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 2005-10-31

00:02:20 Quit DangerousDan ("Miranda IM! Smaller, Faster, Easier.")
00:04:50 Join muesli- [0] (
00:08:47 Join dpassen1 [0] (n=dpassen1@resnet-233-61.resnet.UMBC.EDU)
00:12:33 Nick Kohlriba is now known as Kohlrabi (
00:24:51 Quit muesli_- (Read error: 110 (Connection timed out))
00:27:25 Quit dpassen1 ()
00:34:38 Join LinusN [0] (
00:36:23LinusNRotAtoR: so, i guess it was the playboard clearing that was the problem?
00:39:35 Join arkascha [0] (
00:52:24*preglow strokes his IRAM
00:52:43 Quit Quel|away ("KVIrc 'Marmalade'")
00:53:20MoosHello Linus, preglow :)
00:53:47RotAtoRLinusN: Yes it was, everything works great now!
00:54:30RotAtoRa simple mistake with non-obvious results
00:56:13linuxstbpreglow: What are you doing with your IRAM? :)
00:58:06 Quit muesli- (Read error: 110 (Connection timed out))
00:58:24 Quit ghode|afk (Read error: 110 (Connection timed out))
00:58:39preglowlinuxstb: just looking around in the codecs to see if some of them needs some
00:58:43preglowi'd like vorbis to be faster
00:59:49linuxstbgrep shows that there are still lots of mallocs in Tremor.
01:01:42preglowyes, unfortunately
01:02:38linuxstbMaybe it would be possible to implement ogg_malloc using IRAM.
01:03:01preglowmno, i'd rather just be rid of them completely
01:03:06preglowi'm looking to have the main buffers put in iram
01:03:17preglowgod, i wish monty used more whitespace in his code
01:04:37linuxstbYep - he obviously didn't read the coding guidelines before contributing to Rockbox.
01:05:21preglowthey don't state anything about whitespace anyway
01:05:24preglowi just like whitespace
01:06:48linuxstbHave you had chance to look at the AAC decoder. I've traced it a little on the target, and it seems to be crashing deep inside the decoder when decoding the first frame.
01:06:54 Quit matsl (Remote closed the connection)
01:07:29preglowcoupled with _ONE_ space indent
01:07:37preglowlinuxstb: nah, i just got home again
01:07:43preglowlinuxstb: might have a look in a short while
01:08:45linuxstbThe good news is that it seems to be quite stable on the sim. It repeatedly played my three test files for about an hour before I got bored of them.
01:09:13preglowi'm really curious as to the cpu usage on target
01:09:20linuxstbSo am I.
01:09:56linuxstbLike everything, we will need to get it using IRAM appropriately. This could be hard with all the mallocs going on.
01:10:27linuxstbPlus there is lots of conditionally compiled code - so it's hard to see what's going on at the moment.
01:11:55 Join webguest06 [0] (
01:12:26 Quit webguest06 (Client Quit)
01:15:36linuxstbDoes anyone know the major differences between the sim and the targets? e.g. do the sims have stacks limited to the same size as the targets?
01:15:54 Quit Kohlrabi ("Leaving")
01:15:59linuxstbI'm guessing a stack overflow is one a likely cause for the AAC codec crashing on the target.
01:16:47preglowi think someone's already said that
01:16:55 Nick paugh is now known as AliasCoffee (n=kickback@2001:5c0:8fff:ffff:8000:0:3e03:6822)
01:17:00preglowprobably fuzzie
01:17:24preglowtry placing the codec stack in ordinary ram, and incrasing its size
01:17:31 Quit cYmen ("zZz")
01:17:49preglowplayback.c, line 115
01:18:16preglowthe main playback buffers for vorbis will be 16k of iram
01:18:24preglowthink i'll just try and see if it helps any
01:18:56 Quit ender` (Read error: 104 (Connection reset by peer))
01:19:01linuxstbYes, I think 8KB is a little small for the AAC stack. I'll add a zero.
01:20:36***Saving seen data "./dancer.seen"
01:21:06linuxstbWell it's working.
01:21:18linuxstbOne spurt of sound every 5 seconds... :(
01:21:54 Join XavierGr [0] (
01:22:08XavierGrHi all!
01:22:10preglowi wonder if we can cut out the offending function
01:22:15amiconnWhat codec stack usage?
01:22:17linuxstbI'm now 8 seconds into the first track.
01:22:25amiconnBtw, the codec stack is 9KB by default
01:22:49linuxstbI'm sure I changed it from 0x2000
01:23:00preglow+ default stack size
01:23:03preglowwhich is 1kb, afaik
01:23:09amiconnyup, and default is 0x400
01:23:47linuxstb35% usage - I set the size to 0x20000
01:23:58 Quit arkascha ("Konversation terminated!")
01:24:23linuxstbBut I did see lots of local (non-static) arrays. So we should be able to reduce it a lot.
01:24:35amiconn~45 KB ...
01:24:51linuxstbNow 21 seconds into the track...
01:25:54preglowlinuxstb: any more info about the licensing conondrum, btw?
01:26:19linuxstbNo, no-one seems to have said anything for a couple of months. No change to anything in CVS.
01:26:20preglowshall we even try to support vorbis streams with more than 2 channels?
01:26:43linuxstbNo. I made a decision to limit all the codecs I've implemented to 2 channels.
01:26:55linuxstbApart from AC3 of course.
01:28:51preglowcodecs which include internal handling can keep it
01:28:59preglowand it's a pretty special codec in that regard anyway
01:29:13preglowlinuxstb: but yeah, i can cut a lot of mallocs away if i set max channels to 2, it seems
01:29:33linuxstbJust make it a define - MAX_CHANNELS or something.
01:30:55linuxstb10 minutes later, it's almost played 1 minute of the file. So my estimate is 10% realtime at the moment.
01:31:20linuxstbThis is a standard 128kbps iTunes encoded file.
01:32:27preglowit also seems that we can cut the precision in the window coefs to sixteen bits and still get a more than adequate result
01:32:37preglowputting the pcm buffers of tremor in iram did very little
01:33:34preglowlinuxstb: before we even start optimising it, i think the stack usage should be tamed
01:34:28linuxstbThe codec itself is 373344 bytes....
01:34:45linuxstbIt needs work everywhere.
01:35:03preglowholy shit
01:36:18*amiconn wonders how it does work, then
01:36:32amiconncodec ram is 256KB ...
01:38:19linuxstbI doubled it.
01:38:49preglowhow the flaming hell is it possible
01:39:03preglowi don't even think your build has all the bells and whistles enabled
01:39:05preglowlike sbr
01:39:23linuxstbIt doesn't - only parts of the library are implement in fixed point.
01:40:47preglowdamned be real and their idiot licenses
01:40:56preglowthe helix decoder is three tons better, i believe
02:03:02 Quit Moos ("Glory to Rockbox")
02:05:12preglowoh well, bed time anyway, night
02:13:10 Join bagawk [0] (n=lee@unaffiliated/bagawk)
02:19:21 Quit lostlogic (Read error: 110 (Connection timed out))
02:26:39 Part LinusN
02:29:35XavierGrWhich program should I use for best encoding results in mp3? mp3 Alt-preset standart
02:30:08XavierGrI currently use dbpowerAMP but I don't trust it anymore...
02:33:17linuxstbEAC is the best CD ripper, and that can use lame to encode MP3s.
02:33:40bagawkI recommend EAC+LAME also
02:34:23XavierGrWhich version should I use? I have 0.9 beta 4 installed but I see that there is a new one out.
02:34:51XavierGrI ask this because says the 0.9 version/
02:35:04XavierGrthough it shold be an outdated guide.
02:35:11linuxstbI don't know - but I'm sure the forums at will tell you.
02:35:17bagawkXavierGr, EAC does not do the actual compression, LAME does that
02:35:38XavierGrso it should be safe even with the old version, I guess.
02:35:49bagawkXavierGr, yes
02:36:36XavierGrIt is just that EAC has some neat options, (which I don't know them) and this guid I read has some default configurtion files for best results
02:36:56 Quit Sucka ("a bird in the bush is worth two in your house")
02:37:12XavierGrAnd these configuration files are cd-rom dependant (for best results)
02:39:38XavierGralso does anyone know how to interact with cue files that come with a big flac (or ape) image of the CD?
02:39:58XavierGrCurrently I just encode the flac to wav and then load it as an image.
02:40:12 Quit tvelocity ("Leaving")
02:42:51 Join lostlogic [0] (
02:42:59 Quit lostlogic (Read error: 104 (Connection reset by peer))
02:43:43AliasCoffeeXavierGr: you might want to split the wav with something like cuesplit then recompress in your format of choice.
02:45:18XavierGrwell with the cue file, if I encode the audio file to wav (from default flac) then I can load it as a cd image (audio) and then I have all the tracks saperated.
02:46:09AliasCoffeethere isn't much on linux that handles monolithic rips in any meaningful way. i've not used windows for years so maybe you could find a plugin that can load a monolithic file and it's cue and present the individual tracks. i'd still rather split them, personally.
02:46:17XavierGrI just want to bypass the step of encoding to wav and load it as an image even if the audio file is wav.
02:47:04XavierGrbut splitting will occur one way or the other.
02:47:25AliasCoffeeif you split them then you don't need to do so again. it's a one time operation.
02:47:39 Quit bagawk ("Leaving")
02:47:43XavierGrThe point is that I frequent many files districuted this way. (Cue+flac or ape)
02:48:07XavierGrAliasCoffe: Of course it is one time operation.
02:48:39XavierGrI think you misunderstood me.
02:48:55XavierGrI have a rar file with 2 files in it.
02:49:01XavierGra cue and a flac
02:49:07XavierGrthe flac is the whole cd.
02:49:21XavierGrcue files can be loaded as an emulated drive.
02:50:05XavierGrso if I try to emulate the audio cd (when the audio file is in flac format) a error pops up.
02:50:22XavierGrIf I convert the flac to wav then the image loads normally.
02:51:09XavierGrBut know that I think it again maybe it is intensionally the audio file is converted to flac to save bandwidth in distribution.
02:51:35*AliasCoffee grins
02:51:53XavierGr(crap my later sentence is unreadable)
02:52:33XavierGrs/late/last :(
02:52:54AliasCoffeeyeah i'm familiar with daemon-tools and the like. what i said was that you either need split up the files or find a plugin that can parse the cue and monolithic rip (flac,ape) and present the track list in a useful way
02:54:19AliasCoffeei would split them because it's a one time operation that makes the rip compatible with pretty much everything.
02:54:39AliasCoffeeie: turn a monolithic rip into a normal one with 1 file per track
02:55:46XavierGrok thanks, I will look more to it.
02:55:59AliasCoffeeit's more flexible too. if you want to copy 1 track from the album on to your media player then you can't with a monolithic rip.
02:57:48XavierGryes that is true, though all I do is make it an image. Rip it, and then play the music from the ripped files. i just store the original file for sharing reasons and as a backup. (though the backup is poor because if I loose the mp3s I will have to rerip and retag.)
02:58:43linuxstbfoobar has excellent support for cue files. Just add the cue file to the playlist, and it will present the tracks. I'm guessing it supports using cue files with any audio format.
02:59:14AliasCoffeeif your source is lossless then you can delete the original unless your have weird sharing obligations. if your source is mp3 then.. hmm delete it! :P
02:59:32AliasCoffeety linuxstb
02:59:49AliasCoffeeiirc foobar now uses eac componants to rip too no?
03:00:04XavierGrsource is lossless keep it for sharing reasons and safety ones too.
03:00:04linuxstbI don't use it, but I think so.
03:00:14XavierGrfoobar seems a great player.
03:00:32XavierGrAnd I keep avoiding its usage without any reason...
03:00:35linuxstbIt also works quite well under Linux (using Wine). That's the only place I've tried it.
03:01:42XavierGrdownload the latest lame beta version or stick with a standard version?
03:01:44AliasCoffeei use EAC under wine. i wouldn't like to use it (foobar) as a player under wine for performance reasons though.
03:01:56AliasCoffeetbh, i still make do with xmms ;)
03:02:13linuxstbIs EAC significantly better than cdparanoia then?
03:02:33AliasCoffeesadly yeah
03:03:00linuxstbI've heard rumours that Xiph are going to update cdparanoia. It's been a long time.
03:03:48linuxstb - last updated 27 March 2001.
03:03:49AliasCoffeeyeah. vorbis took priority now it's theora.
03:04:05AliasCoffeethere are some patches floating around that help with various things
03:04:46AliasCoffeei don't think it handles the cache problem tho so on drives with a cache, the error correction isn't so useful.
03:04:57linuxstbYes, I've heard that.
03:05:21AliasCoffeei spent a while trying to improve it but was beaten back by my own stupidity =]
03:05:27AliasCoffeeSCSI :/
03:05:39XavierGrehm sorry to be pushy but which lame version do you recommend?
03:05:40linuxstbSeems I should try and get EAC set up.
03:06:01AliasCoffeeXavierGr, ogg vorbis no good to ya?
03:06:36XavierGrwell unfortunately I chose mp3 back then... so I stick with that.
03:06:59linuxstbXavierGr: No idea about Lame versions, sorry.
03:07:00*ze has mp3, vorbis, and flac on his player
03:07:11XavierGrThough battery reasons and "globalisation" of mp3 are some reasons too.
03:12:05XavierGrAm I the only one here that uses mp3?
03:13:18XavierGroh and one last (irrelevant) question.
03:14:17XavierGrToday I am going to get a new Case for my PC. Pentium 4 3GHz, Any advice on what to look for or be carefull?
03:15:03XavierGrI have seen some case on the net but it all seems to me the same, (if you exclude some idiotic led fans e.t.c)
03:15:40 Join lostlogic [0] (
03:20:37***Saving seen data "./dancer.seen"
03:37:59 Quit AliasCoffee ("Leaving")
03:50:28 Join acidblue2 [0] (
04:04:30 Join JAJDude [0] (
04:06:16 Quit JAJDude (Client Quit)
04:35:58 Quit guillauh (Connection timed out)
04:48:23 Quit acidblue2 ()
04:53:00 Part yosemite
05:20:40***Saving seen data "./dancer.seen"
05:56:02 Quit RotAtoR (Read error: 110 (Connection timed out))
06:34:49 Join webguest74 [0] (
06:36:12 Quit webguest74 (Client Quit)
06:36:29 Join leftright [0] (
06:38:58leftrightjust a quick hello, i'm on holiday in mauritius, enjoying the diving, sea, sand, beach and bitch
06:40:26leftrightXavierGr; the recommended version of LAME is still 3.90.3, Hydrogen Audio will have tons of info about it
06:42:26leftrightor for a definitive mp3 ripping guide go to www.
06:56:00 Part leftright
07:09:51 Join _FireFly_ [0] (
07:20:41***Saving seen data "./dancer.seen"
07:37:38XavierGrbummer and I just ripped 6 CDs using 3.96......!
07:37:47XavierGroh good morning!
07:49:24 Join B4gder [0] (
07:53:44amiconnB4gder: I found that has a bug, probably since the viewer.config changes.
07:54:08amiconnIt puts the rockboy.ovl into /.rockbox/rocks instead of /.rockbox/viewers
07:54:24amiconn(of course archos recorders only problem)
07:55:26amiconnThis way rockboy won't work from a fresh install. An overwritten installation will still work as long as the plugin api isn't changed
08:00:28 Quit _FireFly_ ("Leaving")
08:02:47phaedrus961amiconn or B4gder, can one of you take a look at my convbdf patch? It's only a few lines.
08:03:59phaedrus961it allows for fonts wider than 16 pixels and fixes a bug with the offset table
08:10:35 Quit solexx_ (Read error: 104 (Connection reset by peer))
08:15:37 Join solexx [0] (
08:29:07 Join Kohlrabi [0] (
08:30:33 Quit Kohlrabi (Read error: 104 (Connection reset by peer))
08:38:55 Join _FireFly_ [0] (
08:41:00 Quit _FireFly_ (Client Quit)
08:41:46 Join _FireFly_ [0] (
08:45:44 Join einhirn [0] (
08:46:23_FireFly_TiMiD are you here ??
08:47:35 Join ender` [0] (i=ychat@
08:48:44XavierGrWe are definetely gonna need some sort of screen freeze for the remote!
08:52:06_FireFly_why ??
08:53:07B4gderwhy tell everthing when you can leave out the
08:53:15XavierGrmany users, including me, have faulty remotes (or main units) which can create noise when the screen is rendered. (only on changes)
08:54:14XavierGrB4gder: ?
08:54:29B4gderI paraphrased your comment
08:54:51B4gderyou just said a thing that made no sense without an actual explanation
08:55:24XavierGr_FireFly_: I am suggesting to make the lcd (as an option) render the screen as long as the remote backlight is on, then stop
08:56:03B4gderis that noise audiable always?
08:56:30B4gderI mean music player or radio or whatever
08:56:56XavierGrthat way a user will be able to browse see what's on the playlist e.t.c and when the remote is in rest no ticking.
08:57:30XavierGrWell with tests I've done with iriver firmware (and partially) rockbox you can clearly hear a low volume click every time the screen content changes
08:57:36XavierGrI am not sure about radio.
08:57:40amiconnThat wouldn't make sense. People might want to disable backlight, and still expect the displayed information to be up to date
08:58:06XavierGramiconn: what you suggest then?
08:59:16amiconnWell, if this is really a problem with some remotes, we could have an option that disables the remote lcd when the remote lock switch is on
08:59:48amiconnI mean, blank the display completely, then put it to standby. Even saves some battery
09:00:20amiconnOther than that, I might try to code a "loew-emi" remote lcd update
09:00:33XavierGrthat is a good suggestion too.
09:01:03XavierGryou could test to see if your unit has this or not.
09:01:12amiconnSomething like the "spread spectrum modulated" bios option of some pc motherboards does
09:02:41XavierGrthough I kinda like the backlight usage too, I mean come on who reads the display without backlight, it is too obscure to read without it, at least for me....
09:02:54SlasherHmm, i would like an option that backlight would always come up on button presses even if hold is on
09:07:35 Nick Lynx_awy is now known as Lynx_ (
09:19:44 Join cYmen [0] (
09:20:43***Saving seen data "./dancer.seen"
09:44:42 Join guillauh [0] (
09:49:53 Join DangerousDan [0] (
09:52:31 Join arkascha [0] (
09:52:58 Quit arkascha (Remote closed the connection)
09:54:02XavierGrMy god! These clicks are quite loud with the radio on, on the remote.
09:54:03 Join arkascha [0] (
10:23:56 Quit lostlogic (Read error: 110 (Connection timed out))
10:57:28 Quit _FireFly_ (Read error: 110 (Connection timed out))
11:20:44***Saving seen data "./dancer.seen"
11:20:49 Join lostlogic [0] (
11:37:44 Join ashridah [0] (
11:46:19 Join tvelocity [0] (
11:50:14 Join _FireFly_ [0] (
11:51:29_FireFly_TiMiD ??
11:52:35 Join Moos [0] (
12:38:33 Quit thegeek_ (Read error: 104 (Connection reset by peer))
12:39:22 Join thegeek [0] (
12:39:22 Quit linuxstb (Read error: 110 (Connection timed out))
12:40:32 Join linuxstb_ [0] (
12:40:43 Nick linuxstb_ is now known as linuxstb (
12:42:08linuxstbSome good news and bad news about libfaad. The stack consumption is now normal (17% of the standard codec stack size), but this has increased the size of aac.codec to 467820 bytes.
12:42:38linuxstbPutting the codec stack back into IRAM also means the codec is now running a lot faster - probably around 40%-50% of realtime.
12:43:22preglowi'm wondering if you perhaps should commit it
12:43:43preglowbut MAN, that is a large codec
12:43:45linuxstbYes, I'm planning to tidy it up and commit it this evening. Unless anyone objects to having to increase the codec size.
12:44:09preglowhow much of it is bss and how much is code?
12:44:20linuxstbAt the moment I have associated the ".mp4" extension with AAC. I believe this is one of the standard extensions, but we obviously also need to support .m4a
12:44:38preglowshouldn't be too hard
12:47:08linuxstb206904 bytes are text, 125096 bytes are bss, 81832 bytes are rodata
12:48:05preglowhow do they manage it
12:48:42linuxstbAAC must be a very big standard. Maybe some parts are too rare to be worth supporting. I know absolutely nothing about the different variations of AAC.
12:50:48preglowi know something about it
12:50:56preglowLC is almost everything it's worth to support
12:51:19linuxstbAny idea what the library is called? The CVS module is called "faad2", but the library itself is in a directory called "libfaad".
12:51:22preglowwould be fun to support sbr and ps as well
12:51:40preglowdoesn't matter too much
12:51:51preglowfaad is another library, i believe
12:51:56preglowor a previous attempt
12:52:43linuxstbI think we do support sbr and ps. It's "ssr" and "main" that aren't converted to fixed point.
12:53:20linuxstbAny idea if this is an area where libfaad is better than the helix decoder?
12:57:04preglowi didn't think faad2 even supported fixed point sbr
12:57:09preglowbut yes, i see now that it does
12:57:37preglowi don't think it's better in any area
12:57:47preglowthe resource use of the helix decoder seems lower, and the implementation faster
12:58:32linuxstbI'm sure we improve that a lot though. It seems very very wasteful with RAM.
13:00:08preglowoh yes
13:06:03 Join linuxstb_ [0] (
13:08:16 Join amiconn_ [0] (
13:15:50preglowbut you're working on the last cvs copy from before they went anal with the license, yes?
13:16:06preglowhas anything notable happened on their part since then?
13:18:47linuxstb_There have been a couple of "synced with Nero cvs" commits. But I haven't looked at the details. You can browse their CVS here:
13:19:47linuxstb_In fact, there only seems to be one commit - 8 months ago.
13:19:54linuxstb_That's when the license was changed.
13:20:47***Saving seen data "./dancer.seen"
13:22:32 Quit linuxstb (Read error: 110 (Connection timed out))
13:22:51preglowand no one with the authority to be believed has complained that their recent addition conflicts with gpl yet?
13:24:50preglowseems like the only thing they changed was the license
13:26:27 Quit amiconn (Read error: 110 (Connection timed out))
13:26:27 Nick amiconn_ is now known as amiconn (
13:27:31preglownah, there are some other code as well
13:27:38preglowthere is, even
13:28:25preglowsome of the changes may be of consequence for sbr
13:28:33preglowbut i don't think we'll have enough capacity for decoding that anyway
14:03:39 Nick linuxstb_ is now known as linuxstb (
14:08:14 Quit ashridah ("sleep")
14:13:53preglowlinuxstb: but please, do commit this once you've got time, i'd love to try making it go faster
14:14:26linuxstbNo problem. I'll try and do it this evening.
14:14:34linuxstbToo busy with real work at the moment.
14:14:55preglowno worries
14:15:43linuxstbI don't think it will take much to get it running real-time. I think we should be able to do it by just using the IRAM better.
14:16:01preglowyes, and the emac unit
14:16:21preglowunless you've fixed it, it does a bunch of costly 64 bit multiplies at the moment
14:17:01linuxstbNo, I haven't touched anything like that.
14:28:09preglowdamn, aac has quite a coding delay
14:28:12preglowi wonder how bad the gaps are
14:30:52linuxstbI don't know - but I do know that gapless will be a problem.
14:31:40linuxstbBut I'm pretty sure the m4a container stores the name of the encoder, so maybe that will help.
14:32:12preglowit's remarkable that gapless encoding isn't a design problem these days
14:32:29preglowbut then again, both aac and mp3 were primarily designed to be used to encode one stream, like for a dvd
14:32:53preglowi think i'll start fiddling a bit with speex again
14:33:19preglowit seems to require libogg, but that should not be a problem
14:33:33linuxstbWell, the mpeg elementary streams are designed to be put inside containers. It's just that nobody decided to do that with mp3, and the m4a container doesn't contain the necessary information for gapless.
14:34:35linuxstbBut Apple's whole approach to music (iTunes, iTunes store, iPod) is based around individual tracks. Albums just appear to be an afterthought.
14:35:18preglowyes, a lovely concept
14:35:25preglowfor us who primarily listen to whole albums
14:38:25linuxstbDo you have many AAC test files? My m4a demuxing code is based solely on my experiences with ALAC. But it seems to work perfectly with the AAC files I've encoded with iTunes.
14:40:42preglowi have none
14:41:02preglowi've never used aac in my life
14:41:07preglowthat is, i did once, to compare with vorbis
14:41:15preglowbut i don't think that encoder was all that
14:41:16ender`i could probably demux a few aac streams from the video files i have if you're interested
14:41:27preglowi'm interested in all the test files i can have
14:42:09ender`ok, just a moment
14:42:55linuxstbender`: What format will those files be? Raw ".aac" streams?
14:43:27linuxstbAt the moment, the codec only supports AAC in an m4a container.
14:45:17ender`how do i know if the'yre raw?
14:46:07ender`anyway, i can dcc you the file if you're interested, it's 1,1MB
14:47:18 Join muesli_ [0] (n=muesli-@
14:47:26preglowsure, go ahead
14:47:41_FireFly_ho ;)
14:47:49muesli_ff ;)
14:48:01linuxstbender`: Me too.
14:48:55 Join muesli- [0] (n=muesli_t@
14:48:59 Part muesli_
14:49:02muesli-re 2nd
14:49:07linuxstbender`: What extension did your demuxer give the file? I'm assuming that ".aac" implies a raw AAC bitstream with no container.
14:50:32 Quit ender` (Read error: 113 (No route to host))
14:50:56 Join ender` [0] (i=ychat@
14:51:42ender` can count on Windows to freeze in the most inappropriate moment...
14:52:09ender`anyway, i'm sending the file
14:57:58preglowman, what a great mass of files any automake/autoconf usage entails
15:02:39linuxstbpreglow: If you do start work on Speex, it would be nice to create a libOgg (either our own or the official one) - so we can use the same parser with Tremor, Speex and FLAC. Similar to the libm4a I've created.
15:03:37linuxstbNow that we've got a nice low-level FLAC decoder, I think Ogg FLAC support should be straightforward.
15:07:24solexxFLACs in Ogg containers aren't yet supported?
15:08:52preglowlinuxstb: yeah, i thought about that
15:09:14preglowlinuxstb: i'll probably have to put libogg in here as well, but problem is that tremor probably can't use that without modifications
15:09:58linuxstbsolexx: No, they never have been. And I've never seen anyone request it.
15:10:47linuxstbI think there's a big gain to be had if we start using the libraries at a lower level. Mainly in terms of memory management. That's the big difference with the new FLAC decoder.
15:10:52preglowlinuxstb: the official libogg has more than enough lines of source code for me to even want to attempt coding another one, heh
15:11:06preglowlinuxstb: sure, and a lot of codecs are used at a low level, like libmad
15:11:16linuxstbI'm thinking you could just rip the Ogg code out of Tremor.
15:12:15preglowlibogg is only needed for the encoder and decoder, it seems
15:12:20preglowso i'll be ignoring it for now
15:13:40_FireFly_whenn libogg is neede for both(encoder and decoder) then why do you can ignore it ??
15:13:59_FireFly_or have you mean that libogg is only needed for the encoder :)
15:14:47preglowwhat are .spec files?
15:14:58preglow_FireFly_: only needed for the example source code
15:15:14preglow_FireFly_: speex files can live on their own without a container
15:15:22B4gderpreglow: .spec is RPM stuff I believe
15:15:24 Join ashridah [0] (
15:15:47_FireFly_ah ok
15:16:01preglowB4gder: right
15:19:24 Join bigskinnyflo [0] (
15:20:28bigskinnyflohy, can anyone tell me what kind of hard drive is used by the archos jukebox recorder 20? thx
15:20:48preglowa 2.5" one, no?
15:20:50***Saving seen data "./dancer.seen"
15:20:50B4gdera plain 2.5" 9.5mm one
15:21:06B4gderthe archoses have shipped with various brands/models of disks
15:21:34bigskinnyfloany advise for preffering one to buy in replacment?
15:21:50B4gderyes, low speed, low power
15:22:07B4gderand cache won't matter
15:22:39bigskinnyfloin term of brand on the market cause know nothing 'bout that!
15:22:58B4gderany brand
15:23:07bigskinnyflosorry for y english , anyone speaking french welcome!
15:23:31B4gderand any size (storage wise) too for that matter
15:24:50preglowhow i hate autoconf
15:25:08_FireFly_preglow what's the problem ??
15:29:21preglowno problem, i just think it's bloated, big and troublesome
15:29:39B4gderI believe everyone think that
15:30:20B4gderbut it is the easiest way to multiplatform coding
15:33:20 Quit arkascha (Remote closed the connection)
15:33:39preglowcan i make use of c99 style variable size arrays? :>
15:34:14B4gderI think you can
15:34:27preglowit's either that, or alloca
15:34:38preglowi think perhaps alloca is the wisest in our case
15:35:09preglowbut they're probably the same thing, aren't they?
15:35:19preglowstack allocation
15:39:15preglowdoes remote logf still work with the new remote support, btw?
15:39:45B4gderI don't know, but I hope so
15:42:59 Quit muesli- (Read error: 104 (Connection reset by peer))
15:48:22linuxstbpreglow: Have you tried the latest FLAC codec? The volume should be OK now.
15:50:39preglowlinuxstb: yes i have, and it's just fine
15:56:02amiconnThe plain C memcpy() for coldfire is pathetic... I expect a similar speedup as for memset()
15:56:15preglowthen good
15:56:24preglowtons of unused variable warnings for libpseex
15:57:03 Quit dwihno (Remote closed the connection)
15:57:23preglowa whole friggin herd of them
15:57:36 Join dwihno [0] (n=dw@
15:58:36preglowwhat is the c99 restrict keyword?
16:00:38 Part guillauh ("bye")
16:00:46preglow float * restrict W;
16:00:48preglowlike that
16:02:57amiconnLike that: (lower part)?
16:14:48 Quit ashridah ("Leaving")
16:19:07 Join Maxime [0] (
16:24:35 Quit Maxime (Read error: 104 (Connection reset by peer))
16:25:07 Join Maxime [0] (
16:26:37 Join hshah [0] (
16:28:49 Quit _FireFly_ ("IceChat - Its whips the llama's butt")
16:36:10 Quit hshah ("CGI:IRC")
17:05:00 Join Kohlrabi [0] (
17:15:03 Join _FireFly_ [0] (
17:15:11_FireFly_TiMiD ??
17:18:00 Quit Moos ("Glory to Rockbox")
17:19:33 Join dpassen1 [0] (n=dpassen1@resnet-233-61.resnet.UMBC.EDU)
17:20:53***Saving seen data "./dancer.seen"
17:33:18 Join _DangerousDan [0] (
17:34:40TiMiD_FireFly_: yes ?
17:47:51 Quit DangerousDan (Read error: 110 (Connection timed out))
17:53:03amiconnB4gder/ Bagder: wiki spam again :(( I've deleted it, but this is getting really annoying
17:54:37preglowbloody libm
17:55:17preglowvoid speex_encode_stereo(float *data, int frame_size, SpeexBits *bits)
17:55:21linuxstbIs there a "proper" way to delete wiki spam? i.e. so it doesn't appear in the diffs or page history?
17:55:22preglowlooks promising...
17:56:24linuxstbAre you looking to get the encoder working as well?
17:56:47amiconnlinuxstb: Perhaps directly at the server...
17:57:20preglowlibspeex is chock full of unused variables and float uses
18:05:52 Join RiverFish [0] (
18:06:37 Quit RiverFish (Client Quit)
18:09:38preglowsomeone has posted a big no float patch on the speex mailing list, i wonder if i should apply that
18:25:06linuxstbpreglow: I think so. Did anyone comment on the patch? Maybe it's worth contacting the author.
18:25:49preglowno comments at all
18:26:34linuxstbDoesn't surprise me.
18:27:04linuxstbIt's probably worth asking the author of the patch if he/she has done any more work on it. When was it posted?
18:27:42preglowdoes surprise me a bit, the libspeex author is pretty active on the mailing list
18:29:33preglowi think i'll just try to work things out without that patch first
18:30:29linuxstbI didn't expect Speex to have such a busy mailing list.
18:30:54preglowme neither
18:32:04linuxstbAre you using the very latest SVN version? There seems to be various patches posted to the list in September to fix compiler warnings.
18:32:42linuxstbBut maybe they haven't been applied.
18:33:11preglowi'm not using svn at all
18:33:19preglowi'm using the latest 1.1 branch release
18:33:26preglowlast time i tried svn, it didn't even compile
18:33:33preglowthat is however a long time ago
18:33:55preglowthink i'll just try using this for writing a decoder first
18:36:46linuxstbI've just checked out the current SVN version of speex - I don't know if this is of any use to you:
18:38:34preglowspeex does a lot of malloc :/
18:38:57linuxstbXiph like their mallocs
18:49:37preglowseems like it
18:50:48 Quit bigskinnyflo ("CGI:IRC")
18:58:07preglowlinuxstb: btw, i think me and slasheri agreed one should call seek_complete() when seeking is complete, that'll also zero out the seek_time variable for you
19:06:24linuxstbYes, I remember that function now. I'll add that to ALAC/FLAC/AAC tonight.
19:13:21preglowgah, it decodes data as 16 bit values
19:13:57preglowmight not be easily fixable, speex is coded pretty tightly with regard to the fixed point formats used
19:16:07 Join ghode|afk [0] (
19:16:26 Join hypnoticmonkey [0] (
19:19:17 Quit einhirn ("Miranda IM! Smaller, Faster, Easier.")
19:20:54***Saving seen data "./dancer.seen"
19:30:02 Join arkascha [0] (
19:43:15 Join Cenobit [0] (
19:44:01_FireFly_blub :)
19:44:22CenobitI plan to make my owm MP3 decocer - can be for me Rockbox project usefull?
19:45:06preglowyou need to be more specific. do you plan on making just a decoder library, application, or hardware system or what?
19:45:48Cenobitnow I use for decodig VS1001k (external MP3 decoder
19:45:59Cenobitand I want it replace with my own DPS
19:46:11preglowok, what cpu have you got?
19:46:43CenobitI havent any its just a plan
19:46:57CenobitI use AVR, but its not enough fast
19:47:00preglowrockbox uses libmad for decoding mp3s
19:47:06preglowoptimised for our platform
19:47:12preglowwhich is currently coldfire
19:47:32preglowavr will not be fast enough, no
19:47:33preglownot by far
19:47:48Cenobitcan you tell me more about rockbox architecture (or there any schematic)?
19:48:09preglowthe only arch rockbox does software deocding for, is the iriver h1x0 players at the moment
19:48:34preglowwhich use a motorola coldfire 5249
19:48:58Cenobitthx a lot - I am interested in SW decoding
19:48:59preglowyou can find schematics there as well
19:49:57*preglow cheers on linuxstb
19:51:40 Join paugh [0] (n=kickback@2001:5c0:8fff:ffff:8000:0:3e03:6822)
19:52:24Cenobitin what file in firmware are functions for MP3 decoding?
19:52:40preglowCenobit: it's not there, it's in apps/codecs
19:53:05preglowCenobit: apps/codecs/mpa.c is the mp3 codec plugin
19:53:14preglowCenobit: libmad is in apps/codecs/libmad
19:54:00 Quit arkascha ("Konversation terminated!")
19:54:10 Join arkascha [0] (
19:56:01 Join LinusN [0] (
19:56:42LinusNTiMiD: UTF-8 alert in onplay.c
19:56:54LinusNCopyright (C) 2002 Bj?n Stenberg
19:57:12*preglow celebrates the aac commit with julel!
19:57:36LinusNi suggest you find yourself a text editor that doesn't mess with the files
19:58:17*preglow recommends vim
19:58:20*amiconn will have to find a decent editor that does handle utf-8 for upcoming unicode support
19:59:22SlasherHehe, my pcb just arrived today: :)
19:59:30_FireFly_amiconn: maybe scite for a simple editor with syntax-highlighting
19:59:43preglowSlasher: what is it?
20:00:17Slasherpreglow: a nice bluetooth controller thing, hopefully i will get it working soon :)
20:00:43preglowSlasher: good luck, pcb design for fast digital logic can be a pain
20:00:58Slasheryes, thanks :D
20:04:54preglowany fast way to include additional include dirs for apps/codecs/speex.c, or do i need to add the additional include paths for all the codec plugins, then?
20:04:59preglowi suck at makefiles
20:06:37 Quit Cenobit ("CGI:IRC (EOF)")
20:08:08 Join Aramil [0] (
20:09:41_FireFly_preglow: simple add a line with INCLUDES += after the last INCLUDE=/INCLUDES += line in the makefile
20:09:57_FireFly_with your additonal include-path
20:13:42preglow_FireFly_: yes, that's what i do currently, but that adds a codec specific include directory to all the codecs, and might break some things if two codecs share the same include file name, like config.h or fixed.h
20:14:32preglowtons and tons of linker errors
20:14:37preglowlots of floating point math left still
20:14:57_FireFly_which makefile du you edit? i might be wrong but there are makefiles in each codec directory
20:16:04_FireFly_why don't you edit only the makefile which should be in the dir for the speex
20:16:04preglowi'm talking about speex.c, the codec plugin for speex
20:16:23preglowi can't edit the makefile in the speex dir, that's not the ony bullding the codec plugin
20:16:29preglowthat's what's building the codec library
20:16:31_FireFly_my failure
20:19:03amiconncodec libs should never use config.h, as that collides with the main rockbox config.h
20:19:18preglowdoes it now...
20:19:40preglowi'm also wondering why the codecs/Makefile has dirs such as firmware/ in its include path
20:19:47 Quit Aramil ("Leaving")
20:24:51 Quit tvelocity (Network is unreachable)
20:25:00preglowdidn't change anything
20:45:41linuxstbAll the codec Makefiles needs reviewing - some strange things are in them. Whatever mistakes were in the first one are now duplicated 8 times.
21:03:21 Quit _FireFly_ ("Leaving")
21:09:20 Quit hypnoticmonkey ("CGI:IRC")
21:17:20 Join matsl [0] (
21:20:58***Saving seen data "./dancer.seen"
21:23:49 Quit arkascha (Remote closed the connection)
21:24:18TiMiDLinusN: It's not a problem with my text editor, it's just that my system is set to use UTF-8
21:24:32TiMiDkevin@arrakis ~/rockbox-devel $ echo $LANG
21:24:45TiMiDevery program uses UTF-8
21:24:49LinusNTiMiD: yes, but why is the in bjrn changed?
21:25:07TiMiDbecause I changed the file
21:25:23LinusNso, your text editor decided to change it to utf-8
21:25:37TiMiDit loads it as UTF-8
21:26:12TiMiDand since the value means nothing in utf-8, it messed up
21:26:37TiMiDbecause ö is >127
21:26:49LinusNimho, the text editor shouldn't care at all about the text you don't change
21:27:24LinusNjust like it shouldn't change the line terminations etc
21:27:43TiMiDit depends on how the text editor is done
21:28:13LinusNhence my comment about getting a decent editor
21:28:38TiMiDwhat do you suggest ?
21:28:55TiMiD(I don't loke command line based)
21:29:24LinusNi don't know, find an editor that can deal with it, or make your current editor stop messing with the files
21:30:07TiMiDmaybe kate with vim backend
21:31:34TiMiDI don't know what I should do since I need utf-8 on my system
21:32:03TiMiDand the problem will remain the same with each editor
21:32:11LinusNi doubt it
21:32:47TiMiDI know that vim detects the encoding of the file, so I will try to integrate it in kate as backend
21:32:55 Join linuxstb_ [0] (
21:33:28TiMiDand what about making all the files in CVS UTF-8 ? :D
21:34:41SlasherHmm, kate works fine here
21:35:08SlasherUTF-8 causes problem with cvs and tools so i think it's not an option currently
21:35:25TiMiDI was just joking :p
21:36:50TiMiDkate works welll with utf-8 but you have to specify the encoding of the file manually since there is no auto detection and it takes the default encoding defined in the environment vars
21:37:02Slasherjed is the nicest command line editor i know
21:37:11Slasheror console editor
21:37:45TiMiDI prefer graphical editor
21:38:10 Join Moos [0] (
21:38:14Slasheryep, i use kate too with larger projects
21:38:15LinusNshouldn't it be possible to set the encoding different in the shell from where you start kate?
21:38:58TiMiDthis would be a solution
21:39:06SlasherTiMiD: configure - editor - open/save
21:39:12Slasherthere you can specify the encoding
21:39:23Slasheryou should probably try iso 8859-1
21:39:39TiMiDSlasher: the problem is that the files on my system are in UTF-8
21:39:49LinusNthey shouldn't
21:39:55TiMiDrockbox is the exception
21:40:23TiMiDwhy ?
21:40:35LinusNi thought you meant the rockbox files
21:40:37TiMiDunicode is very convenient
21:40:40TiMiDok :)
21:40:56LinusNi'm too old for unicode :-)
21:41:05SlasherHmm, maybe you could create a kate project for rockbox that has different configuration settings
21:41:13SlasherI think that is possible (never tried myself)
21:43:43TiMiDwell I put kate in iso and I will open my other files with kwrite
21:45:07TiMiD(in kate, the encoding is specified globally :/)
21:45:26TiMiDnot for a particular project
21:46:34solexxstrange. I was just listening to a FLAC file
21:46:41solexxand it was completely distorted
21:46:53 Quit linuxstb (Read error: 110 (Connection timed out))
21:47:09solexxI had switched Replaygain temporarily off so I switched it on again
21:47:24solexxthen everything was fine again
21:47:33 Nick linuxstb_ is now known as linuxstb (
21:47:41solexxthen I switched it off, and it still plays fine
21:47:58solexx(I stopped playback everytime I toggled Replaygain)
21:48:49solexxHm. Cannot reproduce it.
21:50:33solexxThe distortion sounded like all the waves had been cut at the top
21:50:40solexxvery noisy
21:53:28solexxah, there it is again.
21:53:40solexxi skipped from an Ogg Vorbis file to a FLAC file
21:53:50solexxand it sound like my headphones were completely broken
21:54:48solexxReplaygain is off
21:54:53 Part LinusN
21:55:05amiconnTiMiD: The battery meter doesn't update in the menu
21:55:28Maxime(ah, one detail, when song is Paused, and you start another, if crossfade enabled, you have the 'end' of the song before starting the other.. starting the other without finishing the paused song will be great, no?)
21:56:48linuxstbAAC is now in CVS....
22:02:07amiconnAnother observation: In radio mode, the menu lags severely when the remote is plugged (e.g. scrolling through the preset list). This is because the CPU is set to idle mode, and so the remote lcd update is dead slow...
22:02:19 Join LinusN [0] (
22:06:52amiconnlinuxstb: You forgot to change config-h100.h
22:06:52TiMiDamiconn: about the statusbar, have you tried the last build ?
22:08:43TiMiD(I thought I fixed this bug with my last commit)
22:08:51amiconnI'll check
22:10:28amiconnI usually roll my own builds
22:11:11amiconnI can't build atm - linker error
22:11:13TiMiDabout the slowness of the menus, do you think it's related to the fact that the list componment redraws everything each time it draws ?
22:11:19amiconn. /opt/m68k/lib/gcc/m68k-elf/3.4.4/../../../../m68k-elf/bin/ld: cannot find -lfaad
22:11:19linuxstbamiconn: Thanks. Fixed now. I didn't want to change everything because hopefully we can reduce the size back down soon.
22:11:30linuxstbDid you reconfigure?
22:11:49amiconnIs there already an updated configure script?
22:12:05linuxstbYes, I committed it with my last set of changes (I hop).
22:15:03TiMiDamiconn: I just tested and the contextual menu in the radio mode is not slow at all !! how did you get the lag ?
22:15:45amiconnIt's only slow with the remote plugged. I doubt you'll see the effect with the context menu, as that one is short
22:16:18TiMiDI have the remote plugged
22:16:44amiconnGo to radio mode, call the preset list (hope you have a long one), then scroll down with button repeat, and stop after some scroll events, but before the end
22:16:59amiconn-> it'l keep scolling quite some lines
22:17:32amiconnThis isn't your code's fault, as scrolling through looong dir listings doesn't lag
22:17:38TiMiDI don't have any preset since I don't use the radio ;)
22:17:56TiMiDbut do you think my code is significantly slower that the old one ?
22:17:58amiconnThe problem is that the CPU is switched to 11 MHz in radio mode, so the remote lcd update is dead slow
22:18:32TiMiDok so the problem is my code waiting for the remote to update ?
22:18:49TiMiDor to draw
22:19:13amiconnDrawing happens in ram, like with the main display
22:19:27amiconnOtherwise it would also lag with the remote unplugged
22:19:38amiconnThe problem is definitely the update
22:19:50 Join _FireFly_ [0] (
22:19:58linuxstbIs anyone else getting problems compiling the current CVS? My build is stopping at wav2wv.elf
22:20:08amiconnThe remote lcd update definitely needs some work, but we might still have to give up 11 MHz operation while radio
22:21:34 Join len0x [0] (
22:22:34len0xhey, anyone's around ?
22:22:39amiconnIirc, the SPI timing for the remote is quite arbitrary, so it can be optimised, perhaps adapting to the current CPU frequency
22:22:51LinusNlen0x: no, not a single soul :-)
22:22:55preglowlinuxstb: i don't think the codec libs need to include codeclib.h
22:23:14amiconnSome logic analysing should be helpful for getting the timings right within specs ...
22:23:41LinusN/usr/local/m68k/lib/gcc/m68k-elf/3.4.3/../../../../m68k-elf/bin/ld: region PLUGIN_RAM is full (/home/linus/rockbox-all/iriver/apps/codecs/aac.elf section .bss)
22:23:56linuxstbpreglow: I think I did that for some debugging. Please remove it if you don't think it's needed.
22:24:00TiMiDamiconn: else, it's rather normal that my code runs slighly slower that the original one since it draws twice ...
22:24:08len0xLinusN: web client severely lags on me :)
22:24:46preglowTiMiD: does it draw for the remote at all if it isn't plugged in?
22:24:54amiconnTiMiD: The drawing is not the problem, as I already said
22:25:33 Quit _FireFly_ (Read error: 113 (No route to host))
22:25:46amiconnpreglow: It has to, otherwise the code would have to maintain state. The remote can be plugged any time...
22:25:46 Quit len0x (Client Quit)
22:26:04linuxstbLinusN: Which target is that? I was late committing an h100 config file change.
22:26:20TiMiDpreglow: no, it always draw, since it must keep the context for the remote too, it would have been more complicated to take the plug / unplug in account only for drawing
22:26:30amiconnLinusN: You need to rebuild completely
22:26:35 Join len0x [0] (n=len0x@
22:26:41LinusNdoing a clean make now
22:26:43 Join _FireFly_ [0] (
22:26:52amiconnI got the same prob here
22:27:00amiconnmake clean fixed it
22:27:06LinusN512k for a codec, phew
22:27:35LinusNlinuxstb: ignore me, a "make clean" solved it
22:27:56linuxstbSo is CVS compiling cleanly for people? I've got a strange problem, but I guess it's local.
22:28:12preglowis it possible to compile rockbox for h100 without the remote support, then? :>
22:28:42preglowi never use the remote, and don't want it slowing down my unit
22:29:20preglowlinuxstb: great work, i'll see about optimising it asap
22:29:25LinusNpreglow: it shouldn't slow it down noticeably
22:29:33 Quit Kohlrabi (Nick collision from services.)
22:29:37 Join Kohlriba [0] (
22:30:16amiconnpreglow: It doesn't as long as you don't plug the remote
22:32:13amiconnTiMiD: The status bar refresh problem is gone
22:33:55TiMiDamiconn: yes, that was the main purpose of my last commit :)
22:34:24 Join linuxstb_ [0] (
22:35:03TiMiDpreglow: if you don't want the remote, just undefine HAVE_REMOTE_LCD
22:35:13amiconnMy remote doesn't tick
22:35:33TiMiDamiconn: neither does mine :)
22:35:53TiMiDthe sound quality is just fine
22:36:06amiconnI guess it's a question of proper shielding in the remote cable
22:37:09amiconnThe remote lcd is flickering a bit.
22:37:45amiconnMust be a hardware issue, or maybe the frame frequency is selectable, and we set it too low
22:42:45 Quit matsl (Read error: 104 (Connection reset by peer))
22:42:46*amiconn checks the datasheet
22:43:39 Join matsl [0] (
22:48:59 Quit linuxstb (Read error: 110 (Connection timed out))
22:56:14 Quit dpassen1 ()
23:01:18XavierGrlucky guys!!
23:01:35XavierGrMy remote clicks like a mouse!!
23:01:53 Quit ender` (Read error: 104 (Connection reset by peer))
23:03:01preglowhave you tried with different earphones?
23:03:39XavierGrTiMiD, amiconn: About the delay you mention on the radio screen. I thought that radio screen isn't implemented yet for the remote...
23:03:59amiconnNot the radio screen, the preset menu
23:04:02XavierGrpreglow: 3 phones so far. Bundled irivers, some cheap retractable and shure r2c
23:04:10XavierGrr2c= e2c
23:04:11preglowXavierGr: and they all click just as badly?
23:04:36XavierGrwell better phones make this more clear and loud.
23:05:01preglowyes, i was just trying to make sure it's not a sensitivity issue
23:05:07TiMiDXavierGr: its about the preset list lagging when CPU runs at 11Mhz
23:05:14preglowbut it shouldn't be, no
23:05:24XavierGrrendered on the radio or the main screen only?
23:06:06XavierGrI mean does the lists gets renedered on the remote screen.
23:06:31XavierGrOr the lag appears on the main screen with the remote plugged in (but the remote screen idle)?
23:06:37TiMiDyes the list gets rendered on remote
23:06:44 Join webguest21 [0] (
23:06:52XavierGrWhen did you changed radio.c for the remote?
23:06:54TiMiDbut not the radio screen
23:07:12LinusNXavierGr: he didn't
23:07:18TiMiDXavierGr: I didn't changed it, it uses the menu API which I modified
23:07:38 Quit webguest21 (Client Quit)
23:07:55XavierGrI should test it myself then...
23:08:18XavierGrBut I remember that this sort of lag was present in firefly's patch too.
23:08:19TiMiDit only happens hen you have a long list of presets
23:08:41XavierGrThough back then there was a "pop" issue with the PLL
23:08:43TiMiDit's because (according to amiconn ) the remote drawing functions lags
23:08:54XavierGrSo without noticing I removed the cpu = idle and there was no lag
23:09:46amiconnYeah, the remote lcd update is severely slowed down, in order not to exceed the lcd specs at maximum clock. This makes it slow as hell at 11 MHz
23:09:54XavierGrnow that the "pop" is fixed (though main code still has a initialization pop, my radio patch icludes an if to eliminate this) I can get the same old lag
23:10:45 Join tvelocity [0] (
23:11:03XavierGrhi tvelocity.
23:11:49amiconnThe bad thing is that (1) the bits belonging to the remote lcd are scattered across both GPIOs, and (2) we can't assume that the port isn't changed from within an isr like on the archos,
23:12:00amiconnso optimising will be somewhat harder
23:13:18XavierGrAlso I can notice that the remote clicks on my faulty unit is way louder on rockbox than iriver. (Or so I remember when I tested some months ago..>)
23:16:25XavierGrTiMiD: If you are going to make radio work for the remote, please check my file preset patch and see if you can inlcude it. That is if it is okay with the rest of the devs.
23:17:00TiMiDXavierGr: I'm currently not working on it
23:17:38XavierGrOkay then, I just mentioned.
23:17:53TiMiDbut I can't decide to include your patch without the other devs agreement
23:18:21XavierGrthat's why I said "if it is okay with the rest of the devs..."
23:18:31TiMiDwhen I will come to this part of the code, I will take a look anyway :)
23:18:46len0xguys my patch would be probably be the first to go into radio.c :)
23:18:54XavierGranyway if someone objects to the patch I would gladly make modifications to it.
23:19:01amiconnpreglow: I reused your nice jumptable implementation for memcpy :)
23:19:20XavierGrlen0x: which patch?
23:19:47len0xI also have reworked all radio buttons, that are not in the patch yet
23:19:59preglowamiconn: i'm honoured!
23:20:05XavierGrmy patch has nothing to do with radio keymaps
23:20:24XavierGrit just adds file preset management.
23:20:48XavierGrLoading and saving different fmr preset lists.
23:20:48len0xyou can see my patch in the list: 1335463
23:21:02***Saving seen data "./dancer.seen"
23:21:16len0xwell, we are clashing in Preset Mode implementation
23:21:43preglowi just had a look at the jumptable implementation
23:21:46preglowhow the hell can that work?
23:22:05preglow jmp.l (2, %pc, %d2.l*4)
23:22:17preglowshouldn't it be d2.l*2 ?
23:22:23preglowthe jump table entries are 2 bytes
23:22:57preglowforget me
23:23:36XavierGrlen0x: we can adjust this later, my concern is about fmr preset list management
23:23:36preglowi've been drinking again, obviously
23:24:36len0xif I get CVS access eventually, i could merge it in since I'm working in radio.c anyway :)
23:29:32amiconnpreglow: bra.w is 4 bytes long
23:29:37amiconn(one extension word)
23:29:52amiconnI used bra.b, so *I* need *2 :-)
23:31:50preglowamiconn: yes, i realised that after i stopped being a bloody fool
23:33:29preglowanyone got any aac files i can use for testing? :>
23:37:21 Join linuxstb [0] (
23:38:03linuxstbI believe Apple have one or two - $0.99 each....
23:38:11preglowcomplete with drm, i hope?
23:39:04linuxstbAnd free gaps. two per track.
23:39:04preglowin a related note, does anyone know of any good aac encoder?
23:39:34preglowlinuxstb: they always were a company for good bargains
23:39:56linuxstbI only know two encoders - iTunes and Nero.
23:40:20preglowlinuxstb: nero = faac ?
23:40:43linuxstbI think so. But it seems to be quite well respected on HA.
23:41:55preglowdoes itunes cost anything?
23:44:24linuxstbNo. It's a free download. I have an iBook, so I use the Mac version there.
23:47:24 Join elinenbe [0] (
23:47:44elinenbeLinusN: are you around?
23:48:00preglowwell, as long as it doesn't do anything fancy
23:48:16elinenbeI see you've been busy with the H300 port... good luck with it! I'm anxiously waiting it! :-)
23:48:26elinenbeLinusN: "you can do it!"
23:48:56preglowlike hijack all my file extensions and install quicktime
23:49:33preglowdon't think there's a great deal of doubt about that flying around
23:50:17linuxstbI've never installed the Windows version of iTunes, so I don't know what it will do to your setup.
23:50:23preglowargh!! quicktime!
23:50:34linuxstbBut yes - iTunes relies on Quicktime.
23:52:15elinenbeIf I am not mistaken, the itunes installer includes and installs quicktime
23:52:18elinenbeyou don't have a choice
23:52:18preglowmy, how i hate quicktime
23:52:25*amiconn wonders how he will manage to code a proper memcpy() when he doesn't even manage to code a proper test :/
23:53:55elinenbeamiconn: I think you'll be able to do it.
23:54:05elinenbeamiconn: are you re-doing memcpy() in asm?
23:54:05 Quit matsl (Remote closed the connection)
23:54:28amiconnYes, for coldfire this time
23:54:47amiconnI also have a (pending) memmove() for sh1, and will do the same for coldfire as well
23:55:06 Join matsl [0] (
23:55:23elinenbeamiconn: I need numbers... I need percentages...
23:55:29elinenbeamiconn: how much do you expect to save?
23:55:51amiconnI expect a speedup of up to 10x for certain aligments, like with memset()
23:56:32elinenbeamiconn: damn! nice.
23:56:41elinenbeamiconn: what will that effect in the long run though?
23:57:20amiconnI think fast memcpy() and memmove() are useful for the codecs
23:57:34elinenbewhat about something like rockboy?
23:57:58amiconn...and memove() is useful for other purposes too, like scrolling, the lru cache in the unicode patch etc

Previous day | Next day