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 2005-01-10

00:00:23jypIt thought this was temporary so I didn't care too much
00:00:37jypsame as for the entire built process
00:01:16Bagderso why the changes?
00:01:20jypI put my changes in the diff so you can see what I've done... But I'm perfectly aware it must be changed to accomodate
00:02:16jypWell, that not strictly necessary (and I'm not sure at all it is the best method to achieve my goal) ...
00:02:38jypbut since I work with an experimental compiler I often need to check the intermedate ouptuts
00:02:41jyp(.i & .s)
00:02:56jypso I spliced in ad-hoc stuff
00:03:24Bagderfair enough, but that shouldn't be committed...
00:04:37jypThough if you have a clean method to generate intermediate targets that would be cool to provide it
00:05:02Bagderwe don't
00:05:05amiconnRegarding the sizeof(int) problems: What about simply changing the places where 4 byte integers are necessary to use long instead? This wouldn't hurt the other platforms
00:05:09Bagderbut I agree it would be good
00:05:25Bagderamiconn: I think we should do that, yes
00:06:24amiconnint should then be used in places where 2 bytes might be enough, but 4 bytes won't hurt either. The compiler can then decide depending on the platform
00:07:12amiconnIn this case it is often better than explicitly use short, because the code sometimes gets considerably larger.
00:07:28Bagderyes, we should avoid using short when int is fine
00:08:03jypI specified only u32 or s32, and only where that was necessary
00:08:25jypbut still it can be replaced with long
00:08:29BagderI vote for using long and unsigned long
00:08:37amiconnSo it should be easy to replace this with unsigned long resp. long, and all would be fine.
00:09:33Bagderjyp: other minor nits: we use /* C comments */
00:09:43jypOh yes...
00:10:12jypWhen I realized that was too lazy to change before submitting the patch :P
00:10:19Bagderhaha ;-)
00:11:16jypAlso, do you insist on C style, as opposed to C99 ?
00:11:19amiconnBagder: I never really got why this is a requirement, but since this is rockbox style, I stick to it. In fact, this is sometimes useful, in that I put my debugging comments in // C++ style. This way they are easy to spot.
00:11:37Bagderjyp: yes
00:11:55Bagderamiconn: it started as a requirement to make it possible to build with older plain C compilers
00:12:05 Join lImbus [0] (
00:12:10Bagderbut now it is mostly for consistent style
00:12:10lImbusgood evening
00:12:17amiconnhi lImbus
00:12:21jypgo for K&R style :P ;)
00:12:33Bagderlet's not overdo it ;-)
00:15:46amiconn(short or char vs. int) I tried this while developing the grayscale lib, in order to slim down the control structure a bit. The code grew by more than 500 bytes, which was roughly 10% of the whole library...
00:16:41amiconn...of course I reverted it.
00:18:35Bagderjyp: does hwgmini.h really fit in the export/ dir?
00:19:05jyp don't know
00:19:19 Quit pfavr ("ChatZilla 0.9.61 [Mozilla rv:1.7.5/20050105]")
00:19:23jypmaybe it should not exist at and be merged with other files
00:19:35jypThat's something I wanted to ask...
00:19:46jypAnd there's debug stuff in there too
00:19:50BagderI removed the include of it from the cpu.h header file now
00:20:01Bagderdidn't seem to fit the pattern from the other cpu includes
00:21:51Bagderjyp: what's the emu_ functions for in kernel.c ?
00:22:02jyphmmm; that's debug stuff
00:22:18Bagderwant me to ignore that or move it into the tcc370 section?
00:22:21jypagain I put those anywhere...
00:22:31Bagderok, so I remove that for now
00:22:32jypThey are spotted by the emulator
00:23:06jypwhich outputs the message
00:24:05amiconnjyp: Did you update the calmrisc gcc?
00:24:25jypnot perfect but better than the original
00:24:47jypI currently use it to compile my stuff
00:24:48 Quit midk (Remote closed the connection)
00:27:17BagderI spotted a mistake in ata.c
00:27:31Bagderyou spelled the cpu TC730 in the initial #if
00:27:45jypmmm, that shouldn't have made it to the patch
00:28:09jyppurely work in progress ;)
00:28:17jypthanks for the info though :P
00:28:50BagderI'm thrilled by your fast development
00:29:20jypheh ;)
00:29:44jypWe spent a lot of time preparing for a sudden jump ;=)
00:29:47amiconnBagder: I have a small question concerning mp3_playback.c...
00:29:53 Join Ka [0] (
00:30:06Bagderamiconn: try me!
00:30:18jypJoking aside, having a "full fledged" emulator makes things easier
00:31:22gromit`i'm thrilled by jyp :)
00:31:39amiconn...concerning mp3_init(). Is there a reason why for MAS3507, mpeg_sound_channel_config() is called before all other settings (line 1019), while for the other MAS' it is called much later (line 1036)?
00:32:30***Saving seen data "./dancer.seen"
00:32:53BagderI have absolutely no idea! ;-(
00:33:39Bagderthat's Linus and Jörg's area
00:34:33Bagderjyp: in thread.c:create_thread()
00:34:43Bagdershouldn't there be a store_context() for tcc730?
00:35:05jypI have a slightly different approach
00:35:10amiconnI need to mess with that in order to get the variable stereo width in. In fact, I even see no reason why this is a separate function, while all other settings are handled by mpeg_sound_set()
00:36:50amiconnAh no, I just found that this is used both ways...
00:37:12Bagderjyp: any particular reason?
00:38:37Bagderjyp: you've added emu_debug() calls in the middle of generic code
00:38:39jypThe problem faced was that I needed an address register to store the context
00:39:07Bagderperhaps you should use DEBUGF() for your emulator stuff
00:39:33Bagdersince DEBUGF is already used in the cod for debug output
00:39:36jypthat was difficult so I changed it to store on the stack, which in turn led to the other changes
00:39:55jypI'll use DEBUGF then
00:40:18jypSorry, newbie here :@
00:40:26Bagderno worries
00:40:54Bagderplease don't take this as complaints, there are just remarks on how to get your code mixed with the existing in a better way
00:41:14jypHeh, alright ;)
00:50:37 Quit Ka (Read error: 60 (Operation timed out))
00:50:51BagderI don't like the 'DISKLESS' and 'NOPOWERMGT' defines in config-gmini120.h
00:51:04jypOf course
00:51:19jypthat was just to make the stuff compile
00:51:37Bagdercompiling is good! ;-)
00:51:57jyp... a requirement to run ;)
00:59:40BagderI gotta go to sleep now
00:59:47jypme too
00:59:49BagderI committed a bunch of files
00:59:58jypalright, thanks
01:00:02Bagdersome of them slightly edited by me
01:00:31Bagdergood night
01:00:40jypShall I re-submit a cleaned up patch for the rest ?
01:00:49Bagderplease do
01:00:57jypsee you
01:00:59lImbusgood night
01:01:11 Part jyp ("Leaving")
01:02:59amiconnBagder: You broke the sim builds, and caused a warning for Ondio...
01:03:06BagderI noticed
01:03:22BagderI think I fixed at least some of them afterwards
01:04:43amiconnI wonder why it was necessary to take out the led() call from panic.c. led() was empty anyway if HAVE_LED was undefined (Ondio)
01:05:13BagderI think jyp is a bit quick to remove function calls that don't do anything
01:05:37BagderI'll revert that change
01:07:02Bagderok, then it should get all green again
01:07:16amiconnI plan to add some indication for MMC activity to the Ondio version, using the led() calls that are already there to switch a flag. The status bar function would then use this flag to show an icon.
01:07:49amiconnOk, maybe this is unnecessary (and even won't work) from panic.c...
01:08:18Bagdersleep time!
01:08:35amiconnThe MMC-USB bridge is able to signal activity to the controller, so Ondio can also show if there is activity while on USB
01:08:42amiconnNite, btw
01:09:54amiconnI think I'll follow the example. Goodnight.
01:10:10 Join scott666 [0] (
01:10:20lImbusnite 2, amiconn
01:10:23 Part amiconn
01:10:36 Join Ka [0] (
01:10:54scott666do I read the change log correctly? is the gmini120 really supported now?
01:11:34lImbusthere are some people working on it, yeah
01:11:59scott666is it finished? or just new code in preparation for it?
01:12:01lImbusnamely 'jyp'
01:12:22lImbusbut they seem to work fast
01:41:04gromit`he is magic :)
01:42:22 Join midk [0] (
01:58:33 Quit Stryke` (Read error: 110 (Connection timed out))
02:00:04 Quit scott666 ("CGI:IRC")
02:32:31***Saving seen data "./dancer.seen"
02:42:20midkthat was random
02:55:02 Quit midk ("Leaving")
03:06:18 Join lImbus_ [0] (
03:07:14 Quit lImbus (Read error: 54 (Connection reset by peer))
03:07:14 Nick lImbus_ is now known as lImbus (
03:17:29 Join midk [0] (
03:27:04 Join Stryke` [0] (
03:27:30 Quit Stryke` (Remote closed the connection)
04:32:33***Saving seen data "./dancer.seen"
04:36:38 Join ze__ [0] (
04:36:52 Quit ze (Nick collision from services.)
04:36:57 Nick ze__ is now known as ze (
06:20:52 Quit Strath ("Client closed")
06:32:34***Saving seen data "./dancer.seen"
06:45:16 Quit jkarns ("Leaving")
07:16:28 Join TexJoachim [0] (
07:32:52 Join LinusN [0] (
07:33:20dwihnohola señor
07:36:51LinusNbuenos dias
08:04:08dwihnoYou got a fat directory sorting utility? :)
08:07:14LinusNa what?
08:12:21dwihnowell, an utility which will sort the directories by name, as opposed to using old DOS "defrag <drive> /sn"
08:12:52LinusNas a kind of file system optimization?
08:13:00dwihnoyou could say that
08:13:34dwihnoThe iriver firmware won't do the sorting for me (dumb, dumb dumb)
08:14:45dwihnoHaven't found a menu option for it (and I've checked the entire settings menu)
08:15:06dwihnoI _could_ read the manual, but something tells me I won't find anything there.
08:16:31dwihnoyou know what CPU the ifp models use?
08:17:11LinusNi have no idea
08:18:18dwihnoAs a quick test, I've reformatted the unit and now I'm copying the files to the unit chronologically
08:19:41dwihnoDo you think there is such an utility btw?
08:21:28LinusNcheck this:
08:30:17 Quit midk ("poof")
08:32:36***Saving seen data "./dancer.seen"
08:41:03 Join Zagor [242] (
08:56:16 Join Bagder_ [0] (
09:10:35 Join LePoulpe [0] (
09:11:53LePoulpeHi all
09:12:23 Quit Bagder (Read error: 110 (Connection timed out))
09:14:44Zagoroh crap, we'll need to to a monster commit to make all code 16-bit safe to run on the gmini
09:15:17*lImbus braces for commit
09:15:29Zagornot yet :)
09:15:34LinusNyeah, isn't it wonderfulö
09:16:03Zagori'm not sure what I think is the least ugly way to do it
09:16:58lImbustypedefs for types ? or is it more cpu-specific ?
09:17:48Zagori really don't like typedefs
09:18:18lImbusdefines ? *duck*
09:18:19LinusNme neither
09:18:29LinusNi see two options
09:18:36LinusN1) use typedefs
09:18:53LinusN2) use "long" whenever a 32-bit type is required
09:19:48lImbusis there a way to influence padding and packing outside of structs ?
09:20:11LinusNlImbus: gcc has a few command line options for it
09:20:26LinusNi think
09:20:38Zagoras much as I dislike typedefs, 2) is far from obvious when reading the code and would likely result in many bugs
09:21:45LinusNwhen we started the rockbox project, it was such a relief when we didn't have to bother about 16-bit targets :-)
09:22:32LinusNand now we're back in type size hell
09:24:21Bagder_apparently jyp didn't hit many 16bit problems yet
09:24:25 Nick Bagder_ is now known as Bagder (
09:25:04ZagorBagder: but he/we will. so we need to think about how to handle them.
09:26:10BagderI think I vote for using long for 32 bits and int for 16/32
09:28:14*lImbus thinks a bignum-approach for > bit numbers drops out due to lack of cpu-power...
09:28:19Zagori'm not sure what I think yet. i'll need to meditate over this a while.
09:28:23lImbus> 16 bit that is
09:28:45ZagorlImbus: how do you mean?
09:29:20lImbuswell, bignums. used on "normal" architectures to handle numbers > 32 bit
09:29:35lImbusthere, where you don't have __int and so on
09:30:01lImbusAND where you need precision, there where float does not meet your needs as well
09:30:13lImbusbasically the values are stored in strings...
09:30:17Zagorwe have no choice but to use bignums on the 16-bit platforms. we cannot (will not) rewrite all code to use 16 bits only
09:30:28lImbusof course
09:30:50lImbusplus it takes lot of cpu-power...
09:30:52Bagderit has 32 bit longs
09:31:13ZagorBagder: ah, it does? that's good.
09:38:47LinusNusing long will solve the problem
09:42:51pike<- checking your IriverPort page. looking mighty fine. Will everything have to be 100% before it's usable ?
09:43:36lImbuswhat do you call useable
09:43:47lImbusfor music, nearly, but remote control-stuff
09:44:48LinusNthe first version will probably not support the remote lcd
09:45:06pikeplaying a mp3 and listen to it, is remote support a big thing (tm) to make ?
09:45:07LinusN"usable" to me means "mp3 playback works"
09:45:38lImbusto me too
09:46:03LinusNremote lcd is not technically difficult, but it requires quite a lot of changes in the rockbox architecture, since rockbox currently only supports one lcd
09:46:23 Quit TexJoachim ("Bye!")
09:47:11LinusNthe iriver remote is a completely new interface, which requires multiple instances of the dir browser, wps etc
09:47:58pikeokie. which mp3 decoder are you gonna use? Any clue if it is more or less efficient than the iriver fw one ?
09:48:29LinusNwe'll use libmad
09:48:59LinusNand i don't know if it is more efficient than the iriver one
09:49:05LinusNit probably isn't
09:49:08Zagorit will most likely be less efficient, at least to begin with
09:49:34pikebefore you optimize the hell out of it
09:49:58pikethe Foobar mpeg decoder is very much optimized btw
09:49:59Zagoralso it will most likely be better quality
09:50:21Zagorpike: isn't that floating point?
09:50:31pikeprobably, let me check
09:51:59pikeI own a 20gig iriver myself, and lately I've started thinking of upgrading the HD. I filled it long ago and realize I want more space
09:52:39Zagoryou need to wait for the new slim 40 gig drives. current 40 gigers are thicker and don't fit in the h120
09:53:01Zagor5mm vs 7mm iirc
09:53:27pikeyeah saw that. any idea what pricerange such a hdd would land in?
09:53:34Zagorno idea
09:53:47pikeI suspect very expensive unless you buy large quantities
09:54:15pikebut the 1900mAh replacement batteries seems interresting
09:54:53pikealthough 2100mAh seems even more interresting, but they're 1mm thicker
09:58:56pikefb2k's decoder is 32bit FP, figures :p
09:59:21Zagormost decoders use floating point, for obvious reasons
09:59:34pikeexcept portable players
10:00:07Zagorright. and they mostly use proprietary codecs optimised by the processor manufacturer
10:01:00pikeshame motorola didnt opensource theirs :)
10:02:06Zagortheir license for it is rather nice, actually. you get it for free and can distribute all you want. but you only get binaries, no source.
10:04:07lImbusso there is no way to include it in rockbox ?
10:04:22Zagorwe're not very fond of binary-only code
10:05:08lImbusno, I mean there is no way to lib/link it ?
10:05:18Zagordo you mean technically or legally?
10:05:25pikeboth I guess
10:05:33lImbusmhmm. both. I don't know a lot of GPL
10:06:04Bagderit would violate GPL, yes
10:06:30Zagortechnically I would assume it could be done. legally we can't link to it since it is not gpl compatible.
10:06:31pikedamn gpl.. (we got the same rules for XBMC)
10:06:36lImbusso with GPL you're not allowed to distribute even free closed source ?
10:06:59LinusNin this particular case, i don't care
10:07:11Zagordon't slam the gpl. it works like this for good reasons.
10:07:19LinusNi wouldn't want to use the motorola binary even if i could legally
10:07:25ZagorLinusN: i agree
10:08:11lImbusI wouldn't slam GPL as long as I'm not sure I understood it fully. It's just that I thought including free closed source would be ok.
10:08:18LinusNusing a binary black box for the mp3 codec is a no-no if it can be avoided
10:09:10lImbusplus I suppose we don't need the cpu-power saving. what would you want to do during music listeing ?
10:09:30Bagderyou could run the CPU slower
10:09:33LinusNwell, less cpu means longer playtime
10:09:34Bagderto save battery
10:09:35ZagorlImbus: the purpose of GPL is the promote Free Software, meaning software including source code. closed source, even freely distributable, is not Free Software.
10:09:48lImbusah, of course. variable cpu-speeds
10:10:26lImbusI suppose that inclusion would work with lGPL then ? I know the other way round is allowed.
10:11:12Bagderyes, with lgpl it is allowed
10:14:20 Join Lynx_ [0] (HydraIRC@
10:14:55 Join MooMaunder [0] (
10:15:32pikewhat cpu did the iriver have ?
10:15:53lImbusColdfire Motorola
10:15:54Zagormotorola coldfire
10:16:06pikehow many mHz ?
10:16:08 Join amiconn [0] (
10:16:11pikemips etc
10:16:31pikewonder if it's similar to what si used in ie their A925 phone
10:16:45pikesounds like it
10:16:52pikeSymbian UIQ compatible ?
10:16:58LinusNbut we will probably run it somewhere in the 70MHz range
10:17:18LinusNi have no idea if symbian runs on it
10:17:20Zagorpike: i don't even know what that means. but i would assume it's listed on motorolas pages about it.
10:18:34LinusNhi amiconn
10:20:15pikeär Coldfire ARM9 baserad ?
10:20:19pikeoops sorry
10:20:30pikeis Coldfire ARM9 based?
10:21:07Zagorno, coldfire is the core. you could call it a boosted 68k
10:21:36*Bagder spotted some swedish slipping through! ;-)
10:21:58pikeok, then they are not compatible, pardon my swedish, hehe
10:32:38***Saving seen data "./dancer.seen"
10:49:24 Join golot [0] (~chatzilla@
10:49:39golotHey all
10:49:42golotanyone here?
10:51:48golotoh :) whats up?
10:52:09Bagderwe rock our boxes!
10:52:51golotI wanna start in saying how grateful I am for your great effort in making a so-so product (archos rec with archos firmware) into the best HDbased player in the market (archos Rec with ROCKBOX) !
10:53:11Lynx_is there any docs accessible somewhere that describe how the MDB works?
10:53:58golotI wondered if the Archos is capable of charging 2600mah batteries.?
10:54:31golot(some charged only charge up to XXXXmah - is the archos box different?)
10:56:31ZagorLynx_: afaik only the micronas datasheet
10:57:02Zagorgolot: i doubt there is any problem charging 2600 batteries unless they are a different type
10:57:42golottype = like NiMH and so on?
10:57:56golotI mean NiMH 2600...
10:58:00golotok, sounds cool :) thx
10:58:03Lynx_Zagor: is that accessible? i just want to know the general principle out of interest...
10:58:47ZagorLynx_: i was wrong. linus has written a short description here:
10:59:38golotBjorn - do you have any spare Jukebox recorder Front buttons (my On-Off buttons are broken) ? do you know where I can start looking for this?
11:00:08Lynx_Zagor: thx, that's what I wanted to know
11:01:00Zagorgolot: sells spare parts. maybe they have buttons too.
11:02:46 Join [IDC]Dragon [0] (
11:03:10golotoh :( they dont have any spare buttons...
11:03:37Zagorgolot: try emailing them. they might know where to find any.
11:05:20[IDC]DragonI could solder a button off
11:05:48golotwhat do you mean?
11:05:57golotthe button is made of plastic
11:09:25golot[IDC]Dragon: ?
11:09:40lImbushehe, now tell me Dragon is not monitoring this channel via current.txt :-)
11:10:09[IDC]Dragongolot: I thought you bean the switch
11:10:21golotahh cool, thx...
11:10:34[IDC]Dragonthe plastic part I don't have
11:11:14golotanyone here replaced his jukebox light color? I saw someone asking what type to get but does anyone know how to do it? could be cool...
11:11:51[IDC]Dragonyou mean, the backlight?
11:11:58Zagori have white backlight
11:12:06[IDC]Dragonme too
11:12:10golotyeah backlight*
11:12:22golotis it complicated to a total noob like me?
11:12:29Zagoryes :)
11:12:33lImbusgotta go, bbl
11:12:34golotlol :)
11:13:14[IDC]Dragongolot: so what button part do you need?
11:13:15 Quit lImbus (" HydraIRC -> <- *I* use it, so it must be good!")
11:13:18golotwhite gotta look WOOOT!!!
11:13:20dwihnoI've soldered for the first time since my good old school days!
11:13:27dwihnoAlthough I forgot about "flussmedel" :)
11:13:40golotammm I need the On\Off button
11:13:50dwihno[IDC]Dragon: is that really needed?
11:13:57[IDC]Dragonthe switch or the knob?
11:14:03golotand I could use a Play button... but not "must"...
11:14:19golotthe plastic part, (what you actually press on)...
11:14:29[IDC]Dragonthen I can't help
11:14:53[IDC]Dragondwihno: you do, it's included in the solder
11:15:26golotdont you guys have like 2000000000000 jukeboxes lying around just like that? ;)
11:16:04dwihno[IDC]Dragon: it is? aah, nice :)
11:16:16dwihno[IDC]Dragon: In my old school days, we applied it by hand
11:16:42golotlol how old are you - 150 ? :)
11:18:09Lynx_wasn't there a webpage on the backlight mod? Can't seem to find it now.
11:18:24dwihnogolot: ayy! ye olde landlubber! ;)
11:19:05golothehehe :D
11:19:13dwihnosoldering is fun... although you should keep a stable blood sugar level or you will have a hard time soldering while shaking more than a vibrator on steroids
11:20:11dwihnoNow if I could get my hands on a 2.5mm stereo connector
11:20:15golotanyone can reccomend on really good NiMh batteries?
11:20:27dwihnothe GP batteries are good
11:20:32golotoh, I have many of those, all kind of connectors...
11:20:43golotyeah, I have the 2000 GP...
11:21:05golotyou dont know how cool I made my jukebox look
11:21:23dwihnoa portable christmas tree? :)=
11:21:40golotI scrobbed the metal part and its now shinning and looks amazing :)
11:21:41dwihnoZagor, Bagder, LinusN, do you think clas ohlson keep 2.5mm stereo connectors?
11:21:54dwihnogolot: pics! pics! :)
11:21:57golotnot gone that far :D making a Xmas tree
11:22:12Bagderdwihno: I'd say very likely
11:22:24golotMy jukebox is topless now but Ill take some pictures anyway...
11:23:08dwihnoBagder: ah, okay... I'll check after work
11:26:19golotPictures are coming!
11:30:10dwihnoBagder: seems like CO doesn't have them :(
11:34:50golot <−−- look !
11:35:33golotnot so good pictures, but you get the idea... :) its really shinny and looks cool, Im really proud of myself...
11:38:24golotyou should really see it, not in pictures... in real, its so cool... :D
11:38:25dwihnoshinny :D
11:38:54golot2 better pictures are comming in a minute
11:38:57[IDC]DragonI rarely look at the underside of my JB
11:39:37[IDC]Dragonthe Ipod has a shiny back, so you can see the fingerprints better ;-)
11:39:43golotyeah, lol, but its also the sides and the up and down sides (where you connect stuff to it)
11:40:00[IDC]Dragonapart from that, I see no benefit
11:40:14golotfun making it ? :)
11:40:24[IDC]Dragonok, sure
11:40:48golotI wish they would make the front panel and the battery covers out of metal too
11:40:53 Join quelsaruk [0] (
11:40:58golotthen I could have one shinning jukebox :D
11:41:05quelsarukhi, good morning to all
11:41:33golotoh no, the first one is bad
11:42:04golotsecond one is better
11:42:13golotlook at this:
11:46:15quelsarukamiconn, Bagder : Archos is finally going to replace my jukebox :)
11:48:47pikegolot get a better host..
11:49:06golotits down?
11:49:09golotoh :(
11:49:52golotIm uploading to a different account!
11:55:00quelsarukcu l8r!
11:55:24 Part quelsaruk
12:01:00 Join lImbus [0] (
12:01:01golotLook -
12:12:50 Join Kultivator [0] (
12:32:42***Saving seen data "./dancer.seen"
12:33:25Bagder"With the ARCHOS patented retro-recording system, it is possible to record the radio after the song has already started."
12:33:39Bagderpatented being the word
12:34:34HadakaI bet you guys have seen the archos linux based player stuff :)
12:34:53BagderI have
12:35:07dwihnothere is a linux based player now?
12:35:14BagderArchos' latest
12:35:23Hadakawell, to be released still, but yes
12:35:31Hadakaand they are giving out a "full SDK", whatever that means
12:35:53Bagderword is they use qtopia
12:36:07Hadakayeah, they even say it
12:36:15dwihnoI sense NDA
12:36:27Bagderdwihno: not likely since this is all based on linux
12:36:44dwihnoBagder: Was thinking about the SDK
12:36:45BagderI bet they ship a prebuilt gcc version for cross-compiles
12:38:29Bagderor possibly, they'll never deliver their promise :-)
12:38:39BagderI mean, they are Archos after all
12:39:50Hadakayeah, remains to be seen
13:19:28[IDC]DragonI'm getting mailing list postings now from instead of, is that the new way?
13:19:53[IDC]DragonIf so, I have to adjust my mail filter, no problem
13:20:34Zagorsounds like a bug to me
13:21:25BagderI get them from
13:21:35Zagoror, rather, this one person sent his mail to instead of and we didn't bounce it
13:21:52ZagorBagder: look at From:
13:22:38Bagderyou mean the To: line?
13:22:54Bagderthat's because those users mailed to that address
13:22:57Bagderand it works
13:23:21Bagderthe software doesn't care about the To: line
13:23:22Zagorright. but should it work? people are filtering on the To: address, so there is some value in keeping it constant
13:24:13Bagderits more work to prevent it
13:40:41golotanyone here has a GMINI ?
14:18:13 Quit z351 ("Leaving")
14:27:01 Join Nibbler [0] (
14:32:44***Saving seen data "./dancer.seen"
14:45:36 Join golot_ [0] (~chatzilla@
14:45:37 Quit golot (Read error: 54 (Connection reset by peer))
14:52:44amiconnBagder: Speaking about the rockbox mailing list - is there a reason why I get the reminder mails twice?
14:53:12Bagderno idea ;-)
14:53:13LinusNperhaps because you are subscribed to two lists?
14:53:23Bagderno, it should sent one for all lists
14:54:56lImbusamiconn: is it the same password for both mails ?
15:03:41[IDC]DragonI also get 2 each time
15:04:15 Join Strath [0] (
15:04:53Bagderhey Strath
15:05:11Strathhey Bagder
15:05:14lImbusuh, I'm gettin 2 as well. same password...
15:06:28BagderStrath: noticed jyp's recent Gmini work on Rockbox?
15:07:50Strathya, he sure is fast
15:08:08 Join midk [0] (
15:09:30Strathwhen he started coding for the gmini emulator, he had IRQ, DMA, IDE, and UART all working within a couple weeks
15:12:09[IDC]Dragonhow is IDE emulated?
15:12:58[IDC]Dragon(and the other external interfaces)
15:21:38 Quit golot_ (Read error: 104 (Connection reset by peer))
15:25:26 Join R3nTiL [0] (~zorroz@
15:28:29 Join TexJoachim [0] (
15:30:37 Quit midk (Read error: 104 (Connection reset by peer))
15:45:56 Join bogeyman [0] (
15:59:03 Nick mbr_ is now known as mbr (
16:02:58StrathIDE accesses a disk image file, UART uses a named pipe
16:03:48Strathnow there is a gui too, so buttons are well, buttons
16:06:42 Part LinusN
16:11:20[IDC]DragonStrath: very nice
16:11:55[IDC]Dragonso you bring the low level IDE commands back up into an image access
16:17:04amiconnBagder: I'm only subscribed to one list, and the mails look identical..
16:19:39 Nick Strath is now known as StrathAFK (
16:23:42 Quit R3nTiL ()
16:24:03 Join jyp [0] (
16:25:01 Join methangas [0] (
16:32:47***Saving seen data "./dancer.seen"
16:33:25Lynx_I get two mails, too, that started last december, before that i got only one
16:33:41 Quit jyp (
16:34:32NJoinjyp [0] (
16:40:14[IDC]Dragonanybody with an idea if Fujitsu or Toshiba 100GB is better?
16:41:07[IDC]Dragonit used to be Fujitsu, but now the Toshiba specs are a bit more quiet and low power
16:41:51[IDC]Dragondid Toshiba improve something about it?
16:42:57Zagorno idea
16:44:07 Join teuvvo [0] (
16:44:56teuvvoHello. Any news -good/bad - about the Rockbox iRiver?
16:48:56 Nick StrathAFK is now known as Strath (
16:49:46 Join mecraw_ [0] (~lmarlow@
16:57:17 Part Zagor
16:59:23 Join einhirn [0] (
17:08:10 Join mecraw__ [0] (~lmarlow@
17:25:15 Quit mecraw_ (Read error: 110 (Connection timed out))
17:39:45 Join Spida_ [0] (
17:50:46 Quit Spida (Read error: 110 (Connection timed out))
18:04:30 Quit amiconn ("CGI:IRC 0.5.4 (2004/01/29)")
18:12:27 Quit elinenbe (" HydraIRC -> <- Nine out of ten l33t h4x0rz prefer it")
18:32:48***Saving seen data "./dancer.seen"
19:01:12 Join amiconn [0] (
19:02:25 Join webguest19 [0] (
19:02:47lImbus[IDC]Dragon: I tested my broken OndioFM with an external MMC yesterday
19:02:54lImbusit does not work either
19:03:16 Quit webguest19 (Client Quit)
19:15:22 Quit teuvvo ("CGI:IRC (EOF)")
19:15:52 Join TuDo [0] (
19:18:19 Quit MooMaunder (Read error: 110 (Connection timed out))
19:21:25 Part TuDo
19:25:38 Quit [IDC]Dragon ("CGI:IRC")
19:30:28 Join _aLF [0] (
19:36:05crash_ah jyp :) i talked so strath
19:36:41 Quit einhirn (Read error: 104 (Connection reset by peer))
19:37:16crash_i'll start to copy most of the infos from donat to rockbox
19:37:30crash_but only the not too technical stuff
19:37:41crash_at the moment he wants to keep his wiki
19:37:52crash_but he has no problem with puttin it on rockbox
19:37:59jyptoo bad I'm still unable to access it :/
19:38:07crash_it should
19:38:17crash_he said it had been fixed yesterday
19:38:38jypRight, but is broke again today
19:38:47jypup, down, ...
19:41:02jypup, down, ...
19:42:26Strathya.. don't know why
19:42:49crash_i can reach it tough
19:43:18Strathwhere are you crash?
19:43:31crash_at the moment at schlund+partner backbone
19:43:44crash_aka 1&1
19:45:35Strathcould set up an http proxy for jyp :) (assuming he can reach that)
19:46:03crash_just checking if i can rech it via t-online backbone
19:46:06crash_but i dont reach home
19:47:28 Join einhirn [0] (
19:47:43Strathcould it be blocking?
19:48:13crash_from home i dont reach it
19:48:31crash_so the router from here at work seems to be antoher one than at home
19:54:05crash_quite strange yeah
19:54:20crash_unfortunatly i wont reach from home
19:54:23 Nick gromit` is now known as gromit`rip (~gromit`
19:54:29crash_i think i'll have to copy it via my server ;)
19:54:36crash_gotta go
19:54:39crash_see you later
19:54:48jypgood by
20:18:46 Nick Spida_ is now known as Spida (
20:23:18 Join Stryke` [0] (
20:32:51***Saving seen data "./dancer.seen"
20:41:40 Join Tang [0] (
20:45:28TangNothing news with the iRiverport theses lats days?
20:46:46jypI can only give news about the Gmini port :P
20:49:45 Join dcran [0] (
20:50:33dcranOk quick question, I see in todays log that Linus said you were going to use the libmad decoder....any reason not to use the Helix decoder as I imagine that would be better???
20:51:47 Quit Tang ("Chatzilla 0.9.66 [Mozilla rv:1.7.5/20041108]")
20:52:37Stryke`From what I can tell from the website, the Helix decoder is MP3 only (i.e. no layer-I or layer-II support), is written in C++, and is licensed under the Real Networks Public Source License (RPSL). For those reasons, and the fact that MAD is tried and tested, I think we should stick with MAD −− DaveChapman
20:52:41Stryke`from the wiki
20:53:42jypIn what way Helix would be better ?
20:53:54dcraninteresting.......I mean thats fine I would think though that Helix (Open REaL) would be more tested than libmad
20:54:32dcranlibmad may be more tested in rockbox, but I would say overall Helix has been tested on many more platforms
20:55:51*dcran notes that he is not a firmware dev or a audio dev and is just a stray dog with a comment
21:04:34 Join lImbus_ [0] (
21:04:47 Nick lImbus_ is now known as lImbusHome (
21:12:01Spidahi lImbusHome
21:12:29lImbusHometach Spida
21:14:33Spidabist du morgen mit $hardware im hermann?
21:15:06lImbusHomelet's stay in english
21:15:14lImbusHomeyes, I will be there, with my devices
21:15:36lImbusHome(no need anymore to bring with you the mcc-card, I was able to do another test yesterday)
21:43:08jypAny ATA guru online ?
21:44:43 Quit dcran ("dcran has no reason")
22:00:41 Quit Stryke` (Read error: 110 (Connection timed out))
22:03:57 Join teuvvo [0] (
22:13:09teuvvoAny news on the Rockbox iRiver?
22:15:45 Join LinusN [0] (
22:16:04LinusNteuvvo: nope, no news. been busy.
22:17:11teuvvoOkay. Just want You to know how eagerly people are waiting for it, works so fine with the Archos!
22:17:19LinusNjyp: i have some questions about your context switch code
22:17:26teuvvoYou guys are doing a tremendous job here!
22:17:35LinusNteuvvo: thx
22:17:51jypyes ?
22:18:00teuvvoI hope I could do something, but haven't yet found the programmer in me... ;)
22:18:53LinusNjyp: i wonder why you did the initial context differently than the other platforms
22:19:26BagderI asked that too ;-)
22:19:27LinusNchecking a flag in load_context instead of saving a valid context in create_thread
22:20:04*LinusN knows nothing about the cpu in question
22:20:09jypI didn't know where to put the start address in the context
22:20:26jypso I changed it to make the jump explicitly
22:20:44LinusNhow does a function call return in your architecture
22:21:27jypjumps to a register address
22:21:37LinusNand if the functiuons nest?
22:21:56jypthen push it on the stack
22:22:08LinusNjust like the sh7034 then
22:23:13jypThe truth is I didn't realize the code relied on optimization till late
22:23:28jypso I made up my own method...
22:23:39LinusNrelied on optimization?
22:23:55jypIt fails if not compiling with -O1
22:24:16LinusNyou mean the frame pointer removal?
22:24:29Bagderits the inline stuff, isn't it?
22:24:41jypI mean spilling registers on the stack
22:25:32LinusNjyp: it doesn't rely on any other optimization than the frame pointer exclusion
22:26:18LinusN(and the inline of course)
22:26:47jypYou're right...
22:26:58jypMy method does :@
22:27:18*Bagder giggles
22:27:36 Quit lImbusHome (" HydraIRC -> <- :P")
22:27:52 Join lImbus_ [0] (
22:27:55LinusNjyp: you should probably look at the 7034 code
22:28:30LinusNbtw, how does the cpu handle 32-bit operations? does it combine two registers?
22:29:36Strathcombine two registers: exactly
22:29:43LinusNkind of guessed that, since the push operation pushed two registers
22:30:43jypI can revert to saving the context in a memory location, but the cpu like to store on the stack better I think
22:31:05jypAlso, I didn't quite understand what's "volatile context"
22:32:19LinusNthe volatile context is the scratch registers
22:32:46jypcallee saved
22:32:52***Saving seen data "./dancer.seen"
22:33:04LinusNthe non-volatile context are callee saved
22:33:15LinusNthe volatile context is not saved at all
22:33:39LinusNthe context switch saves the non-volatile context
22:33:51jypI meant non-volatile then
22:33:55LinusNi see
22:34:27jyphowever all registers are callee saved on the calmrisc
22:34:28 Quit teuvvo ("CGI:IRC (EOF)")
22:35:15jypAt least all address registers
22:35:26jypmaybe r0-r1 are not
22:36:06Strathmost of the time they are
22:36:42jypStrath: gcc & calmshine don't use exactly the same calling convention...
22:36:56Strathi didn't think they would
22:37:25Strathjyp, does your Makefiles depend on absolute paths?
22:37:50jypSome times yes :)
22:38:13jypproblem is I gave you a snapshot of my working directory...
22:38:25Bagderthe makefiles are generated to use abs paths
22:38:34Bagderat least in the root one
22:39:16jypand unfortunately I changed the generated makefile by hand
22:39:47Strathanyways... i've got a rockbox.o now
22:40:10Bagderjyp: why do you want that?
22:40:20Bagderwhat's wrong with the .a?
22:40:39jypI'm just walking in the dark
22:40:55Bagderthen don't sent it in revised patches to me! ;-)
22:40:57Strathand he's leading me through this first time
22:41:35jypJust ignore everything related to he build process ... sorry ...
22:41:44Bagderok, will do
22:42:19BagderI have a few more questions
22:42:53BagderI didn't commit the export/types.h
22:43:05jypIt shouldn't have been in.
22:43:15jypI thought I removed it
22:43:18Bagderright, but the code in ata.c uses types from there
22:43:32jypIt shouldn't have been in either.
22:43:46*jyp boinks his head
22:47:01LinusNjyp: wanna know what set_irq_level() does?
22:47:25jypI guess I know
22:47:30jypbut in precision
22:47:32LinusNit sets the irq level, and returns the old one
22:48:18jypproblem is there's no such thing as an irq level in my chip
22:48:34LinusNa single irq?
22:49:04jypwe can individually mask the irqs
22:49:19jypbut up till now there's only timer implemented
22:50:26LinusNdo you have a general irq mask bit, that disables all irq:s?
22:51:42LinusNthen that is your irq level
22:52:20LinusNset HIGHEST_IRQ_LEVEL to the mask value of the status register where the bit is
22:52:59LinusNi.e #define HIGHEST_IRQ_LEVEL (1<<bitnumber)
22:53:20LinusNthen have it return the old value of the bit
22:54:42jypthanks for the info
22:55:37LinusNi'm impressed by your work
22:55:49jypthanks ;)
22:56:32jypBut I owe everything to my guide and master, Strath ;)
22:56:52Strathtoo late now suck up :P
23:03:25Bagderjyp: you need to stop using ifdef platform when you want ifdef feature
23:03:35 Quit methangas (" HydraIRC rocks! -> <-")
23:03:51jypI tried to correct it...
23:04:00jypwhere did I sinned ?
23:04:13BagderI've just added a good define to use
23:04:58Bagderand btw, shouldn't just be #ifdef'ed within is the link file, just preprocessed before it is used
23:05:29Bagderto allow #ifdefs etc
23:05:36jypI didn't know the existence of the file till now
23:05:41Strathya, you slice that lazy jyp a new one Bagder :D
23:06:11*Strath cracks a whip
23:06:11jypscrew you guys, i'm going home
23:06:33*jyp leaves
23:08:13Strathgnite lImbus_
23:08:49lImbus_euh, I'm not yet goin.
23:08:55lImbus_it was to jyp
23:09:36lImbus_aah, sorry, it's quite late here around
23:09:54Strathi know... :)
23:14:51 Join Zagor [242] (
23:14:53 Quit courtc (Read error: 104 (Connection reset by peer))
23:15:23Bagderhey dr Z
23:15:32Zagorgood evening
23:16:23LinusNmr Z
23:16:42lImbus_zzzzZZZzZZzzZ :D
23:18:28Bagderjyp: shouldn't the crt1.c contents be in system.c instead?
23:18:43 Join courtc [0] (
23:21:29Bagderand... I don't think __attribute__ is good undontionally in sprintf.h
23:21:38BagderI believe it gets included by the simulator
23:21:48Bagderwhich might be compiled with msvc at times
23:22:08Zagorwhich attribute? icode?
23:22:31Bagder __attribute__ ( ( format( printf, 3, 4 ) ) );
23:22:36jypWell, that's a cool feature to have...
23:22:49Bagderthen ifdef it on some gcc symbol
23:23:04jypThere can be #define __attribute__ () when not gcc
23:23:13Strathwhere is at it called?
23:23:25Strathwrng window
23:23:47Bagderjyp: yes, but it seems a bit overkill
23:23:52Zagordoubleplusungood __attribute__ is a compiler-internal define. messing with those are bad karma.
23:23:52Bagderto me
23:24:10jypNah ;)
23:24:35jypStill __attribute__ is used at a lot of places in your code already
23:24:42jypfor section
23:24:48Bagderyes, in code that isn't used by the simulator
23:24:54Zagorok, i'll rephrase: messing with those will get you spanked by the Code Police :)
23:25:02Zagorsure, but we don't redefine it...
23:25:06Bagdervery true
23:25:44Bagderjyp: and I just can't apply that firmware/SOURCES diff
23:25:51jypNote that __attribute__ is especially designed to be redefined
23:25:56jypwhen not supported
23:26:04jypof course
23:26:36jypI will clean it up as soo as the rest compiles
23:26:41 Join Stryke` [0] (
23:27:28BagderI think you can start posting your updated patches to the mailing list
23:29:02Bagdertry to exclude stuff you know we can't commit
23:39:21Zagorhmm, i've changed my mind. jyp is right. I think we should be using attribute format for printf-style functions.
23:40:19Zagorsplash() is a candidate, for example.
23:40:56Bagdersure we should, we should just not make it unconditional in files used by the sim
23:43:03Zagorthe question is: is it better to ifdef all uses or to ifdef __GNUC__ the definition
23:43:37jypThinking twice ifdef __GNUC__ the definition
23:44:24Bagderyeah, thinking about it more, I lean towards that too
23:51:02 Join webguest55 [0] (
23:52:18 Join mrfry3 [0] (
23:52:18 Quit webguest55 (Client Quit)
23:52:40 Quit mrfry3 (Client Quit)
23:52:47 Join mrfry3 [0] (
23:53:28 Quit mrfry3 (Client Quit)
23:53:35 Join mrfry3 [0] (
23:54:22 Quit TexJoachim ("Bye!")
23:57:38LinusNmrfry3: ?
23:57:52mrfry3can rockbox be used with gmini120
23:58:03jypsomewhat ;)
23:58:05LinusNnot yet

Previous day | Next day