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

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

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

Notice: Only Gecko based browsers prior to FF4 support the multipart/mixed "server push" method used by this log reader to auto-update. Since you do not appear to use such a browser, this page will simply show the current log, and not automatically update.

#rockbox log for 2005-04-28

00:00:58 Quit courtc (Read error: 110 (Connection timed out))
00:01:27 Join courtc [0] (
00:10:44 Quit asdsd_ (Read error: 110 (Connection timed out))
00:16:36 Quit Sucka ("a bird in the bush is worth two in your house")
00:23:44 Quit ashridah ("Leaving")
00:24:44***Saving seen data "./dancer.seen"
00:37:03HCli need to sleep, i'm obviously not awake with all the compiler errors i'm generating :/
00:37:20preglowincrease dosage of stimulants!
00:47:13HCl-rwxr-xr-x 1 root root 3864 Apr 28 00:44 searchengine.rock
00:47:17HClkind of scary.
00:47:21HClthats supposed to work too.
00:47:23HClin only 3.8k
00:47:34HCli'm scared to run it XD
00:52:03 Join Legato [0] (
00:52:17Legatocan i get music in here
00:53:15Legatowhy is it called rock box
00:53:56 Part Legato
00:54:01HClno idea o.o.
00:54:04oddpacman & space invaders written (and playable) in excel:
00:58:03MoosCamarohi HCl
00:58:14 Quit nozomiyume (Read error: 104 (Connection reset by peer))
00:58:28MoosCamaroyour works progress fine
00:58:41HClyea, i'm about ready to test my searchengine.
00:58:44HCli'm scared to run it though.
00:58:50HCli'm always scared to run code
00:58:54HClwhen i spend 4 hours writing it
00:58:56HClwithout testing
00:59:06HCli hate bugs :P
00:59:13HCland code never works right the first time
01:01:01MoosCamarohave you choose searchengine name for your plugin?
01:01:04*HCl should probably add some syntax checking / error reporting first...
01:01:32amiconnI wonder why people don't read the channel messages before asking stupid questions :-/
01:02:43amiconn[00:52:12] <Legato> can i get music in here
01:02:46amiconn[00:53:10] <Legato> why is it called rock box
01:03:08amiconn[23:30:53] -ChanServ- [#rockbox] "Be sure to check out the FAQ before you look for 'general' information!"
01:04:58amiconnApart from the fact that the link should be updated...
01:06:34MoosCamaro20:16 MeTaXo too
01:07:40preglowpeople are stupid, no surprise there
01:07:56amiconnBagder: Who's in charge of the channel?
01:08:11 Join nozomiyume [0] (
01:09:12MoosCamarowe can't filter this chanel, it's impossible
01:09:46amiconnThe FAQ link needs updating... has been removed?
01:10:21preglowworks here
01:11:12amiconnNo, it hasn't been removed. However, the -> transition would be cosmetic, but /docs/faq.html is old static content
01:11:31 Join Jay200 [0] (
01:11:39amiconnThe current link would be
01:12:04Jay200How is the codec side of things coming for iriver rockbox?
01:13:13preglowit's at a stand still
01:13:18Jay200oh right.
01:13:23preglowsince i'm busy with other stuff
01:13:26Jay200I was impressed with the Rockbox logo on the remote
01:13:50preglowhowever, linus is working on playback
01:13:54preglowso all is not lost
01:14:04Jay200what about on the fly playlisting?
01:14:13preglowthat's long since done, man
01:14:18preglowthat's been working for years
01:14:26Jay200i mean for the iriver
01:14:32preglowwell, it works there as well
01:14:34Jay200because of its stupid shuffle in the iriver f/w
01:14:50Jay200in the iriver f/w the shuffle mode plays the same songs in the same sequence
01:14:58preglowbut how does that concern us?
01:15:16Jay200well iriver alledge its a hardware issue
01:15:16HClokay, so now i have proper syntax error reporting..
01:15:23*HCl is still scared to run his code :/
01:15:40preglowthen iriver is bullshitting
01:15:46preglowi can't possibly imagine that is a hardware issue
01:15:47Jay200well there u go
01:15:55Jay200well thats what they said on an iriver forum
01:18:53amiconnThat is definitely no hardware issue. Iirc iriver claims you'd need a hardware clock to implement proper shuffle
01:19:18Jay200well as far as i know the iriver h120 has no clock in it
01:19:40amiconnHowever, rockbox on archos shows that this is definitely wrong. Rockbox does have a decent shuffle, while only 3 of the 6 different archos models do have a hardware clock
01:19:47preglowamiconn: ahaha, do they mean to seed the random generator or something?
01:20:24Jay200i have an H120 btw
01:21:19preglowrealtime clock is most definitely not the only way to get a seed
01:21:27preglowyou can just bloody use time since startup, for one
01:22:31*HCl hardcodes a search query for all songs of the 1980's
01:23:02preglowdidn't bagder do this config.h thing?
01:23:35preglowshall i remove it from todo?
01:23:49HClwell, till its tested and actually used
01:23:54HClits not used yet
01:25:19HClwho feels brave? :P
01:25:48preglowhas anyone volunteered to clean up all the pointer signedness issues yet, btw? :P
01:26:11HCli'm not interested as long as it doesn't boot anyways
01:26:33*preglow itches to see how well it optimizes compared to gcc3.4
01:26:37HClhow big can plugins get on archos again?
01:26:49amiconnpreglow: Guess what rockbox does? seed = current_tick; ;-)
01:27:12amiconnHCl: 32 KB
01:27:23HClsearchengine is 8kb so far
01:27:45HClstarts well XD illinstr
01:27:45amiconnDoes the search engine use the audio buffer?
01:27:52preglowgetting it small enough shouldn't be that big a problem
01:28:02amiconnSo no audio playing while searching the db...
01:28:03HCli'm not gonna debug this now, probably a bug in the pluginapi
01:28:06HClamiconn: nope.
01:28:22preglowthe disk access would be death unless you worked in ram
01:28:31HClyou can still browse it and do plain searches
01:28:32HClpreglow: definately.
01:28:43HClit has a nice load-when-needed structure at the moment..
01:28:44preglowespecially with all the file links
01:28:51HClbut it crashes on an empty function variable thing..
01:28:55amiconnHmm. The db browser does without excessive disk access...
01:28:56HCli'm gonna go to bed now :/
01:29:02HClAI class tomorrow, and i haven't even read up on it.
01:29:05HClits gonna be a disaster :/
01:29:15preglowi'm good at ignoring such stuff
01:29:23preglowclasses are boring anyway
01:29:56*preglow cuddles his lack of need to get up early
01:31:22HClheh :/
01:31:35*HCl goes to reset his iriver..
01:31:52*HCl doesn't quite understand how it managed to get an illinstr 00000000
01:31:52preglownot that i have any classes anyway
01:32:03preglowgcc4? :P
01:32:43preglowuse iram?
01:33:04preglow'cuz i think that's where iram is mapped
01:33:47MoosCamarogood night all
01:33:57preglowthat's where the flash is mapped
01:33:59preglowMoosCamaro: nite
01:34:13 Part MoosCamaro
01:34:23amiconnHCl: Most probably something (incorrectly) calls address 0. Since there is no memory, contents is undefined, hence most probably an illegal instruction
01:34:45preglowthere is memory, but it's probably not instructions
01:35:12amiconnThere is memory at 0x0 ?
01:35:26amiconnAh, ok. RAM would be really bad there
01:35:37amiconnToo easily overwritten
01:35:53amiconnThe archos does also have memory at 0x0, but it's plain ROM
01:35:54preglowwell, yeah, according to crt0.S, it looks like flash is mapped there
01:36:16preglowsram is at 0x10000000
01:36:47preglowbut a jump to 0 is strange
01:36:58preglowalmost sounds like objcopy missed a relocation
01:37:10HClamiconn: yea, i don't understand how though
01:37:13 Quit Aison ("( :: NoNameScript 3.72 :: )")
01:37:14HCli added a few functions...
01:37:16preglowbut that should never happen
01:37:35amiconnHCl: uninitialised function poiter somewhere?
01:37:51HClwell, there are only two functions i added
01:38:07HClwhich should not get called either way since my current tokenstream does not contain strings...
01:38:12HCland tagdb_init
01:38:21HClwhich would only get called if the tagdb isn't initialized.
01:38:22 Quit Jay200 ("CGI:IRC (EOF)")
01:38:33HClwhich shouldn't happen either, since it gets loaded when i boot my iriver
01:39:17HCli'll make a tgz of it
01:39:20HCland throw it on http
01:42:05HCli have to go sleep now.. night..
01:45:59preglowthink i'll indulge in half an hour of lovely assembler
01:46:59 Quit cYmen ("zZz")
01:59:16 Quit preglow (Connection reset by peer)
02:00:44 Join preglow [0] (
02:00:48preglowi love my server
02:24:47***Saving seen data "./dancer.seen"
02:26:22 Quit preglow ("leaving")
02:32:39 Join gromit`` [0] (~gromit`
02:35:46 Join webguest17 [0] (
02:39:40 Quit webguest17 (Client Quit)
02:40:45 Quit thegeek ("( :: NoNameScript 3.81 :: )")
02:41:00 Join thegeek [0] (
02:45:39 Join gromit``` [0] (~gromit`
02:45:40 Quit gromit` (Read error: 110 (Connection timed out))
02:47:01 Quit gromit`` (Read error: 104 (Connection reset by peer))
02:48:36 Join TCK [0] (
02:53:28 Join DMJC [0] (
03:32:41 Join elinenbe_ [0] (~elinenbe_@
03:32:47 Quit elinenbe (Read error: 131 (Connection reset by peer))
03:32:57 Nick elinenbe_ is now known as elinenbe (~elinenbe_@
04:03:12 Quit DMJC ("Leaving")
04:04:11 Join StrathAFK [0] (
04:04:22 Quit Strath (Read error: 104 (Connection reset by peer))
04:05:08 Quit ze (Remote closed the connection)
04:05:12 Join ze [0] (
04:05:27 Join QT_ [0] (as@area51.users.madwifi)
04:09:31 Quit Seed (Nick collision from services.)
04:09:38 Join Seed [0] (
04:11:03 Join G [0] (
04:11:34 Join gromit` [0] (~gromit`
04:11:34 Quit gromit``` (Read error: 54 (Connection reset by peer))
04:11:46 Quit thegeek (
04:11:46 Quit QT (
04:11:46 Quit crash_ (
04:11:46 Quit Hadaka (
04:13:18NJoincrash_ [0] (
04:13:18NJoinHadaka [0] (
04:19:55 Quit Hadaka (
04:19:55 Quit crash_ (
04:20:06NJoinHadaka [0] (
04:20:06NJoincrash_ [0] (
04:20:36 Quit G (
04:20:36 Quit HCl (
04:20:36 Quit ncrawla (
04:20:36 Quit mbr (
04:21:26 Quit QT_ (
04:21:26 Quit elinenbe (
04:21:26 Quit amiconn (
04:21:26 Quit Lynx_awy (
04:21:26 Quit rasher (
04:21:26 Quit bobdbob (
04:22:07NJoinQT_ [0] (as@area51.users.madwifi)
04:22:07NJoinelinenbe [0] (~elinenbe_@
04:22:07NJoinamiconn [0] (
04:22:07NJoinrasher [0] (
04:22:07NJoinLynx_awy [0] (HydraIRC@
04:22:07NJoinbobdbob [0] (
04:22:11NJoinG [0] (
04:22:11NJoinHCl [0] (
04:22:11NJoinmbr [0] (
04:22:11NJoinncrawla [0] (
04:22:44 Quit CoCoLUS (
04:22:44 Quit tedboer (
04:22:44 Quit DangerousDan (
04:22:44 Quit Rick (
04:22:44 Quit [zmaj] (
04:22:44 Quit Bagder (
04:23:04NJoinCoCoLUS [0] (
04:23:04NJointedboer [0] (
04:23:04NJoinDangerousDan [0] (
04:23:04 Join Rick [0] (rick@Rick.user)
04:23:04NJoin[zmaj] [0] (
04:23:04NJoinBagder [0] (
04:24:51***Saving seen data "./dancer.seen"
04:33:27 Join lostlogic [0] (
04:50:20 Quit gromit` (Read error: 110 (Connection timed out))
05:54:24 Join gromit` [0] (~gromit`
06:24:52***Saving seen data "./dancer.seen"
06:29:34 Join matsl [0] (
07:06:29 Quit gromit` (Read error: 54 (Connection reset by peer))
07:06:30 Join gromit`` [0] (~gromit`
07:17:17 Join gromit` [0] (~gromit`
07:17:17 Quit gromit`` (Connection reset by peer)
07:32:52 Quit matsl (Remote closed the connection)
07:36:05 Join webguest75 [0] (
07:49:12 Join einhirn [0] (
07:52:10 Join ashridah [0] (
08:00:03 Join wbgst [0] (
08:09:17 Quit wbgst ("CGI:IRC")
08:21:47 Quit gromit` (Remote closed the connection)
08:21:51 Join gromit` [0] (~gromit`
08:24:56***Saving seen data "./dancer.seen"
08:45:24 Quit edx (Read error: 110 (Connection timed out))
09:12:43 Join Harpy [0] (
09:18:16 Nick Lynx_awy is now known as Lynx_ (HydraIRC@
09:18:46 Quit webguest75 ("CGI:IRC (Ping timeout)")
09:21:25 Join t0mas [0] (
09:28:26 Join cYmen [0] (
09:36:55 Join bobTHC [0] (
09:37:14bobTHCmornin' folks !
09:37:55 Join Zagor [0] (
09:46:23 Quit t0mas (Read error: 110 (Connection timed out))
09:46:25 Join t0mas [0] (
09:47:00*t0mas slaps planet internet
10:03:50*HCl yawns
10:04:08HClits cold :/
10:04:26HCland i'm nauseous and unprepared for ai class :/
10:08:19 Nick QT_ is now known as QT (as@area51.users.madwifi)
10:22:31 Join edx [0] (
10:24:58***Saving seen data "./dancer.seen"
10:40:33amiconnwiki spam :((
10:41:39HCltoo bad for him they didn't become links
10:41:55rasher01.14.50 # <Jay200> in the iriver f/w the shuffle mode plays the same songs in the same sequence
10:41:59rasher01.15.16 # <Jay200> well iriver alledge its a hardware issue
10:42:02rasherworst explanation ever
10:42:22rasherThat's priceless
10:42:26HClpeople have such a distorted view of how computers work..
10:42:44rasherWell that allegedly came *from iriver*
10:42:58HCli highly doubt that.
10:44:04HCl*waits for markun*
10:44:10 Quit bobTHC ("Smoke Weed Every Day !")
10:44:10HClwhere is he XD
10:44:23rasherworking on grayscale patch
10:44:36HClno no, i have class with him
10:44:47HClbut neither of us were very motivated
10:44:49rasheroh yes, you mentioned
10:45:12HClcompiler creation is fun though.
10:45:44HClif he's not gonna show up i'll just "disappear" during a break >.>
10:48:03HClthere he is
11:42:20 Join austriancoder [0] (
11:42:25austriancoderhi all
11:56:26t0masaustriancoder? you are working on the fm-radio code for iRiver right?
12:01:26 Quit austriancoder ("using sirc version 2.211+KSIRC/1.3.12")
12:09:47 Join Aison [0] (
12:10:59 Join MoosCamaro [0] (
12:11:15MoosCamarohello all
12:24:59***Saving seen data "./dancer.seen"
12:25:52 Join preglow [0] (
12:27:45MoosCamarohi preglow
12:41:37 Join webguest13 [0] (
12:42:33 Quit webguest13 (Client Quit)
12:43:51 Join seboo [0] (
12:48:22 Quit seboo (Client Quit)
12:49:52 Join seboowebguest65 [0] (
12:51:44 Quit seboowebguest65 (Client Quit)
12:52:29 Join seboowebguest93 [0] (
12:56:11preglowimdct at least just has MINOR distortion now, suddenly
12:58:09 Quit DangerousDan (Read error: 131 (Connection reset by peer))
13:02:46 Join DangerousDan [0] (
13:10:35 Quit seboowebguest93 ("CGI:IRC (EOF)")
13:19:01*ashridah hands preglow the cookie of success tasted
13:21:49*preglow declines, and instead helps himself to the almost, but not quite yet! cookies
13:28:10 Quit lostlogic ("Going to the moon")
13:30:37 Join F1^Aison [0] (~hans@
13:38:11*HCl tries to figure out what caused the crash
13:49:01 Quit Aison (Connection timed out)
13:50:00HCli hate it when that happens.
13:50:11HClit just makes me uncomfortable.
13:50:14HClbugs just "disappearing"
13:53:58preglowi just had that happening to me
13:54:05preglowthe thing blared distortion
13:54:10preglowchanged one line, bam, nice output
13:54:17preglowchanged it back, bam, nice output
13:54:54 Join ismi [0] (
13:56:51HClthats kind of odd.
13:57:03*HCl is scared to read back the results of his searchengine XD
13:57:18HClit did that search way too fast anyways o.o.
13:57:20preglowbring on the bugs, i say
13:57:32HClwell it didn't give a syntax error..
13:58:09HClbut it went way too fast too. so that worries me.
13:58:42HClin fact, it didn't even read from disk o.o;;
13:59:59HCldumb bug
14:01:25HClthat looked much better
14:01:29preglowthey mostly are
14:01:29HClit stalled to read the database
14:01:32HClthen returned a result
14:01:45preglowdo you test in the sim?
14:01:58HCli actually had the check for the "loadedfiledata" wrong way around
14:02:13HClno, this is target
14:02:44HClan m3u is just a list with filenames, correct?
14:03:29rasherand possible comments
14:03:34rasherstarting with #
14:03:41HClfair enough
14:03:54HCli'll just dump all filenames who matched to the search to an .m3u..
14:05:35HClthe searchengine should return a list of filenames anywho, right?
14:05:52Bagderseems right
14:11:18 Nick F1^Aison is now known as Aison` (~hans@
14:13:51HClwhy isn't this working >.<
14:15:40 Join lolo-laptop [0] (~lostlogic@
14:16:53HClthe moment of truth :3
14:17:18HCli got a few 1's! yay
14:17:27HClit actually seems to be working XD
14:17:39*HCl rewrites it to output filenames rather than boolean values.
14:18:17HClmy id3 tags aren't up to scratch so my hardcoded "all songs for which year >= 1980 and year < 1990" doesn't match much
14:18:20HClbut i have a few matches :)
14:18:31HCland its fairly fast too.
14:18:37HClit just needs to read the entire db in once
14:18:50 Quit ashridah (Read error: 110 (Connection timed out))
14:18:51 Join Lost-ash [0] (
14:24:32elinenbeHCl: how is the realtime db search and playcount doing?
14:25:01***Saving seen data "./dancer.seen"
14:25:06HClhaha, sweet.
14:25:15HClelinenbe: realtime db search is operational
14:25:16HCl :D
14:25:42HClit just correctly made an playlist for all songs in my database that had a year tag between 1980 and 1990
14:26:06elinenbeHCL: you said it was hard-coded however... what about on the fly?
14:26:13HCleasy to adapt.
14:26:19HClhardcoded was just a test
14:26:33HCllet me try whether it compiles for archos sim
14:26:36HCland commit if it does
14:26:40HClyay for 80's music xD
14:26:47HClmy m3u loaded into winamp fine :)
14:27:01HClthis is kind of scary
14:27:04HClaside from compiler errors
14:27:07HCli've had no bugs
14:27:18HClnot counting the databaseloaded flag
14:28:59HClthe search plugin is bound to not work on sims yet since it doesn't fix endianness of the database
14:29:32HClits really appalling how badly my mp3s are tagged xD
14:30:07HClit mostly found mozart and vivaldi, and i'm not even sure whether those should be tagged in that year
14:30:17HClthey're tagged as 198* allright though
14:33:43*HCl commits
14:33:59HClbah, don't you hate it when something is supposed to be a challenge, and it turns out to be fairly easy :x
14:34:13HCli must admit i haven't tested string searching yet
14:34:31HCli'd like people to test though
14:34:41HClthough on archos i have no idea what'll happen if it runs out of ram.
14:35:16Bagderprobably the same thing as if you run out of ram on the iriver ;-)
14:35:31Bagderor? ;-)
14:35:36HClit needs about.. um... filecount * (fileeentrysize+songentrysize+(amount of compares))
14:35:55HClwith optionally filecount *albumlen / artistlen
14:36:01HClif you have a search query with those involved
14:36:19HClit doesn't do rundb info yet
14:36:36 Quit cYmen (Read error: 110 (Connection timed out))
14:36:55*HCl goes to watch the daily build page to see what he'll have to fix after this
14:38:14Bagderthe $Id in searchengine.c is slightly wrong
14:38:23Bagder$Id:$ it should be
14:38:24HCli wasn't sure how to do those
14:38:48HCli still have quite some things to fix...
14:38:59HClall the ints in dbinterface.h should become longs since they have to be 32bits..
14:39:01Lost-ashfor cvs? no it doesn't. at least, i've never used $Id:$ and it's always worked. or are we dealing with a special case?
14:39:44t0mas$Id$ will become some version info and the name of the commiter...
14:39:55 Nick Lost-ash is now known as ashridah (
14:40:14t0maswhat was your comment about then ashridah?
14:41:17HClshould it be $Id:$ or $Id$ ?
14:41:24HCli just changed it to $Id:$
14:41:29HClas well as getting unneeded delays out
14:41:45t0mashm... Bagder said $Id:$
14:41:46BagderI think the colon is needed, at least it works with it
14:41:57t0masI use $Id$
14:41:59HCli took some sleeps out, modified that while i was there
14:42:17*HCl waits for cvs to compile
14:42:18Bagderright, the colon is not required
14:43:08Bagderyou'll get yellows
14:43:13Bagdertoken.h:30: warning: 'spelling' defined but not used
14:43:27HCli'm not 100% sure yet what to do about that one.
14:43:33HClits going to be needed when we add the interface
14:43:37t0masBagder? how's that build thing setup?
14:43:37HClbut at the moment its still unused
14:43:42t0mashomemade script?
14:44:15BagderHCl: you should not add variables in headers
14:44:18t0masin setup loginfo?
14:44:28t0mas*setup in loginfo
14:44:34HClBagder: its an string array with constants...
14:44:45HCli didn't really look at it as a variable, i guess.
14:44:48Bagderit is considered bad practise
14:44:50HCli'll change that
14:44:53HClyea, i know
14:45:49Bagdert0mas: no, the script simple checks for CVS activity every minute and rebuilds everything if something was changed
14:45:51*HCl wants to try one of the more funnier searchquerys in a bit...
14:46:06HCl"give me all songs for which the title contains the albumname"
14:46:10Bagderthe cvs repo is on the same host
14:47:37elinenbeHCl: does it do realtime playcount?
14:47:56HClyes, if it wasn't that the runtimedb isn't added yet.
14:48:20HClat the moment the loadrundbdata just does "yup, its loaded >.>;"
14:48:31HClmaking all runtimeinfo 0 at the moment
14:48:46elinenbeHCl: like could I do a playlist search for all songs played more than 20 times?
14:49:00HCli'm actually considering splitting the searchengine into two parts
14:49:06HClthe actual searchengine
14:49:11HCl(what i finished now)
14:49:15elinenbeHCl: like could I do a playlist search for all songs played more than 20 times, from 1982, that are disco genre?
14:49:16HCland an interface to build searchfiles
14:49:20HClelinenbe: sure.
14:49:24HCl :)
14:49:32elinenbei think you could do a pretty simple widget based search...
14:49:46HClwell, i just need a tokenstream generator
14:49:49HClthe engine is done
14:50:00elinenbemuch like itunes calls their "smart playlist"
14:50:01HClit accepts querys that have already been translated to tokens
14:50:06elinenbeoh nice.
14:50:11HClwe'd pretty much have something like that
14:50:17HCli'm thinking of having something like "search files"
14:50:22HClthat you can open with the searchengine
14:50:25HCland it'll create a playlist
14:50:30elinenbesounds like a great idea...
14:50:30HClbased on the searchquery stored in the file
14:50:50HCldo we have a "current playlist" or queue thing?
14:50:50elinenbethe most important thing i think would be to be able to create those files on the device...
14:50:57HClshouldn't be hard.
14:51:16elinenbesounds good.
14:51:22HCl :)
14:51:24elinenbewe have a "current playlist" thing already...
14:51:35HClyea, i'm kind of wanting it to load to that
14:51:42HClrather than writing to \search.m3u
14:51:50 Join webguest97 [0] (
14:52:31 Quit webguest97 (Client Quit)
14:53:31preglowfinally summer
14:54:13HCli need some people who have 80's songs on their iriver, who can test the searchengine
14:54:18HClit should work fine..
14:54:22*HCl really needs to test strings..
14:54:36Bagderyou should make it work in the sim
14:54:42HCli should do that first :P
14:54:46 Join bobTHC [0] (
14:54:52preglowtesting on target is rather... slow...
14:54:54HClnot hard, just the endianness..
14:55:05HClBagder: are we ready to switch to the new endian defines?
14:55:21 Part t0mas
14:55:26HClcan i just do #include <autoconf.h> ?
14:55:27 Join t0mas [0] (
14:55:32Bagderbut you may need to fix the include path at some places
14:55:52BagderHCl: that works, yes
14:56:10HClwhat should the include path be added?
14:56:43Bagderif you get a problem with that include, you need to make sure "-I$(OBJDIR)" is used
14:56:50Bagderor something equivalent
14:57:08HCleasy enough
14:57:28Bagderyes, it should be
15:03:39preglowhas anyone tried the patcher, btw?
15:04:18t0masI have windows XP here
15:04:19Bagderit still has wrong md5s
15:04:21t0masI'll do it
15:04:22BagderI should fix that
15:04:27preglowas in checked it actually produces valid firmware
15:04:42preglowi have tested it, but that's quite some time ago, and i hardly ever use windows these days
15:04:57 Join austriancoder [0] (
15:05:04BagderI'll build one with the wiki bootloader
15:06:22Bagderok, should work now
15:06:34*austriancoder has a boring englisch lesson
15:06:39preglowwith euro firmware?
15:07:02Bagderyour list
15:07:19Bagder1.65 eu and 1.63 eu/us/k
15:07:57preglow1.65 is probably the only md5 that has used the last bootloader.bin
15:08:15preglowit's also the only checksum i haven't verified
15:08:20Bagderthis version uses the wiki bootloader
15:08:36preglowyeah, i meant last wiki bootloader
15:09:03preglowi just grabbed the sum from the wiki page
15:09:46Bagdersomeone with windows could go over them and check
15:11:34preglowt0mas volunteered! ;)
15:11:52t0masyes, compiling now :)
15:12:08preglowno need to compile, bagder put up a fresh binary
15:12:26t0mashm... isn't it useless to check his bin again on my pc?
15:12:50preglowit's the same result you'll get from cvs
15:13:13preglowi'll check if the md5 is correct, i very much suspect it is not
15:13:42t0maswich firmwares should I try?
15:13:50t0mas1.63 EU + US ?
15:14:11austriancodersee you
15:14:14 Quit austriancoder ("using sirc version 2.211+KSIRC/1.3.12")
15:14:38preglowt0mas: try 1.65 eu for now
15:15:32t0masError: "Checksum doesn't match... etc..."
15:16:20preglowwell, bollocks, it should match
15:17:00t0mashm... I used it on the 1.65 EU hex file linked in the wiki...
15:19:01 Quit ismi ("CGI:IRC (EOF)")
15:19:10preglowgimme a sec, and i'll toss in a new correct md5
15:21:23 Join cYmen [0] (
15:21:44*rasher downloads 1.6{5,3}-{US,EU,K}
15:23:26 Quit Zagor (Remote closed the connection)
15:25:55preglowrasher: wanna make some checksums, pretty please? :)
15:26:03rasherI will
15:26:14rasherI thought they were all in the wiki
15:26:38preglowthey're old
15:26:40preglowand numerous
15:26:44preglowand i don't trust them ;)
15:27:14preglowthey might be
15:27:20preglowi'm not just yanking md5s off a wiki
15:27:24rasherI'll go right ahead
15:27:30preglowthen hooray!
15:27:31rashergood point
15:27:53preglowit takes me bloody forever to get firmwares right now
15:28:00preglowi think the uni gateway is having a very, very hard day
15:28:20rasherthe korean firmware zip contains an exe?!
15:29:21rasherweird shit
15:29:25Bagderself-extracting zip?
15:29:27rasherthankfully I'm on windows atm
15:29:34Bagderah, ok
15:29:46Bagderbecause unzip works on self-extracting exes too
15:30:17rasherit's not a self-extracting exe
15:30:27preglowbut should we even include 1.65 checksums?
15:30:31preglow1.65 and rockbox is clearly broken
15:30:34rasherwhy not
15:30:59rashercould we say "WATCH OUT!" for those checksums?
15:31:08rasherI mean, we'll be needing them sooner or later, probably
15:31:28Bagderno, we'll probably have a modified boot loader then
15:31:37preglowthere's no possibility to warn about some checksums right now anyway
15:31:47BagderI know Linus has some ideas to test
15:31:53rasherI'll make the checksums anyway
15:32:00rasherat least for input files
15:32:09rasheraren't you checking input?
15:32:18rasheror only output?
15:32:55preglowi'm only checking the result
15:33:00rasherof course I run out of diskspace now..
15:33:28*rasher shuffles things around
15:34:47rashergotta make sure to get the names right..
15:35:00rasheralso, I'm *so* not going to test all of these on the target
15:35:12rashermost of themm should match what's in the wiki anyway
15:36:18preglowhell no, don't test them, just make sure the patching is correct
15:40:25rasherdoes it patch in-place?
15:41:25preglowapparently, yes ;)
15:41:29preglowit has several temp files
15:41:44preglowbut the original file will be replaced
15:42:51HCloi its hot
15:45:35preglowtime for another attempt at breaking the speakers
15:47:51rasherMy patching with descramble/mkboot/scramble produced the same result as that table
15:48:00rasherexcept for 1.63-US
15:48:15preglowwould you care to update checksums.h with your results?
15:48:20preglowplease beware it is case sensitive
15:48:50rasherBut I'd love for someone to verify the 1.63-us checksum
15:49:09rasherwacky 5068bf62d68fdbfcb3d8a2eaf2f4fdfe v1.63US 11/02
15:49:12rasherthat's what I get
15:49:20rasherI guess that's alright then
15:50:33amiconnHCl: Is there a reason why the searchengine should only compile for recorders & iriver?
15:50:43amiconnImho it should work on all models...
15:50:47BagderI agree
15:51:03HClit was easiest to add to the makefile as i didn't understand it much :P
15:51:04Bagdereveryone is gonna want it
15:51:05HCland it uses a lot of ram
15:51:15HClthough that can be rewritten
15:51:25HClby simply making dbinterface less efficient and read from disk
15:52:34preglowthat should be an easy define to add
15:52:36preglowwhat amount of ram to use
15:52:37amiconnAll archos models that are missing now (ondios, player) have the same RAM size as the recorders (2 MB)
15:52:45HClpreglow: eh...
15:52:49HClpreglow: not in the way its written no.
15:52:56HClamiconn: oh.
15:52:58amiconnThe only slight difficulty may be the charcell lcd of the player
15:53:16HClwell, the only output the searchengine outputs that doesn't work on the player is debug output anyways.
15:53:40rasherpreglow: the checksums you had matched those I got
15:53:42amiconnOkay, so it shouldn't be a problem at all
15:54:01HCl :)
15:54:14HClthe debug output is fun to read though, when you can
15:54:21HClit shows exactly what its doing
15:54:37amiconnAs for the RAM requirement/ inefficiency when reading directly from disk: Imho it shouldn't be hard to simply let it use all available ram for caching, working in chunks if the db doesn't fit at once
15:55:19rasherpreglow: my korean checksums alos matched those in the wiki
15:55:24rashercomitting now
15:55:32preglowcould you try my patcher for some of them?
15:55:39preglowit worked fine the last time i checked it...
15:56:51rasher :-\
15:56:58rashershouldn't change much
15:58:59HClamiconn: well, i leave that open for someone to implement for archos players, for iriver, 28mb should be more than enough for any database
15:59:03rasherpreglow: are you going to check input as well?
15:59:48rasherpreglow: ihp_120-1.63-EU.hex failed
16:00:08rasherihp_120-1.63-K.hex failed...
16:00:25preglowthey're all going to fail then
16:00:37rasherTotally.. us did as well
16:00:54preglowi'll have to boot windows some time to check if it's the compile or my code
16:01:16preglowit did work fine before i stopped touching it, i'm pretty certain of that
16:01:49preglowbut no time to boot windows right now
16:02:10rasherI have md5s of the input
16:02:16rasherif you're interested
16:02:19preglownot really, no
16:02:28rashernot that it's much work to create :)
16:02:31preglowthere's no point in md5 ing the input
16:02:46preglowif the patcher doesn't make a correct firmware, that's the only thing that can be wrong
16:03:02amiconnHCl: As for the db size: you'll never know if someone has a larger db. Plus, keep in mind that the iHP-110 does only have 16 MB.
16:03:05preglowsince the bootloader is internal
16:03:19rasherIt could help users figure out what's wrong
16:03:35preglowthe firmware is the only thing that can be wrong
16:03:40preglowthere are no other factors
16:04:04amiconnFurthermore, implementing dynamic caching will allow to either do a high-performance search using the audio buffer when not playing music, and also a slower search while playing music, using the rest of the plugin buffer
16:04:17HClokay okay, i got your point :P
16:04:26HClfirst let me fix up the code with code policies etc
16:04:29rasherthat last one is excellent :)
16:04:42preglowapart from a corrupt resource section in the exe, of course
16:05:09preglowbut it's ok by me, i'm not able to code the change now anyway, so someone else has to do it
16:06:28rasherwould be interesting to look at the tempfiles
16:07:22HClamiconn: i dunno how to do dynamic caching, but i made sure to keep dbinterface as generic as possible, so it shouldn't be hard to rewrite the code to make it possible
16:07:28*preglow points to the source code ;)
16:07:28BagderHCl: IMHO, you should use a generic macro to access integers/longs with
16:07:32*rasher apt-get installs mingw32
16:07:43HClBagder: mmm?
16:07:44Bagderintead of having whole code blocks #ifdef endianess
16:07:54HClBagder: i have.
16:08:03HClits just that if the endianness isn't defined
16:08:08HClthere's no point in adding that code anyways.
16:08:18Bagderyou use #ifdef ROCKBOX_LITTLE_ENDIAN [code] #endif
16:08:31HClyes. because it reads directly to those things.
16:08:41Bagderwhen I'd prefer a #define getlong(x) macro
16:08:47HClunless you want to make an readlittleendian(fd,blah,blah)
16:09:14HClit already has such a macro, i'm just not sure whether a=a; assignments get properly optimized out by the compiler
16:09:26Bagderit shouldn't do a=a
16:09:28HClthe BE32() is such a macro.
16:09:33HClwell it will
16:09:36HClif i don't add the ifdef
16:09:39Bagderjust do #define readlong(x) x
16:09:51Bagderor #define readlong(x) magic(x)
16:10:01HClmaybe later...
16:10:13Bagderhaving that special code block will cause you future problems
16:10:42HClmaybe, don't really see it at the moment, it just corrects endianness for the longs read in by the read() calls before the block
16:10:58HClbut i guess i can adapt it.
16:11:00Bagderyes, but one day someone will add a long, or remove one or...
16:11:05HClbut later, not now
16:11:21HClmore focusing on getting it to work for other players, getting it to load its tokenstream from a file..
16:11:52Bagderit is a worthy goal to strive at having the code as similar as possible for all targets. You gain in the long run
16:13:18Bagderalso, you should consider the case when you read a broken db
16:13:28Bagderyou have no error checks
16:15:47*HCl goes to fix the long reading..
16:18:22 Join MADTurnIP [0] (
16:18:51*Bagder has been doing multi-target C programming for very many years
16:20:44 Quit MADTurnIP (Client Quit)
16:21:24rasherhow do I build the bootloader?
16:21:45Bagderusing mingw I guess
16:21:51HClBagder: what do you think, should i read the tokenstream from file directly or read the entire tokenstream file into ram?
16:22:07*HCl is thinking it would waste ram since you only need one token in ram at a time..
16:22:14preglowrasher: i developed it with mingw
16:22:20HCloh wait no :x
16:22:48*HCl goes to make some changes to stop tokens needing to be in ram
16:24:10*rasher sets out to create a 48x48 and 32x32 icon
16:24:40rasherhah, that Just Worked
16:24:53rasherwell except I had to manually copy a bootloader.bin in, but other than that
16:25:01rashermake, BOOM, a fwpatcher.exe
16:25:02***Saving seen data "./dancer.seen"
16:26:23rasherwine expoded
16:28:26 Quit ashridah ("Leaving")
16:32:47HCli keep failing at including autoconf.h...
16:32:53*HCl is confused...
16:33:04HClBagder: why wouldn't -I{$OUTDIR} work?
16:33:20HClthat doesn't work
16:33:23HClobjdir gets redefined.
16:33:32Bagderthen you need to adjust
16:33:44HClyes, well, the original objdir is supposed to get passed by outdir...
16:33:56HCl @echo "MAKE in $@"
16:33:56HCl @mkdir -p $(OBJDIR)/$@
16:34:01HClbut its not working :/
16:34:07HClit did for the sim..
16:34:17HCl :P
16:34:25Bagderit looks ok to me
16:34:40HClyea, my fault
16:34:47HCli hadn't ran the new configure in my build dir yet
16:50:02HClsomething broke :x
16:50:14HCldoh o.o..
16:50:16HCltypo XD
16:50:40HClno, not a typo o.o.
16:50:51HClah well, nm. i probably fixed that.
16:52:23HClBagder: awake?
16:52:44HClis BE16 already defined as something or what?
16:52:55HClapps/plugins/searchengine/dbinterface.c:70: undefined reference to `_BE16'
16:53:11HCli haven't typoed, it must be redefining my BE16 :/
16:54:01Bagderthere's none defined in rockbox anyway
16:54:13HCloh, i think i know why..
16:54:28HCli had forgotten to commit database.h, i commited that just a second ago with some other stuff..
17:02:07Bagdergotta run
17:04:19rasherHCl: searchengine.h:23:22: autoconf.h: No such file or directory
17:05:15HClrasher: rerun your configure script
17:05:21rasherah :)
17:07:36rasherrolo blew up
17:07:53rasherit doesn't react to my keypress
17:09:29HClno idea why
17:09:45rasherin fact, rockbox is not reacting to my keypresses period.
17:10:20HCldid i break something? :X
17:10:26rasheroh, it is now
17:10:52HClhold button?
17:11:24rasheroh.. it started working recently?
17:11:38HClsomeone added a hold button check to the button drier
17:11:40rasherso it did
17:12:04rasherUnable to open search Tokenstream?
17:12:07HCl :P
17:12:12HClits a viewer now
17:12:28HClwith no tool written yet to write its input file
17:12:31rasheroh ah
17:13:27HClgonna play my golf computer game with my flatmate a bit
17:13:29HClthen write the tool
17:33:43 Quit DangerousDan (Read error: 110 (Connection timed out))
17:39:02*HCl ish done..
17:46:51HClwhats a good extension for the search filetype?
17:46:54HCl.pse ?
17:47:02HCl(Playlist Search Expression)
17:47:06HClor pst?
17:47:12HCl(Playlist Search Tokenstream)
17:47:28HClor rps
17:47:33HClRockbox Playlist Search
17:47:41*HCl prods rasher , preglow , amiconn
17:47:53ShyK1 vote for rps
17:48:28t0masrockbox smart playlist?
17:49:48preglowas long as it doesn't have tokenstream in the title
17:50:08HClt0mas: fair enough.
17:52:32bobTHC*.foo Fucking Odd Organization"
17:53:07*amiconn wonders what that file should be good for...
17:53:30HClamiconn: search engine input
17:53:43HCloh, you mean bob :P
17:53:53amiconnno, I mean the file
17:53:59HClsearch engine input
17:54:14HCli'll make a wikipage on the searchengine plugin soon
17:54:18amiconnI wonder why the searchengine needs an input _file_. Seems like an unnecessary step to me
17:54:33HClwell, you can not seperate it, but if you do
17:54:46HClyou can store files that'll create playlists that always have an up to date search
17:55:10HCltheres some gain to store a specific search in a file
17:55:14HCland hardly any drawbacks
17:55:18amiconnThen I'd rather save the search query as plaintext
17:55:19preglowsearch expressions should be able to be saved
17:55:27preglowwell, it's more or less the same, isn't it?
17:55:35preglowplaintext representation would be best, though
17:55:42HCli don't have an lexical scanner at the moment, and i feel they're unneeded too.
17:55:44preglowif i've got a pc handy, i'd much rather type the expression by hand
17:55:46amiconnIf it's plaintext, I'm able to look inside
17:55:53HClwe'll have an editor
17:55:56HClthat can open them and edit them
17:56:01preglowmaking a lexer isn't very hard
17:56:01HCland display them as plaintext
17:56:05HCli know that.
17:56:35HClstill i don't see why we would need a lexical scanner when we can make an editor that will prevent the users to even be able to create invalid playlist searches
17:56:55preglowbecause using the player as an input device is not very intuitive
17:57:02amiconnYes, but that editor runs on the device
17:57:02preglowunless you happen to be really good at gui design
17:57:06rasheris my vote
17:57:47amiconnEntering the query in a text editor on a PC and copying it to the device should be always faster
17:57:59amiconn...if you have a PC around, that is
17:58:30preglowbut that can come later
17:58:49preglowbolting on a lexer is no problem
17:58:56HClpreglow: yes it can be
17:59:02amiconnThe rockbox config files are plaintext too, and I like that...
17:59:18HClyou can limit most identifiers to being able to choose between "year" "rating" etc
17:59:21HClwith up and down
17:59:24preglowa lexer should spit out the same token stream you're using now
17:59:33HClfor strings, you can use the virtual keyboard
17:59:35HClsame for numbers
17:59:41HClbut i can always build a lexer.
17:59:43HCli know.
18:00:02preglowbut anyway
18:00:05preglowignore that for now
18:00:08preglowjust hard code your token streams
18:00:08HClthe searchengine will work on raw tokenstreams for now.
18:00:17HClno, its gonna read file token streams
18:00:41preglowbut i too think the neatest would be it using plain text in the future
18:00:48preglowbut it's not a priority
18:02:44amiconnHCl: We should probably make a separate virtual kb for numbers
18:03:14amiconn(or rather, not separate but just using a different, smaller screen template)
18:03:36amiconn...without several pages and the numbers nicely grouped together
18:07:11HClamiconn: yea.
18:08:25*HCl creates the first .rsp
18:08:26HCl :)
18:08:30HClon my ftp :P
18:08:35*HCl wonders if it'll work.
18:08:49*HCl goes to add searchengine to viewers
18:11:56HClhrm. :/
18:17:30HClah, bug..
18:17:43HClhm... or not..
18:17:48preglowi just had a bumblebee visiting here
18:17:51preglowfreaking big one
18:21:48HClstring searching is broken
18:21:52HClsince i didn't test it :P
18:24:21 Join ferenczy [0] (
18:25:03***Saving seen data "./dancer.seen"
18:30:04HClwhat am i doing wrong....
18:35:09 Nick Lynx_ is now known as Lynx_awy (HydraIRC@
18:40:09HClconfused about file dbinterface.c −− ignoring
18:40:09HClconfused about file token.c −− ignoring
18:40:09HClconfused about file Makefile −− ignoring
18:40:09HClconfused about file searchengine.h −− ignoring
18:40:12HClwhats that?! O.o.
18:40:50 Quit bobTHC ("Smoke Weed Every Day !")
18:43:29*HCl needs a small test database..
18:45:57*preglow needs a beer
18:50:08*HCl needs a break
18:50:21HClmy getstring() is broken
18:50:24HClfor unknown reasons
18:50:31HClmaybe due to malloc failing
18:50:51HClyear works, so i know its loading the songentry properly..
18:52:47t0masyou adopted my filname idea? :D
18:53:12HClwell, the info was there anyways, it was pointless to not add it to the search possibilities :)
18:53:42*t0mas updates his rockbox copy :)
18:56:21HClmaybe i should do a different test..
18:56:33HCl"bla merf blah" contains "merf"
18:56:40HClshould match every song
18:56:51t0mas<HCl> maybe due to malloc failing <−− malloc? in plugin buffer?
18:57:03HClmp3buffer at the moment, not plugin buffer.
18:59:50HClhm, oops. i guess scanf(%s) won't take more than one word.
19:01:19preglowanything else would have been kinda stupid
19:01:23preglowwhat do you want it to do?
19:01:55HCli hate scanf.
19:02:01*HCl moves to fgets.
19:02:31preglowwhat do you want it to do?
19:02:35preglowscanf can do lots of neat stuff
19:04:12HClwhy can't i commit..
19:04:34HClit suddenly says unrecognised request
19:04:41HClwhen trying to commit an update to tokentool.c
19:04:52HClpreglow: already fixed it with fgets..
19:06:45HClbut now i can't commit.
19:07:08HClsomeone modified it o.o.
19:07:10*HCl will bbiab
19:07:17t0masModified Files:
19:07:17t0masLog Message:
19:07:17DBUGEnqueued KICK t0mas
19:07:17t0masscanf sucks >.<;
19:07:32t0masfrom 1.1 to 1.2
19:07:35t0masno diff??
19:07:55HClit malfunctioned o.o;
19:08:07HClit worked the second time.
19:19:04HCli even had to check out half my own change, then recommit
19:19:04 Join Tang_ [0] (
19:19:18Tang_hello guys :)
19:19:47Tang_Hi HCl
19:20:02Tang_still progress for your search engine
19:20:10Tang_but still no name lol
19:21:59HCli named it searchengine..
19:22:11Tang_Ah searchengine okay it's the name
19:22:20Tang_cool one
19:22:51Tang_(even if i liked databox too ;))
19:23:06Tang_mmm no BobTHC here apparently?
19:24:15Tang_mmm i've done the translation of the fornoobs section
19:24:33Tang_anyway i merged the two section (status and genearal)
19:24:52Tang_and i think it would be fine to make same for the english version
19:25:21amiconnHCl: If tokentool.c is meant to be built on host, apps/plugins/searchengine/ seems like the wrong place...
19:25:35t0masthat was my question
19:26:01HClif i place it in tools
19:26:05HClit can't find the header :/
19:26:19 Quit ferenczy ()
19:26:32Tang_okay i will come back later
19:26:43Tang_bye Hcl and t0mas
19:27:00 Quit Tang_ ("Chatzilla 0.9.66 [Mozilla rv:1.7.5/20041108]")
19:27:14CtcpIgnored 1 channel CTCP requests in 0 seconds at the last flood
19:27:14*HCl is confused, strcasestr seems to be malfunctioning?
19:27:18t0masHCl: is the .rsp syntax documented somewhere?
19:27:22HClt0mas: not yet.
19:27:30HCli have it on paper :P
19:27:39HClmarkun saw me designing it
19:27:42HClits in EBNF
19:27:45HClnot very readable
19:28:04HClpretty much
19:28:06HClyou can do things like
19:28:16t0mashm... I tought maybe I could write the wiki page for it :)
19:28:22HCl(year>=1980 and year < 1990) and genre contains "rock"
19:28:32HClit can do pretty much anything.
19:28:58HCland the scary part is, it should work too :P
19:29:27t0maslets test :)
19:29:38HClwhats this.
19:29:41HCloh, nm.
19:29:50HCltokenid :)
19:30:05HCli see why it wouldn't work..
19:30:11HClforgot to strip the \n
19:30:19HClthis search query would indeed not work :)
19:31:01HClstill broken though x.x;
19:31:33HClhm. o.o;
19:31:39*HCl slaps his hex editor.
19:31:46t0masunexpected `40` found?
19:31:55t0masat parseExpr
19:32:06 Join muz [0] (
19:32:48HClthen you fed it invalid input :P
19:33:15HClif you want to input a number
19:33:17muzHCl: is your searchengine thing building yet?
19:33:26HClyou need a NUM token, and put the number as the intvalue
19:33:31muzcos i cant find it :(
19:33:34HClmuz: yes, it works too (for numbers)
19:33:50HClcvs update -d
19:34:05muzwhats cvs update -d?
19:34:36HClcheckout new directories too.
19:34:54t0masHCl: I fed it this: (year>=1980 and year < 1990) and genre contains "rock"
19:35:13HClshould be okay, dunno how you inputted it.
19:35:19t0mascopy + paste
19:35:20muzwhere do i type cvs update -d?
19:35:30HClit doesn't have a lexical scanner at all.
19:35:44HClyou need to use the tokentool
19:35:45t0masah, there's the problem
19:36:37t0mashow does it work?
19:37:01HClwoot it worked.
19:37:07 Join Zagor [0] (
19:37:17*HCl successfully did a "bla merf blah" contains "merf"
19:37:25HClt0mas: gcc -o tokentool tokentool.c
19:37:26ZagorHCl: allocating 500 bytes on the stack is not a very good idea
19:37:31HClthen run it
19:37:32HClZagor: where?
19:37:37t0masHCl: ah, have to do it manually
19:37:38HClyou mean my two tokens
19:37:55HCli can move them from the stack, shouldn't be much trouble.
19:38:38 Join Sucka [0] (
19:39:35HClfor now, we weren't focusing on low memory, but possibility and getting it to work :P
19:39:44Zagoralso (to be my usual boring me) I would suggest using a >3 char extension for new types. we don't support non-vfat systems anyway.
19:40:10 Join sox [0] (
19:40:12ZagorHCl: right, but stacks are small and non-dynamic. so even with 32MB ram you can easily blow the stack.
19:40:12 Nick StrathAFK is now known as Strath (
19:40:20 Quit sox (Client Quit)
19:40:22amiconnHCl: The Makefile is somewhat off, I try to fix that
19:41:58 Join Shagnar [0] (
19:42:02HClamiconn: please do :)
19:42:13HCli have a bug in my tokenstring thing..
19:42:25amiconn(1) Of course we need a linker script for archos too (2) Build for all targets
19:43:27HCloh, i have no clue about linkerscripts :/
19:43:55amiconnThat's no problem for your search engine.
19:44:11amiconnRockboy uses a speciall linker script on archos for only one reason:
19:44:30amiconn...because it is _not_ a plain plugin, but an overlay.
19:44:39amiconn...due to its size
19:45:05amiconnThat's unnecessary for searchengine, we'll just use
19:45:29HClcould it be i'm having stack overflows on my parser? o.o..
19:45:56muzamiconn: how much did u pay for that h140?
19:46:04amiconnsearchengine.rock: 5716 bytes (that's on ondio)
19:46:16amiconnmuz: Still no iriver :(
19:46:29muzamiconn: but how much did u pay?
19:46:31preglowi'd see about getting one another place, where i you
19:46:45amiconnpreglow: yes, I should
19:46:53amiconnmuz: I did not yet pay
19:47:05muzI might want to trade my h140 for an h340
19:47:05amiconnI was supposed to cost ~350,-
19:47:27amiconnmerf. searchengine doesn't compile for player :(
19:47:55amiconn..due to no lcd_getsringsize() and lcd_putsxy()
19:48:11HClamiconn: you can simply redefine the PUTS for player
19:48:14HClto be nothing
19:48:35Zagorwhat does PUTS do that DEBUGF doesn't?
19:48:44HClrun on target :)
19:49:02amiconnDEBUGF() does run on target...
19:49:06HCli never see it o.o.
19:49:17amiconn...if you build a debug build and use gdb :)
19:49:37*HCl ish a step closer to his bug
19:49:42Zagorwhy are you using a macro instead of a function for this?
19:49:46amiconnHmm, iirc there is still no gdb stub for iriver?
19:50:04*Zagor feels like a grumpy old guy, complaining about everything ;-)
19:51:02HCl :P
19:51:47amiconnHCl: When your PUTS() reaches the bottom of the screen, it may do nasty things...
19:51:56HClamiconn: i know..
19:52:23preglownot only is there no stub, it's been removed from the todo list
19:56:16amiconnHCl: I defined a simple PUTS() for player, not doing these nasty things (simply alternating between the 2 possible lines)
19:58:12Shagnarhello everybody
20:02:36HClpreglow: i figured that no one was gonna make it, so i removed it
20:04:41preglowwell, removing it from the todo list doesn't exactly better the odds of it happening
20:04:46*HCl is bummed by his malloc not working properly..
20:04:53HClyou can readd it if you want to
20:05:05preglowno worries
20:05:17HCli don't think many developers are willing to open up their iriver and solder wires to it
20:05:20HCljust to have a nicer debug interface
20:05:30preglowi might, if i get another one ;)
20:05:52preglowbut no
20:05:54preglowwont happen
20:06:16*HCl feels bummed about getstring() not working for stringidentifiers..
20:14:10amiconnpreglow: <- there
20:17:14 Join elinenbe_ [0] (~elinenbe_@
20:17:22 Quit elinenbe (Read error: 54 (Connection reset by peer))
20:17:30 Nick elinenbe_ is now known as elinenbe (~elinenbe_@
20:19:51t0masHCl: How does that tokenizer thing work?
20:21:24t0maswhat would I input for a query like this:
20:21:32t0masgenre contains "rock"
20:22:31HClhold on..
20:23:03HClyou'd input filename
20:23:07HCl17 for stringidentifier
20:23:14HClthen 7 for genre
20:23:19HClthen 10 for contains
20:23:23HClthen 16 for string
20:23:25HClthen rock
20:23:27HClthen 0 for eof
20:23:42t0masthat's not really easy...
20:23:46HClwell duh
20:23:51HClits raw at the moment
20:23:55HClno lexical scanners or interface yet.
20:24:09HCli'm still working on the engine. don't expect end user stuff.
20:24:22*t0mas starts thinking of a perlscript...
20:24:24HClthe interface will be much easier.
20:24:30HCljust wait.
20:24:38HCland be patient :/
20:25:05***Saving seen data "./dancer.seen"
20:26:16HClwhen soccer starts i'll call it a day.
20:27:22 Quit einhirn ("Miranda IM! Smaller, Faster, Easier.")
20:29:42HCli have a concrete error.
20:29:53*HCl will analyze further, brb
20:30:34 Join ferenczy [0] (
20:30:34t0mas17 7 10 16 Rock 0
20:30:55t0masand then I search the player... it gives 0 hits
20:31:08HClthats what i'm working on tomas XD
20:31:12HClstring searches are broken
20:31:14t0masghehe ok
20:31:16HCland i think i figured out why.
20:31:32HCli.. kind of forgot to output the number in big endian in my tokentool >.>;; XD
20:31:57HCland thats no problem with the tokenkind, cause thats just a char
20:32:06*HCl goes to fix :)
20:36:32HClmuch better.
20:36:45HClnow to see whether it actually works if i remove all my debug lines
20:39:04 Join Tang_ [0] (
20:40:03HClhits: 1
20:40:09HClwe have string comparisments :)
20:40:23*HCl did a regular title search for some text
20:40:34HCl"title contains "walking in the" "
20:40:45HClone hit in my 3 song test database
20:40:56HClthe one of which being nightwish - walking in the air
20:40:57HCl :)
20:41:05*HCl goes to remove all the debug info...
20:41:15preglowcomparisments? :P
20:44:26HClyes o.o
20:44:36HClyou compare one string to another string according to a certain way you specify
20:48:39 Quit ferenczy ()
20:49:29HClthere we go..
20:49:38*HCl goes to fix his database and see whether stuff works yet..
20:52:21Tang_I'm back :)
20:52:42Tang_Mmm Bob still away thought
20:55:29HClnow you'll see there'll be bugs in album and artist searches...
20:56:56HClworks :)
20:57:05HClhits: 20 (all albums with "moon" in their name)
20:57:23HClsearchengine done :)
20:57:37HClactually, all songs that have "moon" in their name..
20:57:53HClin their albumname*
20:59:53preglowyou implement stuff like 'string that starts with 'lol'' and 'string that ends with 'lol'' as well?
20:59:53HClit found all the songs of air - moonsafari, all the songs of enya - shepherd moons, and two ddr songs with moon in their dirname :)
21:00:01HClnot yet :)
21:00:20preglowand it's comarisons, not comparisments, btw ;)
21:00:24preglowcomparisons, even
21:00:25HClits not hard to add extra stuff
21:00:38HCleither way, i'm happy that it works :D
21:00:43preglowwell, this sounds pleasing
21:00:52preglowdidn't speen too long a time in coding it either
21:01:05HCli hadn't even considered the "starting with" and "ending with" yet
21:01:22HClgonna relax now, then after that write a wikipage on it
21:01:24preglowplease implement that with ^ and $, like in regex ;P
21:01:27HClwhere people can add stuff
21:01:30HClpreglow: no :P
21:01:38 Join Byron [0] (byron@
21:01:48Byrondoes anyone know where I could buy a new Archos
21:08:11 Quit Tang_ ("Chatzilla 0.9.66 [Mozilla rv:1.7.5/20041108]")
21:09:21HCli love recursive descend parsers :)
21:09:36preglowthey're nice
21:10:34t0mashm... preglow?
21:10:40t0masfor my dictionary idea..
21:10:55t0masis it faster to store words / descriptions binairy?
21:11:11t0masor should I just use a plain textfile?
21:11:37Byronanyone know if the older archos that are supported by rockbox are still being sold
21:11:43preglowi doubt it
21:12:22preglowt0mas: something seekable
21:12:30preglowt0mas: doesn't really matter as long as seeking is fast
21:12:38t0masyes, I have an index file at the moment
21:12:55preglowbinary search would be the nicest
21:13:09preglowbut i guess that would make for a pretty huge index file
21:13:24t0masyes... and I have no idea on how to store the text binairy...
21:13:46preglowall that is required for that is that all entries are the same size
21:13:59t0masthat would waste a lot of space...
21:14:11preglowso all your words would have to be the same length, again, all words would have to be as long as the longest word
21:14:17preglowyes, that would waste an incredible amount of space
21:14:42preglowbut for all other approaches, you pretty much have to read the index to memory for it to be fast enough
21:15:03t0mashuh? why?
21:15:20preglowhow many words do you have?
21:15:21t0masI have a very simple thing like this now:
21:15:27t0masword offset1 offset2
21:15:37t0masso a line for every word... in my index
21:15:49preglowwhy two offsets
21:16:00t0masstart and end of description
21:16:36preglowa special marker for 'end of description' would be more space efficient
21:16:37t0masbut reading that index into mem first wouldn't be very much faster I guess?
21:18:52preglowbut yes
21:18:54preglowwhen i enter a word
21:18:59preglowhow do you look for it?
21:19:29preglowyou can also do binary search if you have a entry sync marker in your index
21:19:38preglowit'll just be quasi binary, though
21:19:55preglowif you don't know what binary search is, look that up first
21:20:01 Join ze__ [0] (
21:21:39 Quit ze (Read error: 104 (Connection reset by peer))
21:21:40 Nick ze__ is now known as ze (
21:22:01*t0mas is going to read some things
21:22:54preglowyou're probably going to have a huge word list
21:23:12preglowso you'll need a strategy to find what word you're looking for as fast as possible
21:23:46 Join webguest28 [0] (
21:26:35*HCl goes to watch soccer :)
21:34:49 Quit webguest28 ("CGI:IRC (EOF)")
21:34:52 Join TCK- [0] (
21:38:26 Part Byron
21:42:46t0maspreglow? I understand that binary search idea... but I don't understand how to get my words into binary data? and how to do a fast compare on them
21:46:23 Quit muz ("CGI:IRC")
21:46:42preglowthe words you save as they are
21:46:55preglowthe only thing you need to save as binary data are numbers
21:47:03t0masthe offsets?
21:48:10t0mashm.. strcmp() gives a bigger than or smaller than thing...
21:48:16t0masis that lenght?
21:48:22t0masor sorted?
21:48:49preglowi don't understand a word of what you're saying
21:49:11t0maswell... strcmp() can return -1 or 0 or 1
21:49:42t0masstrcmp(a, b) then <0 = a is less than b
21:49:44preglowyou can't use strcmp
21:49:49t0masarg :)
21:49:51preglowthat is, yes, you can
21:50:00preglowno buts, you can ;)
21:50:05t0masah :D
21:50:45 Join muesli_- [0] (
21:50:52preglowit returns -1 if string 1 is alphabetically lower then string 2
21:50:54t0maswell... then I only need to sort the words in my index
21:50:58preglowif that is so, you need to seek higher
21:51:03preglowthey must be sorted
21:51:07t0masand get them all into an array
21:51:16preglowand make everything the same case
21:51:35preglowbut how do you plan to do your searching?
21:51:45*amiconn stares at his optimised mmc code and wonders why the binary is _exactly_ the same size
21:52:43t0mashm... can I flood your privmsg?
21:52:47 Quit TCK (Read error: 110 (Connection timed out))
21:53:41preglowwell, sure, but i have to go soon
21:54:50 Join StrathAFK [0] (
21:59:04 Quit muesli_- (Read error: 104 (Connection reset by peer))
21:59:10 Join muesli- [0] (
22:01:32 Join muz [0] (
22:02:19muzok i just got the latest bleeding edge build and i cant see searchengine.rock :(
22:03:28HClits a viewer.
22:04:04muzbut how can i use it
22:04:26HClits not really meant for end users yet.
22:04:41muzo ok
22:04:58*t0mas starts working on his dictionary thing
22:04:59t0mastnx preglow
22:05:14muzanyone seen the new iaudio a2? looks amazin
22:07:10HClnope, i'm happy with my h140 :P
22:08:22*odd just recently spent 15 hours with Tag & Rename organizing the music on his H-140 :(
22:08:46odd is the result... bwahahahahahahaha
22:09:02oddpity that it drove me insane in the process
22:09:52muzomg i love tag and rename
22:10:50muzpshh thats pretty amateur organisation there
22:12:58ShyKfoobar2000's masstagger > *
22:13:06oddi've used foobar's
22:13:13oddtag & rename is far, far superior
22:13:17 Quit Strath (Read error: 110 (Connection timed out))
22:13:19oddmuz: how would you improve it?
22:13:22ShyKodd: really? how?
22:13:31oddShyK: in just about every way imaginable
22:13:47ShyKi can't imagine how something can be far superior to masstagger
22:13:55 Join Patr3ck [0] (
22:14:13muzodd: im a fan of first letter capitals
22:14:31oddi use first letter caps in most of the filenames and id3s
22:14:41oddbut i honestly don't care much about case at all
22:14:55muzoh ok
22:15:39muzfor some strange reason, today i really want to trade my h140 for a colour screen mp3 player
22:17:22muzbut its too thick and video support aint that great
22:18:45ShyKi'll get h320. has a nice big screen, small enough for my taste, and its features seem good.
22:18:57muzif i were u
22:19:04muzid get the i audio x5 instead
22:19:04ShyKbut exeansive
22:19:23ShyKhm audio x5?
22:19:31muziAudio X5
22:19:35muzby Cowon
22:20:55ShyKi just added jetAudio to Musepack program support list today
22:21:00ShyKsomeone sent an email it supports it
22:21:13Gmuz: foobar2k's masstagger is _extremely_ powerful due to the TAGZ interface it provides
22:21:19 Nick G is now known as thegeek (
22:21:32thegeekdoing first-letter caps is not a problem at all
22:21:41thegeekit's not easy to do it
22:21:51muzomg guess wot
22:21:59muzi just sort of played a video on my h140
22:22:17muzi loaded a gameboy rom that actually plays a video
22:22:35muzi can send u it if u want
22:22:56*thegeek realizes irc is not the best media for sarcasm
22:23:01muzok it looks better in greyscale
22:24:11ShyKif RockBox will be supported on iAudio i'll get it ;)
22:24:48muzthegeek: do u want it or were u bein sarcastic
22:25:07***Saving seen data "./dancer.seen"
22:25:37 Quit Zagor (Remote closed the connection)
22:27:31muzi have like 1gb of gameboy roms
22:32:14muesli-happy playing
22:32:31muzzelda's great on the iriver
22:32:57muzdo u reckon we will get gameboy emu at full speed anytime soon?
22:35:18oddwhat about with overclocking?
22:35:30oddi used to have a gameboy emu on my handspring visor PDA\
22:35:40oddbut you needed to overclock the cpu to get it to run right
22:36:11muzi wish rockboy ran perfecy
22:36:34HClnot soon, we need an assembly cpu core
22:37:10*HCl thought up some optimizations for the searchengine
22:38:38muzHCl: u shud try out video gb roms
22:39:32*HCl needs to implement this before he forgets.
22:39:38muzi mean like when ur on a bus or sumthin
22:40:38*HCl immediately bumps into problems when trying to implement a filter..
22:40:46HCli'm gonna need a level variable.
22:45:35 Quit muz ("CGI:IRC")
22:48:10HCli hate it when something compiles immediately without any errors >.<;
22:48:24HClthat usually means there'll be runtime bugs
22:49:38HClan "and" will be dominant on its own level.
22:49:52*HCl hopes thats correct
22:51:59 Join DangerousDan [0] (
22:52:43*HCl wonders whether that breaks things..
22:52:52 Quit DangerousDan (Read error: 131 (Connection reset by peer))
22:54:14HClwrong :/
22:58:03*t0mas wonders who is good at perl here
22:58:15t0mas(and not preglow or some other busy coder)
22:59:45HClnot me
23:03:42 Join DangerousDan [0] (
23:03:45 Quit muesli- (Read error: 131 (Connection reset by peer))
23:04:51*HCl altered his search engine design which will now allow better optimization
23:07:55 Quit Shagnar (Remote closed the connection)
23:12:50 Quit lolo-laptop ("Client exiting")
23:13:51 Quit Sucka ("a bird in the bush is worth two in your house")
23:14:15preglowasm indulge time
23:15:26preglowi finally think i've got imdct_l nailed
23:16:14 Quit TCK- (Read error: 110 (Connection timed out))
23:17:06preglowyes :PPPPP
23:22:20*preglow catches the digital flow
23:25:19oddnice work :)
23:30:43 Join amiconn_ [0] (
23:33:04 Join Querty [0] (
23:33:09HCltoo bad we don't have an easy token file generator yet...
23:33:33*HCl needs testcases to see whether a) his searchengine optimization helped and b) whether it didn't break anything
23:34:03 Join eVox [0] (
23:34:36eVoxquestion: how do i prevent device i/o errors when copying from my ihp120 to my harddrive?
23:34:58HClreplace your hdd? :X
23:35:15eVoxis it the computer hard drive?
23:35:24HCleh, i'd say its your iriver hdd :X
23:35:28HClat least, its far from normal.
23:35:42eVoxwell im due for a replacement ihp
23:35:43Quertydoes the optimisation of imdct_l mean that the MP3 codec is now even faster?
23:36:27eVoxthis 120 kinda got hooked up to a cell phone charger and kinda wont turn on without the reset button and kinda gets really hot now and has really corruped menus and wont play music
23:36:50t0maspreglow: I got the searching thing working on my pc, I'll port it to rockbox tomorrow
23:36:52 Join ashridah [0] (
23:36:54eVoxbut i can still usb connect it and access the files
23:37:02oddevox try running chkdisk on it
23:37:08eVoxit had i/o errors before i fried it
23:37:21oddno, chkdisk
23:37:39oddthen you may want to run a data recovery app to get the stuff off it
23:37:48eVoxdos> chkdisk g:\ ?
23:37:49odda good free one is called Active@ File Recovery
23:38:02t0maspreglow: You know my next problem... the conversion of GNU's cide data to my rockbox-dictionary-format :)
23:38:03oddbut run chkdisk first
23:38:26eVoxhow do i run chkdisk?
23:38:53 Join ismi [0] (
23:39:15 Quit amiconn (Read error: 60 (Operation timed out))
23:39:16 Nick amiconn_ is now known as amiconn (
23:39:29oddRun... −−> cmd −−> chkdsk /?
23:39:55 Join CyberSon [0] (
23:40:00eVoxah its chkdsk not chkdisk
23:40:06oddyeah that was my mistake ;x
23:40:06eVoxoops ran it on my computer hd
23:40:26oddno hard in that
23:40:28eVoxctrl-c for life
23:40:35oddctrl+c ftw!
23:40:40oddbut yeah, chkdsk is your bestest friend
23:40:42CyberSonhi is there any firmware for av400 series? sorry i am new just found the site
23:40:49oddyou may wanna run chkdsk /f on the iriver
23:40:56oddif you suspect that it has errors
23:41:15eVoxit ran in like 3 sec and says no errors
23:41:22*preglow FINALLY gets a good reason to try out a mp3 frame analyzer
23:41:38preglowt0mas: good luck on that
23:42:06t0masyes... that will be the hardest part I guess...
23:42:18eVoxxcopy g: c:\malariabackup /s
23:42:25eVoxerr wrong window
23:42:26odd3 seconds!?
23:42:32eVoxreal fast
23:42:35oddit should've taken much, much longer than that...
23:42:41eVoxespecialyl on usb
23:42:53eVoxit says the right info tho...19,xxx,xxx kb
23:43:02oddalthough actually maybe not
23:43:06oddif it's USB2.0
23:43:13oddand only 20gb
23:43:17eVoxive never gotten usb20 working
23:43:24 Quit ashridah ("Leaving")
23:43:47oddwell, try using a data recovery app
23:43:59eVoxyeah im trying that one u suggested
23:44:02oddirivers are FAT32; you shouldn't have any probs
23:44:07eVoxCyberSon: no idea
23:44:11oddCyberSon: no idea
23:44:20 Join ashridah [0] ( ?
23:44:41CyberSonhehe yea tried it only found this site
23:45:33eVoxwhat do i want to do to backup in active@? create image?
23:45:46eVoxwill i still be able to access the files or is it gonna compress them all into one?
23:49:01oddwell i would recommend creating an image as the first step
23:49:15oddand then once you've got an image, recover files either from the image or from the unit itself
23:49:17HClshould we have a lexical scanner to create search querys or a nicer interface?
23:49:28eVoxwhats a lexical scanner?
23:49:40HCla parser that parses text into tokens
23:49:54*HCl is thinking nicer interface.
23:50:18oddmeh, entering text on one of these things has got to be a pain ;x
23:50:20eVoxtokens o_O
23:50:34HCli can make a smart editor that allows you to only create valid search querys by limiting what you can choose because it can know whats valid and whats not.
23:50:36oddhave you guys considered something like T9 word completion as exists on some cell phones for text entry?
23:50:44HClodd: another good point.
23:50:51HCli'll make a smart editor :)
23:51:11oddmake sure to make it optional, though... many songs use some very non-standard words ;x
23:51:24HClno no. for text you'll just have the keyboard.
23:51:30HCl :)
23:51:32*HCl nods.
23:51:40HClnobody touch the searchengine, its my hobby project :D
23:51:45eVoxerror 255 reading sector 72135 on malaria (g:)
23:51:48oddah i see, selecting letters with directional buttons
23:52:16HCland selecting between a limited set of tokens depending on whats a valid token at that point ^.^
23:52:19eVoxwhat about an autocomplete in a search?
23:52:26oddwow that's pretty 1337, HCl :D
23:52:33 Quit t0mas ("good night")
23:52:37oddthat's essentially what it has, evox
23:52:45oddat least if i understand correctly
23:52:45 Quit CyberSon ()
23:52:50HCli can both make entering search querys fast, and limit it for stupid users to not be able to mess it up :)
23:53:01oddgod, you guys RULE
23:53:10oddi wish i knew anything about programming so i could help out
23:53:17oddis there like a paypal donate or something?
23:53:19*odd checks wiki\
23:53:23HClthere is
23:53:41HCldonate button on the main webpage.
23:53:48preglowprogramming is more fun than donating :P
23:53:50HCli don't think i'll get any of that though xD
23:54:09oddwhy not?
23:54:15eVoxfeatures i want: games (snes emulator?), play mp3s, turn on/off when external power turns on off
23:54:19oddwhat's the system for distributing the donations? ;x
23:54:30eVoxit gets annoying to turn it off whenever i turn my car off
23:54:46oddhow are you gonna get snes on a 4-tone grayscale screen? ;x
23:55:05eVoxer gameboy
23:55:14 Quit Querty ("Leaving")
23:55:14eVoxidk they are smart
23:55:17oddwell that's already in the works
23:55:35HClthe turn on/off when external power turns off/on is at least possible for the off bit
23:55:44eVoxthats what i figured
23:55:46*amiconn is in casting hell
23:56:25eVoxis there a need for partially working 120s?
23:56:32eVoxand remotes
23:56:42HClthe remote you can still sell to people
23:56:47HClthere was a person without a remote the other day
23:56:51HClhe bought a new one for 60$
23:56:57oddremotes pop up on ebay from time to time
23:57:13HCli dunno about people wanting a partly working 120..
23:57:22HClmaybe linus? he already has two irivers though.. an 120 and 140
23:57:23eVoxwell i mean donating it for parts
23:57:42 Join lostlogic [0] (
23:57:43oddi'd take a bum-drive 140
23:57:44oddbut not 120
23:57:54*odd wants to put in the new 1.8" 60gigger
23:58:12oddbut i wouldn't pay much for it
23:58:18eVox3 120s + tape
23:58:19oddafter all, the drive is like 85% of the cost ;x
23:58:59eVoxthere was an mp3 player that people were taking the harddrives out of, hacking them a bit, then selling them for more than the mp3player

Previous day | Next day