:37:50amiconnI never used the high-level file i/o on *nix so far
11:39:13HClmaybe i found a bug in glibc o.;
11:41:50HClerror checking is different
11:43:05 Quit preglow (Read error: 110 (Connection timed out))
11:50:00amiconnHCl: Hmm, I tried searchengine on the player now
11:50:19amiconnDoes also work, but the search time seems to depend non-linearly on the db size
11:50:40HClwhat do you mean?
11:52:02amiconnThe db on the player is 542 KB, the search took 10 sec. The db on the recorder is more than twice as large, 1.2 MB.
11:52:12amiconnHowever search time is only 15 sec
11:53:28amiconnPlayer and recorder have almost the same cpu speed, the player is even a little bit faster (12 MHz, recorder is 11.0592 MHz)
11:54:26HClthe time should be linear to the amount of songs
11:54:29HClnot the db size
11:55:17HClit loads all the file entries, and any data thats requested from that point on
11:56:31amiconnThat's still not linear. Player: 1651 songs, recorder: 2992 songs
11:57:05amiconn(but it comes closer than when comparing db size)
12:00:52HClodd, then o.o
12:03:19*HCl improves his tokentool..
12:06:50HCli think i'm getting close
12:07:16HCli already fixed it so that it automatically enforces the current query to be correct, throwing out incorrect tokens till its correct again
12:20:17 Join MoosCamaro [0] (
12:20:28MoosCamarohey all
12:22:47*HCl sighs at stuff
12:25:53***Saving seen data "./dancer.seen"
12:35:59HCli suck so much.
12:36:18HClindex=tokencount++; gives the *OLD* value of tokencount, duh.
12:37:35HCli hate this..
12:37:47amiconnIt has to be that way..
12:37:51HCli know.
12:38:01HClfor some reason my editor code has become really untransparant and unmaintainable.
12:39:04*amiconn just had an idea how to make sure the mmc was not yet switched to spi mode...
12:40:15amiconnThere are some situations where rockbox assumes it was/was not touched
12:40:35amiconnBut we can ask the card when in doubt, without actually switching it
12:42:41*HCl wtfs @ stuff..
12:42:54HClmy program loops, but where the beep does it loop..
12:43:50amiconnIt does beep??? ;-)
12:44:19HClah i think i got it...
12:45:00HClnow to see what causes it..
12:46:47HClnailed it.
12:46:57HClthis is why i hate not having exceptions in C
12:47:15HCland instead having to have a global variable that every function needs to check and if its 1, return return return
12:49:07ashridahyes, because, you know, exceptions have done such a wonderful job for embedded java's performance :)
12:51:32amiconnHCl: Go ahead and implement setjmp() / longjmp() ...
12:58:00HClamiconn: willing to test a new editor for me?
13:01:48HClwith the request to try to break it (without feeding it an invalid choice)
13:02:34HClamiconn: maybe later..
13:05:32*HCl can't wait to be able to rate songs, and select by rating...
13:07:45HClthe new editor should allow fully editing an expression..
13:08:11HCli'm not 100% sure how well it works :/
13:09:15HClnot too well :/
13:10:33*HCl goes to change...
13:21:47 Join matsl [0] (
13:23:59t0mas[tomas@nigel dev]$ time ./dict fructose
13:23:59t0masFilesize: 5263632 bytes = 146212 words
13:23:59DBUGEnqueued KICK t0mas
13:23:59t0masFound: fructose at offset: 5997434
13:23:59t0masDescription: a simple sugar found in honey and in many ripe fruits
13:23:59***Alert Mode level 1
13:23:59t0masreal 0m0.110s
13:24:01t0masuser 0m0.030s
13:24:03t0massys 0m0.015s
13:29:40ashridah.03 seconds? bah. should be .003 :)
13:30:29HClwho has a build environment
13:30:33HCland is willing to test my editor
13:30:36CtcpIgnored 1 channel CTCP requests in 0 seconds at the last flood
13:30:36*HCl improved it a bit :/
13:32:14HClno one? :/
13:32:28*HCl guesses he'll go shower and make lunch then...
13:33:24 Join Aison [0] (
13:33:44HCltry to compile that, and try to break it (without feeding it invalid input)
13:34:00***Alert Mode OFF
13:34:32rasherand.. what do I feed it
13:34:49HClit'll show what kind of input you can choose...
13:37:26rasherwhat kind of file do I need to give it
13:37:49HClyou can feed it an empty file
13:37:53HClblah.rsp or so
13:38:04HClor a non existing one
13:38:37HClit doesn't have a good interface yet, its mostly to test and demonstrate the editorcore
13:39:07*HCl will bbiab, shower, lunch, etc.
13:41:56rasherI don't get it
13:42:44rasherthere we go
13:49:56t0mashm... haven't seen Linus for a long time..
13:50:04t0massomething wrong with his wife? Or just a vacation?
13:50:46rasherHe got a new car - could be it :)
13:50:54t0masyes, I knew that...
13:51:04t0masbut he isn't driving it around for 2 weeks right?
13:51:13rasherit hasn't been that long has it?
13:51:20t0masdon't know
13:51:23rasheranyway, I saw him replying on misticriver not too long ago
13:51:27t0mashow's that logbot working?
13:51:44t0mas!seen LinusN
13:52:08t0mas[LinusN] idle 96:31:04, signon: Fri Apr 01 06:50:27
13:52:12rasherlogbot: seen linusn
13:52:12t0masnope... 3 days :)
13:52:28t0masmake that 4 ;)
13:52:42*t0mas missed something in math class long ago ;)
13:53:45 Part amiconn
13:54:07HCli need 3 editor modes...
14:10:52t0masHCl? are there any text formatting functions in rockbox?
14:11:00t0masfor formatting a line to fit on LCD?
14:18:18rasheryou mean to wrap?
14:18:45thegeekthe mains speaker output on my h120 is b0rked
14:19:03thegeekwhen I bend the plug a little the contact is sparkly and bad
14:20:58HCli dunno
14:21:20HClt0mas: sounds like you need to open up and resolder it..
14:21:37t0masrasher: yes
14:21:48t0masHCl: I ment wrapping a line of text...
14:21:56HCli meant thegeek *
14:22:04HClstupid nickcomplete
14:22:07rashert0mas: I don't think so.. there's a function to find the width of a string
14:22:14rasheryou could have a look at viewer.c
14:22:19rasherbut it's rather involved
14:22:29rasherbecause it does a bunch of different things
14:23:19t0masI have a word description... in a char array... and I know the max lenght it will ever have... and ofcourse the lenght of the string...
14:23:32t0masI still have to do the wrapping "manually" ?
14:25:56***Saving seen data "./dancer.seen"
14:31:33t0masok, the next time I'm bored I'll write some functions for it... to use from everywhere...
14:40:39thegeekHCl : yea
14:40:40thegeekI know
14:40:46thegeekI opened it
14:40:58thegeekand poked a round a little
14:41:03thegeekI'm scared to solder it
14:41:08thegeekbecause there are components really close
14:41:27thegeekit does not look like any of the solder points are broken
14:41:31thegeekI dunno
14:41:37thegeekit is not that bad
14:41:44thegeekand the optical/analog out works too
14:41:50thegeekthought without eq
15:04:07 Join nobby-BRB [0] (
15:06:03 Nick nobby-BRB is now known as nobby (
15:08:26MoosCamarohi nobby
15:10:09t0mashi nobby
15:11:17t0masBagder? (or someone else?) the new endian things... how do they work?
15:23:10 Join Lost-ash [0] (
15:23:14 Quit Aison ("( :: NoNameScript 3.72 :: )")
15:23:26 Quit ashridah (Nick collision from services.)
15:23:53Bagder#include <autoconf.h>
15:24:12Bagderif the include path is setup right (-I$(OBJDIR) in the makefile)
15:24:25*Bagder runs away again
15:25:09t0masok, tnx
15:25:16 Join uski [0] (
15:25:35t0mashm... more a C question... but can I substract char* pointers?
15:26:21t0masso when I have a pointer to the 'T' and to the 'm' in this string: "Tomas" and I substrct them I get 2?
15:28:56 Nick Lost-ash is now known as ashridah (
15:39:08HCli absolutely hate this work.
15:39:13HClbut i got a new editor
15:39:15HClwith 3 editor modes.
15:39:22HClsomeone please play around with it x.x;
15:41:49t0masok, it works :D
15:41:55t0masand it's fast like hell...
15:42:19t0masbut the output formatting is a little clumsy... as it just cuts in the middle of words...
15:42:24t0masmy dictionary
15:43:10t0masI have 144895 words in the db atm..
15:43:23MoosCamarocongratulations tomas ;)
15:43:29t0masI you can't count the time needed for searching
15:43:36t0masas it almost instantly returns :D
15:44:09HClsorted database?
15:44:17*HCl wonders why he's even asking..
15:44:47*HCl has had his algorhythm class, knows you can't search faster than O(n) if you don't know anything about your list..
15:46:06ashridahheh. you two should see the scenario we're supposed to develop a fast algorithm for for software engineering.
15:46:57ashridahbasically, picture an ultra-basic car booking system. ie, you can place a booking on a car at a particular date. the car is available from a list, which is whittled down by criteria when booking. simple enough, and easy to do.
15:47:25ashridahthen he extended it for the second/third assignments to include only saving the search criteria with a booking. the car is chosen at pickup time from a list, instead of booking time.
15:47:34 Quit nobby (Read error: 60 (Operation timed out))
15:48:10ashridahthis means that a) we have to ensure that all existing bookings can be satisfied by at least one car matching the criteria. b) we have to work out which cars a booking can take at pickup that will not stop another booking from working.
15:48:25ashridahand preferably, we shouldn't have anything that's factorial or exponential search time :)
15:48:30ashridahpolynomial or faster.
15:48:45*t0mas would stick by the select car at booking thing :P
15:48:50ashridahit sounds simple to begin with. but it isn't. there's way too many variables to produce a simple algorithm bar factorial searching.
15:49:04ashridaht0mas: alas, i've already written and submitted that assignment. :(
15:49:24*t0mas has done this kind of thing on nio...
15:49:38t0masthe informatica contest for the netherlands...
15:49:52ashridahi wish i had the mathematics background in order to solve it properly. any hints? :)
15:50:05t0mas(ended up 13th of the Netherlands)
15:50:53t0masbut the problem's weren't as hard as this one... and we had only 1 hour per assignment
15:51:12ashridahi've been trying to work out a way to balance booking/car counts, so that i can jsut easily work out how many of each i have, and then by selecting one (or adding a new booking) see if there suddenly ends up being too many bookings, but i keep finding scenarios where my algorithm fails :(
15:52:15t0mashm... maybe you should work the other way around? set some conditions for a booking.. and check if enouf cars match those?
15:52:21ashridahthe lecturer seems to think it was possible, and simple. i quickly disabused him of that notion, but he won't change the assignment :)
15:52:46t0mashm... has he done it himself? or is he just hoping you can fix it?
15:53:02ashridaht0mas: well, the problem is basically that some bookings can include 10 or 50 cars just with a simple criteria set (like 'must have aircon'
15:53:18ashridahoh, no, he won't have done it himself. he's more interested in our design, than the algorithm :)
15:53:30ashridahit's software engineering after all
15:54:14ashridahif i was getting this in Computing theory, i would have assumed the lecturer had a range of solutions of various speeds in mind.
15:54:34t0masyes, that's what I assumed too...
15:54:52t0masmaybe ask your computer theory lecturer?
15:55:05t0masthere must be a fast algorithm for it...
15:57:07ashridahyeah. i even experimented a little with some goal-seeking ideas, but that kinda relies on me being able to quantify whether a car/booking assignment is 'better', allowing me to hone in on a pattern that'll prove that the car can be added to the list
15:57:10*ashridah shudders
16:03:58 Join Sucka [0] (
16:25:59***Saving seen data "./dancer.seen"
16:38:00thegeekspam subject of the day: Its finally ok to play with your food!
16:38:26rashersounds like a job for
16:50:03 Quit ashridah ("Leaving")
16:59:23 Join ferenczy [0] (
17:13:14 Quit ferenczy ()
17:30:18 Quit Sucka ("a bird in the bush is worth two in your house")
17:33:07 Join Aison [0] (
17:45:07 Join Sucka [0] (
18:10:00 Quit thegeek (Read error: 131 (Connection reset by peer))
18:10:44 Join thegeek [0] (
18:12:02 Quit thegeek (Client Quit)
18:12:06 Join thegeek [0] (
18:26:00***Saving seen data "./dancer.seen"
18:40:07 Join karim [0] (
18:40:12*HCl yawns.
18:58:35 Quit DMJC-L (Remote closed the connection)
19:03:17 Join cYmen [0] (
19:47:18HCli hate having to design gui code...
19:51:14*HCl ponders porting the chess game from ipodlinux to rockbox
19:54:51MoosCamarogood pondering, and poker game?
19:57:13MoosCamaroi supose you like poker
19:59:57 Quit matsl (Remote closed the connection)
20:01:07 Join matsl [0] (
20:02:39 Quit matsl (Remote closed the connection)
20:04:27HClmaybe someday
20:04:44HClipodlinux' api doesn't seem compatible with ours at all, unfortunately
20:05:53MoosCamarohi HCl, what's about your searchengine?
20:06:21HClits finished
20:06:28HClits just that i'm not good at gui design
20:06:41HCland i'm not really wanting to design an interface for it x.x;
20:07:15MoosCamaroexcelent, you're very efficient
20:08:52MoosCamaroi think preglow is good in gui design
20:09:01MoosCamarobut is not here
20:12:00HCli think i'll start to make an attempt...
20:12:27MoosCamarogood luck
20:12:38MoosCamarowe are with you ;)
20:13:51 Join Stryke` [0] (~Chairman8@resnet-241-86.resnet.UMBC.EDU)
20:26:04***Saving seen data "./dancer.seen"
20:28:08HCligh my flatmates are such jerks.
20:28:12HCli should really move :/
20:29:19*HCl hrmz..
20:34:10courtcWhat are we searching through?
20:39:25HClsong info...
20:39:36HClrating, playcount, title, genre, etc.
20:41:22courtchmmm.. I'll have to steal that ;)
20:41:38HClah :P
20:42:09HClwell, its gpl..
20:42:16rashershould start porting rockbox to ipod instead :D
20:42:29HClthey say copying is the ultimate form of flattery.
20:42:48MoosCamaroit's true
20:43:10courtcor podzilla to rockbox (?)
20:43:46Bagderwhat's podzilla?
20:44:08rasheripodlinux' player
20:44:24MoosCamarohi Bagder
20:44:33courtcthats iPodlinux' GUI
20:44:41rasherthe ipodlinux wiki doesn't seem to have a lot of technical details
20:44:57Bagdergetting another GUI doesn't feel like a priority
20:45:04courtcwhat are you interested in?
20:46:32courtc(technical stuff)
20:46:39t0maswell... there are people intrested in a better GUI
20:46:43rasherNothing in particular
20:46:48Bagdert0mas: of course
20:46:51t0masbut that would be iriver only...
20:46:53rasherJust browsing to see how much detail it had
20:46:56Bagdert0mas: why?
20:47:06rashert0mas: bitmap display only
20:47:06t0masbecause the other screens are pretty small?
20:47:08Bagdeririver users think too much on iriver only
20:47:28Bagdert0mas: the iriver one is small too, just not as small as the others ;-)
20:47:43t0masIf I had an archos recorder... I wouldn't want a more graphical interface... as that's wasting space in the tiny screen...
20:47:51t0masin -> on
20:47:59Bagderlots of people with Archoes want that
20:48:02rashertbqh, I probably wouldn't use a graphical interface on my iriver
20:48:03t0masah... ok
20:48:09t0massounds stupid to me...
20:48:23Bagdercell phones had graphical stuff ages ago
20:48:26Bagderwith silly screens
20:48:36Bagderit worked
20:48:44t0maswell... I was thinking of a graphical interface for iriver... and decided I like it the way it is... as that fits more info
20:48:51courtcheh, I said podzilla -> rockbox as a joke..
20:48:52t0masand the screen is pretty big...
20:49:00t0mascompared to archos recorder
20:50:01t0masbut if people want it... it can be done
20:50:12t0mas(has anybody tested bmp loading on archos?)
20:50:21Bagdermy point was more that there's no need to limit it for a particular platform
20:50:36t0masnope... limiting it to bitmap screen's is ok...
20:51:03t0masonly I think the interface still has to be separate someway...
20:51:35t0masbecause if we build some nokia like interface... with silly pictures for menuitems... we need a bigger image for iriver..
20:51:36rasheryou just need different bitmaps
20:51:53rasherI'd say
20:52:03rasherI mean all of the code could be the same
20:57:45*courtc points to Small( as an option to increase portability between ipodlinux and rockbox :D
20:58:35Bagderhow would that help?
20:58:41rasherimplementing a vm?
20:58:45rashersounds like horror
21:00:23courtcheh, specifically example.small
21:00:59Bagderbut... why?
21:01:36Bagderwe spend lots of time optimizing, just to waste that on a vm?
21:02:06BagderI say we have different goals
21:02:41courtcoptimizing, eh? I thought the iriver had a rather powerful proc?
21:02:48Bagderdoes not
21:03:14rasherAnyway, rockbox runs on far more limited devices as well
21:03:26Bagderbesides, the more we optimize the more we gain battery time
21:06:07courtcheh.. I just thought it was kindof nifty. and ironic (the ipod's apple firmware seems to have tried to use Small a little unsuccessfully)
21:07:29BagderI'm sure it was fun to write for those who did it
21:07:51*courtc <-
21:09:23rasherooh, video of spinning cube :)
21:09:28rasherwonder where that came from
21:09:31t0masvm's are usually just a waste of resources... but for porting to different platforms?
21:09:52t0masoh wait... there was already a rockbox driver like idea right?
21:09:53rasherI don't see how it could be worth it
21:10:18rasherif you mean platforms as in ipodlinux
21:10:55rasherand porting to other devices is more or less just doing device specific things as it is
21:11:10rasherrockbox itself is quite portable
21:11:25Bagderit would not be hard to port to ipod
21:11:27rasheras noted by the quick progress as soon as it got running on iriver
21:11:42courtclook familiar?
21:12:49*Bagder reads OpenNeo commits
21:13:51*t0mas google's for OpenNeo
21:13:58Bagderthey openneo guys work offline and commit huge chunks every few months
21:14:03 Join hyarion [0] (
21:14:10Bagderopenneo is a Rockbox fork
21:14:18Bagderfor the Neo players
21:14:18t0masa rockbox fork? :)
21:14:20rasherwhy did it fork?
21:14:30Bagderthey wanted their own
21:14:38Bagderthey never ever tried to be with us
21:15:13 Join coob [0] (
21:15:13t0mashm... they have a more n00b approach on the website...
21:15:21BagderRockbox develops a lot faster than their version does
21:15:30coobbecause you have 2873091278903x more developers.
21:15:45Bagdercoob: yes, probably
21:15:52coobwe have *one* person able to do firmware reverse engineering.
21:15:55coobno, definately.
21:15:55courtccoob- not us..
21:15:59cooboh ok :D
21:16:09*coob says nothing baout guns and jumping...
21:16:24BagderI'm talking about OpenNeo
21:16:45Bagderbut these guys have a malloc() ;-)
21:16:57*t0mas starts reading :P
21:17:08Bagder... and only 256K ram
21:17:12t0masah ok
21:17:33courtchaha.. a little strange
21:17:34Bagderbut they're for in-car use and have their 3.5" disks spinning all the time I believe
21:18:59webmindBagder, do you know if the side prints on the archos are multilayer ?
21:19:35Bagderthe side prints?
21:20:33webmindthe ones that connect to the batteries
21:21:22Bagderoh. Well, I don't know
21:27:23Bagderhey, they (OpenNeo) do check our progress
21:27:35Bagdersees they getting amiconn's memset assembly code
21:27:40*Bagder sees even
21:35:55t0mashm... OpenNeo has support from the manufacterer...
21:36:09t0maserg... manufacturer
21:38:08*rasher cleans his laptop keyboard
21:38:43rashera truly schocking experience
21:39:09*HCl cleaned his room...
21:39:43rashercan't have been worse than this
21:40:13*HCl remembers cleaning one of his keyboards when he hadn't cleaned it for 3 years....
21:40:30thegeekwhat happens if you turn on the iriver without lcd attached?
21:40:31HCli'll refrain from describing it.
21:40:38HClwithout lcd attached?
21:40:43HClwhat do you mean?
21:40:55thegeekas in
21:41:07thegeeklcd removed
21:41:13thegeeklike this
21:41:14*HCl nods..
21:41:17HCli thought so
21:41:18HClno idea
21:41:20HCltry it?
21:41:26HClwhy do you want to know anyways
21:41:45thegeekI just want to see what point is broken on the headphone jack
21:41:56thegeekand I'm afraid of scratching the lcd
21:42:03thegeekor accidentally do anything bad to it
21:46:26 Join webguest00 [0] (
21:47:57 Quit webguest00 (Client Quit)
22:09:19thegeekI think I fixed it
22:09:24thegeeka little bending;)
22:09:31thegeeknot a perfect or good solution
22:09:32thegeekbut it works
22:09:41Bagderthe neuros guys still have no source code available
22:10:38rasherMaybe they're only providing it pr request
22:10:44thegeekthere is one thing I really don't understand
22:10:59thegeekwhy won't iriver/apple provide source for atleast the ui and stuff
22:11:01thegeekI mean
22:11:05Bagderrasher: no, they're having problems getting an agreement with TI
22:11:06thegeekwhat can they loose?
22:11:16thegeekthey loose control over that player's software
22:11:17thegeekbut hey
22:11:34thegeekthey'll hardly going to make a lot of money on a new player with only software updated
22:11:39rasherYes.. they already sold the product
22:11:56rasherwhen their new player comes out with $feature
22:12:13rasherthey don't want people implementing that on $old-player
22:12:20thegeekI know
22:12:22thegeekbut still
22:12:31thegeekthe high-end mp3 player market
22:12:35Bagderyes, that's the argument for NOT wanting people to improve their old units
22:12:36thegeekis not driven by software upgrades!
22:12:45thegeekit's hardware upgrades
22:12:50thegeekand besides
22:12:54Bagderthat's why
22:13:12thegeekthe customer colume would increase
22:13:13Bagderthey get money by selling hardware
22:13:45rasherto be a bit cynic, they only want the software to be "Good enough"
22:13:56rasherso that when they release something new, they might have another sale
22:13:57thegeekmy point is that the benefit of full control over the software does not even come close to the benefit of added customers
22:14:02Bagderyes, since consumers don't care enough for software quality
22:14:04thegeeknot to mention the huge buzz it would create
22:14:18thegeekif iriver opensourced most of their firmware
22:14:18rasherthegeek: it wouldn't
22:14:26thegeekit would attract lots of new people
22:14:27rasheramong a few assorted geeks, sure
22:14:32thegeeknot true imho
22:14:41rasherbut on the whole, I doubt it'd be weorth it
22:14:41thegeekif iriver had a pr campaign
22:14:50thegeekwhere they advertised the player with open source
22:14:55rasher"We are opensource" doesn't sell
22:15:01rasheroutside the geek market
22:15:03thegeekperhaps not initially
22:15:08thegeekbut the word would get out
22:15:12rashernot for another 10 years
22:15:13thegeekand if a large manufacturer did it
22:15:21rasherApple did it
22:15:24rashernoone cares
22:15:28rasherexcept the geeks
22:16:00rasherWe can hope
22:16:01thegeeka company will do it
22:16:04thegeekand if the time is right
22:16:05rasherbut I don't think we're there now
22:16:06Bagderneuros does
22:16:09thegeekit will be a boost
22:16:14thegeekneuros is huge
22:16:20coobexpect you can't get the neuros source.
22:16:29thegeekI considered the neuros
22:16:32thegeekbut the hardware sucks
22:16:33thegeekI mean
22:16:35thegeekcome on
22:16:42thegeekI'm not that desperate for open source
22:16:42 Join matsl [0] (
22:16:45Bagderand if you do get it, you can't afford the dev environment anyway
22:16:48coobwhat proc do they use
22:17:00Bagdercoob: its a TI DSP
22:17:02thegeekI just forgot to add the hd foam piece inside the iriver
22:17:15 Quit courtc (Read error: 104 (Connection reset by peer))
22:17:17Bagderwithout gcc port available
22:17:31rasherthat's a showstopper :X
22:17:58Bagderapparently you can get a 60 day (or something) free-trial version of the TI compiler
22:18:22thegeekthen what is the problem
22:18:59rasherYou just need a lot of talented developers to work in 60-day shifts
22:19:10Bagder"state: planning"
22:19:38thegeekhaha rasher;)
22:24:44rasheropenneo is planning on support for "limitless songs per folder"
22:24:50rashergood luck on that one
22:25:17rasherMulti-user support?!
22:25:23Bagderwell, they did skip lots of rockbox concepts
22:25:55Bagderso limitless could work
22:26:05Bagderplay a song, scan for next, play that etc
22:26:08***Saving seen data "./dancer.seen"
22:26:57Bagderbut it wouldn't be able to sort the list nor do it shuffled
22:27:11rasheroh I mean isn't there a fat32 limit?
22:27:58BagderI would imagine that they didn't mean that limit
22:28:08Bagderwhatever that limit is
22:28:30rasheryes, true
22:28:38 Quit izzy (Read error: 110 (Connection timed out))
22:33:24 Join StrathAFK [0] (
22:33:30 Quit hyarion (Read error: 60 (Operation timed out))
22:35:09 Quit matsl (Remote closed the connection)
22:41:40 Join matsl [0] (
22:42:54 Quit uski (Read error: 113 (No route to host))
22:49:51MoosCamaroHCl: what's happen?
22:51:57 Quit Strath (Read error: 110 (Connection timed out))
23:11:27HClMoosCamaro: meh, just not feeling all that, my flatmates are jerks
23:13:20coobHCl: that's what they invented the flusher for!
23:13:34coobfill a bin with water and lean it against their (inwards opening) door
23:13:39coobthen knock + rujn
23:13:47cooband hope they have electronics on the floor :)
23:14:29rasherI have a hard time imagining that making things *better*
23:14:37*HCl nods at rasher
23:14:57coobshort term success > *
23:15:13 Join izzy [0] (laitinei@
23:15:18HClthe truth is, its not gonna get better
23:15:21HCli have two options
23:15:30HCla) try to ignore them as much as possible and isolate myself a bit
23:15:33HClb) move away.
23:15:39coobc) flush!
23:16:40rasher only several years late
23:35:45 Join [IDC]Dragon [0] (
23:36:09[IDC]DragonBagder, do you read?
23:37:12[IDC]DragonI'm working on "bootbox"
23:37:29[IDC]Dragonbased on Linus' bootloader for the ihp
23:37:43[IDC]Dragonbut for SH this time
23:38:13[IDC]Dragonwould it be possible to have weak linking of the Rockbox firmware lib?
23:38:36Bagderhow do you mean weak?
23:38:44[IDC]Dragonso I could "override" implementations
23:38:52Bagderyou can
23:39:06[IDC]Dragonif that's the correct term, I'm no expert with linkers
23:39:32Bagderthe lib only provides functions that aren't resolved by the object files you write
23:40:09[IDC]Dragonbut if both have it, it complains!?
23:40:58BagderLinus' code overrides lib-provided functions that way
23:41:32[IDC]Dragonto prevent the inclusion ofthe whole .o file, right?
23:42:11Bagderto prevent the inclusion of the functions in general, I'd say
23:42:18Bagdervoid usb_acknowledge(void)
23:42:31[IDC]Dragonthe linker can only use all or nothing of a .o, not on function level?
23:42:32rasheras linus did
23:42:34rasherit seems
23:42:42*rasher shuts up
23:43:13[IDC]Dragonyou're welcome to carry on
23:43:30rasherwell the bootloader main.c contains a bunch of those
23:43:38rasheris all
23:43:40[IDC]Dragonsaw that, yes
23:44:00[IDC]DragonI had to remove most already, because I need the real code
23:45:44Bagderok, so you want some functions from the "real" .o file, but want to override some of the functions from that same .o ?
23:48:43Bagder−−allow-multiple-definition perhaps
23:48:50Bagder(an ld option)
23:49:07Bagder"Normally when a symbol is defined multiple times, the linker will report a fatal error. These options allow multiple definitions and the first definition will be used."
23:49:09[IDC]Dragonand it takes care of preference, too?
23:49:21[IDC]Dragonfirst, bah
23:49:36thegeekI'm ordering an axim x50v
23:49:38Bagderwell, I'd guess that the object files are first, then the lib's files
23:49:39thegeekany objections?
23:50:04Bagderbut of course, it doesn't really say what order it refers to
23:50:59rasherthegeek: only that it's slightly more powerful than the laptop I just sold
23:51:46thegeekit's cheap here right now
23:51:49thegeek25% or something
23:51:53thegeekdell's usual
23:52:08*[IDC]Dragon tries
23:52:30rasherthegeek: how much?
23:53:43thegeekhmm, with 3.551 barebone
23:53:59thegeekgood price for a pretty good pda
23:54:10*rasher converts
23:54:33thegeekyou might want to add ~1000 nok for memory, warranty and a case or something
23:54:42rasher3247 dkk
23:54:56rasherand it's 4099 here >.<
23:55:18thegeekwhat do you mean?
23:55:36rasherThe x50v costs 4011 dkk here
23:55:45rasherand 3551nok is 3247 dkk
23:55:57rasherso you're getting it way cheaper than a dane would
23:56:08thegeekit says 3,060 dk barebone here
23:56:11thegeekfor the x50v
23:56:13thegeekmind the v
23:56:17thegeekonly the v has 20% off
23:56:23rasherI did.
23:56:28thegeekNu fra
23:56:28thegeekInkl. moms og levering
23:56:38thegeekwithout any extras
23:56:54rasher"Fra 40112
23:56:58rasher"Fra 4011"
23:57:10thegeekyou have to "add" it
23:57:18thegeekatleast start the customization page
23:57:24thegeekKonfigurer og kb
23:57:34thegeekthey don't want people buying it
23:57:46thegeekthey just want to lure people in
23:57:47thegeeknot buy
23:57:50rasherwhat the.. now it only says X50
23:57:55thegeekdon't mind
23:57:57thegeekit's just the seris
23:58:01rasherand 3060 :D
23:58:12rasherso danes get it cheaper :)
23:58:15thegeekthat's a damn good price
23:58:26thegeekhow much tax do you have?
23:58:31thegeekwe have 25%
23:58:42rashersame here

Previous day | Next day