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 2007-08-09

00:00:30markunDerPapst: BoS
00:00:56DerPapstheh.. to get Microsoft sam say "Plugins" i have to enter "Plugkins" :D
00:01:02ddaltondoes anyone want an extra setting that says "voice time as?" or just use the one that is called "time format"
00:01:20DerPapstmarkun: isn't that because rockbox already sounds like a bos?
00:01:21n1sddalton: I think you should go with the existing setting
00:01:48n1smarkun: how is the GSoC tts thing coming along?
00:01:56markunn1s: pretty bad
00:02:00rasherddalton: I think it should just follow the regular time format. Even though I'd probably set it to 24 for display and 12 hour for speech if I had the option, I don't think it's necessary to add an option.
00:02:07 Quit matsl (Remote closed the connection)
00:02:13DerPapstwhat is TTSbox using btw? flite or espeak?
00:02:14markunn1s: today I got an email that he has even less time
00:02:29blitheDoes anyone else find that if they set a background by browsing a file that it doesn't remain after rockbox is rebooted?
00:02:42 Join My_Sic [0] (
00:02:42markunI wanted to use espeak, but I figured it was his choice
00:02:46n1smarkun: ah, too bad, it would have been a very nice feature
00:03:00markunn1s: maybe I'll do a espeak por
00:03:11saratogawhat does that text to speech project actually involve?
00:03:14markunbut it requires quite some changes
00:03:22saratogaintegerizing the engine or mostly porting it to rockbox?
00:03:28n1smarkun: how does espeak compare to flite in quality?
00:03:40markunsaratoga: so far it was just turning flite into a rockbox plugin
00:03:59DerPapstso what is better then when espeak sounds like a BoS how does flite sound compared to this... even worse?
00:04:16markunn1s: flite uses real voice samples, espeak is synthesized form formants mixed with samples
00:04:43preglowjhMikeS: think i've got something now
00:04:48ddaltonok I can add that later then but when I put this one up it will just use the setting there now.
00:04:51preglowjhMikeS: will be sure to report success tomorrow :P
00:05:36rasherBagder: committed, yell at me if things catch fire
00:05:37jhMikeSpreglow: what does a "vinyl amp" do?
00:05:42markunDerPapst: I think it's a taste thing. I prefer espeak, but other people prefer flite
00:05:43n1sddalton: I think you should actually avoid adding another setting, too many settings is not good
00:05:52 Quit sneakums (Remote closed the connection)
00:06:06preglowjhMikeS: just do the inverse of the vinyl cutting process, apply an inverse riaa eq filter and gain
00:06:19 Join sneakums [0] (
00:07:11preglowjhMikeS: i've got an amp here that only has line ins, so i'd need a vinyl amp to connect my vinyl deck to it
00:07:12markunpreglow: what are you working on? turning rockox into a phono amp?
00:07:14jhMikeSto play phonograph records through the DAP? *l*
00:07:16preglowmarkun: yeap
00:07:28 Quit robin0800 (Read error: 110 (Connection timed out))
00:07:28 Nick robin_0800 is now known as robin0800 (
00:07:29preglowjhMikeS: vinyl deck -> h120 -> line in
00:07:56 Join dup [0] (i=d9b95459@gateway/web/cgi-irc/
00:08:04jhMikeSwork on something for wax cylinders too...I'm really into that ;)
00:09:52DerPapstwow... at&t sounds really awesome copared to everything else i've heared so far
00:10:13dupHello. I observe the following on my H120 with r14246: when I activate the option "directory cache" (it was disabled before) I get "Please reboot to enable". After rebooting, the disc is spinning and but the end, the message "Filetype buffer full" is shortly displayed. Any ideas?
00:10:36Lloreandup: Ignore it, for now.
00:10:56dupLlorean: so there is no error and it's harmless?
00:11:36LloreanIt depends. It affects viewers.config, the last file type (or last few) will not work in it.
00:12:53dupOk. Also, the shortcuts pluging doesn't work for me, But it's the last entry in the viewers.config, so it's probably related to your last post
00:13:25preglowwhat's the max gain of the line in port?
00:14:07dupWhen I select a dir and choose "Add to shortcuts" from the context menu, nothing happens. The .link file doesn't get cretated.
00:14:43dupIt should be created in the root folder, right?
00:14:59peturpreglow: 24dB?
00:15:15peturthat is before using the decimator
00:15:27peturtotal is 48dB iirc
00:15:48LloreanBagder: Could the voice creation be done in mingw32 with some changes? (I'm not familiar with its limitations at all)
00:15:53DerPapstwow.. at&t german is even better than the english version o.O
00:16:16preglowpetur: should do nicely for me, then
00:16:19BagderLlorean: it probably can, I'm not that familiar with that setup either
00:17:29 Part brent0n
00:17:47*petur gets another file type array full splash and wonders what got added
00:18:26duppetur: I have "link" at the end
00:19:31peturlet's increase the array then...
00:20:16*n1s votes for most useless comment of the day,
00:20:24 Join robin_0800 [0] (
00:20:31duppetur: can't it be increased automatically? E.g. by defining some "end marker" constant and using it as the array size?
00:20:33 Quit sneakums (Remote closed the connection)
00:20:53 Quit saratoga ("CGI:IRC (Ping timeout)")
00:20:55peturdup: it's not that the list grows every day
00:21:13DerPapstbut every 2 it seems :P
00:21:21LloreanBagder: I imagine that'd be a more flexible solution for users of screen readers, though I don't really know.
00:21:40 Join haemmy [0] (n=stefan@
00:21:43 Part n1s
00:21:47duppetur: "petur gets _another_ file type..." (c) petur ;-)
00:21:49Bagderyeah, since the biggest hurdle for them with cygwin is their terrible install thing
00:22:30preglowwould anyone think a vinyl amp plugin worth commiting? :>
00:23:02Lloreanpreglow: I think a lot of people would love to see *any* DSP plugin, just as a sample of how it can be done
00:23:03*DerPapst wonders what that could be...
00:23:34LloreanBagder: Was it pondlife working on getting it working in cygwin?
00:24:02peturdoes the "filetype array full" also come on non sw-codec players?
00:24:03LloreanI wonder if he'd be willing to give it a shot. I neither speak shell nor know how to talk to TTS in Windows at all. :)
00:24:19ddaltonIn settings.h there is a variable called timeformat can I do this in main_menu.c? if (timeformat ==1)
00:24:34preglowLlorean: then jhMikeS should definitely commit his ringmod code
00:24:35Lloreanpetur: I'm kinda surprised we even get that. I rather imagined it could be allocated at boot.
00:25:11 Join sneakums [0] (
00:25:37peturnot worth it. Whoever adds something should check this
00:25:56 Quit Soul-Slaye1 ("Leaving.")
00:25:57dupHas anybody (successfuly) used the shortcuts plugin?
00:26:28duppetur: put this as a comment into the viewers.config?
00:26:31Lloreandup: Many people have. It was tested after all
00:26:56peturbe patient I'll commit after testing
00:27:10 Join maffe [0] (
00:27:12*Llorean definitely prefers espeak of the three options
00:27:14dupLlorean: yes, that I also have assumed. But somehow I fail to create the shortcuts file :-/
00:27:57rasherddalton: if ( global_settings.timeformat )
00:28:39markunLlorean: do you know if flite is compiled to use the 8kHz or the 16kHz voice?
00:28:42dupLlorean: what file should I look for?
00:28:43rasherddalton: if that's true, then you have 12-hour clock. Alternatively, global_settings.timeformat can be either 0 or 1 for 24 or 12-hour clock respectively
00:29:49 Quit Rick ("I… don't need to be here.")
00:29:51DerPapstgood n ight everyone :)
00:29:51Lloreanmarkun: How can I check? Though it's partially how words are pronounced too.
00:29:55 Quit DerPapst ("So Long And Thanks For All The Fish!")
00:30:02rashermarkun: "* Link flite(1) with 16bit voice by default"?
00:30:16ddaltonthanks I will give that a go.
00:30:17rasher(in debian.changelog.gz)
00:30:51 Quit wippeout ("leaving")
00:30:56Bagderok test voice built fine with script
00:31:07Bagderupdated daily build script for tomorrow's build
00:31:38rasherAnd I didn't even need to use my complimentary "oops" commit.
00:32:01Bagderdon't let me spoil the fun!
00:32:14 Join iamben [0] (n=ben@
00:34:46peturdup: new build to test in 3 minutes....
00:35:08peturshortcuts creation and usage works fine here now
00:35:38duppetur: Now? So the fix also fixes this and not only the message?
00:36:28peturthe message was an indication that ther was a problem
00:37:21 Quit miepchen^schlaf (Read error: 104 (Connection reset by peer))
00:37:25dupWow! Cool countdown on the build page!
00:37:38 Join aowzone [0] (
00:37:43 Quit robin0800 (Read error: 110 (Connection timed out))
00:37:43 Nick robin_0800 is now known as robin0800 (
00:37:47peturdup: try now
00:37:49 Quit sneakums (Remote closed the connection)
00:37:53 Join sneakums [0] (
00:38:14aowzonehey looking for some help...last weekend i taped a show, and ran out of battery about 3/4 of the way through the set...i have a partial wav file, but i cant seem to open it...i thought i read something about reparing this sort of thing once or was i imagining it?
00:38:30markunrasher: I maybe I was confusing bit with hz :)
00:38:45markunanyway, time for some sleep
00:38:59peturaowzone: somebody at has written a tool to repair them
00:39:17peturlet me find the thread
00:39:20aowzonedamn...tahts where it saw it! it wasnt at rockbox!
00:39:31aowzonehaha i was tearing apart the wiki...i shoulda know it was TS
00:39:45aowzonethanks petur...perhaps yuo know better what to look for, could you link me?
00:40:04 Quit ompaul ("Leaving")
00:41:52 Quit ctaylorr (Read error: 110 (Connection timed out))
00:44:08 Quit obo ("bye")
00:45:14peturI always see that link, and when you search for it you can't find it :)
00:45:42aowzonethank you!
00:46:13 Join ctaylorr [0] (
00:48:49aowzonedamn...failed to read header from wav file
00:49:42peturaowzone: you can always import it as raw data in an audio editor
00:49:43 Join pixelma_ [0] (i=pixelma@rockbox/staff/pixelma)
00:49:52aowzonei think i tried that too...let me try again
00:50:54aowzoneyeah, soundforge 9 wont open it either
00:51:25peturyou managed to copy the file to pc?
00:51:41XavierGrshouldn't have rockbox shutdown properly the device before shutting down from low battery?
00:51:41 Quit pixelma (Nick collision from services.)
00:51:41 Nick pixelma_ is now known as pixelma (i=pixelma@rockbox/staff/pixelma)
00:51:49aowzoneit should have Xavier...but apparently didnt
00:52:03aowzoneits a 400 something mb it recorded something and tried to save it
00:52:09peturaowzone: then run a chkdsk /f on the drive first
00:52:16aowzoneok petur i will do that!
00:52:53aowzoneconvert lost chains to files?
00:52:59peturXavierGr: there are obviously still issues when recording and encountering low batt or disk full
00:53:00 Quit haemmy ()
00:53:17peturaowzone: yes
00:53:28aowzoneok, done
00:53:30aowzonelet me try now
00:53:30XavierGrpitty that we can't say that it's completely secure
00:53:35Lloreanpetur: Well, there's always a point where the battery is too low to spin up the disk again.
00:53:48 Join saratoga [0] (i=9803c6dd@gateway/web/cgi-irc/
00:54:03 Join BigMac [0] (
00:54:12XavierGrLlorean: I thought that Rockbox was programmed to shut down before that happens
00:54:30XavierGrthough with worse batteries it can come quickly
00:54:37aowzoneExisting header data. Look for the words 'RIFF', 'WAVE', 'fmt',
00:54:37aowzoneor 'data' to see if this is even a somewhat valid WAVE header:
00:54:37aowzone00000000: 52 49 46 46 00 00 00 00 57 41 56 45 66 6D 74 20 RIFF....WAVEfmt
00:54:37DBUGEnqueued KICK aowzone
00:54:37aowzone00000010: 10 00 00 00 01 00 00 00 00 00 00 00 00 00 00 00 ................
00:54:37aowzone00000020: 00 00 10 00 64 61 74 61 00 00 00 00
00:54:41aowzonethats what i get after a fixwav
00:55:00LloreanXavierGr: And you can have a significant amount of time during playback left, even if the battery is too low for a future upcoming spinup.
00:55:12Lloreanaowzone: Please don't make large pastes to the channel
00:55:30peturaowzone: that's ok, it is just that the length isn't filled in
00:55:36XavierGrLlorean: yes it can be another 30 minutes on some occasions
00:55:38aowzoneok, how do i fill that?
00:55:46peturthe repair tool can't handle that?
00:55:52aowzoneah, i shouldnt have exited
00:55:54aowzonei think i see now
00:56:00duppetur: thanks. Everything works now. Except one thing: my link file contains only one entry but when I select ("play") the file this one entry is displayed. But maybe the pluging works this way now. I've read that in this case it should directly go to the dir.
00:56:03 Quit sneakums (Nick collision from services.)
00:56:14LloreanXavierGr: Perhaps recording, though, should be more conservative on when it chooses to shut down (if this is possible)
00:56:31peturdup: with only one link it also shows the link here
00:56:40peturmaybe a bug?
00:56:42Lloreanpetur, dup: Is this the main link file, or an arbitrary other one?
00:56:48aowzonewow! it worked!!!
00:56:50aowzonethanks petur!!!!
00:56:50XavierGryeah but then again voltage drop on worse batteries can be very quick...
00:57:06peturah right, maybe only for link files != main
00:57:14XavierGrso even if it will be changed to be more conservative problems might still occur with very bad batteris
00:57:15peturaowzone: np
00:57:23Lloreanpetur: It is specifically only for link files != main. :)
00:57:27dupyes, this was the main (default) link file
00:57:39dupcase closed then! :-)
00:57:41LloreanFor the main one, he wanted to preserve the option to delete the last link within it, etc.
00:57:44LloreanIf I recall
00:57:51aowzonesomeone should put that info in the wiki!
00:57:53petursounds reasonable
00:58:22dupis it possible to somehow call this link file if the file filter is set to "music"?
00:58:33aowzonethanks guys, goodnight!
00:58:41 Quit aowzone ()
00:59:00Lloreandup: Try "supported"
00:59:10LloreanObviously a link file is not "music"
00:59:57dupLlorean: yes, I see :-) I thought about a menu entry (please don't shoot)
01:00:44dupso that the shortcuts are always at hand
01:01:06LloreanJust put them in a folder in the root of your player.
01:02:22LloreanThere are people who want all kinds of things added to the main menu (podcasts, videos, ebooks, text, pictures, now favourites, etc, etc) if they were all added it'd pretty much remove the point of having them there for the purpose of getting to them quickly
01:02:39LloreanMeanwhile your root folder is rather a flexible place in regard to what you put in it, and what you don't mark as hidden.
01:03:16dupI'll do this. BTW: wouldn't it be nice to be able to have a "readable names" for the folders? Folders names can be very long, and it will be a pain to read such list
01:03:41LloreanI have no idea what you mean by "readable names"
01:04:49dupWe could e.g. add an optional tab separated "short name" for an entry. E.g. the folder is "/blah/hgfjhgf/jhgjhg/jkjhk/sfdsgf/" (long), but it would apper as "tralala" (short) in the list
01:05:38dupThe line in the file would then contain that long folder path<TAB>tralala
01:05:59dupIf there is no TAB then the folder is shown as is
01:06:16LloreanAh, for the favourites plugin
01:06:28dupEhm, shortcuts you mean?
01:06:30 Join basscade1 [0] (
01:06:36LloreanSorry, I'm used to the original name of it. :-P
01:06:55LloreanIt took me ages to stop referring to them as "CVS builds"
01:07:10LloreanAnyway, that sounds reasonable. If you write it, post it to the tracker.
01:07:13dupAnother option could be to show folders as is but just the last segment (or the last N segments)
01:08:06 Quit FOAD ("I'll be back")
01:08:35dupLlorean: I'll think about this. But not today anymore.
01:08:54 Quit dup ("CGI:IRC 0.5.7 (2005/06/19)")
01:10:25 Join JdGordon [0] (
01:10:38 Quit Ishi`` ()
01:12:06 Nick basscade1 is now known as sneakums (
01:12:52 Quit Febs (Read error: 110 (Connection timed out))
01:13:13 Join FOAD [0] (
01:20:05 Quit My_Sic ("Miranda IM! Smaller, Faster, Easier.")
01:20:50bluebrotherDomonoky: for the logs, I started looking into it and was wondering if it's better to extend the ZipInstaller class or to create a new one.
01:21:21bluebrotherbut I just got home today, was quite busy −− so no time to work on that today. Hopefully it will be better tomorrow.
01:21:30 Quit bluebrother ("nite")
01:22:05 Join hannesd_ [0] (
01:23:37ToHellWithGAhey Llorean i'll pick on you since you're here often. do you know if i can safely repartition my ipod? i'd like to break the large part into 50% fat32 for ipod use and 50% ext3 so i can store files larger than 4 GB
01:28:33 Quit bdgraue (Remote closed the connection)
01:30:13 Join perrikwp [0] (n=chatzill@
01:31:34 Quit robin0800 (" HydraIRC -> <-")
01:33:22peturdo we have a location to report flyspray errors?
01:33:58 Join chrisjs169_ [0] (
01:34:18 Quit chrisjs169 (Nick collision from services.)
01:34:22 Nick chrisjs169_ is now known as chrisjs169 (
01:37:47 Quit hannesd (Read error: 110 (Connection timed out))
01:37:48 Nick hannesd_ is now known as hannesd (
01:38:46peturJdGordon: around?
01:40:00 Join Rick [0] (
01:41:47 Join perrikwp_ [0] (n=chatzill@
01:43:56JdGordonpetur: not really, about to head off...
01:44:00JdGordonif its quick.....
01:44:16peturnm, I think I found it myself
01:44:30petursorry to bother
01:44:33JdGordonok cool, no worries
01:44:39 Quit JdGordon ("Konversation terminated!")
01:48:01scorcheToHellWithGA: that should be fine as long as you do that to the end of the partitions...also, there is no need to direct a question to a particular person
01:48:33CtcpIgnored 1 channel CTCP requests in 0 seconds at the last flood
01:48:33*petur wonders who GA is
01:50:04peturbah I hate this menu macro stuff.... *heads over to wiki and hopes JdGordon explained it well*
01:53:15 Join safetydan [0] (i=cbca159f@rockbox/developer/safetydan)
01:54:42 Quit perrikwp (Read error: 110 (Connection timed out))
01:56:22***Saving seen data "./dancer.seen"
01:56:43 Join chrisjs169_ [0] (
01:57:01 Quit chrisjs169 (Nick collision from services.)
01:57:06 Nick chrisjs169_ is now known as chrisjs169 (
01:58:49 Join Febs [0] (
02:00:39 Part pixelma
02:01:16 Join aliask_uni [0] (i=82c20d6a@gateway/web/cgi-irc/
02:06:14 Join tihoc4n [0] (
02:07:04tihoc4nHi, I have a question about vorbis, well .ogg to be specific will this ever be implemented?
02:07:35ToHellWithGAscorche: thanks man. i just figured some of yall were around more than others
02:08:09scorcheyes, but highlighting a specific person isnt considered "nice" when you dont need them specifically
02:08:22scorcheas well, if you screw up, there are MBRs in the wiki
02:09:25 Quit BHSPitMonkey (Remote closed the connection)
02:09:58 Join BHSPitMonkey [0] (
02:12:19 Join spiorf [0] (
02:12:20 Join JoeBorn [0] (n=jborn@
02:15:31saratogatihoc4n: what specifically on that page?
02:16:00 Quit jhulst (Remote closed the connection)
02:19:57 Part maffe
02:20:17 Join maffe [0] (
02:20:42*petur gives the menu code an angry look
02:26:58tihoc4nsaratoga: The .oga extension for vorbis audio, sorry I should have been more specific.
02:30:38 Quit krazykit ("bbiam")
02:32:37saratogatihoc4n: its pretty trivial to add that extension, but it can only be for one format I think
02:33:09saratogado you think it would make sense to enable it for Vorbis only?
02:33:32iambendoes rockbox decide on a decoder based on extension?
02:35:41safetydaniamben, pretty much. Though there's some byte sniffing done for ogg streams to check for Vorbis/Speex/FLAC (maybe FLAC... not sure about that one)
02:37:58 Join miepchen^schlaf [0] (
02:38:25tihoc4nI think it would make sense just to enable it for vorbis only, flac users normally use the .flac extension anyway. and Ghost isn't released yet.
02:38:41 Join chrisjs169_ [0] (
02:39:12saratogatihoc4n: theres also speex
02:39:20saratogado people actually use oga ?
02:40:34*Rondom has never seen a single oga-file
02:40:44tihoc4nI don't think anyone does, because there isn't much support for it yet. Anyway about speex according to that wiki page "Although they share the same MIME type, Vorbis and Speex use different file extensions."
02:40:48saratogaif its just something Xiph suggests but few people actually use, it might be a better idea to wait until we can actually demux the things Xiph says we should
02:43:02Rondomwell, supporting oga as file extension with the things rockbox is already capable to handle is a step into the right direction, imo
02:43:26 Quit petur ("Zzzzzz")
02:43:36RondomI mean this could make more people start to use oga (I still doubt it, though)
02:44:23saratogai don't have a problem with it, but get one of the europeans to say its ok and then I'll commit it
02:45:28saratogaor better yet, if you're a programmer, change line 73 of id3.c and submit a patch to the tracker
02:45:46LloreanJust as an added extension?
02:46:15saratogayeah, just add oga to the list of vorbis extensions
02:46:52LloreanDoes specify what .oga gets to be?
02:47:10 Quit Rondom ("Ex-Chat")
02:47:16saratogayeah it can be any of their codecs but speex
02:47:34Lloreansaratoga: Actually, it doesn't mention Vorbis there
02:47:48LloreanIn fact it clearly states it's *not* for Vorbis
02:47:51 Join krazykit [0] (
02:48:17saratogai'm not sure
02:48:19saratoga"using any of the Xiph audio codecs"
02:49:01LloreanWell look at the Rationale bit
02:49:07saratogaask tihoc4n about it
02:49:10LloreanThough I'm not certain what "Vorbis-II" is relative to what we support
02:49:15saratogahes apparently familar with this
02:49:19LloreanDo we support Vorbis I, Vorbis II, or both?
02:49:25saratogajust I
02:49:29tihoc4nNot that familiar, but I've come across it
02:49:29saratogai don't think II exists yet
02:49:46tihoc4nVorbis II doesn't exist, no
02:49:49LloreanWell, the page explicitly says it's to differentiate Vorbis I files (.ogg) from all other xiph audio files (.oga)
02:49:55LloreanSee the RATIONALE bit
02:49:59saratogai wonder why they want to do that
02:50:13FebsGod, my tolerance for stupid lazy whiny demanding people is just about exhausted today.
02:50:14LloreanVorbis I can only have one logical bitstream, according to that.
02:50:28LloreanFebs: More at ours, or elsewhere? I just got back from dinner and haven't refreshed yet.
02:50:59FebsOurs. Want to take a wild guess at the subject matter of the thread? I'll give a hint. It's in the plugins forum.
02:51:02Lloreansaratoga: Either way, according to, we very definitely should not use .oga for Vorbis-I, should for flac, and *maybe* should for Speex if it falls under that "any of the Xiph audio codecs" umbrella
02:51:15LloreanFebs: Ah. Doom.
02:51:21iambenthat page is kinda confusing
02:51:26saratogaok that sounds reasonable
02:51:36scorche"ok thank you can you tell me if that subdirectory is a folder "
02:52:10saratogareading the whole thing, it sounds like they want Vorbis to have its own file extension specifically to help out hardware players, so i guess we're doing it right :)
02:52:11 Join chrisjs169__ [0] (
02:52:32Lloreansaratoga: Someone should tell they need to write clearer standards. :-P
02:52:41FebsYeah, what you guys can't see is the posted all in capital letters about "WHY IS NOBODY HELPING ME."
02:53:01FebsI can't respond any more, because my next response is to tell him that he needs to change his signature because it's wrong.
02:53:20LloreanFebs: It was very, very hard for me to resist saying something similar.
02:53:43LloreanSuch as "this should be considered basic computer knowledge and is something one should learn while becoming good on a computer"
02:53:48scorcheoh, fine...i wil be the bad guiy and say what has to be said
02:53:53saratogawe should just word filter "DOOM" to "GET OUT NOW"
02:54:23tihoc4nAbout .oga, I see the flaws in that page now, We should wait for more definitive information from them.
02:55:59iambenseems to me like they want ogg speex as .spx, ogg vorbis (I) as .ogg, and all other ogg audio as .oga
02:56:26 Quit JoeBorn ("Leaving")
02:56:28iambenbut their justifications for .spx & .ogg are "for compatibility with hardware players"
02:56:28scorchehrm...perhaps version is a little ridiculing
02:56:44Lloreaniamben: I think .oga will be allowed to include speex. It mentions .spx simply being preserved for compatibility.
02:56:56 Quit aliask_uni ("CGI:IRC")
02:57:00LloreanBut I'm a little uncertain about that
02:57:32*Llorean wonders how many hardware players besides those running Rockbox support Speex anyway.
02:57:32iambensame w/ vorbis really
02:57:45 Quit chrisjs169 (Read error: 110 (Connection timed out))
02:58:01iambenso .oga is ANY ogg audio, .ogg = vorbis only, .spx = speex only?
02:58:15Lloreaniamben: Well it mentions excluding Vorbis-I from .oga, I think
02:58:28iambenits all very unclear =)
02:58:31LloreanThough it may mean "excluding everything else from .ogg", I'm not wholly sure
03:01:08iambenit says the same thing about .ogg = vorbis being kept for back-compatibility
03:01:34iambenim used to mplayer where i can name my files clownpenis.fart and it'll figure it out, whether its ogg vorbis or avi/xvid
03:01:42zewhy can't stuff that cares just look in it and see whats there
03:02:07zenobody tries to make avi files .avd or somejunk to indicate that its divx in thre
03:02:32Lloreanze: Yes, but how many audio-only avi files do you have?
03:02:54LloreanThey're trying to set up a naming scheme for a more universal container, so that users can glance at a filename and have a reasonable idea what to accept
03:03:00zewhy do filesystems have to still be operating on increasingly antiquated paradigms necessitating this silly old "file extension" concept thats getting broken all over the place now
03:03:18LloreanOr so that Rockbox can populate a playlist with *just* music files, and not 5000 oggs containing text and videos that need to be opened and rejected before the next song is buffered
03:03:33zeLlorean: generally i have a pretty good idea what to expect from the file Name
03:03:40Lloreanze: Any metadata can be wrong. What makes a file extension any better or worse than *any* value a human is allowed to edit?
03:03:40zeand yeah that last thing i guess is a problem
03:04:07zebut a problem with data access schemes that nobody's getting past :/
03:04:16zeLlorean: ok, so?
03:04:28Lloreanze: Well, what *exactly* is your complaint against extensions?
03:04:35LloreanThe file "bob" tells me a lot less than "bob.ogm"
03:04:56 Join jurrie [0] (
03:05:02zethat they're effectively a metadata field that's rather hacky and no longer adequete
03:05:18zebut persisting because there's nothing but filesystems still that don't provide for anything else
03:05:37LloreanHow is it "inadequate"?
03:05:43iambenit blew my mind the first time i saw a file extension with more than 3 letters... "can they do that?" i thought
03:06:07zeLlorean: i thought ogg was handily demonstrating how, thus this discussion :p
03:06:26Lloreanze: I don't think extensions are inadequate.
03:06:43LloreanThey provide an easy, user-visible way of tagging a file type through any interface that allows renaming the file, essentially
03:06:53iambenLlorean: i think at this point he is just expressing his disgust for the file extension trend in modern computing, not anything that can/should be handled differently in rb =)
03:07:02Lloreaniamben: And I'm talking about in general to.
03:07:11LloreanWhat makes *any* sort of metadata more adequate than "a file extension"?
03:07:34zewell just look at the id3db style stuff on mp3 players
03:07:37zeincluding rockbox
03:07:43zei kinda thought it was lame at first
03:07:46zebut after getting used to it
03:07:55zei like having multiple ways of accessing data like that
03:08:13zeand things being innately capable of being tagged arbitrarily
03:08:18zeand accessed via such
03:08:42LloreanYou're talking about a semantic filesystem
03:08:47zeand now it starts to look like directories, filenames, and file extensions are just kludgly little stop-gaps on the way to that
03:08:51iambenLlorean: one problem is that we tend to limit file extensions to 3 letters, rather than goodsong.oggvorbis and coolvid.oggtheora, we are stuck w/ .oga & .ogv
03:08:57zeLlorean: is that what its called?
03:09:02LloreanBut even in such a case, there would be a "type" and it would describe what "type" of file it is, and it would be exactly the same concept as an extension.
03:09:24Lloreanze: It's what *I'd* call it, and a quick google provides that there's some people who agree with me.
03:09:32zeLlorean: it'd be silly to put such an arbitrary restriction on it as that
03:09:36iambenthe fact that we have to cram many different filetypes into one extension is a problem
03:09:44Lloreanze: I'm not saying it would be a restriction.
03:10:04zeit could be an "av" type and "divx" type and "ogg" type all the same time
03:10:13zeand maybe thats not what you mean by "type" but does that matter?
03:10:14Lloreanze: I'm saying that there can assumed to be "a text string describing the file type" and that it will be analogous to "file extension" in what it describes, because it does serve a purpose
03:10:25 Quit spiorf ("Read error: 110 (Connection timed out)")
03:10:35 Part maffe
03:10:39Lloreanze: Is there something that prevents you from naming a file blah.av.divx.ogg?
03:10:53zejust that nothing really understands that
03:10:56 Quit chrisjs169_ (Connection timed out)
03:10:58zeexcept a human
03:11:02 Join chrisjs169_ [0] (
03:11:07tihoc4ntar.gz is a real world example :p
03:11:14Lloreantar.gz is a very good real world example
03:11:15iambenand the fact that you can't filter out those needless parts when you just wanna see "blah"
03:11:16zetihoc4n: good call
03:11:32Lloreanze: Your complaint is that "file extensions are an outdated concept"
03:11:40zeLlorean: no
03:11:41LloreanMy point is that they contain exactly as much information as we choose to give them.
03:12:03zeLlorean: not so much an oudated concept per se i guess
03:12:12Lloreanze: You called it an "antiquated paradigm" and described them as "inadequate"
03:12:27LloreanMy point is that they're merely metadata, just as any other metadata we could store in any other way would be.
03:12:28iambenbut we can only fit so much info in them, and you can't sort it out
03:12:32zein so far as the directory/filename/file-extension structure
03:12:32LloreanThey are 1s and 0s on an HD.
03:12:44LloreanThe problem then is not with what they are, but how programs choose to handle them
03:12:58zei'm calling that structure an antiqueated and inadequate metadata system
03:13:35safetydanze, it's a non-trivial problem to solve
03:13:39*safetydan points to the delays on WinFS
03:13:45Lloreanze: It may be, for human use. But it's also simple if you're actually aware of your own organizational structure, and efficient to access.
03:13:58LloreanA semantic file system is much closer to how humans thing.
03:14:09Llorean"What was that green, round thing? Computer, get me all the green, round things I've got."
03:14:16Lloreanthink, not thing
03:15:02zeseems also nicer for programs not to have to traverse a hierarchy seeking something too
03:15:13zebut yeah
03:15:31 Quit jgarvey ("Leaving")
03:15:31saratogai don't see how you're going to avoid traversing a hieracrhy short of only having one file type
03:15:38saratogasooner or later you have to make a decision
03:15:49zeer well
03:15:49Lloreanze: How on earth would you seek without traversing a hierarchy, without specific information in advance? And if you have that specific information, why do you need to traverse?
03:16:10LloreanIf you say "Find me all balls", somewhere a list of balls has to be kept
03:16:25LloreanAnalogous to a folder named "balls", with either files, or links to files, in it.
03:17:05zeand i could have a directory named balls that's full of symlinks to all the ball files on my system
03:17:31zeand a directory called heavy objects thats full of symlinks to all the heavy object files on the system
03:17:41zewhich would have overlap, with heavy balls
03:17:46LloreanAnd something just checks for heavy balls by checking for files present in both, yes.
03:17:49zeand tons and tons of such directories
03:17:51zeand symlinks
03:18:04LloreanWhat do you think a symantic filesystem would look like under the hood?
03:18:10LloreanIt'd be a database with attributes.
03:18:13zeand redundancy to make a filesystem into flexible a metadata type deal
03:18:15zeand a huge mess
03:18:25LloreanSomeone has to label all balls as "balls", one way or another.
03:19:14iambenoff this topic, but on topic, rockbox only handles vfat filesystems right?
03:19:15zei could have all that or a db type deal instead
03:19:20qwmdoes rockbox support speex?
03:19:25Lloreaniamben: Yes.
03:19:41safetydanqwm, yes but not stereo
03:19:42Lloreanze: Yes, but assuming properly written software, you wouldn't be able to tell the difference anyway
03:19:48safetydanstereo speex files that is
03:19:54scorcheyou speak!
03:20:01qwmsafetydan: gonna use it for an audiobook, so i'm fine without stereo. heh
03:20:05iambenscorche: you mean he speex
03:20:11zeLlorean: *shrug* i'm not sure about that
03:20:14*scorche shudders
03:20:27qwmscorche: yeah, i came here to complain about how rockbox is not playing music as well on the nano as it used to.
03:20:31zebut i probably shouldn't be throwing the chan OT anymore
03:20:38qwmlots of glitches and speed variations since i upgraded.
03:20:50scorcheqwm: i havent noticed any difference
03:21:03Lloreanqwm: You probably have one of the mysteriously problematic Nanos
03:21:15qwmbut why has it worked fine before!
03:21:22Lloreanqwm: There seems to be a subset of Nanos (ie: those not owned by me or Scorche) that have IDE timing issues since some clock changes
03:21:37qwmhow frustrating.
03:21:49qwmLlorean: wanna trade nanos?
03:22:00saratogaqwm: don't update until someone figures out what the problem is
03:22:07scorcheespecially that none of the devs/helpers experience the issue
03:22:10saratogaor better yet, figure out what the problem is
03:22:12qwmthat's too late now..
03:22:15Lloreanqwm: Check the bug tracker, there's a related task (mentions firmware 1.3.1, but this is wrong)
03:22:35Lloreanqwm: If you want it fixed, convince amiconn to let you mail it to him. :-P
03:22:56scorcheqwm: actually, i wouldnt mind trading nanos to fix this...but as he said, amiconn is the person to go to
03:22:59 Quit Thundercloud (Remote closed the connection)
03:23:54qwmwhere can i find old builds?
03:24:10scorchecheck them out from avn and build
03:24:17scorchealthough there are other ways...
03:24:36krazykitlike the archived daily builds
03:25:04scorcheif you want older than that, you will ahve to go to svn
03:25:48saratogaanyone know how i seek in the gigabeat sim?
03:25:50qwmi'll give the oldest one there a shot.
03:26:22saratogaqwm: I think someone said the cutoff for the nano issue was around the 26th of last month
03:26:41qwmah, cool.
03:26:50qwmfound the first build i ever tried on my machine:
03:28:15saratogaactually, does anyone know how to seek in any of the sims?
03:28:27saratogai don't really care what the target is, so long as it can run wma
03:28:54 Quit donutman25 ("ChatZilla [Firefox]")
03:29:03 Quit chrisjs169__ (Connection timed out)
03:29:15 Join chrisjs169__ [0] (
03:30:28krazykitsaratoga, i thought you just held "right" on the keyboard, but that may be wrong
03:31:47 Part tihoc4n
03:32:58saratogathanks, didn't realize it needed to be held
03:33:44 Join maffe [0] (
03:39:13qwmsaratoga: tried the july 24th build, seems to be working fine, thanks. :)
03:42:36 Quit chrisjs169_ (Read error: 110 (Connection timed out))
03:42:46 Join chrisjs169_ [0] (
03:56:24***Saving seen data "./dancer.seen"
03:59:30 Quit chrisjs169__ (Read error: 110 (Connection timed out))
03:59:41 Join chrisjs169__ [0] (
04:03:25 Quit XavierGr ("One firmware to rule them all!")
04:05:24 Quit miepchen^schlaf (Read error: 104 (Connection reset by peer))
04:06:00 Join miepchen^schlaf [0] (
04:07:40 Quit scorche (Read error: 110 (Connection timed out))
04:10:11 Quit chrisjs169_ (Read error: 110 (Connection timed out))
04:10:11 Join annulus_ [0] (
04:14:22 Join elinenbe [0] (
04:15:40 Quit dandin1 ()
04:16:33 Join Soul-Slayer [0] (n=Administ@
04:25:59 Quit atsea- (Remote closed the connection)
04:30:40 Join atsea- [0] (i=ariel@gateway/tor/x-c76393063279a79c)
04:31:08 Quit saratoga ("CGI:IRC (Ping timeout)")
04:35:26 Join perrikwp__ [0] (n=chatzill@
04:35:28 Nick perrikwp__ is now known as perrikwp (n=chatzill@
04:50:12*Llorean is somewhat irritated at all the blind users' protestations regarding the voice change.
04:50:24LloreanI mean, I understand being upset by it, but they seem awfully demanding all of a sudden.
04:53:58 Join toffe82 [0] (n=chatzill@
04:54:05 Quit perrikwp_ (Read error: 110 (Connection timed out))
04:55:48*jhMikeS finds it irritating as well and a poor descision when a much simpler design could be done and save even more memory...but oh well, I guess it's entrenched now. :\
04:56:34LloreanjhMikeS: Simpler design?
04:56:55LloreanI don't think anyone would really object to more improvements.
04:57:22jhMikeSyes. I talked about it with Bagder. No target-specific voice files which I think are going to be a major headache.
04:57:58LloreanAfter my first generation of the voice files, it takes me less than 2 minutes to generate them for 5 targets.
04:58:16jhMikeSyou either have to make them or wait for them instead of just having one voice/lang file you load the wanted bits from.
04:58:16LloreanAnd that was, "after my first generation of one voice file", now that the pool is seeded for all of them, it's probably even quicker.
04:58:33LloreanjhMikeS: How do you handle the "voice file is way too big for Archoses" issue then?
04:58:48LloreanJust keep the big file, and only store the appropriate bits in RAM?
04:58:49jhMikeSseems like more trouble than what writing just a few lines of code could make unnescessary
04:59:56jhMikeSonly load what's used and there would be no usused index entries nor order dependence and strings could be deleted arbitrarily when no longer needed by using ids that never change
05:00:22smoo__Hi, this might be a really retarded question but how does one go about getting the firmware for something like an Xclef MT-500 so that I can reverse it? Is hardware used to take it from the eprom? Or is it just a matter of unpacking a firmware update?
05:01:38 Quit weoh (Read error: 110 (Connection timed out))
05:01:55LloreanjhMikeS: Seems pretty reasonable, though with the amount of space I have free on my flash players, I'd really prefer "slightly more complicated to make, but smaller" considering that it can be automated.
05:02:17jhMikeSwhen adding a new string: id = next_id, next_id = id + 1. id will never be used for another string again.
05:04:42jhMikeSA compromise solution could be to still use the ids in order to save more array space. I think archos still has some 300 NULL entries in the table.
05:06:47jhMikeSheck it could even load a master or stripped-down voice file the same way, leaving it up to the user
05:07:09LloreanThat would be a good compromise.
05:07:38LloreanWouldn't that result in a lot of ifdeffing in the code for different talk ids though?
05:09:35LloreanDo current voice files *have* to be target specific? I don't know how it worked, but I assumed the voiced bit was determined by the language string loaded.
05:09:36jhMikeSeach table entry would be char * str[index] = { "string",... }. parallel to str: short[index] = { ID }. strings would be sorted by ID.
05:10:02*Llorean doesn't have any clue how it works, clearly
05:10:19jhMikeSMy understanding is that they do.
05:10:34LloreanThen could a voice file containing all strings work for all targets (that can load it)?
05:10:42 Quit miepchen^schlaf (Read error: 110 (Connection timed out))
05:10:57jhMikeSin SVN or my idea?
05:11:02LloreanIn SVN.
05:11:03 Join miepchen^schlaf [0] (
05:11:13LloreanSince my understanding is your idea is supposed to explicitly allow that.
05:12:47jhMikeSMy understanding is no, but I don't see why there'd be gaps in the table then.
05:13:54jhMikeSthey seem to be gone so I guess that forces them as target-specific
05:14:34jhMikeSdon't know why Bagder implied the NULLs were still there or I misunderstood something
05:15:24LloreanMaybe he meant they were there during the generation, but didn't end up in the final file?
05:16:00jhMikeSno idea now...hehe
05:18:34LloreanIs there some special difficulty that prevents someone from updating the windows script for the new voice format, or is it merely a case of "nobody's around to do it"?
05:18:34 Part maffe
05:18:41 Join maffe [0] (
05:18:43*jhMikeS too preoccupied with the cache problem that gives the C0EDBABEs to try to tackle that problem too
05:19:47LloreanInterestingly enough, C0EDBABE doesn't show up in Wikipedia's article on magic numbers
05:20:06LloreanThough there's a C0DEDBAD, which I rather like.
05:20:13jhMikeSnor does 0xba90f5417 or 0xcacad0d0
05:20:58*jhMikeS forgot the BOS majic number :(
05:21:29LloreanjhMikeS: Did you write the SPC patch to use the COP?
05:21:43LloreanIs that the plan for other codecs too, just leave it entirely up to them how to use the COP?
05:23:00jhMikeSI think they should be able to determine how they make threads even if normally on COP. I'm still waiting for PP5020 test for the better dual-core kernel patch I just made.
05:24:44LloreanAnd what's this about a new dual-core kernel?
05:24:55jhMikeSIt's far better than the whole core lock thing since the processors remain almost completely independent.
05:25:26LloreanIs it functioning so far on any targets?
05:25:30jhMikeSI can send messages core-to-core just fine with my test plugin. Locking is very localized for specific purposes.
05:25:44jhMikeSe200, 5g it's been checked
05:26:05LloreanWell, that covers what I can offer testing on.
05:26:14LloreanWell, Nano, but that ought to be covered by 5G.
05:27:06jhMikeSThere was a problem using backlight fading after using mpegplayer on 5g but I'm not sure it's the patch. There's yet another place that needs a cache flush there.
05:27:46LloreanThe backlight always seems to be problematic in one place or another.
05:28:37jhMikeSI don't think I'd not commit for that since turning off fading had no problems after mpegplayer.
05:29:39LloreanJust claim it's a bug with backlight code and beyond the scope of the patch. :-P
05:31:02jhMikeSI believe it's just a cache issue or that call to mpeg2_close on the main thread without an invalidate first messing things up.
05:37:01 Join scorche [0] (i=Blah@rockbox/administrator/scorche)
05:37:51jhMikeSthe patch is basically a non-change from how things are now. the way threads are woken by IRQs is just generalized to doing it accross cores. it does make mutexes recursive and adds semaphores and events. the last two are used to make codec swaps perfectly atomic.
05:40:04jhMikeSevents might come in handy for the USB driver as well
05:42:14LloreanSo it's just PP5020 testing that's really holding it back?
05:46:07jhMikeSpp5020 testing is holding a bunch of stuff back
05:47:07jhMikeSweren't H10's around for < $90?
05:47:33Soul-SlayerI have an H10 if you need me to do anything?
05:47:46alienbiker99i can test h10
05:47:56jhMikeSSould-Slayer: if you could test a you up?
05:48:25Soul-SlayerSure could. Give me a moment to reboot and I shall
05:48:33 Quit Soul-Slayer (Read error: 104 (Connection reset by peer))
05:49:42jhMikeSpractically everything I looked at on ebay went for like $120
05:50:05alienbiker99they were on a wootoff a while ago for like $25 or less i think
05:50:08LloreanI think they were $90 on Woot, which then ends up with a bunch of them being resold for 120 on ebay
05:50:23alienbiker99oh were they really 90?
05:50:30LloreanWell, they've been on Woot a few times
05:51:00 Join Soul-Slayer [0] (n=jonno@
05:52:48Soul-SlayerJust put it through general use?
05:53:10jhMikeS1) Should boot 2) should play, even with voice 3) mpegplayer if possible ... mostly I'm concerned about the core of course not minor post-mpegplayer glitches
05:53:36jhMikeSIt will need a full update of everything
05:54:09Soul-SlayerBootloader too?
05:54:14jhMikeSoh, no
05:54:24LloreanjhMikeS: Are codecs put one the second core, or is that outside the scope of this?
05:54:29Soul-SlayerAm making from scratch so should be fine.
05:55:05jhMikeSLlorean: this only enables that to be done. one step at a time :)
05:55:42jhMikeSSoul-Slayer: thanks for the tests
05:55:59LloreanjhMikeS: I assumed that would be the case, but I wasn't sure what you mean by "should play, even with voice"
05:56:28***Saving seen data "./dancer.seen"
05:57:01jhMikeSit uses new kernel objects for the syncronizing of swapping but I've had no problems and they've never panicf'ed (there's code enabled to check them right now).
05:58:47 Quit scorche (Connection reset by peer)
05:58:55 Join scorche` [0] (i=Blah@rockbox/administrator/scorche)
05:59:27jhMikeSmutexes being recursive is incompatible with SVN playback.c use
06:01:42Soul-SlayerRight lets see.
06:01:46jhMikeSthe 5020 tests are the most harrowing
06:01:47Soul-SlayerBooted fine
06:02:01jhMikeSscore 1?
06:02:37Soul-SlayerHowever, trying to open a file I'm just getting 'Loading...'
06:02:50Soul-SlayerDisk is spinning but no clicking sound
06:02:58LloreanjhMikeS: In theory is there any way your changes could cause the wrong voice clips to playback, or should I assume this voicefile is broken, and simply check for that later?
06:03:09*Llorean would like to assume the voice file is simply bad
06:03:24*Soul-Slayer reboots and tries again
06:03:28jhMikeSLlorean: I've checked with a new voice file and had no difficulties
06:03:43LloreanI'll assume my ipod-nano specific voice aint quite right
06:04:13jhMikeSthough a voice-free check should be done
06:04:20Lloreanvoice-free check?
06:04:35jhMikeSvoice off and then no voice file at all just to compare
06:04:56LloreanWhat am I comparing in that situation?
06:05:15jhMikeSvoice off = no codec swap. no voice file = no codec loaded at all
06:05:39smoo__window 5
06:05:45smoo__oops sorry
06:05:59Soul-SlayerjhMikeS: Playback seems to fail. I've tried three times, twice it freezes up on the 'Loading...' splash, once it played, I paused it and it refused to resume.
06:06:01LloreanWhat I'm saying is happening is that when I highlight the top line in the system menu, it says "Warn when deleting dynamic playlist"
06:06:36jhMikeSSoul-Slayer: voice present or not?
06:07:27Soul-SlayerMy settings are reset to default by the way.
06:07:28jhMikeSLlorean: sounds like a wrong clip
06:07:45LloreanYeah, there's something wrong with the voice file
06:08:03Soul-SlayerNow it's playing the song, 5th attempt
06:08:09Soul-SlayerI'll see if skipping and so on works
06:08:10jhMikeSSoul-Slayer: so there's no voice file at all
06:08:16Soul-SlayerNot a trace
06:08:30Soul-SlayerRight, IF I can play a song, it functions as expected
06:08:44Soul-SlayerBut it seems to freeze up on the 'Loading...' splash most of the time
06:09:10Soul-SlayerAlthough, now it's letting me choose songs freely..
06:09:24jhMikeSplayback.c is so loaded with timing issues that the problem probably isn't the patch itself
06:10:12Soul-SlayerWell, the player is now responding as expected...
06:10:20Soul-SlayerUntil I restart it I presume
06:10:21jhMikeSwell, this is farther than any dual core patch has ever gotten
06:10:47Soul-SlayerRight, it just randomly froze up on me
06:11:10jhMikeSafter restart?
06:11:20 Quit scorche` (Read error: 104 (Connection reset by peer))
06:11:22Soul-SlayerJust during playback
06:12:06 Join scorche [0] (i=Blah@rockbox/administrator/scorche)
06:12:17Soul-SlayerRestarted, playback is working again
06:15:58LloreanBagder: There seems to be something wrong with at least the Nano-specific voice file, several clips are wrong
06:16:38Soul-SlayerjhMikeS: Hasn't crashed yet... But this is the 6th time I've restarted it
06:17:17jhMikeSSoul-Slayer: well, that actually holds alot of promise. were you able to try mpegplayer?
06:17:47Soul-SlayerWhat would I be looking for?
06:17:52Soul-SlayerWhether it actually works?
06:18:16Soul-SlayerLet me find a test file, bear with me :p
06:18:39jhMikeSall the experience ever was before was "freezes at boot screen"
06:18:57Soul-SlayerIt definitely boots okay :p copy of the site that i was working on is gone...
06:19:30*scorche sends off a mail to redbreva
06:20:36Soul-SlayerJust downloading Elephants Dream... 5 mins or so and I'll be with you.
06:20:44Soul-SlayerDo the other targets you tested on crash?
06:24:30jhMikeSSoul-Slayer: there was a backlight glitch after mpegplayer on 5g. e200 has no problems at all.
06:25:24jhMikeSthe backlight glitch was only when fading was enabled
06:26:24Soul-SlayerjhMikeS: Just about to test mpegplayer
06:27:03Soul-SlayerPlays fine
06:27:04*jhMikeS puts claw marks in the armrests
06:27:28jhMikeSlet it go a couple for a bit
06:27:29Soul-SlayerOpens the file fine, quits fine, backlight seems fine afterwards
06:28:00jhMikeSthe best test before was "plays a few seconds then crashes"
06:28:30Soul-SlayerI wish this movie made sense... :p
06:29:11 Join chrisjs169_ [0] (
06:29:30jhMikeSit's about new ideas vs. old, entrenched ones I think
06:30:19Soul-SlayerShall have to get the proper version some time that isn't on a 128x128 screen and work it all out :p. But the fact remains it's still playing 3 mins or so later...
06:31:30jhMikeSI think I'll have to chalk the playback stuff up to timing changes introduced. mpegplayer is quite atomic thread wise.
06:32:24LloreanSoul-Slayer: But playback is consistently playing now, or do you still have to try a bunch of times like you did right after starting the build?
06:32:46Soul-SlayerLlorean: Let me see, still in the video at the moment. No crash yet though.
06:32:50*Soul-Slayer quits
06:33:04Soul-SlayerPlayback froze it.
06:33:12Soul-SlayerStuck on the loading splash again
06:34:03Soul-SlayerRestarted and worked on playback first time
06:34:19LloreanCan you play a movie, stop, then play another movie?
06:34:20jhMikeSLlorean: not as much as before actually ... it's localized and perhaps traceable
06:34:41 Quit toffe82 (Read error: 110 (Connection timed out))
06:34:44Soul-SlayerCurrently playing a song, shall go straight into mpegplayer, try playing another song, then mpeg again.
06:34:46*jhMikeS also wonders about frequency scaling
06:35:13LloreanWhat about frequency scaling?
06:35:24Soul-SlayerPlayback -> Mpeg = Data abort at 0006C092
06:35:28Soul-SlayerBut the sound is still playing
06:35:36Soul-Slayer(Sound of the video, not playback)
06:35:53jhMikeSthat one's a rather unrelated caching difficulty
06:36:20Soul-SlayerRestarted, again, playback was fine
06:36:20jhMikeShappens in SVN but usually at C0EDBABE
06:36:32Soul-SlayerSeems aslong as I'm quick going into playback it will work
06:36:53Soul-SlayerPlayback -> Mpeg worked this time
06:36:54jhMikeStry setting some extra boost counts in the debug menu and try
06:37:10Soul-SlayerTry what?
06:37:19LloreanSoul-Slayer: out of curiosity, can you feel the disk spin down/up fairly well?
06:37:34Soul-SlayerI can't feel, but I can put my ear to it and hear it
06:37:55Soul-SlayerMakes quite an audible high pitch when spinning, and clicks when accessing
06:37:59Soul-SlayerAnd it just froze on me
06:38:27LloreanI was just wondering if the point where music won't start might align with the point where the disk spins down.
06:39:04Soul-SlayerLlorean: When accessing a song, I'll hear the disk spin up, the splash stays on 'Loading' and the disk continues spinning until I remove the battery
06:39:22Soul-SlayerjhMikeS: What boost count shall I try and what do you want me to try? Playback?
06:39:50jhMikeSSoul-Slayer: just add a whole bunch so it can't unboost
06:39:59jhMikeSand yeah, playback
06:40:02LloreanSoul-Slayer: I meant, during boot the disk spins. Then it spins down shortly after. I was curious if the freeze happens if you let it get to that spindown, or if it's unrelated.
06:40:26jhMikeScould increase the spindown time so it can't for awhile
06:40:30Soul-SlayerLlorean: Ooh, understood, I'll check momentarily
06:40:37Soul-SlayerTrying playback boosted now
06:40:59LloreanjhMikeS: It seems unlikely that it's related to spindown, but I'm a curious person.
06:41:08Soul-SlayerI'll wait for disk spindown and do something else
06:41:31jhMikeSLlorean: me too but the craziest things can be related sometimes.
06:41:52LloreanOr at least provide random clues
06:42:07*Llorean still wants to know what makes the crashing Nanos different.
06:42:16Soul-SlayerSkipping tracks is working ... But like I said, the freeze was very intermittent before, I am not sure how to provoke it
06:43:39jhMikeSLlorean: you ran the patch I take it? you just had a bum voice file?
06:44:02LloreanjhMikeS: Yeah, just a bum voice file
06:44:10LloreanI'm listening to music on a patched Nano right now.
06:44:27*jhMikeS wonders why sansa is such a stable pp target compared to others
06:44:31LloreanI built the voice under the same conditions, even using pooled mp3s, for my Gigabeat and the voice file was good.
06:44:43Soul-SlayerjhMikeS: Does sansa support frequency scaling?
06:44:44LloreanThe Nano's been darn stable until the 80mhz change.
06:45:25jhMikeSSoul-Slayer: yes, but it was never a problem for it except for the FIFO refills during cache flushes causing click when not boosted
06:45:46 Quit chrisjs169__ (Read error: 110 (Connection timed out))
06:45:50Soul-SlayerWell now I've got my H10 boosted it doesn't appear to be crashing
06:46:05Soul-SlayerI'll restart, boost it and see if it crashes, and try it a few times
06:46:06*jhMikeS sees amiconn has more work to do :)
06:47:03Soul-Slayer2nd attempt, boosted it up, went into a song, no problems... Rebooting again and trying
06:47:04perrikwpi was following the current conversation and decide to help test the dualcore patch
06:47:14LloreanjhMikeS: It's in theory an IDE timing issue, and may affect iPod Videos as well
06:47:26 Join chrisjs169__ [0] (
06:47:29perrikwpi have a ipod mini 1gen by the way
06:47:34jhMikeSLlorean: lots of things in RB are not mutexed btw
06:47:49Soul-SlayerjhMikeS: Really seems to be boost related then... 3rd attempt no problem...
06:47:52jhMikeSperrikwp: the more the merrier. did you see the link?
06:48:24LloreanjhMikeS: Was that a "we need more mutexes"?
06:48:29perrikwpi already compiled the build and audio playback works perfect
06:48:38jhMikeSLlorean: quite a few more
06:48:45Soul-Slayer4th no problem... I think we can rest assured it's related to the boosting
06:49:04jhMikeSperrikwp: cool. thanks for checking it.
06:49:28Soul-SlayerYep... 5th... Now I'll try again without the boosting
06:50:09Soul-SlayerI'll test the spindown theory too
06:50:19perrikwpthere is a problem with mpegplayer though, but i don't know if it matters because it is a grayscale target
06:50:31perrikwpthere is audio but no video
06:50:42Soul-SlayerLlorean: Good assumption. I allowed the disk to stop spinning, and yep, it's crashed.
06:50:55jhMikeSperrikwp: hmmm...might be graylib related
06:51:01Soul-SlayerDoes it unboost once it spins down the disk or something?
06:51:41jhMikeSSoul-Slayer: it will be unboosted, then playback will boost and spin the disk pretty much at once
06:52:17Soul-SlayerWell, I just allowed it to spin down, then put up the boost_counter and tried playback, worked first time
06:52:24 Quit safetydan ("CGI:IRC (Ping timeout)")
06:52:40jhMikeSso they combine in an evil conspiracy ... hrm
06:53:14Soul-SlayerIf I don't boost it before trying to open a file, it freezes
06:53:21Soul-SlayerUnless the disk hasn't spun down yet
06:53:28jhMikeSbut only if the disk stops spinning, right?
06:54:45jhMikeSyeah, with sansa, none of that is an issue of course
06:54:55Soul-SlayerYeah, no disk to spin up :P
06:55:23 Quit chrisjs169_ (Read error: 110 (Connection timed out))
06:55:42LloreanDo we attempt disk power down on the H10s?
06:56:35jhMikeSI don't know but something in my some obscure memory says there were issues with ATA and scaling there
06:56:50jhMikeSamiconn should know for sure
06:56:51Soul-SlayerYou aren't kidding :p
06:58:20Soul-SlayerjhMikeS: A backlight issue seems pretty random...
06:58:37jhMikeSSoul-Slayer: you're having one?
06:58:50Soul-SlayerjhMikeS: No no, I was just thinking over the problems
06:59:00Soul-SlayerI can't see how the backlight got involved
06:59:59jhMikeSit shouldn't with nothing changing. the only difference really is that every kernel object has a core lock on every instance.
07:01:21 Join chrisjs169_ [0] (
07:01:58 Join toffe82 [0] (n=chatzill@
07:02:42Soul-SlayerAnyhow, I must sleep. Good job jhMikeS :P Hopefully amiconn will have all the answers ;)
07:04:08jhMikeSSoul-Slayer: thanks and goodnight :)
07:04:37 Quit Soul-Slayer ("Leaving.")
07:05:52LloreanThis patch gonna affect PP5002 targets?
07:07:44jhMikeSLlorean: they don't have dual core enabled at all yet so it'll just have the new kernel objects
07:09:21jhMikeSI don't expect any green deltas from this one either :P
07:10:08LloreanMan, I'm so glad I don't have an iPod Video. I just realized I'm listening at -66
07:10:34jhMikeSwhat's with the Video?
07:10:44LloreanOnly goes to -54 if I recall.
07:10:54LloreanI think it has the least range of our targets.
07:12:06jhMikeSsansa once had that beat
07:13:08jhMikeSmust be quiet over there though
07:13:16LloreanOh, yeah.
07:13:20LloreanI'm wearing open phones.
07:16:24 Join Dylan_ [0] (
07:16:34 Nick Dylan_ is now known as Adikdid (
07:16:43AdikdidHello, all
07:16:53Adikdidcan someone please help me to oput rockbox on my iPod?
07:17:09scorchewhat problem are you encountering?
07:17:20Adikdidhow to put it on my iPod
07:17:27scorchehave you checked the manual?
07:17:30Adikdidi have downloaded it to desktop, now what do i do?
07:17:39scorcheread the manual
07:17:57Adikdidu have read the and it just says
07:18:10scorchedid i say to read
07:19:16 Part Adikdid
07:21:28jhMikeSLlorean: after this there's a pcm rework that cleans up the whole API and puts a start to make it play from whatever core initiates it. that's needed so COP codecs and CPU plugins can both use it efficiently.
07:22:37*Llorean wonders once more why an embedded device was designed dual-core like this.
07:22:39jhMikeSof course pp5020 is holding that up from just going in SVN now
07:23:46jhMikeSsame sort of freeze issues but the tester wasn't as willing to follow through thoroughly so I never really narrowed it down
07:24:43jhMikeStwo really cheap cores instead of one expensive 200MHz one?
07:27:01LloreanBut wouldn't that then lead to like... the arm/dsp setups instead?
07:27:05*jhMikeS wonders when the day will come when there's 4,8 or 16 cores in a device ... probably not far off
07:27:08LloreanSomething less symmetrical.
07:27:38 Quit chrisjs169__ (Read error: 110 (Connection timed out))
07:27:48jhMikeSI think those exist already ... isn't the TMS320 one of those?
07:28:00LloreanThose make more sense to me.
07:28:12LloreanOne core for the UI, and another core more or less designed for the A/V jobs.
07:28:27jhMikeScoldfire with an emac unit beats the pants off 160MHz of ARM7
07:29:29LloreanAnd then you've got the ridiculous Gigabeat.
07:29:38 Join Adikdid [0] (
07:30:16LloreanUnfortunately I doubt I'll ever meet someone who actually knows why any of those interesting decisions was made
07:30:19jhMikeSyeah, and one core that plays video
07:31:03*jhMikeS puts his money on the bean counters
07:31:06LloreanjhMikeS: So, there's a patch in existence for the playback cleanup too, or is that more of a plan?
07:32:03jhMikeSThe MoB work to get buffering separated. It would be nice for that to be done and then getting playback.c in order would be much easier.
07:32:28LloreanYeah, I haven't seen Nico_P since I got back.
07:32:59*scorche wonders why the GSoC students seem to disappear
07:33:01LloreanEverything looked like it was progressing, though. I just don't know if there's 'recent' work
07:33:30jhMikeSI think there's a want for a free lunch in that. there is none. that sort of buffer is complex and any design decision just has conseques in the logic.
07:33:48 Quit Adikdid (Client Quit)
07:34:01amiconnOne "weak" cpu core + dsp isn't such an uncommon design
07:34:23Lloreanamiconn: That's what I was saying though, I'm curious why the choice of identical cores rather than that.
07:34:25jhMikeSarchos is the primest example
07:34:27amiconnTMS320 is just one example (arm+dsp)
07:34:54jhMikeScoldfire even
07:34:55LloreanI'm curious if Apple had a plan for the dual cores, or if they make use of them in some interesting way under the hood or anything.
07:35:17amiconnThere's also TCC720 (calmrisc16 + calmmac24) (the failed Gmini port attempt)
07:35:31jhMikeSI think Toshiba had to have more plans for the Gigabeat and just bailed out to get it to market. It's way overkill for the audio only.
07:35:31amiconn..and the ATJ2085 as an extreme example
07:36:04scorchejhMikeS: the 2nd gen nano is pretty beefy as well
07:36:26 Quit toffe82 (Read error: 110 (Connection timed out))
07:36:39amiconnThe archos is somewhat different, as the 2 cores are two separate chips
07:36:49jhMikeSamiconn: did you follow any of that patch testing on H10? no problems unless the disk spins down while not boosted.
07:37:03*amiconn just got up
07:37:16scorcheas well, i dont think the OF on those nanos even does video
07:37:30jhMikeSwhat's so beefy about the 2nd gen nano?
07:38:34LloreanI thought the second gen was supposed to be a little slower than the 1st. But I'll admit I haven't researched.
07:39:41scorchei think it is a 200 MHz ish arm 9
07:39:48jhMikeSPortalPlayer OF takes a big hit in responsiveness when music is playing. It's kinda odd given it has two cores.
07:40:05 Join ctaylorr_ [0] (
07:41:27amiconnjhMikeS: So the crash Soul-Slayer was observing was with your patch?
07:41:30LloreanjhMikeS: Which player?
07:42:00jhMikeSamiconn: H10, but in just the specific way I mentioned. otherwise no problems booting or running mpegplayer.
07:42:10amiconnIf it works boosted but not unboosted, it's certainly not ATA timing
07:42:26jhMikeSLlorean: I've seen it on iPod video and e200
07:42:48 Join Adikdid [0] (i=7cbac343@gateway/web/cgi-irc/
07:42:58jhMikeSIt worked unboosted as long as the disk didn't need to spin when starting playback
07:44:12AdikdidOK i installed rockbox but it says it has a red error, it says IPOD: File Is Missing
07:44:20 Quit ctaylorr (Read error: 110 (Connection timed out))
07:44:39jhMikeSthe only real difference from SVN is a test_and_set guarding the kernel objects
07:45:08Adikdidcan soneone please help me?
07:45:43scorcheyeah... 200MHz ARM949T and if i am reading this right, it has 256MB of SDRAM
07:46:23amiconnscorche: It's probably the same mistake the ipl people made with some other target
07:46:50amiconnThe RAM is very likely 256M_bit_ == 32Mbyte
07:47:09scorchesounds likely
07:47:36AdikdidHow do i find out what gen my iPod is?
07:48:30amiconnscorche: yep. 2 Mistakes in one section (1.1.2)
07:49:01amiconnGoogling quickly tells that K4M56163PG is a 4M * 16bit * 4 banks SDRAM, i.e. 32Mbyte
07:49:05scorcheamiconn: wouldnt surprise me...tis why i said it with a bit of caution above
07:49:54Adikdiddo i install rockbox to iPod_Control folder or a sub folder in iPoD_control
07:50:01Adikdidit doesnt say in the mahual
07:50:05scorcheyes it does
07:50:34amiconnjhMikeS: Your "only load what's needed" idea has one rather obvious problem - how does rockbox know which IDs _are_ needed?
07:50:49amiconnIt needs a list, and that increases its size
07:50:53Adikdidno scorche it just says extract rockbox to the root folder
07:51:03LloreanIt doesn't say "root folder"
07:51:08LloreanIt says to the root of the device.
07:51:09scorchewhich tells you exactly where
07:51:22LloreanThere's no mention of putting it in *any* folder.
07:51:47Adikdidroot of your player’s drive
07:51:47 Quit Adikdid ("CGI:IRC (EOF)")
07:52:03jhMikeSthat's true though the list is parallel to the index array and each ID item could be 16bits. There might be a way to group things so an ID isn't needed for everything.
07:52:22 Join Adikdid [0] (i=7cbac343@gateway/web/cgi-irc/
07:52:38 Join ackbahr [0] (
07:52:46amiconnThe index array is generated at runtime. The saving when introducing that was several KB
07:52:47Lloreanamiconn: Why not include some metadata in the voice file then? Headers for each string that are stripped during loading?
07:53:14Adikdidso i install it to K:\iPod_Control\Device
07:53:32scorchedid you read what i just linked you?
07:53:37amiconnLlorean: If there's a single .voice (and .lng) file, the core must know which are to be used, and that _is_ target specific, hence it needs to be built into the voice file
07:53:41scorcheat least the first sentence?
07:53:52 Join My_Sic [0] (
07:53:52amiconnErm, into the core
07:54:02Adikdidyes scorche
07:54:16AdikdidOn many Unices, there is also a directory which is named /root. Confusingly, it is not a root directory in the sense of this article, but rather the home directory of the Superuser (conventionally known as "root").
07:54:20Adikdidthat bit?
07:54:24scorche"first or top-most directory"
07:54:28 Quit ackbahr (Read error: 104 (Connection reset by peer))
07:54:56jhMikeSI also just became aware that the tables in the core are gap-free which is different than what I thought I knew.
07:54:58Adikdidso i just instal it to iPod_control then?
07:55:09scorchethat is not the top-most directory
07:55:09Lloreanamiconn: If there's a single voice file, couldn't it determine while loading from metadata in the voice file which lang strings not to load, though?
07:55:14LloreanEr, voice clips
07:55:23amiconnjhMikeS: I explained this multiple times...
07:55:26LloreanAdikdid: It doesn't go in *any* directory
07:55:46AdikdidOh, where do i install it too then?
07:55:47Adikdidim confused now
07:55:59LloreanAdikdid: The root of your iPod. Not in any directories.
07:56:23amiconnLlorean: hmm....
07:56:32***Saving seen data "./dancer.seen"
07:56:36Adikdidso do i just install it to /k?
07:56:55ddaltonso has anyone had a look at p7561? any suggestions on what needs to be added?
07:57:02jhMikeSamiconn: I suppose I was just hearing it all wrong at the time then.
07:57:03amiconnIn fact we could play cheap and just concatenate all voice files, making each core only load the chunk it needs
07:57:21Lloreanamiconn: That'd be a fairly large file, no?
07:57:27amiconnThat's in fact a modification of the idea how to support voice in plugins
07:57:54amiconnLlorean: Large for RAM? yes. Large for a disk? no
07:58:12amiconnBut I admit this might be a bit too cheap
07:58:17LloreanI dunno, a 15mb file on a 1gb player isn't ideal.
07:58:27amiconnStill less than one song
07:58:36amiconnErm, no
07:58:43*amiconn isn't 100% awake yet
07:58:43jhMikeSthough in talking this out some other ideas are congealing and no ID may be nescessary and full order independence preserved.
07:58:56Adikdidi got it!
07:59:21 Join pondlife [0] (
07:59:51amiconnLlorean: But at least better than trying to load it into 2MB ram ;)
08:00:02Lloreanamiconn: Very true, indeed. :)
08:00:20Adikdidwhy cant i get out of a photo when i look at one on my iPod
08:01:57Adikdidman this building database shit is gonna take ages
08:02:06amiconnWell, each clip/string could have associated flags which tell what targets it is used on
08:02:39amiconnBut that puts a limit on the number of supportable targets depending on how many bits we reserve
08:02:52LloreanThis is true.
08:03:35LloreanBut you can add more bits when necessary, anyway, since it's unlikely the file will never need to be rebuilt?
08:04:20Adikdidhey can soemone please tell me how to install themes into rockbox?
08:04:33amiconn32bit would be sufficient atm (we have 27 targets in configure atm, 3 of them are more or less dead)
08:05:35amiconnThis idea would require giving strings which how have different per-target content a separate ID
08:06:06*jhMikeS was thinking something similar. just an index array for the target in the header.
08:06:17amiconnHmm, but it leaves one important problem - the built-in language
08:07:38amiconnThere's another problem - even if the core then knows _which_ clips/strings to load, it still needs to know _where_ in memory to load them
08:08:15amiconnThat doesn't pose problems on any of the disk based targets, but it does on Ondio
08:08:40amiconnOn Ondio we don't load the whole voice file at once, because that'd be too slow
08:08:41jhMikeSon the old idea, the IDs would be sorted and a matching ID put into the slot at that matching key
08:08:41 Join webguest52 [0] (i=c023110a@gateway/web/cgi-irc/
08:08:57amiconnInstead we just load the index array, and clips are then loaded on demand
08:09:36amiconnHmm, that could still work if there's also a table of the clip sizes
08:10:05jhMikeSyou'd need that. I'd want most of the file to be able to be skipped and never loaded.
08:10:51nerochiarohi all, quick question: just wondering if is anyone aside of Linus working on avoiding the source tree appreading 3 times inside the archived daily builds ?
08:10:56amiconnIf we think about that, we should also keep in mind that plugins might also want to use voice
08:11:31 Quit My_Sic ("Miranda IM! Smaller, Faster, Easier.")
08:11:50amiconnIn fact that's not much different from the target selectivity
08:13:31amiconnAlso keep in mind that the file should be loadable without skipping back. Seeking is costly
08:13:48 Join chrisjs169__ [0] (
08:14:09amiconnSkipping forward is okay; should be faster than plain reading
08:14:48 Quit Adikdid ("CGI:IRC (EOF)")
08:15:38amiconnThe built-in language problem could be solved like it is now - by actually leaving out the strings which don't apply to the target
08:15:55amiconnOn disk they wouldn't be left out
08:16:22*jhMikeS just wants this little H10 disk glitch figured out before trying more big projects (oh, pcm rework :))) important those get in SVN soon.
08:16:40LloreanOf course another solution is just to have working TTS. ;)
08:16:55*amiconn wants to get the 1st/2nd gen support out the door
08:17:01amiconnLlorean: On archos?
08:17:04LloreanAh, true
08:17:10LloreanI was just kidding anyway
08:17:26LloreanVoice files should pretty much be kept on no matter what, as prerendered should always be able to sound better, right?
08:17:31amiconnAnd given how festival sounds compared to the voice files I used before, I would still like to have a choice
08:17:53LloreanHave you tried espeak or flite yet?
08:18:11jhMikeS*chant* death to codec swapping
08:18:16LloreanThough they still aren't near the class of AT&T
08:18:17GodEaterflite is just dreadful
08:18:17amiconnI just used the downloaded voice file; no time to fiddle with voice generation on linux
08:18:39LloreanGodEater: flite was better than festival I thought. But I prefer espeak right now
08:19:10 Join My_Sic [0] (
08:19:13GodEaterLlorean: didn't you think flite sounded like someone in slow motion ?
08:19:15*amiconn used a quite outdated at&t voice before.
08:19:29LloreanGodEater: The pitching was a bit funky
08:19:46amiconnThe festival voices are quire slow, even for me as a non-native english speaker
08:19:49GodEaterI much prefer espeak of the three
08:19:52*jhMikeS wonders when 1st/2nd gen support will be out the door enough so real dual core support can be realized _finally_.
08:20:21amiconnjhMikeS: I need to get my head around the backlight thread. I need a way to suspend backlight
08:20:26markunLlorean: I prefer to listen to espeak with -s 200 -p 20 (faster + lower pitch)
08:20:42Lloreanmarkun: 200wpm?
08:20:43markunbut maye that doesn't make sense for voice files
08:20:44amiconnI think I can simplify some things, but I didn't try it yet
08:20:51jhMikeSthe dual core support adds an object to make that possible
08:20:55LloreanOne of the voice files hosted in the wiki was 440wpm
08:21:16amiconnAs soon as I have suspend working, I want to publish a bootloader
08:21:47Lloreanmarkun: Perhaps we should poll the blind users and ask what they consider a reasonable average speed
08:21:48amiconnjhMikeS: I don't want to suspend the thread. It should still process messages in order to avoid queueing them up
08:21:57jhMikeSthe event object would be just the thing since it can be set to unsignalled and keep the thread waiting until it's signaled again
08:22:12markunLlorean: but also non blind people use it (like amiconn)
08:22:42Lloreanmarkun: Yes, we should probably pick on the lower end of whatever range they give.
08:22:43jhMikeSwell, then it can test the object and just discard messages
08:22:46LloreanBut it would help
08:22:51amiconnAnd if that bootloader is out, I need to research cache handling on PP5002. After that I want to bring dual core support on PP5002 on par with PP502x
08:23:34amiconnThat will probably the most tricky part
08:27:33jhMikeSI suppse a while (1) { do { queue_wait(...) } while (flag); switch(ev) {} }
08:28:38jhMikeSThe par should be real dual support PP502x though.
08:29:06 Quit w0rd54 (Read error: 104 (Connection reset by peer))
08:30:45 Quit chrisjs169_ (Read error: 110 (Connection timed out))
08:30:52 Join Rob2222 [0] (
08:31:06 Join TinoM [0] (
08:32:29jhMikeSamiconn: there is an odd problem with caching. it shows as those prefetch aborts using mpegplayer and they happen at the call to mpeg2_init. I can splash a message just before the call but a splash as the first thing in mpeg2_init is not seen.
08:32:56markunLlorean: are the daily voice files made with espeak?
08:34:58 Join aliask [0] (
08:35:06amiconnmarkun: With festival afaik
08:35:22GodEateryeah festival
08:35:38amiconnjhMikeS: I think dualcore support on PP5002 won't be different from PP502x once cache handling is figured out
08:36:27amiconnWe don't know yet how to flush or invalidate properly. But I already found a cache setup function (while looking for something completely different, as usual)
08:38:30amiconnjhMikeS: Do you think a SYS_SUSPEND message would make sense? I now think it does...
08:38:53amiconnAnd that message should be acknowledged, like USB
08:38:55jhMikeSwhy not? it could be handling much like SYS_USB_CONNECT
08:39:11jhMikeS*handled even
08:39:40amiconnAT least 2 threads must do something on suspend. Backlight must switch off backlight regardless of settings, and pcm playback must shut down the codec
08:39:47amiconn(audio codec I mean)
08:39:54jhMikeSthe sansa's cache setup values are not what are set in rb. i know that much.
08:40:17amiconnPerhaps some more cache RE is due also for PP502x?
08:41:01jhMikeSI think so. this error is too odd and no amount of flushing seems to solve it. maybe it does have some separate instruction lines.
08:41:55amiconnI doubt that, but perhaps flushing while running from cache doesn't flush everything?
08:42:16amiconnDid you try flushing from an iram function?
08:42:29jhMikeSnever but could try it
08:42:54amiconnI had a similar problem in my suspend hack - the suspend function sends the dram controller to sleep, but it won't go to sleep while running from dram
08:43:14amiconnNow that I understand how it works, I put the function into iram and it works
08:43:20 Join MySic [0] (
08:43:28amiconn(it's just the OF's asm function hacked to pieces atm)
08:45:37jhMikeSIf it weren't for the little H10 disk glitch I'd clean up and commit this core thing ASAP. I'll do the IRAM tester now though.
08:46:27 Join datasleep [0] (
08:46:28 Quit BobShield (Read error: 104 (Connection reset by peer))
08:46:53 Quit datachild (Nick collision from services.)
08:47:22 Join w0rd54 [0] (i=blackdev@
08:47:45amiconnjhMikeS: I would like to see it tested on other PP5020 targets. Unfortunately I also could test on H10 only
08:47:54amiconnlinuxstb could test on ipod color
08:48:58 Quit Rob222241 (Read error: 110 (Connection timed out))
08:48:59jhMikeSamiconn: well, maybe you'd notice something important and narrow it down.
08:50:44GodEateramiconn: do you still have LinusN's 5.5G ?
08:50:55GodEaterdo you use it at all ?
08:51:05amiconnNo, only for testing
08:51:24GodEaterhmm, then you've not likley noticed the issue I have
08:51:38GodEaterwhich is that when "off" it seems to randomly turn itself on again all by itself
08:51:46GodEaterI've noticed this a few times in the last couple of weeks
08:51:53 Join ender` [0] (
08:51:57 Quit webguest52 ("CGI:IRC (Ping timeout)")
08:52:02GodEaterwhich certainly doesn't help the battery life issue
08:52:06scorcheGodEater: with the hold button on?
08:52:13GodEatereven with the hold button on yes
08:52:21GodEateronly then of course it starts into Apple OS mode
08:52:31jhMikeSargh ... why am I getting "relocation truncated to fit" ? isn't that foced long?
08:52:59amiconnDoes the function have 'static' in front of it?
08:53:21jhMikeSyeah, just tried removing that
08:53:47amiconnUse STATICIRAM instead (which evaluates to nothing if the function must be called far)
08:53:52amiconnThis is a gcc bug
08:53:55LloreanGodEater: A friend of mine with a 5G mentioned similar behaviour
08:54:14GodEaterat first I though I'd not been shutting it down
08:54:16amiconnGodEater: Alarm setting?
08:54:22GodEateramiconn: I don't use it
08:54:29amiconnI never enabled that, and I also never saw that behaviour
08:54:57*GodEater goes to check to make doubly sure, but is pretty certain he's never even been into that menu
08:55:45GodEateryeah - definitely disabled
08:56:02amiconnjhMikeS: If you look at tata.c, you'll see several occurences of STATICIRAM. That's where I needed it for getting rid of sectioned compilation
08:56:58jhMikeSI saw those in the encoder codecs too
08:58:22amiconnAh, yes
08:58:35amiconnBut ata.c is the only place in the core so far
08:59:07 Join B4gder [0] (
09:00:05jhMikeSThis problem occurs after using one cop using plugin and then mpegplayer but not the other way around
09:00:53amiconnWhat other plugin uses the cop?
09:01:45jhMikeSa tester for this. it makes it happen. my patch itself does not cause this problem but running it first can make it happen every time.
09:01:47 Join enyc_ [0] (
09:01:59amiconnWell, the cop has its own cache
09:02:19jhMikeSall thread creation invalidates before running the thread
09:03:41amiconnWhen loading a plugin, both cores need to flush caches before loading the file into ram. While loading, the cop must not touch the plugin ram area
09:03:51jhMikeSlike I said, I can run anything I want in video_thread up to the call to mpeg2_init which sends it off on some address past the plugin buffer or does the C0EDBABE thing...or sometimes DEADBEEE
09:04:32jhMikeSIt doesn't. Threads on dual core are treated like entrypoints always.
09:04:35amiconnIf the cop's cache is flushed when the plugin is already loaded, it will overwrite the plugin if it had some plugin data cached before
09:04:58jhMikeSI'm flushing just before the thread is removed.
09:05:25amiconnThen it still runs in plugin ram, doesn't it?
09:05:34jhMikeSThere is a stall in the CPU thread to allow more than enough time for the thread to be gone into oblivion
09:06:13amiconnYes, but the COP's cache will still have some data cached
09:06:17 Quit My_Sic ("Miranda IM! Smaller, Faster, Easier.")
09:06:24jhMikeSit does and the caches are being invalidated
09:07:03jhMikeSummm, from where? it's flushed _after_ it reenters core code.
09:07:19 Quit enyc (Connection timed out)
09:08:29 Join Jon-Kha_ [0] (
09:09:22jhMikeSthe plugin will not have exited before this is complete. it waits for 1/10 second before exiting plugin_start after getting the signal from the COP thread that it is terminating. the next thing the COP thread does after setting the flag is to call remove_thread
09:09:36 Join enyc [0] (
09:10:34 Join LinusN [0] (i=linus@rockbox/developer/LinusN)
09:10:38 Quit Jon-Kha_ (Client Quit)
09:11:43*Llorean wonders how long will take under Msys and whether it will succeed.
09:12:06GodEaterwth is Msys ?
09:12:14Lloreanmingw32's shell.
09:12:21LloreanOr mingw rather
09:12:37LloreanI'm trying to see if it's possible, without too much insanity, to make voices with it.
09:12:40 Join petur [0] (n=petur@rockbox/developer/petur)
09:13:04 Join JajaComp [0] (i=d521d7aa@gateway/web/cgi-irc/
09:13:11 Quit aliask (Remote closed the connection)
09:13:22LloreanIt's very slow. So very, very slow.
09:13:31JajaCompHow I can make rockmox sources?
09:13:38amiconnLlorean: Try sfu perhaps?
09:13:55Lloreanamiconn: Oooh, forgot all about that
09:13:56*GodEater wonders if it's possible to run it under Microsoft's Services For Unix
09:14:01B4gderJajaComp: do you mean build them?
09:14:01Lloreanamiconn: I'll finish this test, then give that a shot.
09:14:08LloreanGodEater: That's what amiconn just suggested. :)
09:14:13jhMikeSamiconn: odd thing is that the small test_queue plugin has no trouble running after mpegplayer and mpegplayer would certainly have overwritten the test.
09:14:27*GodEater didn't recognise the abbreviation as such
09:14:58LloreanGodEater: I had to google it before it clicked.
09:15:11 Join aliask [0] (
09:15:30LloreanOf course, a part of the slow is probably the whole "msys in winxp in a VM"
09:15:47 Join JdGordon [0] (n=jonno@rockbox/developer/JdGordon)
09:15:53 Join nick89 [0] (
09:16:03JdGordonhey all
09:16:10*Llorean doesn't expect this to be done until after he sleeps.
09:19:13 Quit nick89 (Client Quit)
09:20:39GodEaterJdGordon: did you notice we're still getting reports of "Filetype Array is full" even after your fixes yesterday ?
09:21:09LloreanGodEater: I'm pretty sure that's been patched now, for the time being
09:21:39 Quit enyc_ (Read error: 110 (Connection timed out))
09:21:41GodEatershould someone mention that in the thread then ?
09:21:56JdGordonGodEater: yeah, I was expecting that, im half way through removing the array for good so didnt worry about it
09:22:04JdGordonpetur upped the size though so we should be good now
09:22:44jhMikeShrm...IRAM doesn't make any differece
09:22:46 Quit w0rd54 (Read error: 104 (Connection reset by peer))
09:22:50GodEaterah ok
09:23:26 Quit ldarby (Read error: 104 (Connection reset by peer))
09:23:59 Quit markun (Remote closed the connection)
09:25:34amiconnHow does a 55fps->68fps speedup sound? :)
09:25:48JdGordonnot worth the effort....
09:26:02GodEaterthank god
09:26:06GodEaterI nearly spat my coffee out then
09:26:07jhMikeSlike 13 more fps
09:26:17amiconnThat's on 2nd gen
09:26:26amiconnI also sped up things on mini g2
09:26:31amiconn(but not as much)
09:27:08amiconn3 changes: (1) Don't have the timeout in lcd_wait_write. Provided the controller isn't buggy, this will always work
09:27:36amiconn(2) For mini g2, the register setup doesn't need to be performed for every access. It's sufficient to do it on init
09:27:48amiconn(3) For PP5002, put lcd_write_data in iram
09:30:52*jhMikeS is clueless about this cache thing but ran into similar oddness when experimenting with the dual-core spc codec.
09:34:09 Join davina [0] (
09:34:15 Join markun [0] (n=markun@rockbox/developer/markun)
09:38:49 Join webguest52 [0] (i=c023110a@gateway/web/cgi-irc/
09:41:54 Quit Llorean (Read error: 104 (Connection reset by peer))
09:41:55 Join Llorea1 [0] (
09:42:11 Nick Llorea1 is now known as Llorean (
09:42:19 Join w0rd54 [0] (
09:42:35 Quit perrikwp ("ChatZilla [Firefox]")
09:47:21 Quit webguest52 ("CGI:IRC")
09:49:28 Quit JdGordon ("Konversation terminated!")
09:51:01 Join JdGordon [0] (
09:51:14 Quit JdGordon (Read error: 104 (Connection reset by peer))
09:53:53*Llorean attempts the SFU route since his computer crashed anyway
09:54:25GodEatersfu is a huge download iirc
09:54:45 Join tvg [0] (
09:55:24LloreanGodEater: 217mb looks like
09:55:31LloreanSo far, assuming it doesn't go out and grab other bits
09:55:42tvgCouldn't you integrate my language patch?
09:56:19 Join perrikwp [0] (n=chatzill@
09:56:35***Saving seen data "./dancer.seen"
09:56:57amiconnLlorean: sfu core is pretty much that 217MB, but you probably need some other packages for building rockbox
09:57:01amiconn(at least gmake)
09:57:45amiconnAnd package installation is commandline based, i.e. more difficult than cygwin
09:57:59markuntvg: which patch?
09:58:01*amiconn thinks the cygwin installer is quite nice for its job
09:58:25tvgmarkun: this one:
09:59:30Lloreanamiconn: Is interix make not compatible? It sounds like it comes with its own version?
09:59:56B4gderLlorean: I would guess not, rockbox relies on gnu make extensions
10:00:03amiconnWell, last time I tried the provided 'make' didn't manage to build rockbox
10:00:15amiconnIt doesn't know -C, for instance
10:00:48markuntvg: I'll commit it later today
10:00:51LinusNnerochiaro: i believe daniel fixed it several days aago
10:01:00amiconnIt wasn't significantly faster than cygwin so I didn't follow this route further
10:01:28tvgmarkun: thank you, if there's something wrong, I'll correct soon. Bye.
10:01:36amiconnBut the speed difference might be system dependent
10:01:50Lloreanamiconn: It's going to be slow here anyway, I'm running in a VM
10:01:56markunB4gder: did you chose festival over espeak for the daily voice files because it sounds better?
10:02:07B4gderno, I just picked one
10:02:23B4gdermy plan is to build multiple versions soon
10:02:33 Quit ctaylorr_ (Read error: 110 (Connection timed out))
10:02:33markunok, even better
10:03:21nerochiaroLinusN: ah, let me check
10:04:30amiconnB4gder: There was some talk earlier regarding a possibility to return to a single voice file
10:04:51amiconn(and perhaps also .lng file)
10:05:41B4gdersure, if it is done without too much pain in the targets sure
10:05:57 Join safetydan [0] (n=dan@rockbox/developer/safetydan)
10:06:55 Quit tvg ("Leaving")
10:07:11nerochiaroLinusN: it's indeed fixed. sorry for the noise and thanks
10:08:07LinusNamiconn: concatenating files doesn't sound like a good idea to me, since the file would be ridiculously big
10:08:17amiconnjhMikeS: The table method might be the way to go, and would open up some further possibilities
10:08:39amiconnLinusN: Yes, that was just the most basic example
10:10:09amiconnvoice file header: a struct array with a target id and a pointer to the table for that target further into the file
10:10:24LinusNthe voice file format would be rather complex
10:10:32amiconnThe last entry in that array would be an end marker
10:10:57 Join rvvs89 [0] (n=rvvs89@unaffiliated/rvvs89)
10:11:04amiconnAfter that the tables for the targets are placed, and finally the individual clips
10:11:32LinusNthe file would still be rather big, since each target can have different output for the same lang id
10:12:16amiconnIf there is different output for the same id right now, these ids must be split
10:12:39 Join JdGordon [0] (n=jonno@rockbox/developer/JdGordon)
10:12:44LinusNbut that was the whole point with langv2, wasn't it?
10:12:46 Quit perrikwp ("ChatZilla [Firefox]")
10:12:53amiconnNot necessarily
10:13:21LinusNfor example the button definitions
10:13:41LinusNone target says "SELECT" and another says "ON" etc
10:13:51amiconnThe 2 points of langv2 were (1) to adjust texts per device and (2) save ram and binary size
10:14:12B4gdernot only, but those were 2 of them
10:14:21amiconnBut per-target voice files have a disadvantage that now shows up on the ml
10:14:33LinusNyes, and (1) means that the same lang id can have several voice clip representations
10:14:45amiconnNot everyone can build voice files, and the central build server can't provide all voice files either
10:14:55amiconn(there are dozens of supported languages)
10:15:09B4gderthe central server could indeed collect all voices
10:15:16 Quit Llorean (Read error: 104 (Connection reset by peer))
10:15:18B4gderit wouldn't have to build them all itself
10:15:23amiconnNo it can't
10:15:29B4gderit _could_
10:15:44amiconnOnly for freely distributable voices
10:16:04amiconnLinusN: Yes, but taking this further, nothing speaks against splitting these into several IDs, provided that each target only loads the IDs it needs
10:16:06B4gderyeah, but all these ones we talk about are what people distribute
10:16:12JajaComperror - "Invalid option 'long-calls'" ???
10:16:13B4gderdo they must be distributable
10:16:24JajaCompwhot is?
10:16:38LinusNamiconn: take LANG_CONFIRM_WITH_BUTTON as an example
10:16:43B4gderJajaComp: you have a bad compiler version, go back and read CrossCompiler
10:16:57LinusN(which isn't voiced at the moment)
10:17:26amiconnLinusN: Yes, and? SPlit these into LANG_CONFIRM_WITH_PLAY, LANG_CONFIRM_WITH_SELECT etc
10:17:27LinusNbut when it is, it will have several versions depending on the target
10:17:44B4gderbut yeah, by making a huge voice file with a more complex header and sections the target separation can be done load-time instead of build-time
10:17:48LinusNamiconn: but we just merged them!!!
10:17:57 Quit gtkspert (Remote closed the connection)
10:18:04 Join obo [0] (n=obo@rockbox/developer/obo)
10:18:10amiconnThe point is that the implementation should only load those which are needed on the specific target
10:18:28B4gderI would say that they should not be split in different IDs in the lang file nor in the code
10:18:33 Join gtkspert [0] (
10:18:41amiconnThey key is that an empty id has to be cheap, otherwise it won't work out nicely
10:18:41B4gderbut it would somehow get done behind the schene
10:18:49LinusNB4gder: i agree
10:19:00amiconnHmm, even that is possible
10:19:32 Quit kubiix ("Miranda IM! Smaller, Faster, Easier.")
10:19:37LinusNotherwise we would have tons of #ifdefs in the code
10:19:42 Join pixelma [0] (i=pixelma@rockbox/staff/pixelma)
10:19:50amiconnNothing stops the per-target tables to point to a different clip for the same id
10:20:29amiconnAnd it would be better because there are less empty IDs
10:20:35LinusNthen again, what is the problem with target specific voice files?
10:21:21B4gderit certainly is more kiss like this
10:21:32B4gderin the target
10:22:21amiconnLinusN: The problem is that they're target specific. (1) One needs to download or make the correct version for your target, otherwise some options won't be voiced, or worse, would be voiced wrong
10:23:02 Join Llorean [0] (
10:23:03LinusNnot much different from how it would be with a unified file, is it?
10:23:09amiconn(2) Imagine someone making a voice file with a not freely distributable voice for several of his blind frieds, which have different rockboxable devices
10:23:15LloreanMy windows VM keeps crashing me. :(
10:23:38LinusNamiconn: this person would have to build several files
10:23:42amiconnMaking one file and giving it to all of them would certainly be better
10:24:02B4gderand he wouldn't be allowed to distribute them in the first place so he only needs to build for his own target ,*)
10:24:23pondlifeAssuming he was able to build them under Windows :)
10:24:39amiconnB4gder: That's not quite correct. Giving away the file to persons you know doesn't count as distribution
10:24:42pondlifeI like the idea of self-built via RBUtil best.
10:24:44B4gderyeah, but I guess in the end someone using Windows will step forwards
10:24:54B4gderamiconn: it depends on where you are
10:25:04Lloreanpondlife: RBUtil building them will probably be the best plan
10:25:05pondlifei.e. don't concentrate effort on distributables, more on enabling.
10:26:16B4gderbut the problem is, as often in this project, when there's windows stuff involved
10:26:26B4gderas we quickly provide linux solutions
10:26:26amiconnB4gder: The at&t license states that iirc (that you can give the output to others, as long as you don't make it publicly available)
10:27:09B4gderI still don't care much about the people using those voices
10:27:14B4gderthey are evil
10:27:44B4gderbut that's beside the point here anyway
10:27:59pondlifePeople should be able to use whatever voice they have available, for both .voice and .talk files. That's the basic point.
10:28:25amiconnThe voices sound way better than festival
10:28:39B4gderamiconn: that is entirelly besides the point for all this
10:28:48pondlifeThat implies self-built is the way to go.
10:29:46 Join wippeout [0] (
10:29:55amiconnLinusN, B4gder: Regarding the amount of KISS, I tend to think that it won't change much. Certainly less than some other, recent changes & additions
10:30:08amiconnThe Ondio already has some kind of chunked loading
10:30:48amiconnThe more I think about it, the more I like the idea. Especially since it also provides an easy path towards voice in plugins
10:31:17amiconn(another goal of langv2 we didn't achieve yet)
10:31:29B4gderas I said, U
10:31:33B4gderI'm not against the idea
10:32:00B4gderI am against letting the whiners on the users list set the agenda though
10:32:23B4gderbeing blind is not an excuse for being ignorant
10:33:02B4gderI'll volunteer to adapt the language scripts for a modified system towards unified voice file
10:33:23markunhi wippeout
10:33:24pondlifeI don't see a unified voice file as desirable
10:33:45markunpondlife: why not?
10:33:50pondlifeNot if it's bigger than a dedicated one (i.e. less audio buffer)
10:33:50markuntoo much wasted space?
10:33:59LinusNspace where?
10:34:19pondlifeWhy not just have a policy of build-it-yourself ?
10:34:19 Join bluebrother [0] (i=0g1aBuri@rockbox/staff/bluebrother)
10:34:39LinusNmarkun: it would only load the clips for the target in question
10:34:49pondlifeAh, ok
10:34:58B4gderbut it would still be a megabyte or two wasted in storage
10:34:59LinusNso the buffer usage would be the same as the current files
10:35:08LinusNB4gder: yes
10:35:16pondlifeAnd bandwidth
10:35:53pondlifeParticularly for home users with limited upstream
10:36:07pondlifei.e. those complaining on the ML
10:36:22LinusNnow you lost me
10:36:29B4gderyeah, but supplying 22 voice files is more data than one unified
10:36:47pondlifeYes, but I would only download the H300 file, not them all.
10:36:48LinusNwhat does the upstream have to do with it?
10:37:21B4gderupstream, the ones uploading files, would benefit from a unified
10:37:21amiconnpondlife: The idea is that each target only loads the clips it needs into ram
10:37:30ddaltoncould someone please have a look at ddalton/time.diff"> and tell me where I have done the indentation wrong?
10:37:39pondlifeI know, but how much bigger would the .voice file be? Say 20%?
10:37:50Lloreanrasher: text2wav textfile -o wavfile.wav for festival without using it as a server.
10:37:51B4gderI'd guess at more than 20%
10:37:58LinusNpondlife: something along those lines
10:38:07amiconnThe per-target tables in the beginning won't be big, just an offset and a length per clip used
10:38:21amiconnSo if a target uses 500 clips, that's 4KB
10:38:28pondlifeSo if I download Andre's voice file for my H300, I'm using 20% more of his limited upstream bandwidth.
10:38:37B4gderyeah, its the clips that'll do the size diff not really the tables
10:38:50B4gderpondlife: ah, yes
10:38:54amiconnFor 30 targets, that's 120KB
10:39:21B4gderhey, I can check the total size of all clips...
10:39:26B4gderas a comparison
10:39:27pondlifeIt's not a biggie, but I just think any coding effort should be aimed at making self-build easy for all users.
10:39:51pondlifei.e. spend our time on MakeVoices.vbs and RBUtil.
10:39:57aliaskddalton: Too much indentation.
10:39:58amiconnThe clips would then include all alternatives, but I think the whole file will be <50% larger than one of the swcodec files today
10:40:10pondlifeWhile Rockbox itself is KISSed.
10:40:19B4gderwe have 655 different phrases
10:40:31LinusNthe total voice pool is ~2.5MB
10:40:59amiconnLinusN: And how much of that is included in an average voice file?
10:41:01B4gderwhich then is >50%
10:41:13amiconnBtw, it depends on the voice used how large the pool is
10:41:27LinusNamiconn: between 800K and 1.3MB
10:41:27B4gderlet's do these ones now for comparison
10:41:32JajaCompI can use gcc 4.1.2 to compile RockBox?
10:41:40B4gderJajaComp: read CrossCompiler again
10:42:11JajaCompi don't now how uninstall gcc 4.1.2
10:42:22amiconnWell, let it be twice as large
10:42:36markunJajaComp: why uninstall it?
10:42:53JajaCompit's default compiller in my system
10:43:01pondlifeBTW has anyone tested that my configure patch (for Cygwin voice building) doesn't break Linux voice building (or indeed program building). I'd like to commit it.
10:43:02markunthen it's not a cross compiler
10:43:28markunyou can use gcc 4.1.2 to build the crosscompiler to build rockbox
10:43:29LinusNi'm beginning to like the idea of a unified voice file
10:44:10LinusNconsisting of the entire voice pool and a table per target
10:44:45B4gderI like how it can also fix the plugin voice problem
10:44:51amiconnIt has advantages for our distribution (less overall file size), for users (less confusion of which files to use), and especially for users with different targets, or users downloading voices for other users as well
10:44:51LinusNbut the loading will be complex, involving plenty of seeking
10:44:57ddaltonwhat plugin voice problem?
10:45:06B4gderddalton: plugins aren't voiced today
10:45:16amiconnLinusN: Only forward seeking if done right (except on Ondio)
10:45:40B4gderseeking like mad, yes
10:45:48 Join PaulJam [0] (
10:46:01amiconnThe only disadvantage is that users which only need voice on one single target need to download a larger file
10:46:11LinusNamiconn: true, it could optimize for only forward seeking by sorting the clip indexes
10:46:18B4gderwhich is 99% of the users :-)
10:46:21JajaComphow? −−enable-libssp=no???
10:46:50amiconnB4gder: They would still profit from less confusion, and from the ability to use other user's voice files
10:47:14LinusNi think ease of use outweighs the file size
10:47:16B4gderthey will not get an easier build system
10:47:20JajaCompI see error "cc1: Invalid option 'long-calls'"
10:47:30LinusNB4gder: that is true
10:47:43pondlife97% of users only want files for a single target... another made-up statistic.
10:47:43Lloreanamiconn: Under that system, couldn't users still generate a target-specific file if they wanted?
10:47:51amiconnLlorean: yes
10:47:54markunJajaComp: which instructions are you following?
10:47:55LloreanIf file size was a concern.
10:48:04LloreanFor us people with tiny amounts free, and all.
10:48:13amiconnA target specific file would only include the table for one target and the needed clips
10:48:23B4gderLlorean: ah yes they could
10:48:34LloreanB4gder: It's more or less a best of both worlds deal. ;)
10:48:34LinusNamiconn: it could even include empty tables for the other targets
10:48:52amiconnLinusN: Yes, but that's not needed
10:49:08amiconnIf a target can't find its matching table, it simply won't load the voice file
10:49:34B4gderit would need to scan for its own target section
10:49:34LinusNi like this
10:49:45pondlifeYes, that makes sense.
10:49:57pondlifeBut ALSO work on the self-build stuff... :)
10:49:59LinusNnow over to how to build the voices without a compiler
10:50:31LinusNi guess a rudimentary c preprocessor implementation is needed
10:50:48B4gderI happen to have one
10:50:55LinusNin perl?
10:51:09B4gderno in C, but I'm quite sure it builds fine on windows too
10:51:31B4gderor could be made to without too much pain
10:52:16pondlifeIf we're using a non-standard C program, why not make it do the whole thing?
10:52:37LloreanOr we could hunt down a C preprocessor in perl. I'll bet one or five exist.
10:52:52B4gderpondlife: because I don't wanna do the whole voice process in C
10:53:06B4gderLlorean: indeed
10:53:49pondlifeHmm, I suppose RBUtil will hide any complexity and make it accessible.
10:54:03LloreanB4gder: Exactly what I found
10:54:07LloreanWas looking for a license.
10:54:21LloreanThere's a debian package for it, it seems, so I have to assume it's in the clear.
10:54:31LloreanGPLv2, okay
10:55:04B4gderseems like a fine candidate for this then
10:56:10LloreanI seem to recall there being other situations where configure demanded a compiler and I didn't think it reasonable to expect one. Manual maybe? Could it clear these up to?
10:56:10pondlifeHmm, does this not mean that Windows users will need perl?
10:56:37*pondlife is not much experienced with scripting.
10:57:02pondlifeOr can it be bundled into the RBUtil distribution...
10:57:02B4gderI really don't think building voices without perl is a worthwhile effort
10:57:25B4gdergenlang is perl to start with
10:57:29pondlifeAh, ok
10:58:04B4gderso we've required perl for voices for a long time already
10:58:51pondlifeYes, but is it required for MakeVoices.vbs? I'm fairly sure I've not installed it (aside from inside Cygwin).
10:59:52B4gderif they do it without perl they either re-implemented lots of genlang stuff or built the voices wrong
11:00:25B4gderI've never looked at that script
11:00:26pondlifeWasn't it [IDC]Dragon who wrote it?
11:00:38pondlifeOr amiconn? Will look....
11:00:44LloreanB4gder: Oh, btw, iPod Nano voice files are broken.
11:00:54B4gderback when we only had a single voice for all
11:01:06B4gderso I added an "old format" support for genlang
11:01:09B4gderto feed that script with
11:01:09LloreanOr at least, the one I built was, but the Gigabeat one was fine.
11:01:20amiconnMakeVoices.vbs was written by me
11:01:48B4gderstill, it needs its input as given by genlang -o
11:01:51ddaltonif I download an old revision and add a old patch. then I run svn up and then make a diff will this work?
11:02:01B4gderddalton: perhaps, and perhaps not
11:03:31amiconnYes, it does, since langv2 (first step) went in
11:04:17pondlifeSo this has been a prereq for a while now...
11:04:22amiconnBut pretty much everything that can be done in perl can also be done in vbscript, so once things stabilize, there is nothing that would stop a vbscript version for folks without cygwin
11:04:52B4gderright, everyone is free to do what they want of course
11:04:52pondlifeFolks with Cygwin should just make voice same as Linux users anyway.
11:04:59amiconn(yes, vbscript does have a regex engine, just that not many people use it. MakeVoices.vbs does)
11:05:26B4gderits just that as usual, I expect the linux/cygwin perl method to stay up-to-date and accurate
11:05:30amiconnOne problem atm is the need of a C preprocessor
11:05:31B4gderand all other efforts will lag
11:05:53amiconnThat won't go away with unified voice files either
11:06:16B4gderno, and I prefer the unified voice build way anyway
11:06:19 Quit PaulJam (".")
11:06:30 Join n1s [0] (
11:07:00amiconnBut there's also perl for windows (activestate perl), so that shouldn't stop us
11:07:20amiconnAnd activestate perl integrates with activescripting, so it should even be able to use sapi objects etc
11:07:36B4gdersince people have build voices for a while, clearly perl has not been a major hurdle to them
11:08:48ddaltonWhat is the best way to voice a gui_syncsplash screen?
11:12:25n1sddalton: maybe you can have a look at how it's done in other places
11:16:22pondlifeOK, one more try.. Please can someone who's got Linux (or VMWare) running try out and check it doesn't break the Linux build processes.
11:16:42pondlifeI want to commit, but not break the daily voice build! That would not be popular at this point...
11:17:47 Quit desowin (Read error: 110 (Connection timed out))
11:19:46 Quit courtc (Read error: 113 (No route to host))
11:20:29 Join PaulJam [0] (
11:23:49n1spondlife: ok, I'll try it
11:25:26amiconnB4gder: Iiuc you already implemented one part of building unified voice files (the clip cache)
11:26:01B4gderrasher made that really, I just used his work
11:26:14pondlifen1s: Thanks
11:26:14B4gderbut yes the foundation is there
11:26:36 Join haemmy [0] (n=stefan@
11:26:39amiconnOk, whoever did it, it's there :)
11:26:49pondlifeDontcha just love OSS :)
11:27:42n1shmm, voice crashes seem related to resuming playback on startup
11:28:20*pondlife does the "no voice codec swap" chant too
11:28:33amiconnn1s: If so, I suspect a race in the voce & playback init, combined with the voiced splashes
11:29:35amiconnpondlife: Yeah, just that there is a problem: no voice codec swap means (almost) no iram for the voice codec
11:29:57pondlifeI'm still not convinced that MP3 mono 22kHz playback needs IRAM.
11:30:09amiconnSome targets probably can't cope
11:30:28amiconnNo, pure voice maybe doesn't need it, but what about voice during playback?
11:30:28pondlifeIpod 1G would suffer most, right?
11:31:19pondlifeIt should be easy to experiment with - disable IRAM for the MP3 codec and play some AAC (or WMA) with voice.
11:31:21amiconn1/2/3G, and perhaps all coldfire targets
11:31:34pondlifeCompare with and without.
11:31:39amiconnRemember that coldfire has no data cache
11:32:38pondlifeBuit we're only needing one quarter of the data rate than for normal MP3 playback.
11:33:05amiconnI did an experiment a while ago - running with caches disabled on PP5022
11:33:22amiconnIt was ~7x as slow...
11:33:48n1spondlife: working voice file built with patch, no problem :-9
11:33:51amiconnOuch indeed, but expectable
11:34:04pondlifen1s: You mean using festival?
11:34:15n1spondlife: I used flite
11:34:45*n1s goes to eat pie, bbiab
11:34:47pondlifeHow about building Rockbox itself. I don't think I broke anything but rasher's configure mod also happens.
11:34:53pondlifeOops, too late
11:35:15*pondlife may finish that sentence later.
11:35:35amiconnI didn't check what samplerate the new voice files use. Before per-target voice, it was common to use 12kHz
11:37:09amiconnWe must avoid using 11kHz for archos player. Other than that, all standard mp3 sample rates are possible, with only one further constraint for all archoses: For mpeg2.5 sample rates, we must use lame -B 64
11:37:41amiconn(limit maximum bitrate to 64kbps - voice files are vbr)
11:40:53 Quit maffe (Remote closed the connection)
11:41:42 Join My_Sic [0] (
11:42:51PaulJamhi, i have a little question about the custom icons: with some recent svn change the items in the open_with menu now have an icon, and i was wondering if there is a way to specify which icon will be used for which viewer. i guess the viewers.config specifies the icon, but that file gets overwritten on every update. is there another way to do this?
11:45:19amiconnHmm, a unified voice file would have to deal with bitswapping...
11:45:26JdGordonPaulJam: the icon the plugin is listed with first is the icon shown
11:45:46JdGordonPaulJam: using the .icons file will also work
11:47:44PaulJamJdGordon: how? if i want for example to use icon 0 for the text_editor in the open with menu, what line would i have to put in the .icons file?
11:48:27JdGordonhmm... yeah, you cant..
11:48:30*JdGordon forgot
11:48:40JdGordonthe .icons is for extensions
11:48:56JdGordonyou have to use viewers.config.. for now... im working on fixing this
11:49:25PaulJamok, thanks. then i guess the best solution is to use the same order of the icons like in svn.
11:52:03 Quit obo ("bye")
11:54:26 Quit MySic ("Miranda IM! Smaller, Faster, Easier.")
11:56:40***Saving seen data "./dancer.seen"
12:05:58 Quit nerochiaro ("leaving")
12:12:44 Quit PaulJam (".")
12:17:00 Join maffe [0] (
12:17:06 Join Febs_ [0] (
12:17:21 Quit bluebrother (Read error: 113 (No route to host))
12:18:29 Join spiorf [0] (
12:23:14 Join bluebrother [0] (
12:25:06 Quit rvvs89 ("bai")
12:25:34n1spondlife: rockbox builds fine, both normal h300 build and sim
12:26:54pondlifeOK, I'll commit it now... 2 questions: 1) Are there some special property changes required when adding a file to SVN? 2) Anyone mind me commit .VBS files for the SAPI interface?
12:27:08pondlifes/commit/committing in that last bit.
12:27:18 Quit BigMac (Read error: 110 (Connection timed out))
12:28:15amiconnThe vbscript will probably need an svn:eol-style property
12:28:53pondlifeOK, I'm about to lose power here so will commit a bit later
12:28:53n1spondlife: UsingSVN says you should set the keywords property
12:29:12pondlifeAh, I was looking for that in the wiki, but missed it.
12:29:19 Quit pondlife ("power down")
12:33:16 Join adam2 [0] (
12:33:54adam2is there anybody out there?
12:34:05B4gderI'm in, not out!
12:35:57 Quit JajaComp ("CGI:IRC (EOF)")
12:37:26 Quit Febs (Read error: 110 (Connection timed out))
12:38:53adam2i've got a gigabeat s60 guts only (no hdd or battery and lcd cracked) any use to rockbox developers for tracing cuicuits on the main board or not really?
12:39:41B4gderaliask is one of the gigabeast S dudes, he might have an answer
12:40:19scorchetoffe82 as well...he builds up stockpiles of gigabeat parts
12:40:29aliaskFor hardware things I'd refer to toffe82
12:41:32aliaskBut if someone accidentally flashes their player it could be a saviour
12:41:46aliaskFlashes garbage, that is.
12:42:49adam2you recon i should try puting it on ebay with the battery and hdd? see if i can get some $'s back for the new 1 i got....>?
12:44:25 Part Llorean
12:47:43 Join ctaylorr [0] (
12:47:56 Join adam3 [0] (
12:48:48 Quit adam3 (Client Quit)
12:52:20*preglow finds out about plugin categories
12:52:27preglowis this just a directory thing, or something more fancy?
12:53:07amiconnJdGordon: Why is mandelbrot sorted into games, btw?
12:53:14amiconnIt's surely not a game
12:53:27n1spreglow: dirs with localizable menu to enter them
12:53:39 Join DerPapst [0] (
12:53:46DerPapstgood day
12:54:01JdGordonamiconn: a mistake?
12:55:11preglowat least my vinyl preamp plug seems to work somewhat
12:55:42*amiconn wonders what that plugin does
12:55:56amiconnriaa deemphasis?
12:56:18*DerPapst wondered that too yesterday.
12:57:26preglowamiconn: aye
12:57:58preglowseems i'm doing something wrong, though, there's quite a bit of treble missing
12:59:10 Join amiconn_ [0] (n=jens@rockbox/developer/amiconn)
12:59:19 Join XavierGr [0] (
12:59:20 Quit amiconn (Nick collision from services.)
12:59:22 Nick amiconn_ is now known as amiconn (n=jens@rockbox/developer/amiconn)
12:59:29markunadam2: if the hdd is working you could try to sell it separately
12:59:32markunsame for the LCD
12:59:40preglowthere's a bloody paper on this somewhere, but it's published by ieee, which loves taking cash for them
12:59:56markunpreglow: I can download it for you
13:00:05markundo you have a link?
13:00:16adam2lcd is cracked, hdd and battery is on ebay for auction
13:00:31amiconnpreglow: Now if this could be selected for recording...
13:00:34markunpreglow: at least I hope I can..
13:00:51preglowmarkun: that would rock, i'll dig up the url
13:01:00preglowmarkun: i used to be able to too, but not anymore, apparently...
13:01:05aliaskadam2: Actually, I think I could use it for serial debugging
13:01:30preglowamiconn: i probably need to compensate for the pickup as well, i think
13:01:34aliaskadam2: Do you mind if I PM you?
13:01:40adam2aliask: np
13:01:47preglowmost vinyl preamps have an mm/mc switch
13:02:45amiconnYes, as MM output voltage is roughly one order of magnitude higher than MC output voltage
13:03:17preglowso it's just an amplitude thing?
13:03:32preglowi quite obviously need some kind of extra filtering here, the treble sucks
13:05:22amiconnDerPapst: Do you still have the patch I gave you?
13:08:59preglowamiconn: any idea on how many db gain you'd typically have to apply to gain a turntable signal to line in levels?
13:09:08preglowi'm using 48 db right now, and that works just peachy
13:09:10preglowbut feels like i can do more
13:09:22DerPapstamiconn: yes
13:09:29preglowthink i'll just add a control for that
13:09:39peturpreglow: using 48dB on the iriver will be quite noisy
13:09:48preglowpetur: sounds okish
13:09:54preglowremember we're talking vinyl here :)
13:10:01amiconnDerPapst: Iirc you said the 3rd gen also suffers from blacklevel varying with display content?
13:10:22amiconnYou could try a small modification of my patch
13:10:56amiconnFind the void lcd_init_device(void) line, and within that function, the #elif defined IPOD_3G part
13:11:11amiconn(can't tell a line number 'case my lcd driver contains further patches)
13:11:39DerPapstamiconn: i'm at work currently. so i can do that test at 19:00 when i'm at home again.
13:11:55amiconnThen change the power_reg_h = 0x1500; in the if... branch to power_reg_h = 0x1520;
13:12:10DerPapstok. i'll do that then.
13:12:31amiconnAfter installing this build, the lcd should behave like before, until you call 'View HW Info' once. Afterwards the blacklevel should vary less
13:12:40amiconn(until reboot)
13:13:24amiconnThose bits increase the frequency the voltage multiplier is running at, making the voltage more stable (and needing a tiny bit more power)
13:13:41DerPapstamiconn: sounds good :)
13:14:12preglowpetur: is the adc gaining done purely in the digital domain, or?
13:14:17 Part adam2
13:14:35amiconnThat wouldn't make sense imo
13:14:51petur24dB analog, the rest digital (but on the internal 24(?)bit, not our 16bit
13:17:21scorchea gem: <Colombo> made rockbox is easy.. developers even have no ipods, just writing easy code and its work.
13:17:48*scorche goes about adding that to GoldenQuotes with other Colombo stuff
13:17:56scorchehe is in #ipodlinux if you want to chat
13:18:46*DerPapst enjoys Colombo as well
13:19:01DerPapstB4gder: btw thanks for cURL :D
13:19:04scorcheDerPapst: isnt this entertaining? =P
13:19:11DerPapstindeed it is ;)
13:19:13B4gderDerPapst: you're welcome!
13:20:26 Join Thundercloud [0] (
13:21:02n1shmm, we actually call talk_init() 3 times during a regular boot, feels kind of hackish...
13:22:18scorchehehe...colombo is up to 3 in GoldenQuotes now..
13:22:45 Join Llorean [0] (
13:24:02B4gdern1s: amen
13:24:22ddaltonHow do I build a voice file using p7560?
13:24:26amiconnn1s: On all targets?
13:26:47n1samiconn: on sw-codec at least in init() in main.c, first we call settings_apply() which calls talk_init() then we call talk_init() and then audio_init() which also calls talk_init() hwcodec may only call it twice, havent looked at audio_init() for those
13:27:22 Quit ToHellWithGA (Nick collision from services.)
13:27:30 Join ToHellWithGA [0] (
13:27:36n1sthe good news is that removing the call in settings_apply() seems to fix the crashing (this is not a solution however)
13:28:25amiconnsettings_apply() needs to call talk_init() in case changing the language changes availability of voice
13:30:27preglownote to self: disable idle timeout in dsp plugs :>
13:31:53preglowafter about ten minutes of music i got what i assumed was my speakers blowing :D
13:32:41scorcheB4gder: he seems to be mute for now, but perhaps he shall come back and enlighten us...
13:34:25 Quit safetydan ("Ex-Chat")
13:36:59 Quit spiorf (Read error: 110 (Connection timed out))
13:37:37 Join spiorf [0] (
13:43:31n1samiconn: yes, but should I kill the other call or thell settings_apply to not call the init when we start up?
13:43:51n1sI mean kill the other call and rely on the one in settings apply?
13:44:36amiconnGood question...
13:48:09n1sI still don't know why this causes the hangs though...
13:49:21amiconnjhMikeS: What is the value of DEV_EN on Sansa?
13:49:43amiconn...and how good is battery runtime compared with OF?
13:50:05LloreanI *think* on the Sansa we get about 3/4 of the OF
13:50:31LloreanThe performance gap isn't as big as it is on other portalplayers as far as people seem to say, I haven't done an explicit test
13:50:52amiconnHow is the ratio on Nano?
13:51:24LloreanI think I get about 8, Apple claims 14.
13:51:28LloreanBut that 8 is from a long time ago.
13:51:35LloreanAnd my battery's now a year and a half old
13:52:40LloreanI wish the IpodRuntime page include AppleOS ones
13:52:58 Quit DerPapst (Read error: 104 (Connection reset by peer))
13:53:17 Nick chrisjs169__ is now known as chrisjs169 (
13:53:21amiconnHmm, I would really like to see the DEV_EN value from Sansa
13:53:31amiconnAnyone with a Sansa who could check?
13:53:49LloreanWell, I've got one
13:53:59 Quit haemmy ()
13:54:03LloreanWhat do I need to do?
13:54:05 Join webguest64 [0] (i=57b5e70f@gateway/web/cgi-irc/
13:54:13amiconnView I/O Port
13:55:02LloreanI don't see anything named DEV_EN on that screen? Do I need a newer build, or is it the DEV_0x34?
13:55:43amiconnYou need a newer build then
13:55:53 Join DerPapst [0] (
13:56:08LloreanAlright, gimme a minute.
13:56:09amiconnI committed DEV_EN display on Aug 3, so an Aug 4 daily should do
13:56:31LloreanYeah, mine's from before my trip, so July 27 or so
13:56:41***Saving seen data "./dancer.seen"
13:57:22amiconnDEV_0x34 is DEV_TIMING1
13:57:26webguest64Hi - jemand aus deutschland hier , der mir sagen kann, ob rockbox auf meinen zen xtra 30gb drauf geht?
13:57:49B4gderwebguest64: this is an english channel
13:57:52amiconn(but I do not yet understand what it actually does)
13:58:50 Quit ddalton ("I was using BOFHNet IRC version 1.2 by fmillion - get your copy today from !")
13:59:39*amiconn suspects there's a secondary DEV_EN on PP502x, like there is on PP5002
13:59:42webguest64ok thx - so anyone kann tell me if rockbox works on my zen xtra 30 gb ?
13:59:58Lloreanamiconn: C440597F
14:00:01Lloreanwebguest64: No.
14:00:09Lloreanwebguest64: It only runs on the players listed on the site.
14:01:48amiconnLlorean: thx.
14:01:50 Join pondlife [0] (
14:02:14amiconnC... and ...97f seems to be common on all PP502x targets
14:02:22webguest64what a pity
14:02:30webguest64 thx
14:02:34amiconnBut the inbetween bits differ
14:03:02GodEaterC2C1197F here
14:03:02 Quit webguest64 ("CGI:IRC (EOF)")
14:03:09GodEaterwhich is 5.5G of course
14:05:07amiconnI know the G5.5 value...
14:05:09LloreanAnd magically today my Nano can dump its ROM.
14:05:22amiconnAh, that reminds me
14:05:34GodEatersorry amiconn, was just letting Llorean know in case he was curious
14:05:45LloreanGodEater: Same value as my Nano, anyway
14:06:05*GodEater looks at the two values...
14:06:19amiconnMini G2 sets the same as G5.5, and H10/6GB sets 0xC240197F
14:06:49DerPapstLlorean: your nano is wired. ;)
14:06:55LloreanDerPapst: Very
14:06:57amiconnSome of the bits are documented in the ipl wiki, but at least 2 of them must be wrong
14:07:11LloreanNow when I say "Dump Rom" there's no UI delay at all. Not even the slight delay it had before while writing.
14:07:30GodEater*gasp* - ipl be wrong?!?!
14:07:32GodEatersurely not
14:08:13DerPapsti doubt that too.
14:08:14amiconnWell, they say bit 29 is IDE0 and bit 14 is ATA. Apart from that it is rather unlikely, both bits aren't set on any ipod
14:08:38amiconnBut then bit 14 is set on Sansa which has no ATA
14:10:40*preglow tries to figure out the menu macro system
14:10:53amiconnEhm, seems I can't read
14:11:07amiconn<bit 25 is IDE0
14:11:43amiconn...and that _is_ set on ipods and H10, but not on Sansa
14:12:19amiconnThat still leaves several set bits unexplained
14:13:53dionoeawin 16
14:13:57amiconnBits 0, 1, 3, 4, 5, 8, 30, 31
14:14:16preglowwhat's up with declaring menu variables with a macro?
14:14:22amiconn(and, not set on ipods, bit 14, 21 and 26)
14:16:16preglowsometimes rockbox goes unresponsive after starting when i insert usb during startup, anyone had this before?
14:17:23amiconnYes, that's a known problem
14:17:27preglowhrmph :/
14:17:29 Quit B4gder ("It is time to say MOOO")
14:17:34amiconnHappens since active usb detection was introduced
14:17:43preglowwho introduced that?
14:17:43 Part LinusN
14:18:18amiconn...but it doesn't happen every time. It's more likely to happen when USB is already plugged at boot though
14:18:19*Llorean sighs.
14:18:32*elinenbe sighs too.
14:18:36LloreanI really don't think "expecting users to read the development list if they want to know what's going on with development" alienates users.
14:19:27LloreanUSB detection on iPods is funky anyway. The "hold menu to prevent reboot" hardly works right now
14:19:43amiconnIt works. YOu just need to hold Menu long enough
14:19:44dionoeayou have to hold it for a few seconds
14:19:49amiconn>10 seconds
14:19:49GodEaterwell it does - but you have to hold it a LOOOONg time
14:19:51preglowLlorean: nah, i'd say that is fair
14:20:11Lloreanpreglow: There are a lot of upset blind people on the list regarding the voice file changes.
14:20:20*elinenbe thinks 10 seconds is just too long. Maybe it should be changed to 3 seconds.
14:20:26LloreanAnd I'm kinda of the opinion "it's been in the works for a year and a half, and nobody kept it a secret"
14:20:38Lloreanamiconn: I tried twenty seconds on my Nano
14:20:41LloreanIt's just not reliable.
14:20:49 Join Ishi`` [0] (
14:21:00GodEaterelinenbe: it's not intended to be 10 seconds
14:21:00 Quit miepchen^schlaf (Read error: 104 (Connection reset by peer))
14:21:20preglowLlorean: so i see
14:21:23amiconnIt works for firewire. I still think we need to change state of the USB controller if we want to stop it
14:21:23elinenbeGodEater: ah... so that's the problem!
14:21:35GodEaterelinenbe: yes!
14:21:46preglowi don't really think we had to bring out the big banners for that
14:21:53preglowwe're in development anyway
14:21:53elinenbeGodEater: Do you REALLY eat god?
14:21:58amiconn(and that probably would solve the undetected usb device problem, and also be a bit linux friendlier with its sensitive usb stack)
14:22:19GodEaterelinenbe: off topic....
14:22:39Lloreanpreglow: That's my opinion. "If you download development builds, expect things to change. If you want to know what's going to change, pay attention to development. If you don't want things to change, don't change builds."
14:22:59rasherpreglow, Llorean: It would probably be less of an issue if Rockbox did releases more than every couple of years
14:23:03amiconnBut we want users to try new builds...
14:23:11elinenbeGodEater: not THAT off...
14:23:13Lloreanamiconn: I'm all for users to try new builds.
14:23:28 Quit TinoM (Read error: 113 (No route to host))
14:23:30LloreanBut I think the blind users are overreacting about the voice change.
14:23:35 Quit Febs_ (Read error: 110 (Connection timed out))
14:23:51LloreanVoice has been *broken* for longer than this.
14:24:14LloreanAnd now we're providing voice files (though they need some fixing still)
14:24:45Lloreanrasher: I'm still in favour of attempting a 3.0 release as soon as humanly possible. My vote is for "after MoB and Coprocessor support"
14:24:55elinenbeLlorean: people are afraid of change. Once things settle down, everyone will be happier
14:25:00elinenbewhat is MoB support?
14:25:05LloreanMetadata on Buffer
14:25:14GodEatercurse your faster fingers Llorean
14:25:18GodEateryou even got spaces in
14:25:23preglowrasher: of course it would, but we're not there
14:25:26elinenbeMoB should lead to supporting embedded art?
14:25:32LloreanNot necessarily embedded
14:25:55LloreanThat might depend on how the discussion on 'jpeg in core' goes.
14:26:06preglowsome time now we really need to start thinking about 3.0, though
14:26:17preglowif only to try
14:26:23Lloreanpreglow: Well, there is one thing we can do now.
14:26:25rasherpreglow: I don't know. Simply releasing the current state (once voice has been sorted) as 2.6 would be preferable to the current situation, in my opinion
14:26:39*amiconn refines the unified voice file suggestion, and now suggests 2 voice files
14:26:40LloreanWe can set up a "road to 3.0"
14:26:51LloreanA list of things that, when they're all done, we can enter 3.0 feature freeze.
14:26:54pondlifeLlorean: Ovrereacting maybe a little, but it's very unpleasant when voice doesn't work reliably. I'd suggest we advise all blind users to stick to builds before the change (on the front page).
14:27:04markunLlorean: or clean up the old wiki page
14:27:05preglowLlorean: would as always be helpful
14:27:07*pondlife visits the typo class
14:27:14pondlifeamiconn: 2?
14:27:17amiconnThe mechanism allows a voice file to support an arbitrary subset of targets, as long as the clip format is the same
14:27:25preglowbut i think 3.0 might be viable now that ipods are starting to shape up a bit
14:27:32*preglow has high hopes for amiconns work ;)
14:27:36Lloreanpreglow: I think with a list of things out there saying "We should get these done so we can freeze for 3.0", people might look in those directions a wee bit more
14:27:44preglowLlorean: and i agree
14:27:53amiconnFor archos we want the clips preswapped, for swcodec they should be in normal bit order
14:28:10pondlifeOK, so MASCODEC and SWCODEC
14:28:18amiconnSo one voice file for archos, another for swcodec
14:28:48amiconnActually it's not a hwcodec vs. swcodec issue, but due to the SH1 SPI and the MAS hookup
14:28:56pondlifeAh, ok
14:29:08amiconnAV300, while being hwcodec, will also use non-swapped clips
14:29:24amiconn(if that port ever gets done)
14:30:02*preglow has doubts :P
14:30:14LloreanWhat's the Rockbox 3.0 wiki page?
14:30:30amiconnOf course we could bitswap the clips on load and just use a single file, depending on how much perceived delay that produces
14:30:33rasherLlorean: ReleaseTodo?
14:30:47LloreanAh, right
14:30:58LloreanSo, should pre-freeze qualifiers go there to?
14:31:01amiconnOn Ondio that wouldn't be a problem at all, but maybe on the disk based archoses
14:31:49amiconnSwapping 1MB takes ~1 second with my optimised bitswap
14:32:09preglowwhat would the nicest current non-iriver target be for recording?
14:32:27amiconnThere is no current target that can record afaik
14:32:51LloreanI mean, the Sansa can record, and some iPods have a line in.
14:32:56amiconnHmm, wrong. The G5.5. can record
14:33:08LloreanThe Sansa's rather limited in quality though, iiuc.
14:33:09amiconnLlorean: Yeah, but sansa recording is a joke imo
14:33:21amiconnMaximum sample rate of 24kHz
14:33:48preglowso i'm stuck with irivers
14:33:51LloreanWell, it's designed for voice really.
14:33:53preglowi'll need to get another h1x0
14:34:00LloreanMemos 'n such
14:34:01preglowthis one is getting really, really scruffy looking
14:34:11Lloreanpreglow: Or start a new port. ;)
14:34:22amiconnThe iAudios can record too, but for line you need the subpack, and gain range is way lower
14:34:27 Quit DerPapst ("So Long And Thanks For All The Fish!")
14:34:48preglowLlorean: for what? :/
14:34:56amiconnOn the other hand, they are able to record with up to 96kHz
14:35:02Lloreanpreglow: Something that can record?
14:35:06pixelmathe H10 can record too, no?
14:35:07LloreanGet the iFP running!
14:35:19preglowi don't think i can maintain interest if i have to work on a port alone
14:35:22amiconnI doubt that the iFP can record
14:35:22preglowthe ipod port was more than enough
14:35:30Lloreanamiconn: It certainly can
14:35:33LloreanLine in and inbuilt mic.
14:35:40amiconnnow, really
14:35:57LloreanI meant that the player is physically capable, I doubt it can in Rockbox.
14:36:05LloreanI was suggesting Preglow finish it up for us. ;)
14:36:10peturpreglow: there are some (3) iHP120 on ebay atm
14:36:36preglowpetur: i want an h140 this time around
14:36:47preglowand i'm strapped for cash now anyway
14:38:19rasherLlorean: perhaps it's time for a round of release-talk on the -dev list? It's really awful that there hasn't been a release for any swcodec target yet.
14:38:32preglowit's very high time now
14:38:34preglowlike it or not
14:39:06preglowi certainly don't like it, but it IS a shame we don't have a release
14:39:22rasherI'm more or less of the opinion that a rushed and buggy release is better than none, at this point.
14:39:33preglowjust getting a wiki page up with stuff we (and other people) think we have to do might help organize a release very much
14:39:47preglowit's not like a release have to be bug-free
14:39:51preglowbut the bugs should be documented
14:40:20*amiconn thinks that we're currently further away from a release than a year ago
14:40:35preglowamiconn: and i agree, but that doesn't change the fact that we now need one more than we did a year ago
14:41:04*GodEater goes to read up on voxin
14:42:35Lloreanamiconn: I'm not sure if we're further than we were pre-freeze though
14:42:48LloreanWe entered that freeze with some unnecessarily large things on the table
14:42:48 Join Adikdid [0] (
14:43:36 Quit Adikdid (Client Quit)
14:43:50 Join barrywardell [0] (
14:45:41 Quit My_Sic ("Miranda IM! Smaller, Faster, Easier.")
14:46:10preglowi'm not convinced we'll fare much better this freeze either, but we pretty much have to try
14:46:20preglownot even _trying_ for a release around once a year or so is a shame
14:46:36LloreanWell this time the freeze should really be bugs-only
14:46:41LloreanNot "get these features done"
14:47:05rasherAnd at the end, release whatever you've got.
14:47:07preglowbut why does my vinyl amp have so little treble :/
14:47:28preglowrasher: i might agree with that
14:47:41preglowas long as nothing truly embarassing/disastrous is left
14:47:52LloreanYeah, last time we had some considerable playback issues.
14:47:53preglowand at least now we have excellent installer tools
14:48:30rasherIt's pretty much the only way a release will happen. And yes, disastrous will need to get fixed. But "high battery-consumption" or "sometimes freezes during playback" isn't disastrous.
14:48:39preglowrasher: agreed
14:49:16LloreanI don't think we need to release for iPods.
14:49:19rasherAs long as it works most of the time, for most people and isn't actively dangerous for the rest, I'd say it's ready for release (not an optimal situation, of course).
14:49:29preglowLlorean: i don't think we _need_ to either, but i think we should
14:49:33rasherLlorean: I really don't see why not.
14:49:51Lloreanrasher: Skippy playback when you use the equalizer, relatively terrible battery life?
14:49:57LloreanI'd like at least "features" to all work.
14:50:08 Quit XavierGr (Nick collision from services.)
14:50:09preglow"skippy playback on eq" is hard to solve
14:50:10 Join XavierGr_ [0] (
14:50:11LloreanMaybe if coprocessor playback is done in time.
14:50:13preglowwould put a release months back
14:50:21preglowbut cop support is also going to put us months back
14:50:23rasherLlorean: Personally I don't think those things are important enough if documented.
14:50:27preglowit'll introduce tons of bugs, i can guarantee you that
14:50:32Lloreanpreglow: My understanding is actually that Cop support is fairly nearby right now
14:50:40preglowi can still guarantee you tons of bugs
14:50:44preglowi'll eat something nasty if that doesn't happen
14:50:51preglowand even take pictures
14:50:58Lloreanrasher: But why "release" on the iPods that have poor battery life?
14:51:03 Join Febs_ [0] (n=chatzill@
14:51:04 Nick Febs_ is now known as Febs (n=chatzill@
14:51:09preglowhey, give amiconn some credit :P
14:51:12preglowthat'll be solved in weeks now :P
14:51:14LloreanWe have the H100, H300, X5, M5, and Gigabeat.
14:51:36LloreanI don't see why we "should" release on those targets.
14:51:42rasherLlorean: Because the ipods need a release as badly as the rest, if not more (considering it's the largest userbase by far)
14:51:56LloreanI don't understand what you mean by "need" a release.
14:51:56preglowbut we should be a bit careful, though
14:52:01preglowa bad ipod release won't help us much
14:52:07amiconnImo we could also release for 3rd gen, and perhaps 1st/2nd gen - if we want to release
14:52:15LloreanWhat you could do is issue a "Stable development build" for the iPod
14:52:18GodEater<devil's advocate> What is a "release" going to give the end user</devil's advocate>
14:52:19preglowamiconn: how is codec performance on those guys now?
14:52:27LloreanParalleling the release, but making it absolutely clear it's not release-ready yet
14:52:33amiconnBut that's also a bit tricky - even treble/bass can make medium bitrate oggs skip
14:53:02amiconnWe can increase CPUFREQ_MAX to 90MHz - I think that's safe now that the extra power drain is solved
14:53:13preglowwe need cop support :/
14:53:30rasherGodEater: nothing tangible, and we know that. But users like releases. It gives them peace.
14:53:44amiconnThe difference between playing mp3 at line level and playing mp3 with some treble/bass is clearly noticeable
14:53:47pondlifeamiconn: Is it not the 75 -> 80 MHz increase that's upsetting Nano's (timing or heat, or whatever)?
14:54:03amiconnThe latter increases buffering time _a lot_
14:54:07preglowmight be, the nano is pretty tightly packed
14:54:10rasherBesides, many users are pretty much expecting release-quality from the dev-builds right now anyway. Which is not acceptable either.
14:54:12*GodEater wonders how many end users will see the difference between a release and current build
14:54:18amiconn(but it doesn't skip with preset standard mp3)
14:54:20preglowrasher: and that is a good point, i think
14:54:28preglowa release will make it clear that dev builds are dev builds
14:54:31markunrasher: why not realease now then?
14:54:40amiconnpondlife: I doubt that. Then it wouldn't be a problem at boot
14:54:41pondlifeHas anyone given a positive reason for a "release"?
14:54:49 Join JajaComp [0] (i=d521d7aa@gateway/web/cgi-irc/
14:54:57pondlifeAs opoosed to "we should have one"
14:55:03rashermarkun: personally I don't think it's such a stupid idea (except for the current voice issues which are fairly important)
14:55:18pondlifeMaybe before COP or MOB are integrated.
14:55:29amiconnpondlife: I guess its either ata timing, or something completely different on nano that I overlooked so far
14:55:30pondlifeThose will both cause major bugginess I expect.
14:55:35markunJajaComp: managed to install the crosscompiler?
14:55:53Lloreanrasher: Why not release for the good targets (the ones I listed) and a "Stable development snapshot" for the iPods, so it's clear it's not at the same level, but offers users a stationary version?
14:55:53rasherpondlife: Sounds likely. Is anything big on the table besides those two?
14:55:57*petur wonders why the hurry...
14:56:03amiconnI don't have a nano ROM dump
14:56:07JajaCompi compile RockBox, but when i reboot ipod see error message "bad chacksum"... whot is?
14:56:16Lloreanamiconn: Want one?
14:56:36preglowpetur: because we've been developing swcodec for years now with no release?
14:56:38amiconnWell, now that you can make one... yes
14:56:39rasherLlorean: Maybe we need to make clear some way that Ipods are second-class citizens right now. Naming the "release" something else might do it, I guess.
14:56:49pondlifepreglow: Another year won't hurt, then ;p
14:56:59amiconnrasher: It's not only ipods, also H10
14:57:08GodEaterand sansa
14:57:09 Join TinoM [0] (
14:57:24rasheramiconn: of course, yes.
14:57:28amiconnSansa doesn't suffer that much regarding battery runtime
14:57:37pondlifeSeriously, we are making decent progress at the moment, why stop for a freeze? Either immediately before or a while after MOB would be my choice.
14:57:38peturpreglow: indeed, so what's the bit extra time going to make? And anyway, who is waiting for a release? I'd rather have some of current bugs fixed first...
14:57:39Lloreanrasher: I think it's pretty important that you make it clear that they're second class. People don't read release notes, and might thing that the shortened battery life is "regular"
14:57:51Lloreanrasher: I frequently see people explain it as 'Of course you get worse battery life... Rockbox does more'
14:58:12JajaCompwhat is error "BAD CHECHSUM"????
14:58:21GodEaterLlorean: well it cleans my fridge for me :)
14:58:22pondlifeIt would be better if the standard was "Rockbox extends your battery life".
14:58:32markunJajaComp: something wrong with your rockbox.ipod file
14:58:46preglowpetur: me too, but with no release talk, there'll always be tons of new stuff thrown in that will forever make bugs
14:58:46GodEaterJajaComp: possibly you built for the wrong target ?
14:58:47amiconnpondlife: It does that... just not on PP502x yet
14:58:48rasherLlorean: Fair enough, a differently-named release for the ipods+h10 is a good solution I think.
14:58:56pondlifeamiconn: My point exactly.
14:59:02amiconnLlorean: 403
14:59:13amiconnOn PP5002 it does :P
14:59:20pondliferasher: Or no release.
14:59:21peturpreglow: feature freeze didn't help last time...
14:59:22Lloreanpondlife: Which is why I suggest we only mark the Coldfires, Archoses, and Gigabeat as "Release", and the PortalPlayer as something second class, but "available" for use
14:59:27Lloreanamiconn: Silly me
14:59:45LloreanI always forget that if I copy the file from this computer, I need to redo permissions
14:59:59dionoea"Bata" for second class (most people would understand what that means)
15:00:03Lloreanamiconn: Try again?
15:00:13 Join Nico_P [0] (n=nicolas@rockbox/developer/NicoP)
15:00:13GodEaterdionoea: I like "Bata" it's more original ;)
15:00:18pondlifeI've still not heard a decent reason for any kind of release. "We're about to break music playback" is probably a good one, but would that need a branch too?
15:00:43amiconnLlorean: thx
15:00:50Lloreanpondlife: I think a "release" will help people understand the current build should be considered "in-development"
15:00:59rasherpondlife: I really don't think the current situation is good. It makes users expect things of the svn builds that they shouldn't. An analogy would be firefox. People know not to expect release-quality from nightly builds, and most stick with the releases
15:01:01LloreanSpecifically, that if they're updating constantly, they should expect things to change and possible break
15:01:10peturpondlife: maybe test a bit more before committing, then?
15:01:18 Join XavierGr [0] (n=xavier@
15:01:22 Join My_Sic [0] (
15:01:36Lloreanamiconn: I make no promises that it's any good. I can't assure my Nano's sanity.
15:01:36preglowpetur: didn't help, no, but it might now, who knows
15:01:43preglowif we are to keep in pretending we do releases, we have to try doing some
15:02:07pondlifeMany users would continue to update daily (or regularly) anyway, there's always new goodies.
15:02:11rasherpetur: I don't think the mission now should be a bug-free release (as it was last time), but a release that doesn't kill anyone.
15:02:19Lloreanpreglow: Instead of "Releases" we could just have "Stability-enhanced Rockbox" with a 1-month feature freeze once a year for bug-focus. :-P
15:02:30pondlife"Rockbox - no longer fatal!"
15:02:57amiconnLlorean: At least it looks like a typical ipod rom dump
15:02:58*Llorean wants to know who to talk to about voice files having the wrong clips in them.
15:03:02GodEaterpondlife: do you know of a case where rockbox has killed someone then ? :)
15:03:02rasherpondlife: but then they'd a) have a release to fall back to once they hit a bug. and b) know not to expect much
15:03:04preglowRockbox - no broken ears in the default install for 1 year
15:03:14Lloreanamiconn: Alright, good to hear.
15:03:19pondlifeGodEater: Ask rasher ^^
15:03:49pondlife"Rockbox - bricking iPods since 1995"
15:04:05rasherLlorean: I'm considering blaming genlang or perhaps voicefont
15:04:36*Nico_P is reading the logs but just wants to mention that a stable branch is far better than a feature freeze
15:04:36Lloreanrasher: All I know is that if I build a Gigabeat voice, it seems fine, and if I build a Nano voice, it's got the wrong clips in several of the menus
15:04:37markunJajaComp: the compilation went fine and then you did "make zip" ?
15:04:43pondlifeMaybe we just need to be able to mark particular daily builds as "good ones" ?
15:05:00peturpondlife: just thinking the same
15:05:10markunJajaComp: no to what?
15:05:10pondlifeNot sure how we'd decide. Most are pretty good.
15:05:18Nico_Ppondlife, petur: why not a stable branch ?
15:05:33pondlifeMore overhead.
15:05:48pondlifeWhen time is limited, I'd rather not be administrating multiple branches.
15:05:49peturmain brabnch should be stable
15:05:52 Join leftright [0] (i=d9e1f7cd@gateway/web/cgi-irc/
15:05:59*petur joins typo class too
15:06:01rasherA lot more, and it'd quite likely be neglected as devs work on the dev branch
15:06:13markunJajaComp: maybe you should read the wiki again:
15:06:21LloreanI think a stable branch would just stagnate.
15:06:33Nico_Prasher: I don't see how it could be wrose than the feature freeze
15:06:33pondlifeWe do need more bug fixing, but the main problem is that our current audio engine is not well structured and over delicate.
15:06:59rasherNico_P: because it'll be lower priority as a branch, because you can keep working on the dev branch
15:07:08pondlifeCurrently only a few people can really fix the bugs (and I don't count myself in that group).
15:07:24Nico_Prasher: at least it's stable, so users who don't like change can rely on that
15:07:27peturhonestly, here at work the checked in code should be well tested, because a snapshot can be taken at any time for testing+release. So why can't we keep up quality commits at Rockbox?
15:07:28dionoeaWhat kind of files do you play to make it hang/crash? I play oggs all day long on my iPod video and i've never had any of those.
15:07:30pondlife"SWCODEC - kills developers, dead!"
15:07:33*GodEater suspects playback.c is pondlife's personal bugbear
15:07:38amiconnLlorean: I still think that the nano problems might have to do with ata timing, but the fact that just going back to 75MHz with the new code doesn't seem to help doesn't fit in well
15:07:50pondlifeIt is. Looks at the bugs in it. Race conditions all over the place!
15:07:56GodEaterfix it
15:07:59amiconnHowever, the facf that the nano ata flash doesn't like going to sleep would support this theory
15:08:02pondlifeNo, it'll kill me.
15:08:12pondlifeLike it did to poor lostlogic.
15:08:18pondlifeI value my mind.
15:08:21GodEaterre-write it
15:08:25Nico_Ppondlife: btw, seen my latest work ?
15:08:29Lloreanamiconn: What's interesting is that sometimes the music plays, but corrupted, and sometimes there's a crash, suggesting to me that the loaded data is being corrupted. Which would line with the ata issue, right?
15:08:38amiconn(after sleep it might come up in a slow pio mode, not being able to properly communicate with the ata controller)
15:08:54pondlifeNico_P: Not yet, no. As long as it's got structure and comments, I'm happy :)
15:08:57leftrightbecause there is no current stable release everyone gets the latest because the assumption is that its better, then when theres a majpr issue you guys just say, "use a previous that works", as long as you dont have a stable release folks will always think that the recent current is the best and most stable
15:09:13pondlifeleftright: Generally the most recent is best.
15:09:15amiconnWe had a similar effect on archos with some old HDD models iirc
15:09:15Lloreanleftright: The vast, vast majority of the time it is the most stable.
15:09:21amiconnLinusN should know
15:09:24Nico_Ppondlife:;h=no_rb_in_rb ... heopfully you'll be happy
15:09:24pondlifeThe recent voice stuff is an exception.
15:09:56amiconnleftright: The most recent is almost always the best choice
15:09:58LloreanThe recent voice stuff isn't really a case of stability. It's a case of "it took a few days to finish working on it"
15:10:03markunleftright: would you prefer to use an older version of rockbox?
15:10:18leftrightif it works well why not
15:10:20amiconnEven 2.5 for archos has some major bugs which are fixed in svn
15:10:23LloreanSometimes things just simply don't become apparent until you go live.
15:10:39pondlifeMaybe we need a system where users can rate a particular daily build out of 10..?
15:10:55markunleftright: then you could also just not update rockbox if it runs well
15:10:59leftrightthen you need a different policy/method for dealing with major changes
15:11:10markunah ok
15:11:26markunjust release if there are major changes and it's stable
15:11:28markunmakes sense
15:11:34Lloreanleftright: Anyone updating their daily build regularly instead of picking one and sticking with it should consider themselves a tester.
15:11:44 Quit XavierGr_ (Read error: 110 (Connection timed out))
15:11:59 Part pixelma
15:12:02rasherLlorean: But they don't, unless they know that there are "releases" or something like it.
15:12:24Lloreanrasher: Partially because they don't read the manual, and partially yes because we don't have a clear release version
15:13:33leftrightmarkun: exactly, when instituting a major change only release it when its stable
15:13:41rasherYou shouldn't expect users to read the manual. It's a given that the mast maority of users won't read more than what's needed to get running.
15:13:44preglowpetur: even a 60 dB gain sounds perfectly ok :P
15:14:21Lloreanleftright: We haven't released it.
15:14:29 Quit XavierGr (Nick collision from services.)
15:14:31 Join XavierGr_ [0] (
15:15:00Lloreanrasher: If we have a release version, it'll just mean we get a new type of confused-user question.
15:15:06leftrightrelease is perhaps incorrect with your current method, perhaps not make it available to the public is better
15:15:13Lloreanrasher: Such as "I heard Rockbox could play NES games, I downloaded the latest release but it doesn't" or whatnot
15:15:56Lloreanleftright: So, password-protect the build page any time there's a major change so that only people we consider "testers" can get to it, thus decreasing the likelihood of people finding bugs?
15:15:56rasherLlorean: I think it's more manageble "It's under development and will be availble in the next release, or right now, if you install a *development build*"
15:16:04*amiconn leaves for a while
15:16:21leftrightmaking buggy builds availbe to the public is not a good idea
15:16:42*dionoea agrees with rasher. That's what most other software projects do anyway.
15:16:50leftrightor rather super bugy builds
15:16:56rasherleftright: Making buggy builds available is not the problem. The problem is that it's not apparent that they may be buggy
15:17:09Lloreanleftright: How the hell are people supposed to test development versions if they're not made available?
15:17:20LloreanThe problem isn't that the build's there. The problem is that people don't clearly know it's in-development
15:17:21rasherleftright: At least to the casual user who doesn't understand how development works
15:17:40leftrightnot all folks want to debug your software, they just want something that works
15:17:53markunbut we DO want them to debug it :)
15:17:54LloreanThen they should use the retail firmware, or wait for release versions.
15:18:05FebsOr find a version that works and stick with it.
15:18:10LloreanIf they're unwilling to accept the terms upon which the software is currently offered, they don't have to accept them.
15:18:19markunFebs: which could be called a stable release
15:18:28FebsIdeally, yes.
15:18:44LloreanI think at least a semi-stable release should be tried for soon (before the end of the year)
15:18:47 Quit aliask ("ChatZilla [Firefox]")
15:18:55bluebrothermaybe we should rename the "current build" to "development build"?
15:19:02leftrightor tag the currents as unstable somehow
15:19:11GodEaterbluebrother: that's a very good idea
15:19:12bluebrotherand "archived build" to "archived development build"?
15:19:21markunleftright: but it's usually not that unstable
15:19:22rasherBut when it's the only thing available, people won't have anything else to pick from
15:19:34Febsmarkun: but right now, a build may work perfectly for someone and yet still have bugs affecting features that person doesn't use.
15:19:50bluebrotherdisplay a splash message when no user cfg is present: this is a development version, use at your own risk
15:19:52Lloreanleftright: "The current build is built at each source code change to the Rockbox SVN repository, and represent the current state of Rockbox development. This means that the build could contain bugs, but is most of the time safe to use."
15:19:53leftrightbut every now and then the wheels fall off, so you need to advise the foks about that somehow
15:20:15rasherThe problem is that the current build is *all* we offer to most users
15:20:16bluebrotheror something similar ... the only problem is that this will increase binary size for no functional benefit.
15:20:39leftrightflag the table somehow
15:20:54rasherSo they see it as the regular build in spite of the warnings
15:21:10bluebrotheradd a big red blinking "warning, you are looking at unreleased stuff" at the page? ;-)
15:21:20Lloreanrasher: Well, if more users contributed proper bug reports and tested them, we might advance toward a stable version more quickly
15:21:28LloreanI mean, it's a two-way street, 'n all.
15:21:33FebsI think that it is a reasonable suggestion to send announcements to the user list when major changes are implemented.
15:21:48FebsAnd also post in the "Announcements" forum.
15:21:58markunbluebrother: but having to wait for years for a stable release while many 'unstable' builds have worked fine for many users is also a bit unfair
15:22:00rasherFebs: true. This should happen whenever something major happens
15:22:22bluebrotherwe could also add information about which player has a release on the front page.
15:22:48LloreanI don't see why major changes need to be replicated in several places.
15:22:49bluebrotheras the current list somewhat implies that all players are "done", and "additional models are in development"
15:23:02LloreanInvariably someone will forget one, or multiple, of the places.
15:23:22 Join XavierGr [0] (
15:23:27rasherLlorean: then have a few people responsible for putting out such warnings
15:23:29bluebrothermarkun: I never said users should wait for years for a stable release −− they just should be aware about the not-yet-released state
15:23:38Lloreanrasher: Why can't people read the MajorChanges page?
15:23:39leftrightsome targets are more mature than others, right now the assumption by joe soap is that build table is green so all is good
15:24:17GodEaterSoap knows the score!
15:24:19 Join pondlife1 [0] (
15:24:42rasherLlorean: that page is very brief, and don't really explain in detail what happened. Besides, it doesn't give a warning in advance.
15:24:43bluebrotherhmm, is there a way to export the MajorChanges page to simply html / txt? We could display that information in rbutil
15:25:36preglowhave anyone seen the german rockbox article someone posted about?
15:25:59rasherbluebrother: Certainly a page on could parse majorchanges and output it in a "clean" way
15:25:59leftrightLlorean, people will only read docs if they abosolutely have to, its a very frustrating humsn trait, folks want to click and everything must just work, they dont understand development work
15:26:17Lloreanleftright: Then tell them to stop updating.
15:26:18rasherbluebrother: the wiki markup is there for the server to parse afterall
15:26:48markunpreglow: which one?
15:26:52Lloreanrasher: Well there's also always the changelog. I just don't see why people should be expected to write explanations of everything they do.
15:26:54leftrightI would, but to what build do i point them to
15:27:09Lloreanrasher: I think it's the user's responsibility to know what they're downloading, otherwise people will get even more upset the one time someone forgets to do so.
15:27:14FebsLlorean: or simply advise them when there is a major change so that they don't update then if they don't want to . "We have made a major change today. You may experience instability with current builds, or feature X may not work as it has in the past." Then people can decide for themselves whether to update.
15:27:24Lloreanleftright: Tell them to find one that works for themselves, and stick with it?
15:27:46FebsBut at least they're put on notice that there has been a major change.
15:27:48preglowmarkun: pc magazin
15:27:54 Quit pondlife (Nick collision from services.)
15:27:56LloreanFebs: Again the problem with people then growing to expect that if they haven't been told "expect instability" then there will be none.
15:28:00rasherLlorean: you're expecting too much of the users. In the end, we want to make it easier for them. If not because we're nice, then because it makes our lives easier as well.
15:28:15LloreanFebs: I think it's better to nurture a knowledge that *all* development builds can be expected to be unstable.
15:28:27 Quit My_Sic (Connection timed out)
15:28:32 Nick pondlife1 is now known as pondlife (
15:28:56Lloreanrasher: Then why not get a release put up, and tell people who want stability to stick with it until the next release, rather than creating a false sense of security by randomly warning about instability sometimes, and completely missing it other times?
15:29:01FebsLlorean: of course. But that doesn't mean that we can't advise of major changes when we *know* that there will be a higher likelihood of instability, or at the very least a change from what they've become accustomed to.
15:29:16LloreanFebs: Changes from what they become accustomed to, sure.
15:29:32leftrightyou will be saving yourselves major headaches by having a stable build somewhere that you can point to
15:29:34LloreanAnd "there is a near certainty X will be broken for at least the next few builds" makes sense to.
15:29:47rasherLlorean: I'm saying we should do both. Once the releases are there, missing a "new feature" announcement isn't as big a problem, because people will know they are dev builds.
15:29:49Lloreanleftright: Feel free to start fixing bugs for a release then. I look forward to seeing them start vanishing from the tracker.
15:30:02leftrightnot a release, a stable build
15:30:04*GodEater thinks no-one will use a stable build, and everyone will still keep using the current one
15:30:13Lloreanleftright: Okay, either way, bugs need to be fixed
15:30:33LloreanFebs: I think people using the development version can subscribe to the developer list.
15:30:35rasherGodEater: but it'd make things a lot simpler. "Use the stable build if you just want something that works"
15:30:46LloreanFebs: I see no problem with sending out a message on the dev-list warning of upcoming large patches, we do that anyway most times.
15:31:04markunGodEater: depends on how ofter we release I think
15:31:16FebsThere's no reason such an announcement couldn't be sent to the user list.
15:31:43morrijror have an announcement list just for err, announcements :)
15:32:13pondlifeThe user ML is a bit underused, aside from the blind community.
15:32:14LloreanFebs: I think it is better if it's just for the dev list, though, to keep it clear that what you're doing is participating in development
15:32:23LloreanFebs: It's part of the whole "testers" vs "users" concept.
15:32:23FebsI disagree.
15:32:36pondlifeAll users are testers, no?
15:32:39FebsBy that logic, there should be no user list right now except for v2.5.
15:32:41rasherLlorean: depends how you're wording the mail, in my opinion
15:32:41 Quit JajaComp ("CGI:IRC (EOF)")
15:32:54*GodEater agrees with Febs
15:32:56leftrightsome people dont want to be testers, they want smething that works
15:33:01rasherpondlife: as things stand now, yes. But some users don't know it, or dont want to be testers.
15:33:04pondlifeWell, OF normally works.
15:33:10LloreanFebs: I'm basing my statements on the assumption that we're going to try to have a release.
15:33:17LloreanOr regular releases even.
15:33:27pondliferasher: OK we need more warnings before they download maybe?
15:33:39LloreanWe don't need to be suggesting to people that these are "Updates", but rather "Patches being rolled out for testing"
15:33:47LloreanAnd I think that's very definitely a "Development" topic.
15:33:52rasherpondlife: Won't work as long as they have to download that build anyway.
15:34:17Lloreanleftright: People who don't want to be testers need to wait to download Rockbox until a release happens.
15:34:34leftrightyes but when
15:34:38pondlifeI don't see the point in having releases unless we also branch and fix them - and I'm not in favour of doing that at all.
15:34:44Lloreanleftright: How long will it take you to make Rockbox stable?
15:34:56rasherpondlife: with freezes prior to release, that's effectively the same
15:35:06Lloreanleftright: Saying "Rockbox needs to have a release" is all well and good, but you and others need to work on it.
15:35:13GodEaterrasher: no it's not - what about fixes post release?
15:35:14pondliferasher: No, because the more serious bugs never get fixed.
15:35:27*GodEater high fives pondlife
15:35:31Lloreanpondlife: That's why you need a pre-freeze checklist.
15:35:47Lloreanpondlife: Things that need to be done before a freeze can be entered, so it's hanging over everyone's head.
15:35:53 Quit daurnimator (Read error: 104 (Connection reset by peer))
15:35:55GodEaterthat still doesn't help
15:35:56rasherGodEater: the assumption is that any bugs left in the release is not serious enough that people can't wait for the next release
15:36:07 Join daurnimator [0] (n=fake@unaffiliated/daurnimator)
15:36:10leftrightLlorean: I'm not saying a release, to help the user and make your life easier you should always have a "Stable Build" that you can refer people to
15:36:13GodEaterthat's an awfully big assumption
15:36:25GodEaterleftright: that's the same thing with a different name
15:36:33Lloreanleftright: Alright, so either tell me which build is that stable build, or get to work eliminating bugs until there's a stable enough one
15:36:36dionoeas/bugs/known bugs/ ?
15:36:36pondlifeLlorean: I would love a stable release, but do we have people who are interested in that and have the time needed to understand the problems and fix them, within a reasonable period?
15:36:42Lloreanleftright: Producing a stable build is the same problem as producing a release.
15:36:48 Join DataGhost [0] (
15:36:50dionoeaand then you'd have a stable branch where you only fix bugs (and not add features)
15:36:52rasherGodEater: I don't think so. Any bugs so major that it absolutely needs a fix can't hide for weeks without being hit.
15:37:10leftrightwith software there is only more stable and less stable
15:37:12GodEaterso what about the bugs still in 2.5 ?
15:37:20Lloreanpondlife: Depends both on the level of stability you define as a release, and on whether you decide you want "A release by X date" or "A feature freeze of X days *after* these key bugs and features are resolved"
15:37:26rasherGodEater: That's a different situation, since it's so old now.
15:37:39preglowanyone know what the impedance of the h120 input is? :>
15:37:41GodEaterrasher: but we'd still end up in the same situation with any release we made now
15:37:43pondlifeLlorean: It would have to be the latter IMHO
15:37:48 Quit XavierGr_ (Connection timed out)
15:37:57Lloreanpondlife: That's what I've been advocating since about the day the 3.0 feature freeze was cancelled.
15:38:05GodEaterIMO to do a release properly you MUST branch
15:38:07 Quit XavierGr (Nick collision from services.)
15:38:08pondlifeStability level - doesn't need the pin to reset it once a week.
15:38:09 Join XavierGr_ [0] (
15:38:21pondlifeOnce a month is ok :)
15:38:22Lloreanleftright: Yes, so someone needs to work on making Rockbox more stable.
15:38:48leftrightuhuh, but you wont get that if you add new feautres all the time
15:39:07rasherGodEater: the 2.5 situation is entirely different, and can't be compared to a new release, because the new release should come with a promise of a next release as well.
15:39:13bluebrotherhmm. How about a bugfixing month?
15:39:17GodEaterjust like 2.5 did....
15:39:35Lloreanbluebrother: Major new features need to be either finished, or put on hold before first commit, before that can happen
15:39:48pondlifeI think we need to work on spreading the knowledge about how lower-level stuff works. Rockbox cross-training....
15:40:11bluebrothersure ... but such a period could be used to stabilize and fix bugs in the current state
15:40:18Lloreanleftright: Just because features are being added doesn't mean you can't fix bugs.
15:40:24bluebrotheryou need to have some kind of feature freeze of course
15:40:29Lloreanleftright: But yes, a feature freeze will be necessary at some point.
15:40:31GodEaterbranching at each release doesn't lose us anything, and it saves us headaches if we don't / can't stick to some intended release schedule
15:41:06bluebrotherbranching after the release has finished or before it?
15:41:09rasherGodEater: if you mean branching and then working on in the dev branch, I think that's a very bad idea. If you mean "feature freeze, release, then branch"
15:41:16GodEaterbluebrother: at the point of release
15:41:21rasherGodEater: then I'm all for it, to give the possibility of backporting fixes.
15:41:23dionoeayou usually branch before the release hits final
15:41:28rasherGodEater: then I think we agree anyway
15:41:32dionoeaelse people can't work on new features anymore
15:41:38 Join miepchen^schlaf [0] (
15:41:41dionoeaand that just slows down the dev process
15:41:42GodEaterrasher: your second example, not the first
15:41:46GodEaterthe first would be silly
15:41:53dionoeayou can't force people to only do bugfixing
15:41:58bluebrotherso you will have a branch wich will only get used for fixes after the release has made? Maybe.
15:42:03rasherGodEater: it's not unheard of to let development continue during a freeze
15:42:12rasherGodEater: but I don't think it'd work well for Rockbox
15:42:16GodEaterrasher: then have multiple branches
15:42:18pondlifeFreeze, then release/branch.
15:42:28Lloreandionoea: But you can force people not to commit any new features in the hopes that they'll do bugfixing.
15:42:30GodEaterone at feature freeze start, and one at release, and keep a dev branch
15:42:32rasherpondlife: sounds reasonable. Release and branch happens at the same time
15:42:41pondlifeGodEater: No, only one current and one stable branch.
15:42:47leftrightfeatures and improvements will always happen with software, but you need to develop a system where, a stable release will only be released to the public after a new feature which has been incorporated is stable
15:42:47GodEaterpondlife: why ?
15:43:03rasherGodEater: I don't think you should keep a dev branch open during the freeze. It takes a lot away from the point of the freeze
15:43:09Lloreanleftright: Yes, that's called "release versions" or "stable versions" and we've covered that.
15:43:12dionoeaLlorean: and then you end up freezing most work on trunk for months at a time. (at least that's what happened for VLC before we decided to have a stable branch)
15:43:18Nico_Pa typical pattern is to have a stable branch from which releases come from
15:43:22pondlifeWill anyone fix bugs at all? The more branches, the more dilution of our Rockbox time.
15:43:24Nico_PI don't see why we couldn't do that
15:43:47Lloreandionoea: If the dev branch is kept open during the freeze, a large majority of people will just continue working on it.
15:43:50rasherIf you branch at the point of the release, you lose nothing, but gain the possibility to backport fixes (easily)
15:43:51GodEaterrasher: well that's the only way I can see to be tidy about allowing dev to continue during a freeze
15:44:02*pondlife assumes that the stable branch will only be worked on while there's no fun elsewhere.
15:44:11Lloreandionoea: Even when we didn't have a dev-branch, and froze for bugfixes last time, we never reached 'stable' because everyone kept working on their own thing, and asking us to end the freeze so they could commit
15:44:25pondlifeGodEater: Dev does continue, fixing bugs!
15:44:26leftrightPondlife: will anyone fix bugs at all...... does anyone have pride in their work,
15:44:26dionoeawhich is exactly why you need a stable branch
15:44:33Nico_PLlorean: that was because of the lack of branches
15:44:41dionoeaso fixes can get commited on that branch but not the new features
15:44:43LloreanNico_P: How can we have a stable branch until we *get* stable?
15:44:52peturwe need another tracker cleaning week, and focus on the bugreports...
15:44:57Nico_PLlorean: with branches people have a choice: fix bugs or work on new features... hopefully they'll do a bit of both
15:45:05rasherGodEater: I don't want to allow dev to continue during a freeze. I'm expecting devs to fix bugs and work on features locally, and as a secondary thing. Of course this doesn't work if the freeze is several months. It needs to be no more than a month I think.
15:45:14Nico_PLlorean: we branch and then stabilize
15:45:22Nico_Pby not adding too much new features
15:45:25dionoeaLlorean: stable branch basically means "don't commit new features(aka bugs) here"
15:45:32pondlifeNo, stabilize first, so both branches benefit.
15:45:35GodEaterrasher: so let the devs manage their own local branches
15:45:50rasherGodEater: exactly. No official dev branch during the freeze.
15:45:51pondlifeFix/freeze for a month, then branch/release.
15:45:57dionoeapondlife: you usually fix bugs in trunk and backport to the stable branch
15:46:01*GodEater has got too used to the git mindset already
15:46:07dionoea(unless the stable branch and trunk are too far appart)
15:46:08Lloreandionoea: If that happens, very little work will ever focus on the stable branch.
15:46:18Nico_PLlorean: I'm not sure about that
15:46:20peturHow about: please test better before committing?
15:46:22dionoeavery little is better than none IMO
15:46:22Lloreandionoea: I suspect the "too far apart" will happen, and then most people "won't have the time"
15:46:29markunGodEater: I'm getting used to it myself :) But don't actually use git yet
15:46:32leftrightduring all feature freezes, its a tiny handfull of dev's that rool their sleeves up and fix others bugs
15:46:44GodEaterLlorean: but NOT doing it means you can't make any fixes at all to the "stable" code
15:46:47 Join BigMac [0] (
15:46:49rasherpetur: it's a nice thought, but in practice you can't beat the testing of all users.
15:46:54pondlifepetur: The main problems are more fundamental, and in old code...
15:47:01dionoeaLlorean: and the few times i've experienced that method it worked really well
15:47:10GodEatermarkun: I used it properly all the way through developing my shortcuts plugin
15:47:12GodEaterit was great!
15:47:15dionoeabetter than just freezing the main dev tree
15:47:19Nico_PLlorean: dionoea's right, it works for many projetcs
15:47:21LloreanGodEater: I'd much rather a stability-only (no features at all) push for a period (1month? 2months?) *then* a branch, whether we release or not
15:47:39LloreanA kick-start as it were.
15:47:44GodEaterLlorean: that would be nice - but it's not gonna happen ;)
15:47:46*Nico_P has been reading Karl Fogel's "Producing OSS" :)
15:48:04LloreanNico_P: The only thing that really ruined our feature freeze last time was that it wasn't a feature freeze.
15:48:15LloreanIt involved the addition of "Database" and the attempted playback engine rewrite.
15:48:15leftrightthere seems to be a fundamental problem where programmers push products "over the wall" which aren't ready and expect others to do the dirty works of debugging xyz
15:48:19markunNico_P: don't give away the ending yet, it's too much fun to figure it out on our own
15:48:28LloreanWithout either of those, there's a decent chance we'd have a 3.0 out for a year now
15:48:50Lloreanleftright: 1) This is open source. 2) No "product" has been "pushed"
15:48:52GodEaterleftright: until you become a dev, you don't get to complain
15:48:56pondlifeleftright: I don't think there's too much of that here.
15:49:07Nico_PLlorean: the playback rewrite was necessary for the release
15:49:13LloreanNico_P: No, it really wasn't.
15:49:19LloreanIt was wanted for the release, which is different.
15:49:29pondlifeWhich release? Last year's non-release?
15:49:43pondlifeYep, playback suffered through that a bit.
15:49:46markunonly 1 year ago? Looks like much longer
15:49:48rasherThere were problems with the old playback engine, but they were mostly known, and could be released with known bugs.
15:50:05Lloreanmarkun: Actually it's nearly two, I think
15:50:17pondlifeTime flies when you're debugging playback.c...
15:50:20Lloreanrasher: Or just those bugs could've been mitigated or worked around for the release
15:50:54LloreanEven if it was a dirty hack just to make it seem to work right, from a user perspective it would've worked.
15:51:00Nico_P it was planned for 1st May 2006
15:51:55Nico_Panyway I think we desperately need branches
15:52:24Nico_Pat least a stable one from which users can expect minimum reliability
15:52:29pondlifeCurrently there are 39 music playback bugs... some of which are not really playback bugs, but most of which are non-trivial to fix properly.
15:52:56LloreanNico_P: As I said, I've nothing against a stable branch, but I think a hard freeze should happen first in an effort to get as much fixed for that branch as possible.
15:53:07rasherThere's no one saying a release has to have 0 bugs, or even close.
15:53:17rasherEven known bugs.
15:53:18GodEaterthat be insanely unrealistic
15:53:28JdGordonNico_P: maximum reliability from stable branhc you mean....
15:53:46Nico_PJdGordon: I meant "a minimum level of reliability"
15:53:55LloreanJdGordon: I think he means "there's a certain minimum level that reliability never dips under"
15:54:07*JdGordon tired and confused
15:54:12pondlifeNo, but known issues such as "Starting playback directly after a recording ends in a data abort." are a bit off.
15:54:26GodEaterpondlife: fix it!!!
15:54:33pondlifeI don't have an iPod...
15:54:41Nico_Pwe also need to use flyspray's release management features better
15:54:50pondlifejhMiksS knows the cause though...
15:54:54rasherpondlife: but if it's not easily fixed, I don't think it should hold up a release
15:55:08pondlifeThat was just one (maybe not great) example...
15:55:21Nico_Prasher: agreeed, but a workaround should be implemented
15:55:46 Part leftright
15:55:49JdGordonworkarounds are bad
15:55:54pondlife"Bug fixing fortnight" should happen, certainly.
15:55:56LloreanJdGordon: Not for "release versions"
15:55:56Nico_P...better than nothing
15:56:08LloreanIf it's a release version, the key is *working* not elegance in code.
15:56:15LloreanYou can revert all the nasty hacks right out immediately after if you must.
15:56:19JdGordonnot really.. it means that the bug appears fixed and is ussually forgotten about
15:56:28LloreanJdGordon: That's why you revert the hacks
15:56:34pondlifeHacks upon hacks....
15:56:43JdGordon_if_ you can be bothered and remember about them
15:56:45rasherActually, that's why you put the hacks in the stable branch, and fix it properly in the dev branch
15:56:48***Saving seen data "./dancer.seen"
15:56:54Lloreanrasher: Same thing.
15:56:57Nico_PLlorean: actually you don't revert the workarounds, you commit them in the release branch :)
15:57:08pondlifeHah, hacks leading to stabliity...whatever next.
15:57:08LloreanNico_P: Yes, yes. I was simplifying the situation. :-P
15:57:19LloreanHacks can lead to stability.
15:57:25LloreanMaybe at a loss of performance, or maintainability though
15:57:27JdGordonhmm.... this starts to sound ok
15:57:50bluebrotherloss of maintainability?
15:57:55LloreanBut as far as I'm concerned, with a "release" or "stable" version, you concern yourself with what the user sees, not what coders see.
15:58:02JdGordonwho decides when to update the stable branch?
15:58:03bluebrotherI don't think that would be a reasonable price.
15:58:04Lloreanbluebrother: Harder to fix the code later if something new goes wrong.
15:58:10rasherbluebrother: not as important if it's in the stable branch
15:58:21Lloreanbluebrother: The whole point is that they aren't left in the "real" code. They're only there for the user-version.
15:58:29Nico_PJdGordon: maybe a kind of approval would be needed for makor changes to go in the stable branch
15:58:37Nico_Pand thourough testing
15:58:49JdGordonso /me would never commit to that branch :p
15:58:53pondlifeWould users use the stable branch? It seems there are plenty who choose a custom build rather than SVN as it is.
15:58:56rasherNico_P: I'm still of the opinion that the stable branch should be branched from the dev branch once in a while
15:59:10bluebrotherwouldn't a freeze period which is only used to fix bugs help development as well?
15:59:15 Join Jon-Kha_ [0] (
15:59:16JdGordonrasher: multiple stable brnaches?
15:59:25rasherJdGordon: well, then you close the old stable branch
15:59:34pondlifeOnly the most recent would get any fixes, I suspect.
15:59:40Nico_Prasher: I don't hink so, but the stable branch would need to stay in sync
15:59:41JdGordonrasher: whats to say the new stable is actually more stable than the old one?
15:59:47bluebrotherafterwards there could be a (of the nearly-release state) branch and include only workarounds for known bugs
16:00:27JdGordonif we had branches, we should have a fair few... how about an AA branch? and a custom list position/scroll/width branch?
16:00:27rasherJdGordon: there's no guarantee, but you freeze and branch to attempt to avoid it
16:01:08 Join low_light [0] (i=c730180b@gateway/web/cgi-irc/
16:01:12*Febs runs screaming from the support morass that so many branches would become.
16:01:18LloreanJdGordon: Why not just have that code be finished in a commitable manner?
16:01:23LloreanI think three branches is good.
16:01:31LloreanOld-stable, New-stable, and Dev
16:01:57LloreanNew-stable only existing if for some reason Old-stable becomes unmaintainable, and then old-stable is closed once the new-stable is deemed "good enough)
16:02:12Lloreanunmaintainable == some change is too difficult to sync
16:02:16 Join Dylan7262 [0] (i=7cbaf29e@gateway/web/cgi-irc/
16:02:29Dylan7262Hey people
16:02:49Nico_PLlorean: Dev == trunk, and stable branches could actually be release branches
16:02:50Dylan7262just wondering if someone can help me please? i have a 5th gen iPod video
16:03:02LloreanNico_P: Yes, dev == trunk.
16:03:16LloreanNico_P: I guess in my head, trunk is arguably still a branch in a manner of speaking.
16:03:17JdGordonwell, branches are cheap arnt they? why not allow everyone to create branches for their vairous WIP's? (everyone is obviously the current commiters.. not literally everyone)
16:03:28pondlifeWe only need to maintain one stable release, right?
16:03:52bluebrothereveryonce can setup his personal branch at home ...
16:03:53Lloreanpondlife: Yes, but if you can't sync it for some reason, then you need a place that's not trunk, but also not that, to work on getting a new stable release rolling along.
16:03:53Nico_Pthe stable branch I was mentioning should be somthing like "release_3.0.x"
16:03:57bluebrotherjust think of git-svn
16:04:03pondlifeLlorean: Sync it?
16:04:14bluebrotherbranches are cheap, but merges?
16:04:20Nico_PJdGordon: us git fot that
16:04:27Lloreanpondlife: If some significant bug-fix or stability enhancement in trunk can't be backported effectively, and it's decided that it's better to re-branch.
16:04:31Nico_Pbluebrother: merges are not so cheap in SVN
16:04:33pondlifeI assumed that it would be feature-frozen.
16:04:43bluebrotherNico_P: that was what I wanted to say ;-)
16:04:48Lloreanpondlife: You freeze the features, but backport bug-fixes.
16:05:06Nico_Pbluebrother: yeah, I think we agree :)
16:05:07pondlifeSerious bug fixes that can't be back-transported would be a reason for a new release from the trunk.
16:05:10bluebrotherand while having a feature freeze development of new features can go on the same way: in local trees.
16:05:12Lloreanpondlife: You can also backport features if they're considered really bug-free, in preparation for the next release.
16:05:39pondlifeStable = bug fixes only, and careful ones at that.
16:05:45 Quit Toki (Read error: 104 (Connection reset by peer))
16:05:48pondlifeDev = all else
16:05:54Lloreanpondlife: Then how do you put together "next release"?
16:06:01rasherLlorean: branch from trunk?
16:06:05rasherafter a freeze
16:06:09Nico_Pwith a new release branch
16:06:20Lloreanrasher: But couldn't you also have "stable", "next release" and "trunk"?
16:06:29Nico_PI say the freeze should affect the release branch after it's created
16:06:38LloreanStable gets only bug fixes, next release gets bug fixes and "non-buggy" features.
16:06:42LloreanTrunk is everything else
16:06:42bluebrotherwhy a "next release"? That would require merging all the time
16:06:44JdGordonyeah, freeze never happens on trunk
16:06:44rasherLlorean: I think you should only create "next release" at the point stable gets unmaintainable or a new release is wanted
16:07:08 Join BobShield [0] (
16:07:16 Join DerPapst [0] (
16:07:35rasherDylan7262: you should just ask your question
16:07:42Lloreanrasher: If you have an in-sync next release, you can publish it every 3 months as the new current-release, and simply apologize that few features have made their way into it rather than having to do hard freezes or anything
16:07:48bluebrotherI think it would be better to just branch upon a release, and only bugfixes for the released version should get there
16:07:50LloreanOr even publish it monthly
16:07:53Nico_Prasher: with releases close enough from one-another, the stable branch wouldn't be necessary... it would just be the current release branch
16:08:07rasherNico_P: that's my thinking as well
16:08:15rasherLlorean: but that'd require keeping "next release" in sync
16:08:17LloreanI just think that there's no reason that once a feature has been "tested" it can't be backported to some "stable-ish" version.
16:08:17 Quit Dylan7262 ("CGI:IRC (Ping timeout)")
16:08:36LloreanRather than keeping it mixed in with all the unstable features
16:08:37 Quit Jon-Kha_ (Read error: 131 (Connection reset by peer))
16:08:38bluebrotherbut backporting can be quite painful ...
16:08:59bluebrotherso how realistic ist that someone will do it?
16:09:10JdGordondepends on the bug
16:09:15rasherJdGordon: not bug, feature
16:09:28LloreanWell, someone has to either backport "good" features, or fix/remove all "bad" features before a release.
16:09:31JdGordonhang on, features dont get backported
16:09:44rasherJdGordon: in Llorean's world they do!
16:09:46LloreanJdGordon: Depends on how you look at it.
16:10:07JdGordonnew features should mean a new stable
16:10:28bluebrotherthen we would need to release much more often.
16:11:21 Part wippeout
16:11:44rasherMy proposal: Once a release is wanted/needed (old release is no longer maintainable), a new branch is created from trunk, and work focuses on bugfixing that branch. At some point it's released, and after that, work continues on trunk, and bug-fixes (and only bugfixes) are backported to the release-branch and the release is updated with a point-release.
16:11:46bluebrotherwell, I wouldn't object in releasing more often
16:12:36rasherAnd yes, releases should happen more often, possibly a couple of times per year
16:12:39bluebrotherbut that would need to not add multiple features at once, and possibly add a much shorter fixing freeze after a new feature comes in
16:12:39rasherif not more
16:12:44bluebrotherfollowed by a release.
16:13:02Nico_Prasher: I think I agree with your proposal
16:13:08JdGordonwell.. how often does a major new feature go in? not very...
16:13:19JdGordonbranhcing to a new stable on each big one would be fine
16:13:32bluebrotherI'm not sure.
16:13:33Nico_PJdGordon: it would go in the trunk, then in the new major release
16:13:54JdGordonNico_P: yeah, i skipped a step,...
16:13:57bluebrotherat least we would need to feature freeze trunk after a new feature goes in for, say 2 weeks and concentrate on fixing
16:14:09JdGordononce the branch happens no moer bug fixes should need to be put in
16:14:34JdGordonthen rebuild all the stable builds on each enw commit to than branch instead of the dailies
16:14:54Nico_Pbluebrother: why ? the feature goes in the trunk, then gets tested/fixed until the next release branch
16:15:05Nico_Pthen in the release branch it gets fixed even more
16:15:20JdGordonno, it sholdnt go into release untill its bug free
16:15:28JdGordonthats the whole point of the release branch
16:15:32austriancoderpetur: ping
16:15:34rasherJdGordon: this is a utopian ideal
16:16:02Nico_PJdGordon: the release branch would be only for fixes to the features it contains
16:16:06JdGordonrasher: obviously... but if no fixes get commited for a few days or a week it can be considered bug free and branched
16:16:21JdGordonNico_P: yes, but not to the feature that caused the branch
16:16:25Nico_PJdGordon: obvioulsy, we'd need to agree on what changes to include in a release
16:16:58bluebrotherI don't think having multiple branches will help much
16:17:11Nico_PJdGordon: the way I see it, a feature doesn't cause a branch. It's a release decision that causes a branch
16:17:12bluebrotherit rather looks to me as it would create more hassle ...
16:17:20rasherI really believe very strongly that the model I proposed would work well. Using "a huge feature went in" as a reason to branch also works in my model.
16:17:33Nico_Psay we decide to release 3.0 soon, then we create a 3.0 branch and nothing goes into it but fixes
16:17:38rasher"a huge feature that we want to release", that is.
16:19:51peturaustriancoder: bad timing... work stkov... what's up?
16:21:13Nico_Prasher: maybe it would be good to post a proposal on the ML
16:21:42 Quit idnar (Nick collision from services.)
16:21:45 Join idnar_ [0] (i=mithrand@unaffiliated/idnar)
16:21:47rasherNico_P: I'll put something together. Basically re-stating what I just said.
16:22:04Nico_Pcool :)
16:22:05austriancoderpetur: I am working in integrating the current stack. I need to call usb_stack_start() and usb_stack_stop() to start and stop it, but usb_enable seems to be never called with argument true. So at the moment I am doing the start call in usb_detect, which leads that start() is called very very often.
16:22:21austriancoderpetur: but its not that important and we can talk about it later the day
16:23:33peturaustriancoder: I'm having real life obligations tonight, it could be quite late when I pop in again...
16:24:02austriancoderpetur: no problem
16:24:09peturaustriancoder: who calls usb_enable?
16:24:56austriancoderpetur: usb_slave_mode() in firmware/usb.c and the bootloader
16:25:21peturso whenever a connection is detected?
16:27:17austriancodernope - usb_tick checks for the connection and notifys others: queue_post(&usb_queue, USB_INSERTED, 0);
16:28:14austriancoderpetur: cant find a place where usb_slave_mode(..) is called in the source
16:28:31peturthat would explain it, no?
16:28:52preglowamiconn: you wouldn't happen to know anything about the effects of not properly matching the impedance of the cartridge at the amp input stage?
16:29:04peturI would expect to start the queue on USB_INSERTED
16:29:20 Join Adikdid [0] (
16:30:12Adikdidhi can someone help me?
16:30:48Adikdidhow do i get album art to shoy up on my 5th gen iPod
16:31:08austriancoderpetur: so for the moment I will enable and disable the stack in usb_detect - later we can cleanup the usb stuff
16:31:24peturAdikdid: not supported officially, go see the forum
16:31:43AdikdidAh, damn
16:31:59markunAdikdid: here's some info
16:32:07peturaustriancoder: why are you getting multiple usb_detect?
16:32:45austriancoderpetur: i dont get multiple detects... i have the problem that after a connect usb_enable is _not_ called, where i enable/disable the stack
16:33:24peturhmmm, you said above "So at the moment I am doing the start call in usb_detect, which leads that start() is called very very often."
16:33:40 Join nerochiaro [0] (
16:34:15austriancoderpetur: correct... in usb detect i have something like:
16:34:35austriancoder if (connected == false) { usb_stack_stop(); return false; } /* do we want autodetection? */ if (usbcore.mode == AUTOMATIC) { logf("automatic mode"); UDC_OTGSC |= 0x5F000000; if ((UDC_OTGSC & 0x100) == 0) { usb_controller_select(HOST); } else { usb_controller_select(DEVICE); } } /* TODO correct place? */ usb_stack_start(); return true;
16:34:35 Quit low_light ("CGI:IRC (EOF)")
16:34:38peturanyway, I don't know the relation between usb_enable and usb_detect. Do you want to start the stack at boot or at connection?
16:34:49peturaustriancoder: use pastebin please
16:35:30austriancoderpetur: so if i do the enabling in usb_detect, I dont know for what usb_enable exists!!
16:35:57austriancoderpetur: I want to start it under a running rockbox, when i insert the usb cable
16:36:10peturaustriancoder: don't assume the current code is correct... check it...
16:36:23austriancoderpetur: okay..
16:36:38peturstarting at connection would be best
16:37:29 Join Domonoky [0] (n=Domonoky@
16:37:36peturyou probably will need to stop playback and stuff to claim the iram
16:38:04 Quit JdGordon (Remote closed the connection)
16:38:10*austriancoder will spend some hours later the day to understand fully rockbox code
16:38:51peturdon't forget to check the tx problem too ;)
16:39:12peturwork always piles up after holiday :p
16:40:42Nico_Paustriancoder: It looks like we're at about the same point... I'm starting to try integrating my MoB work in playback.c :)
16:40:52austriancoderpetur: jep... I only need to write selection of usb device driver and then I want to do the first commit. WIll have only two dummy drivers, but for testing its okay.
16:41:42 Join Adikdid_ [0] (
16:42:21Adikdid_hey where do i actually download the album art patch?
16:42:57austriancoderpetur: and maybe i get this also working: idea: all drivers should be something like rockbox plugins... rockbox usb drivers.the app code scans for usb drivers and register them automaticly in the stack.We only need one *.c file for a driver and we dont need any *.h files with onemethod in it: usb_<drivername>_init(void);
16:43:12Nico_PAdikdid_: are you able to build rockbox from source ?
16:43:22Adikdid_what do you mean?
16:43:25DomonokyAdikdid_: from the patch tracker (flyspray) but you have to compile.. or use an inofficial build (see forums)
16:43:53Adikdid_i cant domonoky, it wont let me sign upto the forums
16:44:01Adikdid_can someone please send it too me over here?
16:44:07peturaustriancoder, Nico_P: I don't think it is important that commits are done soon enough, make sure the code is well tested and has no bad effects for users...
16:44:28DomonokyAdikdid_: you dont need to signup for the forum, to just see some thread..
16:44:32pondlifeYes, and make patches available in the meantime.
16:44:39Nico_Ppetur: I have a git repo on which I plan to do my work
16:44:56*austriancoder has only online patches
16:45:06peturNico_P: good... nobody presses for commits atm
16:45:07Nico_P... I've already started the rest of the work on another repo
16:45:16pondlifeNico_P: I'd like to see a Flyspray entry with patches against SVN too... git can do that quite easily, right?
16:45:23Nico_Paustriancoder: I can give you push access to my public repo if you want
16:45:52Adikdid_is this thread right?
16:45:53peturdon't mix the developments please
16:45:58Nico_Ppondlife: currently I have only one commit in my mob branch for rockbox git repo...
16:45:58austriancoderNico_P: I am not very familiar with git.. at the moment I dont see a need for it
16:46:18austriancoderNico_P: do you need a linked list in your project?
16:46:33Nico_Paustriancoder: believe me, it's great ;)... I loved it for offline work
16:46:40Nico_Pyes I have one
16:46:50DomonokyAdikdid: thats the patch tracker.. so if you want to use a patch from there, you have to be able to compile rockbox yourself..
16:47:06bluebrotherhi Domonoky
16:47:10Nico_PAdikdid_: seen ?
16:47:11peturNico_P: I will set one up at home too when I get the time :(
16:47:17Domonokyhi bluebrother :-)
16:47:36bluebrotherI was wondering if it's a good solution to extend ZipInstaller to also be able copying files instead of extracting them.
16:47:41Nico_Paustriancoder: are you going to need a linked list ?
16:47:45 Join My_Sic [0] (
16:47:49bluebrotherthat way we could reuse it for the voice files. What do you think?
16:48:17austriancoderNico_P: yep.. and I am using this one:
16:48:24Nico_Ppetur: is good start in case you don't know about it
16:48:28Adikdid_Nico_P - i cant find where to downoad it?
16:48:34Domonokybluebrother: sounds good, just make a parameter to skip the unzipping..
16:48:36*petur knows
16:48:51Nico_PAdikdid_: a patch is a tect file you apply to the source code
16:49:10bluebrotheryeah −− I thought about a setUnzip() which is true per default −− if it's false just copy the downloaded file to the target location
16:49:25Adikdid_ok, so how do i do this, then?
16:49:27Domonokythats good..
16:49:31bluebrotheronly "ZipInstaller" might be a bit ... off after that ;-)
16:49:36bluebrotherbut I can live with it
16:49:50bluebrotherah, and what do you think about adding doxygen comments to the sources?
16:50:13Nico_PAdikdid_: one of the first lines of says this : "If you don't know what a patch is or how to apply one, you should have a look at the SimpleGuideToCompiling."
16:50:25Nico_Pwith a link to
16:50:57Domonokybluebrother: we could rename the ZipInstaller class :-) doxygen comments would be nice.. but not too much comments.. doxygen comments tends to comment usesless things :-)
16:51:01Nico_Paustriancoder: looks nice... I wrote quite a simple one
16:51:19Adikdid_yes, but i dont get it
16:51:27austriancoderNico_P: I was lazy and tooked the liked list impl from linux kernel
16:51:41Nico_PAdikdid_: then you should head to the forum and download an unofficial build
16:51:45bluebrotheruseless comments? I'm usually quite brief in doxygen comments, what could be useless there?
16:51:57bluebrotherIt just makes browsing the code nicer imo :)
16:52:03Adikdid_well thats what i have been talking about fro the start
16:52:13Adikdid_a link to a unoficcial build. haha
16:52:35Domonokybluebrother: often the comment just tells the same as what the function name implies.. but we can do better :-)
16:52:53bluebrotheruse cryptic function names? *g*
16:53:01 Quit Adikdid (Read error: 110 (Connection timed out))
16:53:04austriancoderNico_P: we need to choose one linked list impl, which we both uses
16:53:14Nico_PAdikdid_: sorry, I didn't know that... I can't do much for you but point you to the forums
16:53:21*bluebrother read something about "ugly variables" once
16:53:26Nico_Paustriancoder: that sounds like a good idea
16:53:34Adikdid_can you give me the link please?
16:53:39Nico_P... a reasonable one too
16:54:01DomonokyAdikdid_: go to the forum, and choose an unofficial build, its not that hard..
16:54:01Nico_PAdikdid_: is the website *that* unuseabe ?
16:54:21 Join low_light [0] (i=c730180b@gateway/web/cgi-irc/
16:54:39Adikdid_no i can use that one
16:55:06bluebrotherif the website is that unusable, how unusable Rockbox must be? :o
16:55:13Nico_Paustriancoder: I'm quite happy with my implementation... it's quite short and simple and it suits my context well
16:55:30Nico_Paustriancoder: but maybe it wouldn't be appropriate for yours
16:55:53peturif only the small one is needed, go for that...
16:56:03austriancoderNico_P: I only need it to have all usb drivers in list.
16:56:07austriancoderpetur: mine is also small
16:56:32austriancoderso it would be fine, when we commit one linked list soon
16:56:42Nico_Paustriancoder: a possile problem with mine is that it's not separate from the rest of the code and it acts directly on my data struct
16:56:43peturthen get an agreement between you too ;)
16:56:51austriancoderNico_P: can I see your impl
16:57:16Nico_Paustriancoder: sure :);f=testplugin.c;h=41a248ab7b4722f92e2acb510ab13fedf8d13453;hb=no_rb_in_rb#l130
16:57:46Nico_Pit would need to be made more generic though... and it's quite basic
16:58:50Adikdid_does anyone here download with torrents?
16:59:16austriancoderNico_P: mine is generic
16:59:18DomonokyAdikdid_: please stay on topic.. only rockbox talk allowed here..
16:59:45Adikdid_surely your allowed to talk about other stuff?
17:00:09rasherAdikdid_: please see
17:00:38bluebrotherAdikdid_: try #rockbox-community
17:01:31Adikdid_there is knowone there
17:01:47Nico_Paustriancoder: I agree it's pretty appealing
17:02:11markunnice sarcastic quote from slashdot: "Remember: you should be able to do whatever you want with information, except if its the GPL! Then you have to follow the GPL!" :)
17:02:37bluebrotherhmm, can't I do git diff <files>? Doesn't seem to work :(
17:02:45 Part Adikdid_
17:04:41Nico_Pbluebrother: you want to diff only some particular files ?
17:04:42 Join ian_hawdon [0] (n=ian@
17:05:02bluebrotheryes. Is that possible?
17:05:06 Part ian_hawdon
17:05:33 Join dandin1 [0] (
17:05:59Nico_Pbluebrother: it seems to work for me
17:07:19Nico_Pbluebrother: "git diff file1 file2" shows the diff between the working copy and the repo for file1 and file2
17:08:00bluebrotherhmm. Strange, that didn't work for me.
17:08:41Nico_Pwhat version are you using ? are the files in the repo ?
17:09:04bluebrotherah ... seems I was in the wrong folder.
17:09:36bluebrotherbut it's somewhat strange that I didn't get an error message (and git diff showed me the diff)
17:11:00austriancoderNico_P: so shall be both use the linux linked list?
17:11:48Nico_Paustriancoder: I'm all for it
17:12:19Nico_Paustriancoder: though I'll probably focus on other things and port my code to that list when I get thet time
17:12:19 Quit BobShield (Read error: 104 (Connection reset by peer))
17:12:33austriancoderNico_P: okay
17:12:34Nico_Pbut it seems to be a good and reliable choice
17:13:06Nico_Phave you made sure it compiles and works nicely in rockbox ?
17:13:41 Join Adikdid [0] (
17:13:49Adikdidhi me, again
17:14:45austriancoderNico_P: I will do some extreme testing tonight and/or tomrrow
17:15:24Nico_Pcool :) let me know if you need help
17:15:42Nico_Pmaybe there are other linked lists in rockbox we could port too...
17:15:49 Join BobShield [0] (
17:16:37rasherI believe database uses linked lists, but I'm not sure.
17:17:39Adikdidi downloaded a theme called Purpbox
17:17:56Adikdidhow come the actual images in the background dont show up?
17:18:08Adikdidand why doesnt it look like it does in the preview on the rockbox website?
17:18:17Nico_Pbluebrother: is the tracker cleanup "week" still on ? and what results did it yield ?
17:19:30rasherAdikdid: Have you installed the fonts package? Also, purpbox requires an unsupported build. Any problems you have with it, you should take up with the author of the WPS or the maintainer of the build you're using.
17:20:38bluebrotherNico_P: it ended last sunday. We got down to less than 800 open tasks (over 1000 before it)
17:20:56Nico_Pnice :)
17:21:06Adikdidok rasher, does anyone know of any themes that actually look like what they do on the rockbox site?
17:21:25bluebrotherIMO we should repeat it in a couple of month ;-)
17:21:48rasherAdikdid: They should all look like the screenshot, assuming you use the build they recommend, and have the required font installed.
17:23:46Adikdidrasher, i'm on the page and i have a 5gen 30gig vid iPod
17:24:16Adikdidso should i choose iPod Video from the "identify your player" page?
17:24:45rasherAdikdid: Yes. What else would you chose?
17:25:53Adikdidok so what do i do now?
17:26:05Adikdidlike i mean i have installed the purpbox theme too my iPod
17:26:37rasherAdikdid: Then you read that the purpbox theme needs "Senab's build" (an unofficial rockbox version) and install that.
17:26:47GodEaterAdikdid: you still haven't said whether you've installed the fonts either
17:26:56Adikdidyes i have
17:27:21rasherAdikdid: Again, this is an unofficial version of Rockbox, and we will not help you if it has bugs or other problems.
17:27:42Adikdidi installed the fonts package fronm the "fonts page"
17:27:59GodEaterso if you've done that, and you've installed senab's build
17:28:04GodEaterthen we can't help you any further
17:28:06GodEatergo ask senab
17:28:17rasherAdikdid: you should read the forum thread dedicated to Senab's build:
17:28:46Adikdidon the senab d/l page it says there is two versions "normal" & "cop"
17:29:01Adikdidwould it be best to go with normal?
17:29:13GodEaterAdikdid: we don't know - ask senab
17:29:24GodEaterwe've no idea if there are any other differences between them
17:29:29GodEateronly senab knows
17:31:36AdikdidOK, so i installed that Senab build
17:31:40Adikdidand the fonts pack
17:31:43Adikdidanything else?
17:32:27rasherIf the theme still looks wrong, it's a problem with the theme.
17:32:34rasherOr with Senab's build.
17:32:46rasherNot really something we can help you with.
17:33:44Adikdidno i meant is there anything else i need to install before i dosconnect my iPod and see if it has all worked ok
17:34:21rasherSame answer.
17:34:26 Quit DerPapst ("So Long And Thanks For All The Fish!")
17:35:45GodEaterWe apologise for the inconvenience
17:36:14Adikdidsall good but damn it sasy it has to rebuild my database
17:38:08GodEaterpetur: it's amazing, that actually appears to work as D. Adams said it would ;)
17:38:09 Join Soul-Slayer [0] (n=Administ@
17:39:00Adikdidis it easy to make a theme?
17:39:20GodEaterread the wiki page on the subject and decide for yourself
17:40:16*petur looks at GodEater with his left head
17:40:40petur(my computer has dual head)
17:40:41GodEaterwhile the right one shovels pan galatic gargle blasters down its throat no doubt
17:43:59 Quit Adikdid ("ChatZilla [Firefox]")
17:44:05 Quit My_Sic ("Miranda IM! Smaller, Faster, Easier.")
17:44:27Soul-SlayerI found a program that is charging people to use it, and it's core element is ffmpeg... And it isn't being distributed with it's source code, and not even acknowledgement of the license... Surely thats breaching the GPL?
17:45:00 Join XavierGr [0] (
17:45:19rasherSoul-Slayer: If they're distributing ffmpeg without an offer of source or acknowledgement that it's GPL, yes.
17:47:17*Soul-Slayer wonders what he should do about it
17:47:27GodEaterreport it to the EFF
17:47:38peturor mail ffmpeg guys
17:47:54rasherI'd say contacting ffmpeg is at the top of the list. Let them handle it however they want.
17:49:11Soul-SlayerI'll see what they have to say in IRC
17:54:29Nico_PSoul-Slayer: what's the program ?
17:55:14Soul-SlayerConverts youtube videos into other formats, but it doesn't even mention it's using ffmpeg
17:55:38Soul-SlayerEven in their FAQ, the legal aspect only covers whether it's legal to download from Youtube, nothing about ffmpeg
17:55:49 Quit XavierGr_ (Read error: 110 (Connection timed out))
17:55:53Nico_Phow do you know it uses ffmpeg ?
17:56:15 Join polluxx2006 [0] (
17:56:25Soul-SlayerThe windows version uses ffmpeg.exe
17:56:31Soul-SlayerAnd the mac version is using ffmpeg
17:56:41Soul-Slayerwith no extension, but no license bundled with it
17:56:50***Saving seen data "./dancer.seen"
17:57:11 Quit pondlife ("disconnected has pondlife")
17:58:15GodEaterif they're using ffmpeg like that (i.e. calling it externally), then isn't that a valid use of the LGPL ?
17:58:41*GodEater notices the time and has to run
17:58:49Soul-SlayerI'm not sure, that's why I'm wondering
17:58:50*petur too
17:58:55rasherWell, they need to bundle the license and a written offer for source-code, at the very least.
17:58:57Soul-SlayerSurely they need to mention the license somewhere though?
17:59:03 Quit petur ("work->home->real life")
17:59:38*bluebrother has voice file installing in rbutil
18:02:15 Quit polluxx2006 ()
18:05:13Domonokynice :-)
18:13:45 Nick idnar_ is now known as idnar (i=mithrand@unaffiliated/idnar)
18:23:30Domonokybluebrother: is the voice file really named english.lang ? i thought it should be english.voice
18:24:04 Join Siku [0] (
18:24:47rasherDomonoky: You're right
18:25:58bluebrotherand I just updated my windows binary :'-(
18:27:15*bluebrother wonders who did the most oops commits
18:27:47*Domonoky thinks about changing the talkfile dialog, to move the configuration of tts and codec to the config dialog... it gets more and more to configure :-)
18:28:07bluebrotherhehe ;)
18:28:25 Join PaulJam [0] (
18:28:50bluebrotherok, windows binary updated once more. For those who dare to test ;-)
18:29:03bluebrothernow with accessibility plugin compiled in.
18:29:48PaulJamhi, i have a little question: with the plungins in subfolders, where do i have to put the autostart.rock? still directly in .rockbox/rocks/ ?
18:29:51Domonokynice, we should get some blind users to try it.. so we get response how good/bad it is to use for them..
18:30:06 Join DerPapst [0] (
18:30:13PaulJamthank you.
18:30:26bluebrothererr, /.rockbox/rocks/apps
18:30:42bluebrotherdamn, I'm error-prone today :(
18:32:06bluebrotherhmm, looks like RBInstaller doesn't use the ZipInstaller class
18:32:58 Join pixelma [0] (i=pixelma@rockbox/staff/pixelma)
18:32:59Domonokybluebrother: install.cpp (which does install rockbox) uses ZipInstaller
18:33:25bluebrotherhmm −− and what is installrb.cpp doing?
18:33:46bluebrotherlooks like it isn't used anymore.
18:33:48Domonokybluebrother: you are in rbutil and not in rbutilqt :-)
18:34:20Domonokyno, i dont have this file
18:34:34Domonokyso its not in svn
18:34:53Domonokythere is installbl.cpp which installs the bootloader
18:34:54bluebrotherah. I have this file but it isn't in svn
18:35:42Domonokyinstallbl should be renamed to installbootloaderwindow.. :-)
18:37:43bluebrotherwell, I won't object :)
18:44:37*Domonoky renamed it.. :-)
18:47:04 Join My_Sic [0] (
18:48:04*bluebrother added some plugin ideas to the RockboxUtility page
18:48:08Domonokybluebrother: should i make a combined TTS Engine and Encoder configuration Tab (with title like Talkfile) or seperate Tabs for encoder and tts ?
18:49:22bluebrotherhmm, a combined tab would get quite crowded, wouldn't it?
18:49:39Domonokyyes, thats likely..
18:50:48bluebrotherI think I'd use two tabs
18:52:01 Join haemmy [0] (n=stefan@
18:54:38PaulJamhmm, my h300 just turned off during operation and i wasn't able to power ot on again until i connected the charger. the time and date was then reset. does this mean my battery is faulty or could there be another reason for this behaviour?
18:55:02PaulJam(the battery was at least half full)
18:56:59*alienbiker99 needs a new h300 battery
18:57:36*PaulJam propably too :(
18:58:07alienbiker99actually i think i need to just solder the wire back on, it ripped out
19:00:17*Nico_P wonders why FS #821 hasn't been rejected
19:00:30 Quit BigMac (Remote closed the connection)
19:04:51 Quit TinoM (Read error: 110 (Connection timed out))
19:09:40amiconnpreglow: Input impedance for MC or MM systems shouldn't matter, as long as it is high enough (a few dozen kOhm). Only piezo pickups (?) want a really high impedance (>=500kOhm) (and they need no riaa deemphasis)
19:09:56 Join dan_a [0] (n=dan_a@
19:12:57 Join Jon-Kha_ [0] (
19:17:08 Quit PaulJam (".")
19:22:17 Join sergey [0] (
19:28:37 Join chrisjs169_ [0] (
19:28:45preglowamiconn: yeah, i tried soldering on a resistor of 47kohm and a 100 pf cap, and it didn't do anything
19:28:55preglowamiconn: so the problem with too much bass/too little treble persist :/
19:29:07 Quit chrisjs169 (Nick collision from services.)
19:29:08 Nick chrisjs169_ is now known as chrisjs169 (
19:31:53 Join jgarvey [0] (
19:35:35 Join Quelsaruk [0] (
19:36:30amiconnpreglow: Sounds like you're applying too much un-riaa'ing
19:36:48*amiconn prefers to stay away from vinyl and other analog relics if possible
19:40:30dan_aamiconn: Would you be interested in the value of DEV_EN on a G4 greyscale? If so, C2C3197F
19:41:43DerPapstamiconn: i wonder... don't you have the nano flash rom dumps i gave you anymore?
19:41:48low_lightis sansapatcher the only way to get the rockbox bootloader on a sansa?
19:42:17amiconnDerPapst: Flash update file != flash rom content
19:42:20 Join desowin [0] (n=desowin@
19:42:42DerPapsti thought they were the same
19:42:47 Quit Nico_P (Remote closed the connection)
19:42:52amiconndan_a: So we have another unknown bit, bit 17
19:43:19 Quit sergey (Read error: 110 (Connection timed out))
19:43:32amiconndan_a: Would be nice if you could give my other patch a go on G4 grayscale
19:43:43*DerPapst just remeberd to do the test...
19:43:55amiconn(the one that shows 2 LCD registers and tries LCD suspend/reenabling
19:44:37dan_aamiconn: Where can I get it / how far back through the logs do I need to look">
19:45:32dan_aamiconn: I'm just trying to get USB detection on the G3, then I'll do that on the G4
19:45:56amiconnWith that build, first enter "View I/O Ports" and note the 2 LCD register values. Then enter "View HW Info". That should put the LCD into suspend for 2 seconds, then wakeup
19:46:41amiconnRe-check the LCD registers afterwards (they might be different if I made a mistake in my RE research)
19:47:11amiconnIt's already tested on G3 by DerPapst
19:47:23amiconndan_a: Did you find something regarding USB on G3?
19:48:02amiconnAfaiu, apple does *not* use the built-in USB controller of the PP5002 (which is USB1.1 only)
19:48:28dan_aamiconn: Only which GPIO pin lights up when it's plugged in, but IIUC that should be enough to put it into disk mode - the same as the behaviour on the other iPods
19:48:46DerPapstamiconn: correct.
19:48:54 Join sergey [0] (
19:48:55amiconnThe GPIO pin isn't enough to distinguish USB from pure USB power
19:49:43amiconnOn PP502x we initialize the built-in USB controller "a bit", enough to make it tell us whether it's a real USB connection
19:49:46 Quit haemmy ()
19:50:02dan_aAh. I'll skip that test then!
19:50:35amiconnAnd in fact you could also test that LCD patch on G3, in case yours has a different LCD panel (there seem to be 3 versions according the the OF's tables)
19:51:09amiconnWell, you could make it reboot into USB mode when detecting USB power. Maybe that's better than not rebooting at all
19:52:00amiconnIt will take some disassembling to do it properly
19:52:06DerPapsti think you only connect a 3G to usb when you want to transfer files. otherwise there is really no reason to connect it to usb
19:52:10amiconn(i.e. initialising the USB controller)
19:55:06amiconnIf your G3's LCD panel is different *and* you also observe unstable blacklevel (black becomes "less black" when large black areas are displayed) you could also test another idea that DerPapst will test soon
19:55:33DerPapsti'm installing the build already
19:55:34amiconnI tested it on 2nd gen and it makes the display significantly more stable (although not perfect)
19:56:05dan_aI do have the unstable blacklevel issue. OF isn't perfect either. I'll tell you about the panel once I've done the G4
19:56:51***Saving seen data "./dancer.seen"
19:58:23 Join TinoM [0] (
19:58:31amiconnI know that OF isn't perfect, but I found a controller setting that improves things (and the OF doesn't use)
19:59:44DerPapstwoo... the contrast increased a huge bit...
20:00:31DerPapstfrom contrast setting 52 to 42
20:00:38DerPapstthey look similar
20:00:52amiconnWhat about the changing blacklevel?
20:01:14DerPapstdosn't change that much anymore
20:01:30amiconnYeah, that's what I observe on 2nd gen as well :)
20:01:43amiconnSo I'll add that setting to our LCD init
20:02:40DerPapstok. but make sure to lower the default contrast too.
20:02:57DerPapst40 to 42 is ok then
20:03:01amiconnThe default is 42 for 3rd gen iirc
20:03:05preglowamiconn: yeah, i know, but the filter is completely within specs :/
20:03:27DerPapstamiconn: ok then nvm ;)
20:04:34dan_aamiconn: G4 R2=0075, R3=120C. LCD goes blank with the backlight on when I go into View HW Info, and the registers are unchanged afterwards
20:06:27DerPapsti got R2: 0055; R3: 150C on my 3G yesterday
20:07:38jhMikeSamiconn: USB detection is part of the AS3514 on e200 complete with debouncing and seems to actually function (via AS3514 IRQ). The USB-controller GPIO is not a detection line for sure.
20:09:51 Quit Soul-Slayer (Read error: 104 (Connection reset by peer))
20:10:39amiconndan_a: Goodie, so suspend also works
20:10:50amiconnThe register values are those expected for a G4
20:11:39dan_aGlad to hear that... I'll check whether my G3 values match DerPapst's now
20:11:46amiconnOn G3 you should either get 0055 and 150C, or 004A and 110C
20:11:59 Join robin0800 [0] (
20:12:00amiconn(the latter would be a type 8 LCD)
20:12:02 Quit My_Sic (Connection reset by peer)
20:12:47DerPapstthe other is either type 7 or 9
20:13:04austriancoderDomonoky: how do i compile Dominik Wenger rbutilqt? I have found no makefile/CMakeLIst.txt
20:13:32DerPapstaustriancoder: run qmake
20:13:39amiconnDerPapst: correct
20:13:42DerPapstthat will create a makefile
20:13:49austriancoderah okay
20:15:10amiconnType 7 and type 9 should be distinguishable by the contrast you need to set to make it look right (without my blacklevel improvement patch)
20:15:19austriancoderDerPapst: which version of qt do I need?
20:15:24amiconnIf it's around 52, it's type 7, if it's around 42, it's type 9
20:16:34 Join Soul-Slayer [0] (n=jonno@
20:16:46Domonokyaustriancoder: qt 4.03
20:16:55amiconnIt might be that the difference is fixed by my blacklevel patch
20:17:17Domonokyif you have it installed just type "qmake && make" when you are in the rbutilqt folder
20:17:20*amiconn is curious whether the OF just uses whatever its loader (or our bootloader) sets
20:17:54amiconnIn that case a rockbox bootloader with the blacklevel patch should also fix it in the OF (as long as the OF doesn't send the LCD to sleep)
20:18:02austriancoderDomonoky: I have 4.3.0 and compile failed
20:18:29dan_a55 and 150C on my G3 - and it suspended correctly
20:19:10Domonokyaustriancoder: it should work with 4.3 ( there was a 0 to much ) :-)
20:19:21Domonokywaht error do you get ?
20:19:40austriancoderDomonoky: nope 4.3.0 - updateing to 4.3.1
20:19:44Domonokyaustriancoder: and linux or windows ?
20:20:15amiconndan_a: Thanks for testing :)
20:20:16DerPapstwould ther be a ways to test that? i would like to have a lowered contrast for the OF
20:20:33Domonokyi am running 4.3.0 on windows and it is working.. bluebrother compiles with linux.. dont know which version he uses..
20:20:33amiconnNow a test on a mini G1 would complete the test
20:20:44dan_aamiconn: You're welcome. Thanks for your brilliant PP5002 work!
20:20:48DerPapstyou have a mini G2?
20:20:58amiconndan_a: A runtime test on G3 is still missing....
20:21:01amiconnDerPapst: Yes
20:21:13amiconnAnd mini G2 can't read back LCD registers
20:21:21amiconnBut suspend is tested and works
20:21:38Domonokyah.. thats because your Qt version has no accessibility support..
20:22:13 Join My_Sic [0] (
20:22:31austriancoderDomonoky: can you do some #ifdef magic to get it compile without accessibility support?
20:22:38 Quit barrywardell (Remote closed the connection)
20:23:18Domonokyi will take a look.. *starts googling*
20:23:29*amiconn should also repeat his test now that DEV_EN2 setup is in place
20:24:30bluebrotherthat is because of a few accessibility properties are set in rbuiltqtfrm.ui
20:24:40bluebrotheryou can safely remove thos properties.
20:24:51 Join RockingD [0] (
20:24:58 Join BigMac [0] (
20:25:17amiconndan_a: Around half of the gain is due to implementing disk poweroff on 1st..3rd gen - and that could have been done months ago. The port pin was known and documented in the ipl wiki
20:25:22bluebrotherand for the question earlier, I use qt 4.3.0 on linux and the same version (statically built) on windows
20:25:49dan_aamiconn: Hint taken about the runtime test ;) Do we have a list of what the bits in DEV_EN and DEV_EN2 represent? The IPL wiki isn't always accurate
20:26:07amiconnThe other half is due to switching off clock for unused components within the PP5002. Found that by coincidence due to a slight flaw in the 1st/2nd gen firmware
20:26:42amiconnI know that the ipl wiki isn't always accurate
20:27:08DerPapstsadly... and no one fixes it :-/
20:27:15amiconnUnfortunately we don't have a list of DEV_EN(2) bits - and we can only try to figure out those bits which remain set by now
20:27:24 Quit ender` (" Cynics regarded everybody as equally corrupt... Idealists regarded everybody as equally corrupt, except themselves. -- Rober)
20:27:49amiconn2 bits I know for sure (and one of these is not set, and wasn't set before)
20:27:57 Join son1 [0] (n=megsona@
20:28:25 Part son1
20:28:31amiconn...the bits that enable the built-in LCD controller (not used) and the LCD bridge controller (used in all grayscale ipods)
20:31:53 Quit RockingD (Read error: 104 (Connection reset by peer))
20:32:57 Join WGC [0] (n=chatzill@
20:33:12 Join ender` [0] (
20:37:39 Quit sergey (Read error: 110 (Connection timed out))
20:40:31 Quit tchan (Read error: 110 (Connection timed out))
20:40:32 Join datachild [0] (
20:40:52 Nick alien___ is now known as majeru (
20:43:13WGCI just installed Rockbox
20:43:51 Join RockingD [0] (
20:44:11WGCI can't access the MP3 player via windows now
20:44:16WGCSays Unknown Device
20:44:29WGCIt runs ok otherwise
20:44:37ThundercloudVersion of windows?
20:44:47 Quit robin0800 (Read error: 110 (Connection timed out))
20:44:54ThundercloudShouldn't happen
20:44:59ThundercloudXP ships with UMS drivers.
20:45:01pixelmaI guess you have a Sansa?
20:45:01dan_aWGC: At a guess you'll need to boot into the original firmware to transfer files
20:45:14WGCHow do I do that?
20:45:25dan_aWhich device do you have?
20:45:31 Quit My_Sic ("Miranda IM! Smaller, Faster, Easier.")
20:45:31WGCsansa e250
20:45:36 Join tchan [0] (n=tchan@lunar-linux/developer/tchan)
20:45:52dan_aHold |<< as you boot (this is in the manual)
20:46:06WGCI hope they can work the kinks out eventually
20:46:23WGCI was so eager to install Rockbox that I'm a hospital waiting room.
20:46:34WGCMy mother has an operation today
20:46:35WGCNot me.:P
20:50:46 Quit datasleep (Connection timed out)
20:51:08WGCI generally think that the Rockbox firmware uses up less battery than the default one, it doesn't have that glossy colorful look
20:52:51 Join kubiix [0] (
20:53:45 Join My_Sic [0] (
20:54:13 Part maffe
20:54:28 Join maffe [0] (
20:56:14 Part maffe
20:56:26 Join maffe [0] (
20:56:54 Join WGC_ [0] (n=chatzill@
20:59:05 Part maffe
20:59:20 Join maffe [0] (
21:01:11 Quit My_Sic ("Miranda IM! Smaller, Faster, Easier.")
21:01:50 Quit spiorf (Read error: 110 (Connection timed out))
21:03:24WGC_I should eventually organize my MP3 collection into genres
21:03:38WGC_20 gigs of just mp3s in one folder is a tad much
21:04:36 Join spiorf [0] (
21:13:05 Quit WGC_ ("ChatZilla [Firefox]")
21:13:44 Join perrikwp [0] (n=chatzill@
21:14:42 Join |desowin| [0] (n=desowin@
21:14:43 Quit WGC (Read error: 110 (Connection timed out))
21:15:38 Quit desowin (Read error: 113 (No route to host))
21:16:28 Quit low_light ("CGI:IRC (EOF)")
21:19:46 Join My_Sic [0] (
21:19:57austriancodercan we have some marcos like __init and __exit (linux kernel modules) in rockbox? __init is defined as __attribute__ ((__section__ (".init.text"))) - I have the following wish: in the usb stack i have host drivers and device drivers and everybody has a function like usb_<drivername>_init and need to be called. First i thought I could make something equalent to plugins -> e.g. serial.usbd and put them in a dir and load of them at stack init. But I d
21:19:57austriancoderr loaded at the same time. And this is needed, when I connect e.g. second soundcard and keyboard. Has anybody an idea?
21:22:00Domonokyhm.. but you will have to limit the amout of drivers which can be loaded, because of no dynamic memory..
21:23:18austriancoderDomonoky: only one device driver can active and host drivers will clean out when they not needed
21:24:16*preglow listens to his wma file
21:24:53Domonokybut you want more then one host driver ?
21:24:58 Join Buschel [0] (
21:26:13*Domonoky would just limit this for either one host or one device driver aktiv, atleast for now.. we can extend this later, when its fully working :-)
21:27:06austriancoderIf I connect rockbox device to usb hub I could have more then 1 host driver active. The design of the stack allows this. If I have all drivers in a dir e.g. .rockbox/drivers/usb/ and a new device connects to host, I load every driver, ask him: can you handle new device, if not go to the next, if yes use the driver
21:27:10preglowBuschel: hiya, still hanging in there? :)
21:27:29 Join saratoga [0] (i=98039a34@gateway/web/cgi-irc/
21:28:07Domonokyaustriancoder: thats nice.. but it will not be easy to have multiple "usb plugins" running..
21:28:47austriancoderDomonoky: if I compile them in e.g. rockbox.mi4 then it is no problem
21:29:26Domonokyif you compile it in, then its not a problem.. its only a problem if you load them dynamically
21:29:29 Quit |desowin| ("use linux")
21:30:02 Join desowin [0] (n=desowin@
21:30:13austriancoderI know... hmmm
21:30:44Domonokyso i would compile them in for now.. we can change this, when the stack is working..
21:38:03 Quit RockingD ("using sirc version 2.211+KSIRC/1.3.12")
21:39:53rasherBagder: had any chance to look at the repeated LANG_BOOL_SETTINGS_YES from genlang?
21:50:54nerochiarosaratoga: hi, do you have a minute ? i'm working at porting WMA decoder and having some issues
21:51:05pixelmaperrikwp: I think I read you have a Mini 1st gen?
21:54:32saratoganerochiaro: sure whats up
21:55:34DerPapstamiconn: pixelma fund a volunteer :)
21:55:56pixelmaperrikwp: sounded like amiconn wanted to test something on a 1st gen mini
21:56:14perrikwpok i would be glad to help
21:56:38pixelmaor let someone test... hope that gets his attention now ;)
21:56:38DerPapstperrikwp: can you compile rockbox?
21:56:53***Saving seen data "./dancer.seen"
21:57:05DerPapstgreat... wait a sec...
21:57:09nerochiarosaratoga: well,since i'm using a different ASF parser, i need to make sure setup the data for the WMA decoder correctly. does the decoder need all the payloads from ASF packets to be lined up in a single continuous buffer ?
21:57:36DerPapstperrikwp: ally this patch">
21:58:12nerochiarosaratoga: that's what the rockbox codec seems to be doing, however i'm not sure i'm doing it properly since the decoding of the first block is failing
21:58:29DerPapstperrikwp: <amiconn> With that build, first enter "View I/O Ports" and note the 2 LCD register values. Then enter "View HW Info". That should put the LCD into suspend for 2 seconds, then wakeup
21:58:36nerochiarosaratoga: does it need only the payload data, or also the payload header ? or some other info ?
21:59:00DerPapstperrikwp: <amiconn> Re-check the LCD registers afterwards (they might be different if I made a mistake in my RE research)
21:59:33saratoganerochiaro: yes, i believe linuxstb's parser does that
21:59:50saratogaat least his comments imply that hes memcpying each packet into a continuous buffer
21:59:51merbanannerochiaro: you need to init the codec with the extradata
21:59:52perrikwpi don't have time to compile it right now as i have to leave for an hour or two but i can test after i'm back. is that ok if i come back with results later?
22:00:15saratogamerbanan: you're the guy in the ffmpeg channel?
22:00:35merbananlast time I checked
22:00:40saratogawhich wiki
22:00:51 Join chrisjs169_ [0] (
22:00:59nerochiaromerbanan: what extra data ? i'm not sure how am i expected to pass it to the wma codec. the init function requires only a context struct point. should i fill anything in that context ?
22:01:01saratogai don't follow ffmpeg i only modify their code
22:01:44 Quit chrisjs169 (Nick collision from services.)
22:01:48 Nick chrisjs169_ is now known as chrisjs169 (
22:02:01nerochiarosaratoga: i have seen these comments, but i can't understand what parts of the packet he's actually copying
22:02:05saratoganerochiaro: as i recall, linuxstb's parser loads a few things into a struct, stuff like bitrate, length, etc thats needed by wma_init
22:02:40nerochiarosaratoga: that would be the waveformatext structure. i fill that too
22:03:03merbanannerochiaro: read the beginning of the wma decoder source in ffmpeg, there you will understand what I'm talking about
22:03:22DerPapstperrikwp: sure
22:03:24nerochiaroalbeit yours has two extra fields in respect to mine, but you never use them anywhere. but i need to double check that structure
22:03:33nerochiaromerbanan: ok, i'll look
22:03:48saratoganerochiaro: which asf parser are you using anyway?
22:04:01perrikwpok i'll be back in a hour or two hopefully
22:04:07merbanannerochiaro: and what are you porting too ?
22:04:14nerochiarosaratoga: one made by one of the xmms2 folks.
22:04:22saratogaoh ok
22:04:29saratogaours is based on libasf very losely
22:04:44saratogai've been meaning to ask about that one, do you know if it supports seeking correctly?
22:05:05nerochiarosaratoga: appearently it does, but i have not yet double checked with the author
22:05:43nerochiaromerbanan: i'm trying to port the rockbox WMA codec to an xmms2 plugin that will be used on a Neuros OSD
22:09:05 Quit XavierGr (Nick collision from services.)
22:09:05 Join XavierGr_ [0] (
22:10:10 Join chrisjs169_ [0] (
22:10:13nerochiarosaratoga: i'm also trying to use the freestanding decoder you have pointed me from your homepage, but i am not sure what asf libary should i use with it
22:10:23saratoganerochiaro: i don't think libasf can seek correctly in VBR files FYI
22:10:31saratogaat least the code doesn't look like it can
22:10:52saratoganerochiaro: the zip has the asf parser built in
22:10:56nerochiarosaratoga: and the one in rockbox can ?
22:11:02saratoganot yet
22:11:08saratogai'm working on it now
22:11:24saratogawe currently parse the time stamp info, but our seeks are inaccurate for some reason
22:11:43saratogaas far as i can tell, i'm doing the same thing as ffmpeg, but it doesn't work for me
22:11:51nerochiarosaratoga: the one in rockbox seems pretty mixed up with the rest of the plugin code though, it doesn't look very weel separated
22:11:52 Join MySic [0] (
22:12:03saratogaits not quite the same as in rockbox
22:12:08saratogaoriginally it was a seperate app
22:12:10nerochiarosaratoga: and i don't see any asf source in that zip file
22:12:18saratogawhich zip are you looking at?
22:12:58saratogalet me make a new zip
22:13:46nerochiarosaratoga: thanks
22:14:15nerochiarosaratoga: that would help a lot in finding out what the wma expects
22:14:45nerochiarosaratoga: also, how old is that source ? it seems pretty easy to retrofit from the rockbox one, at a quick glance
22:15:03nerochiaroer, with the rockbox on
22:15:24saratoganerochiaro: i resynced it to rockbox last week
22:15:45saratogathough the functions are closer to ffmpeg then rockbox (it stilld ecodes one superframe at a time)
22:15:51saratogatry the link again
22:15:55saratogashould include the parser now
22:16:14saratogasorry the code is a mess, i use that one specifically for dumping decoder state data to disk
22:16:23saratogaits the only reason i keep it synced with rockbox
22:16:59nerochiaroit makes sense, i will use it for that purpose too
22:18:09nerochiarosaratoga: it seems still to not contain asf sources
22:19:40 Quit MySic (Read error: 104 (Connection reset by peer))
22:20:00 Join MySic [0] (
22:20:32saratogastupid caching
22:21:11nerochiarook, looks good now
22:22:00saratogasorry i'm evasive about what the decoder expects, to be honest i didn't really look to carefully at that
22:22:29saratogathat stuff didn't need much effort to make it fixed point so i didn't pay much attention to how it worked
22:22:57nerochiarosaratoga: no problem, maybe i can find out a way to factor out your current asf parser and use that
22:23:16saratogait shouldn't be hard
22:23:16nerochiarosaratoga: or find out from the freestanding decoder what it really needds
22:23:25preglowdid anyone look at the codec relink problem yet?
22:23:35 Join ender [0] (
22:23:41saratogawhen i switched from ffmpeg to our asf parser a while back it took me about 30 minutes of trial and error to do it
22:23:46nerochiarosaratoga: no offense but the libasf seems much cleaner thought
22:23:46saratogatheres only minor differences
22:24:10saratogatalk to linuxstb about that
22:24:16nerochiarohehe, will do
22:24:21saratogaour parser continues to have issues, but it also uses massively less memory
22:24:32saratogawhich i think was the reason linuxstb wrote it
22:24:42saratogapreglow: what issue is that
22:24:50nerochiarohow massively less ? i mean, roughly speaking
22:25:25saratogai think libasf buffers someting like 128k worth of data or something equally ridiculous
22:25:29saratogamaybe more
22:25:32 Join Juice^ [0] (n=Juice@
22:25:56 Join low_light [0] (i=c730190b@gateway/web/cgi-irc/
22:26:20nerochiarosaratoga: ah well, that's probably not going to be a problem for me, but i understand
22:26:27preglowsaratoga: that you need to "make" twice for codec to fully build after changes
22:26:33 Quit MySic (Success)
22:26:33 Quit My_Sic (Read error: 104 (Connection reset by peer))
22:28:02 Quit chrisjs169 (Read error: 110 (Connection timed out))
22:32:43 Join stripwax [0] (
22:33:23 Join Nico_P [0] (n=nicolas@rockbox/developer/NicoP)
22:34:39 Quit Siku ()
22:36:36 Quit Buschel ()
22:37:13 Quit Nico_P (Remote closed the connection)
22:38:05 Quit dandin1 ()
22:38:30nerochiarosaratoga: still around ?
22:40:16 Quit ender` (Read error: 110 (Connection timed out))
22:40:59 Join webguest49 [0] (i=50d81efc@gateway/web/cgi-irc/
22:41:09 Join ompaul [0] (n=ompaul@freenode/staff/gnewsense.ompaul)
22:42:33 Join salty-horse [0] (n=ori@pdpc/supporter/active/salty-horse)
22:42:38 Join haemmy [0] (n=stefan@
22:44:18saratoganerochiaro: yeah i'm here
22:45:28nerochiarosaratoga: do i understand right that the rockbox decoder basically splits the superframe decoding in several steps, while the freestanding one does it all in one go ?
22:45:47nerochiarobut using the same code
22:45:49salty-horseshouldn't the mandelbrot plugin be classified as "demo" instead of "game"?
22:46:00 Join ender` [0] (
22:46:18saratoganerochiaro: thats correct
22:46:25saratogasuperframes can potientially be very large
22:46:45saratogai think 10s of thousands of samples
22:46:55saratogaso we do one frame at a time
22:47:15nerochiaroone frame can still be lots of samples though ?
22:47:26saratogaone frame is usally 4096 samples
22:47:34saratogathough i'm pretty sure that can vary
22:48:09saratogathe unit the decoder itself deals with is the block
22:49:24 Join n00b [0] (i=c00c4efa@gateway/web/cgi-irc/
22:50:05n00bthis question has probably been answered a million times, but can someone explain to me how to configure shuffle so that it randomly chooses a directory and file each time you skip a song?
22:50:50n00bi tried shuffle=yes and autochangedirectory=random and repeat=off but it doesn't seem to descends the directories in order
22:51:21Bagdermake a single big playlist, enable shuffle. smile
22:52:05n00bso you have to build a playlist each time you add new files?
22:52:31Bagderyou don't have to do anything
22:52:36Bagderit was a suggestion
22:52:53n00bso there's another way?
22:53:59rasherThat's how I'd do it as well.
22:54:24n00bhow do you build a recursive playlist from a certain root directory?
22:54:50rasherThere's a setting for recursively adding files from subdirectories
22:55:04rasher(Either always or ask each time)
22:56:07*bluebrother points to the manual at and smiles :)
22:56:41n00bi tried the isn't perfectly clear what combination of flags you need to do what I'm trying to do.
22:57:00n00bin fact, no combination of flags will do what I want...I have to build a playlist and shuffle the playlist
22:57:03bluebrotherwell, it's part of the concept: as Rockbox is playlist based you'll always shuffle a playlist.
22:57:18n00bthanks..thats clear now
22:57:29bluebrotherif you play a folder it's a playlist too (this dynamic playlist thingy)
22:57:52bluebrotheryeah, that is common to cause confusion.
22:58:52bluebrotherDomonoky: around?
22:58:59n00brockbox is sweet ...I donated to show my love...but that is one feature of the original firmware I'll miss...being able to just shuffle the whole disk sans playlist
22:59:01webguest49austriancoder: I'm looking a disassembly of sansa e250 firmware (v1.00.12 EU).. I don't see any references to 0xc5000600 (USB_CONTROL)
22:59:52austriancoderwebguest49: thanks for this info :) Can you figure out the init of the usb controller? 0xc5000xxx
23:00:10n1sn00b: that's easy in rockbox. if the file browser is in the root, go to the main menu -> playlists -> create playlist, then turn on shuffle, done
23:00:25stripwaxn00b - you can play all music (with shuffle) from the database..
23:00:36stripwaxdon't need a playlist
23:01:10n00bthanks the database update automatically when I move files on and off the drive?
23:01:26bluebrothern00b: you can enable auto-update.
23:01:27austriancoderwebguest49: can you tell me what tools I need to search myselfe?
23:01:33n1swow this voice crash bug thingy is really odd, I could reproduce it reliably this morning by just starting up with resume playback and entering the menu and it would hang, now it's entirely reliable...
23:01:46 Quit ender (Read error: 110 (Connection timed out))
23:01:56bluebrotheryou'll need to put these files to play again −− resuming the old playlist won't find new files (as the resume info is the dynamic playlist)
23:01:58webguest49austriancoder: the adresses that have references (directly) are: C5000140,C5000148,C5000154,C500015C,C5000184,C50001A4,C50001A8,C50001AC,C50001B0,C50001BC,C50001C0
23:02:29webguest49austriancoder: I'm using IDA
23:02:34n1sdoes anyone know why or where we stop playback when we change language?
23:02:35low_lightBagder: is sansapatcher the only way load a rockbox bootloader onto a sansa? I'm trying to figure out how to get a bootloader on the c200.
23:02:41*DerPapst wonders who webguest49 is....
23:03:12n00bthanks for tips folks
23:03:13n00bover and out
23:03:15Bagderlow_light: that's the only way we offer these days, but you can just put the mi4 to the unit in recovery mode
23:03:15 Quit n00b ("CGI:IRC 0.5.7 (2005/06/19)")
23:03:19austriancoderwebguest49: okay.. I own IDA, but ever failed to find references to 0xc500xxx
23:03:40Bagderlow_light: bootloader mi4 I mean
23:05:20 Join homielowe [0] (n=chatzill@
23:06:01webguest49austriancoder: I use some IDC scripts to 1) make a "usb" segment 2