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

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

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

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

#rockbox log for 2010-04-05

00:02:34gevaertshm, one possibility is that things go wrong if you change a font used by scrolling text
00:07:42 Quit Bagder (Quit: It is time to say moo)
00:10:16kugel is my supposed to be final application for GSoC. I would welcome if some people would have a look, I plan to submit tomorrow
00:11:18 Join stripwax [0] (
00:16:19jhMikeSgevaerts: that was what I had in mind earlier
00:16:45jhMikeSturn off scrolling then?
00:17:02gevaertskugel: looks good I think
00:17:17gevaertskugel: do you know if the sbs code always redraws text?
00:17:48kugelI think it does but I don't know exactly
00:18:06gevaertsIf it does, just disabling scrolling won't help much
00:18:37gevaertsI think the font code should use some locking
00:19:34linuxstbkugel: For a *summer* of code, isn't your timezone GMT+2?
00:20:23kugelshould that info contain the summer time?
00:20:30jhMikeSkugel: there's grammatical errors in case you're worried about that
00:20:51kugelI bet there's a lot of them :p
00:20:58pixelmajhMikeS: there *are*? ;)
00:21:03 Quit adnyxo (Read error: Operation timed out)
00:21:30 Join adnyxo [0] (
00:22:04jhMikeSpixelma: there are errors of the grammatical nature, indeed. :) what are you saying? " there *is* grammatical errors?"
00:22:29jhMikeSnever mind, lol.
00:22:37 Quit jordan` (Read error: Connection reset by peer)
00:22:55jhMikeSthere's coloquialisms and all sorts of junk
00:28:00 Quit stripwax (Quit:
00:28:25 Quit flydutch (Quit: /* empty */)
00:32:39 Quit togetic (Quit: WeeChat 0.3.0)
00:34:32pixelmaspeaking of which... AlexP: I just discovered an old patch again that changes sub-diretories to subdirectories in the manual. What do you think?
00:35:12 Quit ender` (Quit: All great truths begin as blasphemies.–– George Bernard Shaw)
00:35:39AlexPsubdirectories looks funny to me
00:35:43AlexPlinuxstb: ?
00:36:31linuxstbYes, sub-directories would probably be how I would write it. I wouldn't be surprised if both are "correct" though.
00:36:37pixelmahmm... makes me wonder if the vice versa case woulf be needed too.
00:38:43pixelmaif I remember correctly, the patch was provided by bascule (more active in the forums back then) who was from Scotland I think
00:39:49 Quit moos (Ping timeout: 264 seconds)
00:42:34 Join phanboy_iv [0] (
00:45:55 Quit phanboy4 (Ping timeout: 252 seconds)
00:46:14AlexPpixelma: I wouldn't actively argue too hard either way, but I prefer with the -
00:46:37 Quit merbanan (Ping timeout: 245 seconds)
00:48:14 Join Blue_Dude [0] (
00:51:26Blue_DudeHey, quick question. A couple of days ago a dev made a change in the hotkey code I disagree with. Any problem with changing it back? There used to be a confirmation splash when setting a key but he took it out. I miss it.
00:51:51AlexPYou should discuss it first
00:52:02AlexPOtherwise we get into a commit war
00:52:09AlexPfml did IIRC
00:52:21AlexPAnd the reasons should be in the IRC logs
00:52:39gevaertswell, they should be in the commit message too...
00:52:55AlexPshould be, don't know if they are
00:53:13Blue_DudeActually, it was part of another patch. The comment didn't address the change at all.
00:53:14pixelmaAlexP: I haven't checked but I *guess* there are currently both ways of spelling
00:53:31AlexPI know he changed the lang files as they way it was constructed initially (add ? to the question) doesn't work for some languages
00:53:43gevaertsBlue_Dude: r25457?
00:53:46Blue_DudeThat's the one. He also took out the splash screen.
00:54:04AlexPHe did talk about it here first - I'd have a look at that, then mail the dev list
00:54:12Blue_DudeWoops, brb...
00:54:13AlexPI have a feeling it was discussed here
00:54:34 Join jordan` [0] (~jordan@
00:55:20gevaertsBlue_Dude: maybe ask him. The way he did it leaves line1 unused I think, so I suspect the removal of the splash was a mistake
00:55:35gevaertshm, or maybe not
00:55:57kugelBlue_Dude: the yesno screen has a confirm message itself
00:56:23 Join mirak [0] (
00:56:44kugelif we add a confirmation splash after each yesno we can just as well remove the ability of the yesno screen to do it itself
00:56:51 Join saratoga [0] (~9803c20d@gateway/web/freenode/x-jxkcvhosvnuovdsj)
00:57:18saratogamaybe you can help out that guy
01:01:33 Join Darkknight512 [0] (
01:02:11 Quit Darkknight512 (Client Quit)
01:02:30***Saving seen data "./dancer.seen"
01:08:51niekieHeh, I just killed my iPod.
01:09:03niekieI wanted to see what the shiny "Delete bootloader" option did in ipodpatcher.
01:09:10niekieApparently, it deletes the bootloader :)
01:09:14LloreanYou can reinstall the bootloader, though
01:09:16niekieDon't worry, I got it running again.
01:09:40niekieLlorean: nope, ipodpatcher didn't recognize it as an iPod anymore and refused to install the bootloader :(
01:10:15LloreanWere you booted into disk mode?
01:11:04LloreanWhich iPod. It should be able to detect it just fine if all you've done is remove the bootloader (which also shouldn't kill an iPod, just cause it to be OF only)
01:11:17niekieiPod Nano 2G.
01:12:18 Quit JohannesSM64 (Ping timeout: 260 seconds)
01:12:55linuxstbniekie: What's the problem exactly? You first said you killed your ipod, then you said it's running again....
01:13:40 Join jeffp [0] (
01:14:08jeffpwhoever runs the website: the sansa fuze (and others) manual has been offline for a few days
01:14:18jacekowskinot a big loss if he killed ipood
01:14:22jacekowskiget n900
01:14:39 Join Adubbb [0] (~aldubuc@
01:16:10 Quit GeekShadow (Quit: The cake is a lie !)
01:16:40 Quit Adubb (Read error: Connection reset by peer)
01:17:17 Quit mirak (Quit: Ex-Chat)
01:17:30niekielinuxstb: yes, used iTunes on someone else's PC to restore it.
01:17:59 Quit anewuser ()
01:19:53 Part jeffp
01:21:09 Quit Adubbb (Read error: Connection reset by peer)
01:21:36 Quit bertrik (Ping timeout: 265 seconds)
01:22:19niekieAlso, I tried to install a bootloader created from source.. so that might have helped in messing it up :)
01:22:31niekieAnyway, it's all up and working again.
01:24:02Blue_DudeWhew, sorry about that. I had reading duty. Bedtime for my 6 yr old. :)
01:25:31Blue_Dudekugel: Yes, the yesno screen has a confirmation, but it goes fast and it's hard to see. Maybe I could get rid of the yesno message and go with a splash instead. It gets the point across more quickly. Besides, I stole the code from bookmark.c and *it* has both confirmation and splash.
01:25:57kugelI'd welcome if that's fixed instead of double confirmation
01:26:04 Join Adubb [0] (~aldubuc@
01:27:06kugelIMO it's not worse or better visible than the question. and the visibility probably depends on the theme
01:27:37 Join Strife89 [0] (
01:32:06 Join JohannesSM64 [0] (
01:32:45Blue_DudeHm, well the yesno dialog is regular text on a normal background, placed in the upper left of the screen and it goes away quickly. By the time you register that there's information there, it's gone. Splash is reverse video in the center of screen and lasts two seconds, long enough to read it.
01:33:13 Quit Adubb (Ping timeout: 264 seconds)
01:37:15 Join Adubb [0] (~aldubuc@
01:42:07 Join Chronon [0] (
01:42:52 Quit Schmogel (Ping timeout: 265 seconds)
01:43:32pixelmashouldn't the yesno only go away on a button press? Not sure at the moment, just my first thought
01:44:25*pamaury proposes a yesno to confirm a yesno choice
01:44:57pamauryuser choose to reboot when he/she is sure of his/her choice :)
01:45:38CIA-5New commit by 03mc2739 (r25472): Fix Clip keymap (manual) so that Clipv2 and Clip+ manuals build properly
01:46:18kugelBlue_Dude: we could increase the timeout of the confirmation
01:46:19 Quit geertvdijk (Quit: ChatZilla 0.9.86 [Firefox 3.6.3/20100401080539])
01:46:35kugelI generally dont like splashes too much because they are so different to the rest of the ui
01:46:52Blue_DudeI'm fooling with a patch right now. I'll see what the results are and brb.
01:48:27 Quit xiainx (Ping timeout: 276 seconds)
01:50:36Blue_DudeOh garn, it's going to be a few minutes...
01:50:55 Quit piotrekm (Quit: piotrekm)
01:53:08kugelBlue_Dude: also you need to consider that splashes are only clearly visible on color targets, which is not always the case on greyscale/mono targets (where plain text draw is just as, or even better, visible)
01:53:43Blue_DudeSo you're saying... *both* would be preferable. :)
01:54:32kugelno, I'm just trying to invalidate your argument that splashes are better visible :p
01:54:44LloreanDoes the "yes no" disappear on its own, or only when a choice is made?
01:54:56LloreanI don't see why you need a splash *after* the confirmation request if it requires user input to confirm.
01:54:57Blue_DudeIt only goes away after a keypress.
01:55:03LloreanSo why have a splash?
01:55:11LloreanIf the user wants to read it before they confirm, they will.
01:55:24Blue_DudeWell... I liked it.
01:55:25LloreanIf they're going to ignore the yes/no, they'll ignore the splash too anyway.
01:55:28 Quit pamaury (Quit: Page closed)
01:55:36LloreanWhat purpose does it serve, though?
01:55:42Blue_DudeI liked it.
01:55:51*Blue_Dude is starting to pout by now.
01:56:37Blue_DudeI don't know exactly why it's there, except that the code I stole from also had it.
01:57:29Blue_DudeAnd it seemed like a nice clean dialog. Pretty.
01:57:48Blue_DudeOther than that, no reason I guess.
02:01:01 Join kramer3d [0] (~kramer@unaffiliated/kramer3d)
02:03:52 Quit kramer3d (Client Quit)
02:03:59 Join xiainx [0] (
02:07:52 Quit grndslm (Ping timeout: 252 seconds)
02:10:32Blue_DudeSo realistically, what are we up against? Is the splash a bad idea? Useless idea?
02:15:53 Part froggyman
02:17:31kugelalright, final version of the application: I'll submit that unless there are any comments (which are still welcome of course)
02:23:17kugeldoes anyone know how much applications we've got already?
02:24:27saratoga4 as of a couple hours ago
02:24:38saratogalast year though we got a lot on the last day or two
02:25:32saratogaoh 6 now
02:27:39kugelsaratoga: are you going to be the raaa mentor? I saw you on the list of possible mentors and nobody else
02:27:59saratogakugel: i'll probably do one of the codec projects
02:28:12saratogaone of the people whos done more with the core would make sense for RAAA
02:29:16saratogaalthough I suppose if you're going to do it you can probably just talk to the rest of us directly and your mentor becomes more of a formality
02:33:42 Part r2k000
02:37:49Blue_DudeI was just trying out something for a forum user. I tried deleting a file within Rockbox while it was playing. You would expect that the file would stop playing, the file would delete, the playlist would update (without the deleted file) and playback would resume with the next file. This isn't what happens.
02:38:39Blue_DudeInstead, the file may or may not delete, it never stops playing (even if deleted!) and the playlist doesn't update. Seems very odd.
02:41:01kugelwe had this topic before
02:41:22kugelwhile it's currently buggy, stopping the playback (or skipping to the next song) is not the prefered solution
02:42:03Blue_DudeThe current situation is better?
02:42:29Blue_DudeIn that case, why have delete in the WPS context menu at all?
02:42:36kugelit's buggy as I said
02:42:47Blue_DudeSo what is the desired behavior?
02:43:05LloreanBlue_Dude: If the song is buffered, you can delete it while still finishing playback
02:43:09kugeldelete, and play out what's left in the buffer
02:43:26LloreanThis means that if you want to, for example, remove a podcast after listening, once the last buffer occurs you can delete it, then continue listening to the next one with no further worries.
02:43:27Blue_DudeYeesh, that's a mess. Why do that?
02:43:36LloreanThe bug comes in what happens if you delete it before all of it's in the buffer.
02:44:01kugelnot deleting only happens if the song is not fully buffered because the file handle isn't closed; I identified that problem once but fixing it introduced audio glitches for some reason
02:44:14Blue_DudeOK, I can see that. But if it's not completely buffered, why not go with the ebehavior I described?
02:44:45Blue_DudeWhere were the glitched/
02:45:02kugelat the end of a file
02:45:07LloreanBlue_Dude: Well the behaviour definitely shouldn't change depending on when you delete it.
02:45:35kugelI experienced it on my fuze where it's rather likely that not even a single song fits into the buffer
02:45:42Blue_DudeSo then, play out already buffered material, then track change?
02:45:52kugelit looked like closing the handle and reopening on rebuffer inserts the glitch
02:45:53LloreanIdeally it should either delay the delete until the next time the file can "safely" be removed, or refuse to delete with a notification.
02:46:19 Quit cjcopi (Read error: Operation timed out)
02:46:27LloreanIn my opinion, at least.
02:46:37Blue_DudeOw, that hurts my head. Why not just kill playback immediately, track change and rebuffer with the new track?
02:46:58Blue_DudeI mean, delete means delete, right?
02:47:20LloreanThere are already people who make use of the fact that "delete" doesn't also mean "unbuffer"
02:47:22Blue_DudeBTW, splash! −−>
02:47:22kugelit probably shouldn't delete before the next rebuffer anyway, or do you want a disk spinup only for an immediate deletion?
02:47:26LloreanYou'd be removing existing functionality.
02:47:57Blue_DudeLlorean: I'd make the argument that it's not a function, just the exploitation of a bug.
02:48:36Blue_Dudekugel: treat it internally as a track change, but with a file operation thrown in.
02:48:44LloreanI'm pretty sure it's been said to be intended behaviour, so calling it a "bug" is untrue.
02:48:46Blue_DudeAnd a playlist update.
02:48:51kugelthere's no argument, playing out is the intentional behavior
02:49:06LloreanYou can say it's intended behaviour that needs changed, but at the moment it's definitely a feature.
02:49:28Blue_DudeSorry, I just don't see the virtue of that. Trust the user to know what he's doing.
02:49:42kugel(but I do think it should not delete until the end of the file has been buffered)
02:49:53LloreanBlue_Dude: Can't we trust the user to delete *and* skip if they don't want to allow it to finish?
02:49:55kugeland not before the next disk spinup
02:50:24Lloreankugel: I more or less agree there. Delay the delete until it's "safe" if the file is in the current buffer and not wholly buffered. Or maybe even until it leaves the buffer / playback is stopped
02:50:31Blue_Dudekugel: the disk will need to spinup anyway to start buffering the next track. What difference does it make?
02:50:34kugelBlue_Dude: achiving what you want manually is trivial, while the opposite is impossible
02:50:53saratogai think the best option would be to advance the track and then immediately delete the file
02:50:56kugelBlue_Dude: huh?
02:51:13kugelthe next track might be buffered as well
02:51:14Blue_Dudekugel: it's trivial with extra keypresses and unexpected behavior.
02:51:18kugelor the next 5 tracks
02:51:22Lloreansaratoga: Why can't the user choose whether or not to advance the track?
02:51:41saratogawell first because I dislike having options that aren't overwhelming important
02:51:48LloreanAs it stands, there's an existing feature that you're advocating removing without really gaining anything from removing it.
02:51:53saratogabut also because thats not going to work well on low mem targets
02:51:53kugelsaratoga: what option?
02:52:13Lloreansaratoga: Several people (including non-devs) have explicitly stated they like the ability to mark a podcast for deletion before they're entirely finished listening to it.
02:52:13saratogaon something like the Clip the current track can basically never be buffered entirely
02:52:17Blue_DudeLlorean: even so, the current behavior should update the playlist once the track is advanced. At present, you can back up to play the "deleted" track out of the buffer.
02:52:27LloreanIdeally it'd be an actual "mark for deletion" rather than our current method, but that's just an improvement.
02:52:28kugelit's considered as a feature, if at all it lacks documentation
02:52:49kugeland as long as skipping the track is easy to do manually I wouldn't want it to change
02:53:19Blue_DudeIf marked for deletion, when would the deletion take place? After stopping playback?
02:53:23LloreanBlue_Dude: It shouldn't necessarily update the playlist (there's no reason to ever write to a user's playlist unless they modify it themselves) but it should prevent seeking back into it if the file's gone, just like any other skip to an invalid playlist entry
02:53:31saratogaon something like the clipv1 i imagine its basically impossible to delete the currently playing track?
02:53:42kugelBlue_Dude: probably at the next disk spinup, via ata_idle_notify
02:53:45saratogaand probably for flac files on lots of players
02:53:57Blue_DudeLlorean: I meant dynamic playlists, but that's what I mean.
02:53:59LloreanI'd say deletion should occur either on stopping playback or on when none of the file remains in the buffer, whichever happens first.
02:53:59kugelimmediate deletion isn't very important is it?
02:54:12saratogawell its less complicated
02:54:18LloreanBlue_Dude: Dynamic playlists should behave as much like static playlists as possible though.
02:54:21saratogado we have a system for queuing file operations?
02:54:36Blue_Dudesaratoga: I don't think so.
02:54:47kugelsaratoga: what do you want to queue?
02:54:54saratogathe track delete
02:55:02saratogalets add another 10 special cases to buffering.c to handle this!
02:55:11Blue_DudeGod I need to look into the buffering code now. "Abandon all hope, ye who enter here." Drat.
02:55:28saratogaif you're really bored you could write that test driver i always wanted
02:55:37kugelsaratoga: why special case? a delete after buffering flag could be quite generic
02:55:42saratogaso we can test buffering with the build system
02:56:08saratogaeverything in the buffering system ends up being really complicated, i'm skeptical this will be different
02:56:14saratogabut i am only guessing
02:56:29kugelI'm expecting this to be a relatively simple change actually
02:56:32Blue_DudeBuffering rewrite!
02:58:40Blue_DudeOK, a "Mark For Deletion" context menu item, which flags the file for deletion upon stopping playback, or end of file, whichever happens first. Possibly with an optional automatic track change?
02:59:08kugelwhy add something?
02:59:09Blue_DudeAlso, before I lose track: OK, or not OK?
02:59:37kugelBlue_Dude: Why do you want to push the splash just because you like it?
02:59:40Blue_Dudekugel: not everyone wants to keep going. Why not make it easy for him if he wants ti?
03:00:05Blue_DudeI guess I can take it out of bookmark.c too...
03:00:10kugelBlue_Dude: this is too minor to increase bloat imo
03:00:37Blue_Dudekugel: what's too minor, the mark for deletion or the splash?
03:00:48kugelthe former
03:01:04LloreanThe mark for deletion just needs to have the existing bugs worked out. It doesn't need a setting to automate a single additional button press.
03:01:42kugelthe current item can be changed to mark for deletion if you want, it doesn't make sense to me if you add another one parallel to that with an extra option for something as minor (really really really minor IMO) as an automatic track skip
03:01:51Blue_DudeWorking out the bugs is possible only if you define the correct behavior. I can't even get the file to delete consistently.
03:02:06kugelwe could simply fix the bug instead of overengineering this
03:02:29kugelBlue_Dude: intended behavior *is* defined
03:02:31***Saving seen data "./dancer.seen"
03:02:48kugelplay out what's in the buffer
03:02:54saratogaits defined you just can't predict it without looking at the buffering debug screen
03:03:02 Join grndslm [0] (
03:03:03Blue_DudeI assume that the correct behavior is to always delete the file though.
03:03:07*Llorean really thinks splashes are only needed to tell users things they might not already know, and shouldn't follow a choice they explicitly made.
03:03:26kugelplaying out the track (with possibly buffering the rest of it), i.e. mark for deletion, would be an improvement imo
03:03:27LloreanBlue_Dude: Yes, but your assumption is actually wrong.
03:03:41Blue_DudeLlorean: ok, ok. I'll take out the one in bookmark.c too when I have the chance.
03:03:44LloreanWell, it will always be deleted, yes, but deletion != "skip forward in playback"
03:03:54LloreanBlue_Dude: What's the one in bookmark.c for? Another yes/no?
03:04:33Blue_DudeLlorean: No, I mean I tried the behavior in the sim, and I couldn't get the file to delete all the time. Sometimes it would and sometimes it wouldn't/
03:04:44LloreanBut that's simply a bug.
03:04:54kugelBlue_Dude: I already told you twice that this is a bug
03:04:55Blue_DudeLlorean: yes, another yes/no. The bookmark delete function.
03:05:15LloreanI don't remember bookmark deletion having a confirmation, though I haven't done it in a while.
03:05:19kugelyour proposed fix isn't a fix though because it changes the intended behavior
03:05:20LloreanDidn't it used to be a single button press?
03:05:21Blue_DudeI know, but what's buggy, the fact that it won't delete?
03:05:46LloreanThe file should *always* be removed.
03:05:47kugelI actually explained where the bug is too
03:05:51Blue_DudeLlorean: no, I don't think so. I didn't change its behavior.
03:06:09LloreanAnyway, if there's always a yes/no dialogue, then there should be no splash.
03:07:32Blue_Dudekugel: To sum up: close the file handle, delete the file, keep playing whatever is buffered... is that it?
03:07:47kugelsplashes should only used for important things (because they're so different to the rest of the ui), confirming a decision just made is not imporant
03:08:21kugelBlue_Dude: basically, but as I mentioned there's a glitch when closing the handle early
03:08:31kugelwhich is why I didn't fix it yet
03:09:14Blue_DudeIt sure would make it easier if you just invalidated the buffering handle at the same time.
03:09:42Blue_DudeI know that's not going to happen.
03:10:21kugelthe bug is that buffering doesn't close the file handle when it notices that the file to buffer doesn't fit into the remaining buffer space
03:10:49kugelso, until the next rebuffer the file is basically blocked
03:11:17saratogaFWIW i really dislike the idea of continuing to play after you attempt a delete because its so target specific
03:11:32saratogaor at least the way we do it now
03:12:13saratogahmm i take that back
03:12:13kugelwhich is why I consider mark for deletion as an improvement
03:12:37saratogayeah i see what you're getting at now
03:13:47Blue_DudeI know next to nothing about the buffering system. Is there a way to set flags on buffered items?
03:19:08Blue_DudeAnyway, I'll look into it. Eventually. I'd like to fix the bookmark stuff first.
03:20:58CIA-5New commit by 03Blue_Dude (r25473): Fix capitalization in hotkey dialog
03:24:49 Quit Blue_Dude (Quit: ChatZilla 0.9.86 [Firefox 3.6.3/20100401080539])
03:32:24*linuxstb should really dig out his patch to remove gratuitous use of Title Case
03:35:00kugellinuxstb: yes you should ;)
03:35:28kugeltime's working against you
03:39:43 Quit linuxstb (Ping timeout: 248 seconds)
03:40:20*flyback thinks the caps on this mb might finally be going, firefox keeps seg faulting
03:42:39Mode"#rockbox +o Llorean" by ChanServ (ChanServ@services.)
03:42:46 Join froggyman [0] (~me@unaffiliated/froggyman)
03:43:02Kick(#rockbox flyback :You've been warned about this being an on-topic channel.) by Llorean!~DarkkOne@rockbox/user/Llorean
03:43:10Mode"#rockbox -o Llorean" by ChanServ (ChanServ@services.)
03:43:48 Join flyback [0] (
03:45:00flybackso all the versions of sansa clip are supported?
03:45:30LloreanTargets are listed on the front page of the site.
03:46:06 Join linuxstb [0] (~linuxstb@rockbox/developer/linuxstb)
03:53:27 Quit Strife89 (Quit: Bed.)
04:03:30 Quit adnyxo (Ping timeout: 252 seconds)
04:05:15 Quit TheSeven (Disconnected by services)
04:05:30 Join The_Seven [0] (~theseven@rockbox/developer/TheSeven)
04:05:40 Nick The_Seven is now known as TheSeven (~theseven@rockbox/developer/TheSeven)
04:08:05 Join adnyxo [0] (
04:15:36 Quit kugel (Remote host closed the connection)
04:16:08 Quit adnyxo (Ping timeout: 276 seconds)
04:16:55 Quit CaptainKewl (Quit: ( :: NoNameScript 4.22 :: ))
04:22:16 Part froggyman
04:29:00 Join Rob2223 [0] (
04:32:39 Quit Rob2222 (Ping timeout: 265 seconds)
05:00:33 Quit Barahir_ (Ping timeout: 260 seconds)
05:00:34 Join RandomInsano2 [0] (
05:01:51RandomInsano2Hiya. I'd like to add some scans I've done of the internals of an Insignia NS-DV2G to the Telechips Info page.
05:02:17 Join Barahir [0] (
05:02:33***Saving seen data "./dancer.seen"
05:02:43RandomInsano2I needs Wiki edit priviledges for username RandomInsano
05:02:45saratogaRandomInsano2: sure, whats your name
05:03:01RandomInsano2Edwin Amsler, but I chose a different handle.
05:03:11saratogaoh, well fix that then i'll give you access
05:03:50 Quit n17ikh (Ping timeout: 245 seconds)
05:03:58RandomInsano2Is that possible, or should I just create a new user?
05:04:43saratogai think you need to make a new account
05:05:02RandomInsano2I'll go do that then. Delete the current then please
05:06:59RandomInsano2How... do I log out?
05:08:36 Join n17ikh [0] (
05:08:42RandomInsano2Registration complete for 'EdwinAmsler'
05:10:58 Join funman [0] (~fun@rockbox/developer/funman)
05:13:47saratogaRandomInsano2: ok added
05:14:27RandomInsano2Much thanks! I'm still a little unclear how I logout of the wiki so I can log back in
05:14:52funmanI have another theory for CGU_PROC / CGU_PERI on as3525v2
05:15:08funmanCGU_PROC divides PLLA to give fclk
05:15:15funmanCGU_PERI divides fclk and not PLLA to give pclk
05:19:04 Join CaptainKewl [0] (
05:19:11 Join arbingordon [0] (~w@unaffiliated/arbingordon)
05:19:23 Join mikroflops [0] (
05:19:23 Quit RandomInsano2 (Quit: CGI:IRC (EOF))
05:23:09 Quit mikroflops_ (Ping timeout: 264 seconds)
05:26:38 Quit TillW (Quit: This now concludes our broatcast day.)
05:26:55 Join fejfighter [0] (
05:29:46S_a_i_n_t_Reading through the logs, Re: the guy that "killed" his iPod, he mentioned he installed an SVN bootloader (though I have no idea why ipodpatcher refused to detect his iPod, *unless* he was messing with advanced-iPodpatcher install options and fucked it up...I' guilty of doing this myself). I haven't checked in the last week or so, but the SVN bootloader for Nano2g just blackscreens the player.
05:29:57S_a_i_n_t_...May have been a contributing factor.
05:32:10*funman slaps buschel for not testing r25464
05:33:56S_a_i_n_t_Errrr, perhaps slightly unclear. "Blackscreen" being: turn on, Apple logo, indefinite black screen needing a hard reboot.
05:34:58 Join kramer3d [0] (~kramer@unaffiliated/kramer3d)
05:41:29 Quit scorche (Disconnected by services)
05:41:39 Join scorche` [0] (~scorche@rockbox/administrator/scorche)
05:42:17 Quit stavrob (Ping timeout: 258 seconds)
05:42:58 Join stavrob [0] (
05:43:31 Quit Minataku (Ping timeout: 276 seconds)
05:50:35 Join Minataku [0] (~Ed@unaffiliated/payphoneed)
06:03:07 Quit jd (Quit: Ω)
06:03:44 Join jd [0] (
06:03:44 Quit jd (Changing host)
06:03:44 Join jd [0] (~jd@Wikipedia/HellDragon)
06:05:58 Join notlistening [0] (
06:08:02notlisteningHey bluebrother just saw that your trying to do some wine detection when people are running under linux etc. Domonoky did some some silimar work when we put open-sapi in rbutil if you have done it no worries but there was a solution that worked quite well when we did that
06:14:57funmanfuzev2 works fine at 24MHz but the display is noticeably slower
06:15:21S_a_i_n_t_Do you have a runtime for 24MHz yet?
06:15:33funmannope i just got it right in my build, bench tonight? ;)
06:16:11funmanprevious bench for Clip+ (16h30) was with CPU at 24MHz / peripheral clock at 6MHz, and 10 times faster when boosted
06:16:14funmannot sure
06:20:28 Join TillW [0] (
06:26:47notlisteningfunman do you sleep :D
06:27:17notlisteningIs it normal for the Clip+ to have a yellow/orange bar at the top of the screen?
06:27:23S_a_i_n_t_"can't sleep... must. fix. red." :P
06:28:00notlisteninglol now as someone is talking about red I see it all the time on commits what is red etc etc?
06:28:25S_a_i_n_t_Red = Bad, broken. etc.
06:28:42funmannotlistening: yes the top quarter of the screen is yellow on all clips and the rest blue
06:30:28notlisteningahh but S_a_i_n_t_ to put a commit as fix red and thats all is a bit non descriptive?
06:30:36notlisteningthanks funman
06:31:49S_a_i_n_t_notlistening: yes, it is...but it seems to be a habit now.
06:31:54S_a_i_n_t_Or, for some time rather.
06:32:50notlisteningfair enough, they guys here at rockbox are mega good at not breaking things on commits so it is not the biggest issue
06:33:10*S_a_i_n_t_ isn;t so sure about that sometimes ;)
06:33:35 Nick S_a_i_n_t_ is now known as S_a_i_n_t (S_a_i_n_t@
06:34:07funmansomeone with an e200v1 could give me the results of test_fps with both boosted & unboosted CPU ?
06:35:59notlisteningonly got a V2 sorry
06:38:37saratogafunman: beyond the ones on the wiki:
06:38:45saratogaif so I can compile that plugin
06:40:29funmanah nope, it's enough, thanks
06:41:53funmanpixels swapping for fuzev2 display slows down things a bit
06:43:30funmanwithout swapping, and pclk==60MHz, performance is a small bit under fuzev1 (pclk==62MHz)
06:44:59funmanwith swapping (and thus correct display) I get 73fps for 1:1 updates unboosted
06:45:06 Join Eugenpaul [0] (
06:45:17 Part Eugenpaul
06:47:57 Join n1s [0] (~n1s@rockbox/developer/n1s)
06:48:40CIA-5New commit by 03funman (r25474): test_mem: fix r25464: button_get() can't be used with actions
06:48:45CIA-5New commit by 03funman (r25475): as3525v2: set PCLK correctly ...
06:51:20funmanI wonder if it's possible to change pixel format with the Fuzev2 LCD
06:55:50saratogafunman: what was pclk at before when unboosted?
06:56:44funman6MHz on Clipv2/+ and 15MHz on Fuzev2
06:57:11funmanno changes in playback because mclk is based on PLLA which didn't change
06:57:17funmanand timers are based on the 24MHz crystal
06:57:47saratogamclk is the DRAM?
06:58:06funmannot it's the i2s clock
06:58:10funmanfor pcm/recording
07:01:00 Quit CaptainKewl (Remote host closed the connection)
07:02:35***Saving seen data "./dancer.seen"
07:13:30 Quit panni_ (Quit: ( :: NoNameScript 3.81 :: ))
07:26:01 Quit drostie (Remote host closed the connection)
07:26:45 Join kramer3d_ [0] (~kramer@unaffiliated/kramer3d)
07:29:59 Quit kramer3d (Ping timeout: 264 seconds)
07:30:31 Quit mc2739 (Ping timeout: 268 seconds)
07:30:45 Nick kramer3d_ is now known as kramer3d (~kramer@unaffiliated/kramer3d)
07:36:11 Part veeloc
07:41:54CIA-5New commit by 03funman (r25476): Fuzev2: write pixel swapping in assembly for a some speed up ...
07:44:35 Join Horscht [0] (~Horscht2@xbmc/user/horscht)
07:45:59 Quit Horschti (Ping timeout: 246 seconds)
08:01:05 Join r2k000 [0] (~r2000@
08:23:01 Quit JohannesSM64 (Quit: WeeChat 0.3.2-dev)
08:27:12 Quit CGL (Remote host closed the connection)
08:28:10amiconnfunman: Why don't you just use swap_odd_even32() from system-arm.h? It doesn't need a separate mask and also just 4 cycles on arm <= v5
08:28:47funmanoh didn't know it existed
08:29:01funman(also i don't know the armv5 isntructions)
08:30:02amiconnIt uses plain armv4/v5 instruction, nothing special
08:30:25funmanah i was looking at v6 then
08:30:34amiconnOn armv6 it's a single instruction, but that's irrelevant on the ams sansa
08:33:14 Quit r2k000 (Ping timeout: 258 seconds)
08:35:25CIA-5New commit by 03funman (r25477): Fuzev2: don't reinvent the wheel: swap pixels with existing swap_odd_even32 ...
08:37:21 Quit pixelma (Disconnected by services)
08:37:22 Join pixelma_ [0] (quassel@rockbox/staff/pixelma)
08:37:42 Nick pixelma_ is now known as pixelma (quassel@rockbox/staff/pixelma)
08:37:55 Quit amiconn (Disconnected by services)
08:37:57 Join amiconn_ [0] (quassel@rockbox/developer/amiconn)
08:38:19 Nick amiconn_ is now known as amiconn (quassel@rockbox/developer/amiconn)
08:41:26 Join JanDo [0] (~JanDo@
08:41:40 Quit n17ikh (Read error: Connection reset by peer)
08:42:00 Join n17ikh [0] (
08:44:24 Join bmbl [0] (~Miranda@unaffiliated/bmbl)
08:45:08 Join Boldfilter [0] (
08:48:01funmanamiconn: do you want to modify the fuzev1 write_yuv() functions to add this swapping ?
08:49:23funmangreen part needs to be split in 2
08:50:02funmanr(5)g(6)b(5) = r(5)g1(3)g0(3)b(5), needs to be g0(3)b(5)r(5)g1(3)
08:52:20 Quit kramer3d (Quit: Leaving)
08:53:43 Quit n1s (Ping timeout: 264 seconds)
08:55:08*amiconn wonders why the fuze doesn't use RGB565SWAPPED as the pixel format if the lcd controller needs it swapped
08:55:15funmanif i go the stupid way there is 3 more instructions per pixel (2 to extract g1 & g0, and 1 shift for each componnent)
08:55:26amiconnIt would save the swapping in the update (not for yuv of course)
08:55:26*funman discovers new things everyday
08:58:20funmanin fact i had looked at lcd-16bit.c but didn't see anything
09:02:07amiconnOf course not, since lcd-16bit.c doesn't have to care about the actual bit order within the 16 bits
09:02:36funmanok i'm just testing it
09:02:40***Saving seen data "./dancer.seen"
09:02:48funmanwhat about yuv assembly, are you interested?
09:02:56amiconnAll colour PP ipods (Color/PHoto, Nano G1, Video) use RGB565SWAPPED
09:03:12funmanyep I see bmp2rb -f 5 option is documented as being "Ipod"
09:09:20 Quit notlistening (Quit: Leaving)
09:10:03CIA-5New commit by 03funman (r25478): Fuzev2: use RGB565SWAPPED (pointed out by amiconn) => 91fps
09:11:05funmanamiconn: ?
09:12:50saratogais 100 fps still the limit like on the v1?
09:18:05 Part Boldfilter
09:18:13funmanI don't remember how the limit was calculated
09:18:30funmansupposing the DBOP hardware didn't change it should be the same afaiu
09:21:11funmanamiconn: reading lcd-as-video.S i don't see where the 16 bits rgb value is swapped?
09:21:44saratogaaren't they computed in swapped order?
09:21:51funmanah ipodvideo lcd isn't swapped
09:22:07 Join n1s [0] (~n1s@rockbox/developer/n1s)
09:22:13funmanonly color & nano1g
09:23:51funmanand they just use swap16()
09:25:31funmanthe vibe500 has asm but the LCD register is 8 bits so it just swaps the order of writing
09:28:18 Quit n1s (Ping timeout: 268 seconds)
09:29:46amiconnfunman: Eh, sorry, the video does not need swapping
09:30:20 Quit JanDo (Ping timeout: 276 seconds)
09:30:29amiconnColor and Nano still use the old C mess
09:30:55*amiconn needs to finish his PP colour lcd bridge work :\\
09:33:09 Quit phanboy_iv (Read error: Connection reset by peer)
09:36:33 Join JanDo [0] (~JanDo@
09:47:33CIA-5New commit by 03funman (r25479): Fuzev2: YUV output adapted from Fuzev1
09:49:01 Join ender` [0] (
09:51:47*funman "fixed" the bmp2rb yellow
09:56:46 Nick scorche` is now known as scorche (~scorche@rockbox/administrator/scorche)
10:03:08*funman starts benching Clipv2
10:06:14saratogai guess the fixed pclk should mean less boosting?
10:08:14funmansince memory is faster yep
10:08:25funmani have no previous bench for clipv2 though
10:08:50funmanand it crashed earlier so i hope it'll work for the whole 15 hours or so
10:09:14saratogafrom what i understand the clock changes made the sd driver fairly unstable
10:09:35funmanyep especially µSD
10:09:45funmanit could be better now
10:10:41 Join stoffel [0] (
10:12:16saratogathe low clock on the amsv2 is nice, finally gives me a reason to look into making mp3 faster again
10:13:00saratogabet we could shave 1 or 2 MHz off the filterbank on arm9e
10:15:08funman(clipv2 crashed)
10:26:31 Join flydutch [0] (
10:29:41funmanI started the Clip+ with the same album I had used previously, let's if i have better luck
10:30:10funmani also have put backlight always on on the clipv2 to see if it panics
10:32:57funmanclip+ crashed too :(
10:41:52 Join Kitr88 [0] (
10:41:57 Quit Kitar|st (Read error: Connection reset by peer)
10:46:44S_a_i_n_tIs the "lamp" plugin supposed to time out at all, after "X" period, or after a completely random "Y" period?
10:46:52S_a_i_n_tUsing it, its kinda hard to tell.
10:46:56 Quit Kitr88 (Ping timeout: 276 seconds)
10:47:37funmanit doesn't timeout
10:47:51S_a_i_n_tSomtimes it appears to "time out" (the period is random), and others it just crashes the player.
10:48:05funmanwhich player.
10:48:16S_a_i_n_tNano1/2g I get the results I've just mentioned.
10:48:48S_a_i_n_tIt looks as though the battery dies, but when I start it back up the battery is always fine.
10:49:22S_a_i_n_tbut when it appaers to "time out" (ie. not crashing the player) it always seems to be a different time limit.
10:49:42S_a_i_n_tBut now you've said it *doesn't* time out, its even weirder still.
10:50:01funmanaccording to lamp.c it will exit if you press any button other than left/right and scrollwheel
10:50:24S_a_i_n_tYes, this is just starting it up and leaving it though.
10:50:33S_a_i_n_t(I tried to use it to read last night)
10:50:40S_a_i_n_t...much to my dismay.
10:50:58funmanhow much time approximately? i just started it on fuze
10:51:20S_a_i_n_tIt seems to be anwhere from 1~5minutes on my Nanos
10:51:42 Join Kitar|st [0] (Kitr88@
10:51:55S_a_i_n_tAhem.."anywhere from 1 to 5 minutes."
10:54:21S_a_i_n_tfailed after 2 minutes an Nano1g just now.
10:54:33S_a_i_n_tNano2g still going.
10:54:43 Join Lynx_ [0] (~Lynx@
10:54:49S_a_i_n_tBatter on the 1g reads 92%
10:55:19S_a_i_n_tNano2g failed, batter 44% (lamp.rock on for 3mins approx)
10:57:36funmanno crash on fuzev2
10:57:49 Quit Rob2223 (Ping timeout: 265 seconds)
10:58:05funmansomething happen if you just leave the backlight always on and sit in a plugin ?
10:58:18S_a_i_n_tHmmm, well. The Nanos don't seem to be able to handle it for some reason.
10:58:37S_a_i_n_tand no, it can sit in a plugin with the backlight on indefinitely.
10:58:42S_a_i_n_t(as far as I know)
10:58:44ThomasAHfunman: I had a damaged .ogg that crashed the clip+ on Rockbox. Then I tried OF and it froze completely -> 10sec power was the only way out. So RB was much better :)
11:00:00 Quit bluebrother (Disconnected by services)
11:00:01 Join bluebroth3r [0] (~dom@rockbox/developer/bluebrother)
11:00:50S_a_i_n_tfunman: I'm letting it run "plasma.rock" with backlight set to "on" to see if it fails there too. (I figure plasma will be a bigger battery eater than lamp, but I may be wrong)
11:01:08ThomasAHfunman: I still have the file if you think it is worth fixing the code to make it cope with this type of damage
11:01:24 Join Rob2222 [0] (
11:02:05S_a_i_n_tFunman, wow..actually, you were right.
11:02:12funmanThomasAH: Clip+ crashes randomly for me so unless you can reproduce the crash on another target i won't have any use for your file
11:02:24S_a_i_n_tPlasma.rock with backlight on kills it even faster than lamp does.
11:02:40S_a_i_n_tbut when it starts back up, the battery still reads the same.
11:02:41***Saving seen data "./dancer.seen"
11:03:00*S_a_i_n_t is slightly confused.
11:03:48ThomasAHfunman: hmm, with svn from today I just had random crashes, too ...
11:03:48S_a_i_n_tWait, if a plugin is running..."idle timeout" shouldn;t matter, should it?
11:04:00ThomasAH(internal memory)
11:04:05saratogaThomasAH: if you can put it online, you might as well file a bug report
11:04:24saratogaassuming it crashes repeatedly in rockbox
11:04:41ThomasAHsaratoga: repeatedly in roxbox and OF at the same second
11:04:51ThomasAHfamily interrupt ...
11:05:05saratogathats probably a bug the vorbis decoder, the OF and rockbox use the same one
11:12:03 Join bertrik [0] (~bertrik@rockbox/developer/bertrik)
11:16:31 Join Highlander [0] (
11:19:11S_a_i_n_tHmmm, a little testing points to the idle timeout not being applied correctly. With idle timeout set to off, all is well in lamp.rock-land, but when its set to timeout at 1min lamp.rock continues running anywhere from 1~5mins.
11:20:07S_a_i_n_tIMO lamp.rock shouldn;t timeout at all.Even if there is a timeout period set, But that's just me.
11:21:21saratogaas long as you're screwing around with that, maybe add an option to leave the backlight on in in test_codec too
11:21:41saratogaor just leave it on all the time, the time out is annoying and seems to cause problems on the clipv2
11:22:01 Quit saratoga (Quit: Page closed)
11:34:43 Quit funman (Quit: free(random());)
11:46:42 Part JanDo (" ")
11:48:59 Join GeekShadow [0] (~Antoine@reactos/tester/GeekShadow)
11:54:50 Quit flydutch (Quit: /* empty */)
11:58:04 Nick YPSY is now known as Ypsy (
12:01:43 Join merbanan [0] (
12:08:02CIA-5New commit by 03Buschel (r25480): test_boost: fix r25464: button_get() can't be used with actions
12:13:00 Join pamaury [0] (~c2c7a50a@rockbox/developer/pamaury)
12:18:43 Join kugel [0] (~kugel@rockbox/developer/kugel)
12:50:04 Join geertvdijk [0] (
12:56:57 Quit kugel (Ping timeout: 264 seconds)
12:59:39 Join kugel [0] (~kugel@rockbox/developer/kugel)
13:01:55 Join Xerion [0] (
13:02:44***Saving seen data "./dancer.seen"
13:03:13 Join mc2739 [0] (~mc2739@rockbox/developer/mc2739)
13:13:52*kugel thinks the devcon could well have a few more participants
13:16:12AlexPOr do you mean should have?
13:16:12kugelwhoever wants
13:16:38kugellast year we were 2 more
13:18:02kugelhm, I see Zagor in the participants list but B4gder in the bed room list
13:18:15kugelI guess both come?
13:18:16AlexPyeah, they are both coming though
13:18:37AlexPIt is just one isn't sleeping and the other isn't participating :)
13:20:08pixelmaI'll try to but don't know yet... and there were a few who decided quite late the last years too
13:20:34kugelthat was me ;)
13:23:08*gevaerts thinks they're both sleeping
13:23:26gevaertsOnly one of them is participating though
13:33:11 Nick Ypsy is now known as YPSY (
13:48:01 Join JanDo [0] (~JanDo@
13:51:57gevaertsIs there something wrong with the theme site? The admin page seems to insist on showing me all themes
13:56:46 Join froggyman [0] (~me@unaffiliated/froggyman)
13:59:30 Part froggyman
14:01:46CIA-5New commit by 03kugel (r25481): Fuzev2: Reduce code duplication by reusing Fuzev1 code.
14:02:21kugelhm, git tricked me
14:02:26 Join stripwax [0] (
14:02:58CIA-5New commit by 03kugel (r25482): Add forgotten file (git was supposed to rename!).
14:04:50kugelinteresting, bootloader reds. I don't think we need yuv blitting in the bootloaders
14:08:31pixelmathere's a Packard Bell Vibe500 build in the 8MB-Recorder build download - hence the delta
14:09:50pixelmastuff like this happens from time to time in the last month(s), I wonder what's going on there
14:12:45mc2739pixelma: Zagor says this is a build server bug that he has not been able to track down
14:13:09*gevaerts has a look at the code
14:13:20gevaertsIt's perl though, so I might not find much
14:17:13CIA-5New commit by 03kugel (r25483): as2525(v2): We don't need yuv blitting/greylib support in the bootloader so don't compile it.
14:17:16pixelmaif it's only happening sometimes, I can only imagine a timing problem (maybe if two builds are received at the same time or so)
14:17:33 Quit stoffel (Ping timeout: 246 seconds)
14:17:53 Join kugel_ [0] (
14:17:54 Quit kugel (Ping timeout: 265 seconds)
14:17:59 Nick kugel_ is now known as kugel (
14:18:11 Quit kugel (Changing host)
14:18:11 Join kugel [0] (~kugel@rockbox/developer/kugel)
14:18:35gevaertsthey should have unique names by then
14:19:42 Join adnyxo [0] (
14:20:36 Quit Adubb (Read error: Connection reset by peer)
14:23:31 Join robin0800 [0] (
14:23:32CIA-5New commit by 03kugel (r25484): Fix yellow. Another function unneeded in the bootloader.
14:25:10 Quit Highlander (Quit: Quitte)
14:26:34gevaertsnow it's some other builds that go wrong...
14:32:09 Join froggymana [0] (
14:35:58 Quit stripwax (Quit:
14:37:54 Join mt [0] (~chatzilla@
14:42:46 Join mikroflops_ [0] (
14:45:15 Join dfkt [0] (dfkt@unaffiliated/dfkt)
14:46:09 Quit mikroflops (Ping timeout: 240 seconds)
14:47:22 Quit mt (Remote host closed the connection)
14:51:04 Join mt [0] (~chatzilla@
14:55:20 Join stripwax [0] (
14:55:32 Join yorick [0] ( player is consistently refusing to play my music in the database order
14:57:17S_a_i_n_tis "shuffle" on?
14:57:27yorickI think not
14:57:32S_a_i_n_tsorry...I know it sounds stupid, but you'd be surprised.
14:57:53yorickbut would that show the behavior of "select file" -> "play entirely different one"
14:58:08S_a_i_n_tAh, no. It wouldn't.
15:00:30yorick"play track 1" "screw it, I'm playing track 3 and then continuing randomly, entirely skipping track 2"
15:00:56gevaertsthat does sound like shuffle to me...
15:01:09bertrikmaybe it's skipping those tracks because of a codec error
15:01:15gevaertsthat's possible too
15:01:29S_a_i_n_tbodgy FS?
15:01:43 Join jgarvey [0] (
15:01:52bertrikwhat player is this anyway? and what rockbox version?
15:01:59yorickI can rewind back to track 1
15:02:03linuxstbyorick: Have you checked that shuffle isn't enabled?
15:02:10yorickit's an apple ipod video 30GB
15:02:35yorickwith rockbox 3.5.1
15:02:40S_a_i_n_tShuffle can be switched on pretty easily by mistake with the quickscreen...
15:02:47***Saving seen data "./dancer.seen"
15:02:47S_a_i_n_t?me has done this *many* times.
15:03:02yorickah great now it's stuck on "Scanning disk..."
15:03:32 Quit geertvdijk (Ping timeout: 264 seconds)
15:03:55yorick"Shuffle No"
15:03:58yorick"Repeat Off"
15:04:03yorick"Show Files Supported"
15:04:39yorickhmm, it played the same track 3 times now
15:04:58yorickbasically, I took my ipod video, with 25 GB of music on it
15:05:03yorickand then installed rockbox on it
15:05:07yorickand initialized the database
15:05:20yorickthe original firmware still works fine
15:05:30pixelmagevaerts: haven't looked inside the wrong Ipod3G one but the other two which went wrong had builds in them that were done by n17-roolku (the build table listed some different clients for the actual builds so I looked at the builds that were actually included). Two results isn't a very safe bet yet though
15:05:31yorickone album is even entirely doubled
15:05:44S_a_i_n_tI'm ussuming that the original (Apple) firmware plays tracks free from error?
15:06:11linuxstbyorick: After Rockbox initialised the database, how did you restart Rockbox? i.e. what buttons did you press?
15:06:30yorickI did menu+select
15:06:38yorickbecause it wasn't shutting down with long play
15:06:42gevaertsyorick: maybe try checking the filesystem. Filesystem corruption has been known to cause all sorts of weird issues
15:06:54yorickhow am I supposed to check the filesystem
15:07:04S_a_i_n_tdepends on your OS
15:07:11yorickubuntu 9.10
15:07:32mtyorick: Are all the tracks the same format ? Or are the ones being skipped of different format than the ones being played correctly ?
15:07:36gevaertsfsck.vfat then
15:07:55yorickmt: all the same format
15:08:01yorickCBR 320kpbs mp3
15:08:06yorickit happens on every album
15:08:22gevaertspixelma: at least one of today's wrong builds came from GodEater's system
15:08:39linuxstbyorick: Do you normally eject/unmount your ipod before unplugging it, or do you somethings just pull the cable out?
15:08:56yorickI do safely remove :)
15:08:56pixelmagevaerts: I meant the build that was actually put in the wrong zip
15:09:23gevaertspixelma: yes, that's what I mean too :)
15:09:45yorickhmm I think I fixed it
15:09:59pixelmagevaerts: was that the round with the wrong 3G Ipod build?
15:10:15gevaertspixelma: no, the one after, with the wrong yh920
15:10:19yorickit was the centerart theme
15:10:21gevaertsoh, wait, no
15:10:28yorickswitching to the rockboxed theme fixed all my problems
15:10:32gevaertsThe one before, with the wrong recorder
15:11:13gevaertsyorick: that really shouldn't change anything...
15:11:14 Quit antil33t (Read error: Connection reset by peer)
15:11:17S_a_i_n_tyorick: Generally speaking, a theme *shouldn't* affect playback.
15:11:19 Join antil33t [0] (
15:11:22yorickwell it did
15:11:37pixelmagevaerts: then we are talking about different things. I mean - the recorder build had a Vibe500 in it (which was build by n17-roolku). The last round has an Onda v777 in it (which was build by n17-roolku)
15:11:54gevaertspixelma: hm, maybe I wasn't looking right...
15:11:58linuxstbyorick: I would still check the filesystem for errors. And probably re-initialise the database.
15:12:17pixelmagevaerts: in the Samsung zip
15:12:36yorickok...switching back to CenterArt -> broken again
15:12:45gevaertsYes, I know. I mean, maybe I didn't look at the r25482 case properly
15:12:46 Join evilnick_B [0] (~0c140464@rockbox/staff/evilnick)
15:13:10linuxstbyorick: The fact that long-play didn't work isn't a good sign. Was Rockbox itself working fine (i.e. you could move around in the menus), or had it frozen? How long were you holding play for?
15:13:31S_a_i_n_tyorick: Many people use that theme..I'm sure there would be more people complaining if it was the theme alone that "broke" playback.
15:13:59yoricklinuxstb: every part of rockbox worked otherwise
15:14:06yorickexcept the database, which told me to reboot
15:14:19yorickcould the charger being plugged in have anything to do with it?
15:14:33linuxstbYes, Rockbox won't shut down if the charger is plugged in.
15:14:40yorickS_a_i_n_t: I know it's strange, but it's true...
15:15:02S_a_i_n_tthere is probably something deeper going on here...*probably*
15:15:16 Join kugel_ [0] (
15:15:18S_a_i_n_tI can't reproduce it in the sim, the theme works fine for me.
15:15:24gevaertswell yes, the filesystem is still a reasonably likely candidate...
15:15:25*yorick will reinitialize the database
15:15:30 Quit kugel (Disconnected by services)
15:15:33yorickor check the filesystem first?
15:15:36 Nick kugel_ is now known as kugel (
15:15:40 Quit kugel (Changing host)
15:15:40 Join kugel [0] (~kugel@rockbox/developer/kugel)
15:15:47linuxstbyorick: As we've all been saying for 10 minutes, check the filesystem...
15:16:24yorickhmm it switched back to some kind of other screen after connecting
15:16:42S_a_i_n_tdefine "other screen"
15:16:44linuxstbProbably Apple's disk mode screen.
15:17:22yorickempty screen showing battery symbol, flashing forbidden sign, and "Do not disconnect."
15:17:37S_a_i_n_tApple disk mode, this is fine.
15:17:40S_a_i_n_tAnd normal.
15:17:59yorickhmm lets see: fsck.vfat /media/ipodmountpoint
15:18:11gevaertsno, the device, not the mountpoint
15:18:34gevaertsafter unmounting first
15:19:22yorickunmounting automatically ejects it
15:20:04 Quit froggymana (Quit: CGI:IRC)
15:20:38 Join froggymana [0] (
15:20:42 Join n1s [0] (~n1s@rockbox/developer/n1s)
15:22:37gevaertsI know umount doesn't do that. Maybe ubuntu's gui tools do, but I don't know those...
15:22:54gevaertsDoes that affect the ipod though?
15:23:22yorickipod says "OK to disconnect"
15:23:23linuxstbyorick: Type "mount" at the command-line, and that should tell you the device - e.g. "/dev/sdb2"
15:24:03gevaertsOK, as soon as it says that you can run fsck on it
15:24:10yorick/dev/sdb2 on /media/myipodname type vfat (rw,nosuid,nodev,uhelper=devkit,uid=1000,gid=1000,shortname=mixed,dmask=0077,utf8=1,flush)
15:24:24linuxstbOK, so first do "sudo umount /dev/sdb2"
15:24:33*linuxstb doesn't know the fsck syntax...
15:24:58yorickI did umount /media/ipodname
15:25:06gevaertsfsck.vfat -a /dev/sdb2
15:25:10yorickit said "it's now safe to remove ipod"
15:25:23yorickand ipod is still on "Do not disconnect"
15:25:44linuxstbWhat said "it's now safe to remove ipod" ? The umount command should just say nothing.
15:25:54yorickthe umount command
15:26:31 Join lpereira [0] (
15:27:24gevaertsok, so it's not filesystem corruption
15:27:42yorickhmm now how do I disconnect it
15:27:47 Quit mikroflops_ (Ping timeout: 268 seconds)
15:27:55yorickit did it in 5 seconds?
15:28:05gevaertshm, good point
15:28:32gevaertswell, it might
15:28:39gevaertsAnyway, just disconnect it. It's not mounted, so it's safe
15:29:05yorickit says "Do not disconnect"
15:29:30gevaertsif you really want to, run "eject /dev/sdb" first then, but it's not needed in this case
15:29:41gevaertsfsck time usually depends on the number of files, not on the size of the disk
15:29:55gevaertsand 3000 files isn't *that* many
15:30:16S_a_i_n_tdepends on the size of said files
15:30:25yorick20 GB
15:30:29 Join mikroflops [0] (
15:30:53yorickI'll reinitialize the database
15:31:03gevaertsS_a_i_n_t: a bit, but that's all in the FAT, so it's one block of consecutive data, which is fast
15:31:29 Quit ender` (Ping timeout: 252 seconds)
15:31:30S_a_i_n_tAhhh...good point. Still, damn fast, impressive.
15:32:03yorickhow do I reinitialize the database then
15:32:24 Join ender` [0] (
15:32:55 Quit JanDo (Quit: goes home)
15:33:05yorickwhen I do "Initialize Now" it says "Updating in Background"
15:36:26 Quit mikroflops (Ping timeout: 252 seconds)
15:37:28 Join mikroflops [0] (
15:43:38 Join geertvdijk [0] (
15:45:48 Join Schmogel [0] (
15:49:31 Join CGL [0] (~CGL@
15:50:22yorickhmm...centerart is the only theme I installed using rockboxutility
15:50:28yorickcould it be it's broken somehow
15:50:37yorickand breaks playback somehow
15:50:50yorickbecause every other(preinstalled) theme works
15:51:26gevaertsstill sounds unlikely
15:52:02yorickthen what
15:52:21gevaertsDid the problem stay after rebooting?
15:52:29 Quit froggymana (Quit: CGI:IRC)
15:52:33yorickand after reinitializing the database
15:53:48S_a_i_n_tperhaps the sim is doing something different (though, I am using an SVN sim, and not 3.5.1), but the theme works fine for me.
15:54:08yorickmaybe it's a 3.5.1 bug?
15:54:25S_a_i_n_t*unlikely*, but possible.
15:54:34gevaertsS_a_i_n_t: for this sort of weird issues, it's not that unlikely that the sim behaves differently
15:54:39S_a_i_n_tI'm sure it would have been mentioned by now though.
15:54:56yorickcenterart uses custom menu, it says
15:55:24S_a_i_n_tA theme can't alter the menu setup.
15:56:12 Nick fxb__ is now known as fxb (
15:56:20yorickcenterart doesn't seem widely used, maybe only the database is bugged?
15:56:42yorickdownloaded 1886 times
15:56:57S_a_i_n_tthats fairly *widely used*
15:57:00gevaertsyorick: just to clarify, this is CenterArt and not CenterArt v2.0?
15:57:49yorickthis is v2.0
15:58:21S_a_i_n_tWell, *that* would have been nice to know from the start lol...
15:58:35yorickI guess so
15:59:10gevaertsWell, neither of them look very suspicious to me
15:59:43*gevaerts will try to reproduce this as soon as his ipod doesn't complain about a very low battery anymore
15:59:56S_a_i_n_tv2.0 behaves as it should in the sim
16:00:01 Quit n17ikh (Ping timeout: 245 seconds)
16:00:05S_a_i_n_t(or seems to at least)
16:00:07yorickI'll install another theme and see if that also fails
16:00:10 Quit fejfighter (Ping timeout: 258 seconds)
16:01:04GodEatergevaerts: "wrong" builds?
16:01:09GodEaterdo I need to mend something?
16:01:45gevaertsGodEater: no. There's a bug somewhere, probably in the server-side scripts, that sometimes puts an uploaded build in the wrong place
16:03:07yorickok...I installed another theme...that worked
16:03:12yorickthen switched back to centerart
16:03:18yorickthe center art thing is gone
16:03:21yorickbut the rest works fine
16:03:38S_a_i_n_t"center art thing is gone"?
16:03:44S_a_i_n_tas in, no album art/
16:03:56yorickand also no playlist
16:03:59S_a_i_n_tdoes the track *have* album art?
16:04:08yorickbut it didn't before
16:04:12S_a_i_n_twell, there ya go ;)
16:04:19yorickand it displayed some nice playlist instead of album art
16:04:23yorickand now it doesn't do that aswell
16:05:38 Nick fxb is now known as fxb__ (
16:06:07 Join n17ikh [0] (
16:09:25 Join Adubb [0] (~aldubuc@
16:11:28*kugel runs test_codec on fuzev2
16:12:02 Join TabalugaFX [0] (
16:13:00kugelit's 25%-35% faster throughout all testable files
16:14:17TabalugaFXhi @ all
16:14:29 Join panni_ [0] (
16:15:17TabalugaFXI have a question about the themes submitting process which I can't find in the wiki / docs sections
16:15:31S_a_i_n_t...ask away.
16:15:53TabalugaFXso I want to ask how and if it's possible to submitt my thmes which i have "updated"
16:16:00 Quit jordan` (Read error: No route to host)
16:16:07AlexPUse the dame details as before and it'll be replaced
16:16:22AlexPyour name, your e-mail and theme name that is
16:16:28S_a_i_n_t"re"submit it using the same author details and email address, it will ask for confirmation about replacing it.
16:16:57TabalugaFXgreat thanks. this information would be useful inside the wiki / docs
16:17:08kugelleave the fields for the .zip and screens empty, you'll need to refill them when you're asked for updating
16:17:17AlexPTabalugaFX: The wiki is a wiki :)
16:18:37TabalugaFXyes i know, so if i have more time i'll edit and add some informations
16:19:40TabalugaFXand i have another question with an error or perhaps an unwanted bug
16:19:56gevaertsMost of our bugs are unwanted
16:20:05 Join jordan` [0] (~jordan@
16:20:28TabalugaFXyesterday i copied the full HVSC archive (SID) onto my player
16:20:50AlexPWe can't fix that bug :)
16:21:15S_a_i_n_tgevaerts: *most*?
16:21:45gevaertsS_a_i_n_t: well, we do have some debate about whether some things are bugs or features
16:21:52 Join stoffel [0] (
16:22:03S_a_i_n_tAhhhh, right. Gotcha.
16:22:51TabalugaFXand after a long waiting time of the database update process I want to listen to my SID music but after i switch from one track to another rockbox hangs up (occurs after the 6th or 7th switch)
16:23:18gevaertsyorick: do you have the font pack installed?
16:24:00TabalugaFXor sorry i wasnt mean
16:24:39AlexPTabalugaFX: Does this happen if playing through the file browser?
16:25:03TabalugaFXno, while i'm using the wps
16:25:16TabalugaFXbut I created the playlist inside the file browser
16:26:10TabalugaFXand i already turned off the repeat feature to sbypass the subsongs of every sid file
16:26:45AlexPI meant if you select the song through the filebrowser
16:26:55AlexPas opposed to the database
16:27:11 Quit stripwax (Quit:
16:27:14TabalugaFXyes indeed i selected the folder i want to listend and add them to the playlist
16:27:39AlexPSounds like a bug then
16:27:40yorickgevaerts: dunno
16:27:59yorickgevaerts: I did standard installation
16:28:10AlexPCould you add it to flyspray, including all the details (player, Rockbox version, settings etc.) and say how to reproduce it?
16:28:19gevaertsyorick: does the CenterArt theme look correct, i.e. does it have proper line drawing?
16:28:39TabalugaFXok i'll
16:28:45yorickgevaerts: yes
16:28:49AlexPTabalugaFX: Thanks
16:28:51TabalugaFXbtw thanks for your help
16:28:58AlexPno problem :)
16:29:03TabalugaFXah btw
16:29:12 Join paulk_ [0] (
16:29:20TabalugaFXis it possible to install rockbox on my dingoo a-320?
16:29:32AlexPOnly the players listed on
16:29:55TabalugaFXyes i know, but dingux (dingoo linux) uses the rockbox bootloader
16:30:06AlexPDo you have a link?
16:30:11TabalugaFXhmm wait
16:30:16AlexPEither way, full Rockbox would have to be ported
16:30:27gevaertsyorick: can you check the font settings to see if the 14-Terminus-Bold font is there?
16:30:38AlexPIt sounds like it should be possible, but it is a lot of work
16:32:12TabalugaFX (named "Rockbox mode")
16:33:40yorickgevaerts: it is
16:34:55yorickooh it's broken again
16:34:57yorickafter rebooting
16:35:04AlexPTabalugaFX: I don't see any mention of Rockbox though, just something called Rockbox mode
16:35:31yorickgevaerts: it just asked me if it was OK to remove dynamic playlist
16:35:40S_a_i_n_tRockBox == Dualboot?!? (apparently)
16:36:35S_a_i_n_tyorick: confirming deletion of a dynamic playlist is a system setting, this can be turned off.
16:36:43 Quit TabalugaFX (Quit: CGI:IRC)
16:36:47yorickS_a_i_n_t: yes but I don't remember making one
16:36:50gevaertsyorick: can you put the contents of .rockbox/config.cfg on a pastebin?
16:36:51yorickit asked me this after a reboot
16:38:26 Join TabalugaFX [0] (
16:38:59TabalugaFXre. sorry my connection is very slow and hangs from time to time :-/
16:39:40AlexPTabalugaFX: Don't worry - see for logs to see if you missed anything
16:39:51*S_a_i_n_t wonders why there is a blank line between each setting.
16:40:11TabalugaFXthanks. yes they only mentioned the dual boot feature
16:42:35gevaertsThis is getting weirder and weirder...
16:43:23S_a_i_n_tgevaerts: You can reproduce?
16:44:02gevaertsI installed CenterArt2 on a fresh 3.5.1 install, and started playing an album from the database. The WPS shows that it's playing track 4 in that album. I then switch to cabbiev2, and suddenly the WPS shows track 1
16:44:16*gevaerts decides to go and get some earbuds
16:45:19yoricktold ya
16:46:40TabalugaFX@ gevaerts. ckeck (id3) tags, perhaps an *corrupted* vaule inside, or id3v1 = tack numb 4, id3v2 = track numb 1 ? or something inside the wps cfg
16:47:27yorickTabalugaFX: this happens on every track I play
16:47:40yorickI think it's more like the difference between the database entry and the file list entry
16:47:43 Join Eugenpaul [0] (
16:47:50TabalugaFX@ gevarts or try an "updated" rockbox version
16:47:53TabalugaFXand update database again
16:47:57 Part Eugenpaul
16:48:00AlexPTabalugaFX: gevaerts is a core developer :)
16:48:01gevaertsTabalugaFX: I'm trying to isolate a bug here...
16:48:10AlexPAnd he is trying to help yorick with a bug
16:48:12TabalugaFXor sorry, i don't know
16:48:13gevaertsgah, battery dead again...
16:48:24yorickgevaerts: connect charger?
16:48:34gevaertsyorick: it's connected...
16:48:57yorickopen it up and short out some diodes :P
16:49:09S_a_i_n_tyeah, that'll help.
16:50:20 Join toffe82 [0] (~chatzilla@
16:50:32gevaertsAnyway, since I can reproduce it with 3.5.1, let's try a current build next
16:51:25yorickbut how come there are no bug reports on it
16:51:52*gevaerts decides to apply FS #8802 for this build
16:52:52paulk_I've made some personnal modifications to rockbox :
16:52:52paulk_* I've set-up the BUTTON_REC of my sansa's keymap to ACTION_FM_RECORD
16:52:52paulk_* I've set radio.c to launch recording screen with mic as input source when ACTION_FM_RECORD is pressed
16:52:52DBUGEnqueued KICK paulk_
16:52:52paulk_* I've made some minor modifications to francais.lang
16:52:52paulk_I think that someone could be interested by those changes, so can I upload them to svn ?
16:53:17AlexPno, as you don't have access
16:53:25gevaertspaulk_: no, but you can upload patches to our tracker
16:53:25AlexPYou could put a patch on flyspray though
16:53:40S_a_i_n_tAnd since there is now HOTKEY which can already do this...I doubt it even further.
16:53:44kugelpaulk_: I don't think we're interested in the 2nd one
16:54:16AlexPpaulk_: Maybe you should e-mail the dev mailing list and see what people think
16:54:32yorickgevaerts: FS #8802 works here
16:54:40TabalugaFXnext question: what does the battery time (menu settings > rockbox info) behind the precentage means
16:55:09TabalugaFXis this the "remaning" battery time?
16:55:25paulk_not with mic as input source but with radio as input source :
16:55:38paulk_Ok, I'll probably do a patch :)
16:55:48AlexPTabalugaFX: yes
16:56:10gevaertspaulk_: don't make one patch for all of those, unless you want to make sure it's rejected...
16:56:51TabalugaFX@alexP: but my battery only have a limit of 8 hours (already make an battery bench and read out the txt file) and a value of 104h 57m is shown here
16:57:02AlexPThat means it hasn't been calibrated
16:57:15AlexPA stupidly high value is shown to make it clear that it is wrong
16:57:30paulk_No : one patch for francais.lang and one for radio.c/keymap
16:57:57AlexPIn recent builds I *think* it shows 0 or something like that if it hasn't been calibrated
16:58:02TabalugaFXah, now i understand. and how can i calibrate it
16:58:12AlexPWhat player?
16:58:26TabalugaFXiriver h10 6gb ums
16:58:40AlexPWe need a series of battery bench marks to get an idea of the discharge curve, then a developer can apply it and newer builds will use it
16:59:00TabalugaFXi can submit mine if you want
16:59:42AlexPTabalugaFX: See
17:00:09AlexPfor the conditions to do one, and then yes if you could submit one that would be helpful :)
17:00:56gevaertsOk, the bug is still there in trunk, and it doesn't depend on the database (and CenterArt is half-broken on the current build, *and* I hate touchwheels...)
17:00:59AlexPIt should be with reset settings too, to make sure things like the EQ are off
17:02:13TabalugaFXi already have done this bench with the listed conditions (charging up, repeating playback, don't touch it while playing, writing down end time)
17:02:29AlexPThat'd be useful then
17:02:49***Saving seen data "./dancer.seen"
17:02:51AlexPCould you attach it to
17:05:12 Quit grndslm (Ping timeout: 276 seconds)
17:06:42 Join grndslm [0] (
17:08:43 Join funman [0] (~fun@rockbox/developer/funman)
17:09:21gevaertsyorick: my impression is that it only goes wrong if you switch to CenterArt after booting, i.e. if it was already loaded on boot it's fine. Does that match what you're seeing?
17:10:08TabalugaFX@alexp: sorry i can't sttach my battery bench cause "access denied"
17:10:16kugelfunman: ping
17:10:26kugelwe have 1MB iram right?
17:10:52funmanright, i only tested on Clip+ though
17:10:54kugel(on as3525v2)
17:10:56AlexPTabalugaFX: You need to ask in here for write access after registering
17:11:19AlexPWhat is your wiki name?
17:11:43funmanClipv2/Clip+ are quite unstable now, we should try changing pclk/fclk by small steps so pclk doesn't go too low
17:11:48yorickgevaerts: no
17:12:27funmani'm running a bit with boost forced (so there's no switching) to see if it improves (I only saw a stkov ata/sd on Clip+ with this)
17:12:43AlexPTabalugaFX: You should be able to now
17:12:45CtcpIgnored 1 channel CTCP requests in 0 seconds at the last flood
17:12:45*yorick reboots
17:12:55TabalugaFXok, thanks, i'll try again
17:13:00FlynDicefunman: We're running PCLK at 24 MHz for clip+/v2 and 60 MHz for fuzev2 <−−−−is this correct?
17:13:12 Join Eugenpaul [0] (
17:13:26funmanalthough it's not exactly 24MHz when we switch frequency
17:13:38yorickgevaerts: nope, untrue
17:13:50kugelbtw, my fuzev2 shows 240MHz after boot even if unboosted
17:13:54 Part Eugenpaul
17:13:56gevaertsyorick: ok
17:13:58funmanyeah i didn't modify debug menu
17:14:14kugeltest_fps also assumes 240MHz
17:14:30FlynDiceI just modified debug page and will commit it soon
17:14:48kugelfunman: I think I verified 1MB on my fuzev2 (by modifying test_mem a tad bit)
17:15:06yorickgevaerts: just switched mid-track after switching theme
17:15:19gevaertsyorick: yes, I've seen that too
17:15:22kugelat least (512k-65k)*2 fits into the plugin iram. 0x20000 bytes is for the core
17:15:35kugelboth together is pretty much 1MB
17:15:48funmanwhich is faster: iram or dram?
17:15:49kugeland i don't get a crash or something
17:15:59TabalugaFX@ alexp: ok, ive attached it
17:16:06kugeliram is a bit faster, not so much when unboosted but about 50% when boosted
17:16:37kugelfunman: my idea was to move the entire codec buffer into iram. the current one is 1MB but it's in dram
17:17:00AlexPTabalugaFX: Thanks, hopefully someone will have a look
17:17:10kugelmaybe saratoga shares his test_mem changes for more accurate numbers
17:17:25TabalugaFXyes this would be great
17:17:51kugelwe could probably still reserve 128kB or so for the core, 1MB is hardly needed for codecs
17:18:08TabalugaFXbtw is there a way i can calibrate it mysel to show the right remaining time
17:18:18funmankugel: i think the core+plugins don't even need 128kb
17:18:41kugelplugins share codec iram
17:19:16kugelthe core iram is hardly used on fuzev1 indeed the biggest blob is the framebuffer, but then there only like 6k used additionally
17:19:33 Join Eugenpaul [0] (
17:19:44TabalugaFXand when my rockbox voice tell me "battery level 50%" is this the right time / value instead)
17:19:50kugel(I always meant to fill the rest with the dma buffer of the sd driver to free some dram and speed up)
17:20:29AlexPTabalugaFX: You would have to edit one of the source files then recompile
17:20:40funmanFlynDice: nothing in the SD driver depends on pclk ?
17:21:14TabalugaFXphew. this would be my 1st rockbox cimpiling steps. but i'll try it
17:21:34AlexPIt isn't too bad really :)
17:21:41TabalugaFXbtw you need to update your doom wad file archive, cause a new freedoom version is out there
17:22:00TabalugaFXand if i have time i want tu update your icons archive
17:22:02AlexPDoes it work properly on Rockbox?
17:22:14AlexPWhich icons archive?
17:22:16TabalugaFXcause some of them won't work right
17:22:19kugelfunman: I also did a test_codec run, 60MHz looks like a big waste. do we have a choice for something between 24 and 60 (although 24 seems ideal)
17:22:38CIA-5New commit by 03funman (r25485): as3525v2: show the correct freqs in debug menu, CGU_PERI uses fclk as source
17:22:40TabalugaFXextras > icon setgallery
17:22:40kugelthe most common codecs (mp3, ogg, wma) nee about 20-25MHz
17:22:43AlexPThe gallery?
17:22:50AlexPYeah, that is user contributed anyway
17:23:10funmankugel: check if display is fast enough, I didn't tested after switching to RGB565SWAPPED pixel format
17:23:21FlynDicefunman: It seems to be so, I think that CGU_MS goes to the sd controller bus interface and CGU_SDSLOT goes to the controller card interface and we need CGU_IDE but I'm not sure exactly why..
17:23:39S_a_i_n_tTabalugaFX: the icons thing is a known bug.
17:23:44funmanperhaps there's 1 clock for the controller and 1 clock for the card ?
17:23:44S_a_i_n_tI found it a while ago.
17:23:48kugelfunman: I actually meant that the CPU frequency debug menu shows 240MHz
17:23:49funman1 for each card rather
17:23:54AlexPS_a_i_n_t: Is it on flyspray?
17:23:58TabalugaFXno, you only need to adjust the cfg files right
17:24:09AlexPgoody :)
17:24:27S_a_i_n_tViewer icons apply incorrectly (or similar)
17:24:28funmankugel: ah i've seen that too
17:24:41TabalugaFXcause afer that i can use all of them (expect the larger ones look very large for my small display)
17:24:43S_a_i_n_tdont remember the FS# offhand sorry
17:24:45funmanif i press right it shows a good value
17:24:46kugelsansafuze.h still defines CPU_FREQ to 240MHz if that matters
17:25:05TabalugaFXyes you need to crate *.icons file
17:25:13TabalugaFXand all goes right
17:25:24S_a_i_n_tTabalugaFX: even then, some just *don't* show
17:25:40S_a_i_n_ttrust icons file has ~250 filetypes ;)
17:25:50TabalugaFXplease share with us
17:25:58TabalugaFXbut why so many
17:26:03TabalugaFXor much
17:26:34S_a_i_n_tits inside the "Symmetry" theme for iPod Nano 1/2g...but you probably won't like it as it only points to 1 icon
17:26:39S_a_i_n_t(goes with the theme)
17:27:00TabalugaFXok i'l take a look inside
17:27:23TabalugaFXnxt theme i want to "remake" is the pen&paper for h10
17:27:26S_a_i_n_tyou could easlily edit it to point to whichever icon you want though.
17:28:35TabalugaFXok thanks for the details. btw why the transparency vaule won't work for some icons
17:29:01S_a_i_n_tjust displays a black square?
17:29:15kugelfunman: hm, not unbearable slow anymore but still noticeably slower (and no fun). it would be great if we could go for something between 30 and 40 MHz I think
17:29:31funman36 or 48?
17:29:38TabalugaFXhad some troubles with the tango s/w icons which shows the icons but not the transparency background so i saw the icon inside a white box
17:30:03funmanhm no
17:30:12 Join efgpinto [0] (
17:30:21funman30, 34.29, 40
17:30:22 Nick YPSY is now known as Ypsy (
17:30:24kugelfunman: 36 would work nice I think. the fps should at least be above 50 imo
17:30:28TabalugaFXsorry i mean black & white for (s/w) )and not the background. so I reated my own tago greyscae icons
17:30:58S_a_i_n_tblack & white has no transparency IIRC
17:31:27bertrikyou might reconsider the highest frequency used in ams v2 players too, the audio playback rate is now more than 1% off
17:31:36funmankugel: it must be a fraction of 240 (down to 240/16 == 15MHz)
17:31:48funmanbertrik: we don't know how to change the PLL freq
17:32:25TabalugaFXbut the rockbox default icons are black&white and transparency works here
17:32:45kugelfunman: does clock target handle fractional frequencies?
17:33:05 Quit pamaury (Quit: Page closed)
17:33:08funmanyes but they're reduced to integer
17:33:36 Join fabioalmeida [0] (
17:33:36 Quit efgpinto (Read error: Connection reset by peer)
17:33:40kugelhow can I double check the divider that's calulated?
17:34:12kugelthere's no #define for it?
17:34:33FlynDiceyou can look on the debug page if you can boot it ok
17:34:36S_a_i_n_tTabalugaFX: the "mono" icons are for non-colour targets, which (AFAIK) don;t support transparency...but I haven't played around much as I only have colour targets.
17:35:21 Quit fabioalmeida (Quit: Ex-Chat)
17:35:25TabalugaFXok, so i need to change the mono icons to color and set the transparency value by myself
17:35:45S_a_i_n_tthat seems sane, so yes, probably.
17:35:45 Join pamaury [0] (~c2c7a50a@rockbox/developer/pamaury)
17:36:25funmanhm 24MHz is just not enough for mp3 :)
17:36:27 Join efgpinto [0] (
17:36:32TabalugaFXok, than i have done this already inside my "simple theme grey" which includes my created tango greyscaled icons
17:37:05S_a_i_n_tWell, I think the mono targets *do* have transparency, but the "magic" colour (the transparent colour, magenta in the colour targets) is a different value
17:37:19S_a_i_n_tso, doesn't show as transparent on colour targets.
17:37:34pixelma1-bit mono bitmaps use foreground and background colours on colour targets too
17:37:39kugel34285715 seems to work
17:38:33yorickhmm the matrix effect is a bit silly here
17:38:41yorickit's moving on the beat of the music
17:39:01pixelmathis works on greyscale too (in the WPS where you can have different foreground and background shades with viewports
17:39:23kugelfunman: I think I could live ok with 34.3MHz, let me try 40 too
17:39:42S_a_i_n_tI knew a non-colour target themer would chime in eventually ;)
17:39:55funmankugel: the e200v1 framerate is awfully fast at 24MHz :(
17:40:08TabalugaFXso is it possilbe to make a theme with wto ore more colors (e.g. red, green, blue and so on) and package all into one "theme"
17:40:09kugelthat's because the e200v1 cheats
17:40:30pixelmaS_a_i_n_t: what? I use this in my c200 WPS too
17:40:41kugelit memcpy's the framebuffer to another one where the lcd driver auto-reloads from automagically
17:41:07S_a_i_n_tpixelma: yes, but you also understand how the mono targets work better than I ;)
17:41:30AlexPTabalugaFX: Not yet
17:41:39 Join Kitr88 [0] (
17:42:56TabalugaFXthis would be great 1st: select your theme; 2nd: select your color; 3rd: select / edit your custom color
17:43:13 Join lImbus [0] (
17:43:22 Join fabioalmeida [0] (
17:43:43AlexPTabalugaFX: It is in the vague "It'd be nice to have multi-colour packs when someone gets round to doing it" area
17:44:04 Quit Kitar|st (Ping timeout: 264 seconds)
17:44:23 Join Kitar|st [0] (
17:44:33gevaertsyorick: I've submitted bug FS #11175 ( about this. If you find out more, feel free to add notes
17:45:20kugelfunman: ok, with 40MHz I cannot really tell the difference between boosted and unboosted (UI-responsiveness wise). fps is improved by 20% over 34.MHz (51 vs 60)
17:45:35TabalugaFXoh great, than I'll see what the future brings. curently i've created the "iamp h10 remakes" both in normal and inverted colors, cause my friends say the night mode look great too and i can't devide which one i sould submit, so i submitted both
17:45:54 Quit Kitr88 (Ping timeout: 240 seconds)
17:46:14kugelwhile the clock speed is only increased by 16.6%
17:46:36yorickgevaerts: low?
17:46:39funmandisabling switching doesn't crash (at least less quickly), but i'm not sure how to switch nicely
17:46:52yorickgevaerts: that makes my favourite theme unusable :p
17:47:04kugelmaybe you checkout the frequencies too, I'd almost vote for 40MHz, but I wouldn't die with 34.3
17:47:08funmanthe OF crudely iterates over dividers to keep pclk very approximately close to the desired value as fclk changes
17:47:14CIA-5New commit by 03amiconn (r25486): Improve SDL detection, so that path order doesn't matter if both native SDL and cross-mingw32 SDL are present. Reduce code duplication as well.
17:47:21gevaertsyorick: well, I left those at the default values. People tend to ignore them anyway
17:47:54yorickgevaerts: tend :p
17:47:59bertrikfunman, I suppose they first reduce the pclk, then increase fclk, right?
17:48:11kugelI'd say 34.3 feels about the same as unboosted on a e200v1 ui-wise
17:49:42funmanbertrik: < they modify both in a loop
17:50:00gevaertsyorick: well, have a look at the tracker and you'll notice that we only have two bugs that are not severity: low
17:50:01funmanin this function they are boosting
17:51:01 Quit jd (Ping timeout: 265 seconds)
17:51:18AlexPyorick: Even if they were used, a particular theme not working is low anyway in my book :)
17:51:18 Part lImbus
17:51:21TabalugaFXok, now i need to say goodbye. thanks again for the great support, technical help and the answer of my questions and ideas. and thanks for your vision and your development of my beloved opensource firmware which rocks through my ears every day (and during some sleepless nights)
17:51:32AlexPHeh, you are welcome :)
17:51:55TabalugaFXand sure, I'll help where can
17:51:56S_a_i_n_tcya TabalugaFX
17:52:09yorickAlexP: it's probably some feature the theme uses
17:52:13TabalugaFXto support and promote
17:52:26TabalugaFX...logging off..
17:52:38 Quit TabalugaFX (Quit: CGI:IRC)
17:52:45kugelfunman: what do you think about my codec buffer idea?
17:52:46AlexPyorick: Yeah, but it still isn't critical or high, like playback not working
17:53:13funmankugel: sounds good if the iram is not too much slower, we'd use memory effeciently that way
17:53:13 Join jd [0] (
17:53:13 Quit jd (Changing host)
17:53:13 Join jd [0] (~jd@Wikipedia/HellDragon)
17:53:25kugelit's faster :p
17:53:37S_a_i_n_tAs long as rockbox_defalut loads, music plays, and plugins work...all is well with the world ;D
17:53:37kugelbut really only when boosted it seems
17:53:50 Quit Xerion (Read error: Connection reset by peer)
17:54:03gevaertsAlexP: I consider the fact that this bug is even possible to be a bit frightening though
17:54:23kugelwe should also decrease the pcm buf size, 90% of it unused
17:54:29 Quit jd (Read error: Connection reset by peer)
17:54:38funmankugel: try with ape first
17:54:39S_a_i_n_tgevaerts: very true...looking at the code its VERy hard to pick what the heck is going on there.
17:54:41AlexPIt is all academic anyway as we don't use the rating thing
17:55:24S_a_i_n_t(code for the theme that is)
17:55:42kugelfunman: nobody uses ape...
17:55:49FlynDiceclip+ is getting random freezup for everyone and not just me correct?
17:56:14funmanFlynDice: yep, keeping it boosted (or unboosted but you need to patch) should work around it
17:56:24funmankugel: perhaps but we still support it
17:56:37kugelfunman: smaller pcm buffer doesn't mean ape will not work
17:57:00FlynDiceI'm testing some delays after changing the divs and so far so good but I'm only at 10 mins or s
17:57:08kugeljust that it *maybe* boosts more often which is the case with ape anyway
17:57:14*FlynDice keeps fingers crossed
17:57:21funmandepends if filling it takes more time than what's available in the buffer
17:57:58kugelape data aborts apparently
17:58:17funmanFlynDice: i think either we need to add a delay, either pclk goes too low for some peripheral even if we're not using it at the moment we switch (pclk goes to 2.4MHz)
17:58:28 Join jd [0] (
17:58:28 Quit jd (Changing host)
17:58:28 Join jd [0] (~jd@Wikipedia/HellDragon)
17:59:25FlynDiceI'm going to let it play for awhile here but it's doing well so far with an i=40 delay similar to OF
17:59:59 Quit yorick (Quit: Poef!...)
18:00:13 Join lImbus_ [0] (
18:00:51 Nick lImbus_ is now known as lImbus (
18:01:22kugelfunman: should I commit the 40MHz or do you want to wait and see yourself?
18:01:37 Quit xiainx (Ping timeout: 252 seconds)
18:01:58funmanjust commit
18:03:13 Nick fxb__ is now known as fxb (
18:03:22 Join yorick [0] (
18:03:32yorickbattery 95% 195h 0m
18:03:55yorickI think it's supposed to be charging
18:04:01yorickit was 96% just then
18:04:09 Part lImbus
18:04:10yorickand the charger icon is showing
18:04:57 Join komputes [0] (~komputes@ubuntu/member/komputes)
18:05:10FlynDicecrap, just froze :( Seems to be an SD access issue as the LED icon is illuminated
18:05:19S_a_i_n_tyorick: Is there a question in there?
18:05:37yorickS_a_i_n_t: how do I fix the 195h
18:05:51S_a_i_n_tits nothing you can "fix"
18:06:03S_a_i_n_tsubmit a battery bench using the criteria on the wiki
18:06:03funmanwith 24/240MHz we can switch while staying between 13.3MHz and 30MHz
18:06:20kugelany reason we use the DRAM_FREQ for defining the cpu freq?
18:06:22S_a_i_n_tit will help toward calibrating the battery for your player.
18:06:56funmankugel: we use PCLK (which is equal to DRAM_FREQ, unless you set bit 6 of CGU_PERI on amsv1)
18:07:41CIA-5New commit by 03amiconn (r25487): Set options so that the sim actually builds on Opensolaris. The build still throws hundreds of warnings.
18:07:46FlynDiceback later..
18:09:41yorickat best it's only 10 times too much
18:09:49CIA-5New commit by 03kugel (r25488): Add T for plugins to the advanced build options to build all test_* plugins.
18:09:53CIA-5New commit by 03kugel (r25489): Fuzev2: Use 40MHz as unboosted frequency. The lcd speed and ui responsiveness is good at this freq.
18:11:51CIA-5New commit by 03kugel (r25490): Correct comment.
18:12:40 Join xiainx [0] (~iain@socs-7.CS.McGill.CA)
18:19:29 Join lImbus [0] (
18:20:07 Part lImbus
18:20:39funmanFlynDice: i'm testing on Clip+
18:21:55linuxstbkugel: Shouldn't more of those test plugins have #ifdefs around them in SOURCES? e.g. SWCODEC for test_codec, HAVE_ADJUSTABLE_CPU_FREQ for test_boost.
18:22:22kugelhrm, probably
18:22:43kugelI didn't go through all possible combinations and I may have forgotten something
18:23:05funmani suppose the peripherals running off pclk are DMA and i2c
18:24:02funmanbut for dbop & ssp it shouldn't matter
18:25:20funmani don't see clock requirements for dma
18:25:44bertrikI often thought it would be nice to have multiple boost levels
18:26:26linuxstbkugel: Also, s/unbearable/unbearably/ (but I wouldn't use subjective words like that in comments - what is unbearable to one person may be bearable to others.)
18:26:34funmani2c runs at 400kHz and iirc the requirements is "between 100 & 400" so we should let pclk only go lower (up to 4 times lower), not higher
18:26:47 Join flydutch [0] (
18:27:41*S_a_i_n_t thinks multiple screen refresh rates for different targets that don't mess with the scrolling speed in lists would be nice also...
18:27:42funmanbertrik: saratoga suggested the same thing
18:27:47amiconn400kHz is already i2c "fast mode"
18:27:57funmanamiconn: yes so we shouldn't go faster
18:28:01amiconnStandard mode is <= 100kHz. Afaik there is no lower limit
18:28:24amiconnBut basically all i2c devices support fast mode nowadays
18:28:50funmanas3525 datasheet quotes "supports standard (100 kbps) and fast speed (400kbps)"
18:28:55amiconnThere's one notable exception among the rockbox targets: The m3's button controller (a pic) doesn't
18:29:05amiconnIt also has a rather buggy i2c implementation
18:30:39 Quit TheSeven (Ping timeout: 265 seconds)
18:30:56 Join tomers [0] (
18:32:35 Join stripwax [0] (
18:33:20ThomasAHfunman: do you expect the crashes to go away with
18:33:49funmannope it's buggy (see above)
18:33:58funmanpclk must not go higher than 24MHz
18:40:44 Join Antibuddha [0] (
18:40:57Antibuddhahow hard would it be to rockbox an mp3 player like cowon q5w?
18:41:18Antibuddhai just got one off ebay but im afraid it will be crap without rockbox
18:42:19yoricknot easy
18:42:21S_a_i_n_tWell...its either a LOT of work, or you could hold your breath. Possibly indefinitely.
18:42:29stripwaxAntibuddha - working from the assumption that all mp3 players are different, the answer would be 'as hard as all the other ports'. see the NewPorts page on the wiki if you are able to help!
18:42:59Antibuddhahmm forget that then, anyone know how to test if an spdif is resampled to 48khz?
18:43:15stripwaxAntibuddha - (wiki link by the way is
18:43:27Antibuddhaif my new cowon q5w can spdif rca output unresampled 44.1khz, i will be happy
18:44:07 Quit tomers (Quit: ChatZilla 0.9.86 [Firefox 3.5.8/20100214235838])
18:50:25 Join phanboy4 [0] (
18:50:29stripwaxAntibuddha - I personally don't know, but this is #rockbox, so it's possible you won't find your answer here ..
18:50:36 Join captainkewllllll [0] (~2669ecc2@gateway/web/freenode/x-wavbesbpygsqmcfc)
18:52:35 Quit paulk_ (Quit: Ciao)
18:52:54 Join Strife89 [0] (~michael@
18:55:01yorick:o @ gapless playback in practice
18:55:15funmanFlynDice: ThomasAH: only 1 intermediate step and fixed the numerous bugs
18:56:00 Join TheSeven [0] (~theseven@rockbox/developer/TheSeven)
18:57:23funmanwe might even calculate the intermediate value with awful macros :)
18:59:49funmanPANIC read multiple blocks failed (Clip+ µSD)
19:00:20 Quit stripwax (Quit:
19:00:50 Quit antil33t (Read error: Connection reset by peer)
19:00:55 Join antil33t [0] (
19:01:59 Join stripwax [0] (
19:02:46 Quit geertvdijk (Ping timeout: 246 seconds)
19:02:53***Saving seen data "./dancer.seen"
19:06:15funmannot sure why µSD is affected :(
19:07:22 Quit stoffel (Remote host closed the connection)
19:07:24 Join moos [0] (moos@rockbox/staff/moos)
19:08:23 Join Strife89|PalmTX [0] (~cstrife89@
19:08:28gevaertsmoos: is boot speed back to normal for you now?
19:09:20moosgevaerts: hi, didn't had the chance to test yet. I still have to fix my beast to do various set of tests :(
19:09:38 Quit scorche (Ping timeout: 240 seconds)
19:14:21 Join scorche [0] (~scorche@rockbox/administrator/scorche)
19:16:04 Quit Strife89|PalmTX (Quit: Leaving)
19:16:44 Join freddyb [0] (
19:20:16 Join freddyb_ [0] (
19:21:46 Quit freddyb (Quit: CGI:IRC (Ping timeout))
19:21:57 Nick freddyb_ is now known as freddyb (
19:23:27funmanClipv2 still plays though
19:23:44 Quit Strife89 (Quit: Changing buildings.)
19:24:25 Join geertvdijk [0] (
19:24:46funmanas3525 datasheet says "Clock gating takes effect immediately!" (for CGU_PERI though)
19:25:03funmanhm but it refers to enabling/disabling individual peripherals
19:28:09 Part fabioalmeida
19:31:26 Join Strife89 [0] (~michael@
19:31:27 Quit grndslm (Ping timeout: 264 seconds)
19:32:56funmanlinux for as3525(v1) doesn't show any delay after modifying either CGU_PROC either CGU_PERI
19:35:11freddybfunman: is there a different datasheet for Sansa v2's?
19:35:56funmanhum well there is a datasheet for the Sansa *AMS* v1, not for the Sansa *AMS* v2
19:36:33funmanAMS v1 were previously called "Sansa v2" so just to confirm, you're speaking about Clip+/Clipv2/Fuzev2 ?
19:37:32freddybYes, I have a Fuze v2 also and I was thinking about digging in to see if there's anything I could help finish up.
19:38:17 Quit phanboy4 (Read error: Connection reset by peer)
19:39:46funmanFlynDice: ThomasAH
19:41:32funmanfreddyb: afaik they have an as3525 with modified CPU/peripheral clocks (i'm looking at it now), and the CPU, SD controller, USB controller, i2c (audio/PMU) registers; of an as353x; also a bit more IRAM (320kB -> 1MB)
19:42:33funmanyou could look at recording, it used to work (although there is not write support so you can store the records on disk) but I broke it with dynamic CPU freq switches
19:42:46funmanyou're already familiar with that :P
19:43:17freddybDid you break it by loafing in the higher interrupts? :p
19:43:59freddybI'll take a look.
19:44:36funmannope i'm not sure what happened
19:45:05funmani just looked on my fuzev2 and it seems to work.. (not using latest rev though)
19:45:39funmanusually i would just stand in the menu and a push error come after 1 second
19:49:16 Quit CGL (Remote host closed the connection)
19:51:26 Quit TillW (Quit: This now concludes our broatcast day.)
19:56:44funmankugel: re r25489, if you want to change CPU_FREQ in config/*.h, fine, but do it for all targets and start CPU unboosted in system_init(). IMO you should rather revert it (did you grep for CPU_FREQ ?)
20:00:37kugelfunman: cpu_frequency is initialized to CPU_FREQ, that was the reason the debug menu showed the wrong frequency
20:00:49 Quit Antibuddha (Quit: ChatZilla 0.9.86 [Firefox 3.6.3/20100401080539])
20:01:10 Join hebz0rl [0] (
20:01:24kugelsomething wrong with it?
20:03:52 Quit Horscht (Quit: Verlassend)
20:05:01funmanhm ok
20:05:13funmanyeah we initialize CPU to boosted
20:05:28funmaniirc I had seen a call to set_cpu_frequency(CPUFREQ_NORMAL) in init()
20:06:05 Join Horscht [0] (~Horscht2@xbmc/user/horscht)
20:06:11funmanwell it's the first thing after system_init()/kernel_init()
20:06:59kugelit can be reverted if it's wrong, I thought that causes the wrong freq to be displayed
20:07:47funmanwhatever you prefer, if it's in sync with system_init() and with other targets for consistency and
20:08:29funmanbtw AMSv1 still say 250MHz in their config file but debug menu is fine
20:09:30funmanso mind = blown
20:09:57kugelI think it worked before your today's fix actually
20:10:39funmani had seen it yesterday iirc
20:10:48funmanyou meant r25475 ?
20:11:59 Join kadoban [0] (
20:12:16kugelbut I don't know for sure
20:14:46funmanworks fine if I set CPU_FREQ to 240M and DRAM_FREQ to 40M (I have the last diff I pasted here applied though)
20:15:15funmanah .. but this diff uses freqs for Clips :o
20:15:17 Join wolfsmond [0] (
20:16:50 Join TillW [0] (
20:18:19funmankugel: ok for ? it shows 40MHz in the debug screen after boot
20:18:58kugelthat's simply reverting...why does it work for 40 but not for 60?
20:19:38kugels/unbearable/unbearably/ as linuxstb while you're at it :P
20:19:44funmanalready done :D
20:20:14 Quit n1s (Quit: Lämnar)
20:20:20funmanhm indeed with 60 it doesn't work
20:20:33 Join hebz0rl_ [0] (
20:21:56 Quit hebz0rl (Ping timeout: 276 seconds)
20:22:55 Quit robin0800 (Remote host closed the connection)
20:28:53 Quit wolfsmond (Ping timeout: 240 seconds)
20:30:42 Quit TillW (Quit: This now concludes our broatcast day.)
20:30:44 Quit hebz0rl_ (Read error: Connection reset by peer)
20:34:15funmanif i comment 1 or 2 cpu_boost() in main.c i see correct values
20:37:04 Part efgpinto
20:37:59 Quit flydutch (Quit: /* empty */)
20:38:50 Quit linuxstb (Ping timeout: 276 seconds)
20:40:05funmanbtw last patch looks fine, played 1 full album from µSD on Clip+
20:42:00 Join mandred [0] (
20:42:26funmanmind = ultra blown. 60MHz freq is showed in the debug menu if i enable boost logging
20:42:59kugelyour mind is clearly suffering lately :)
20:43:24 Join linuxstb [0] (~linuxstb@rockbox/developer/linuxstb)
20:43:59 Quit pamaury (Quit: Page closed)
20:45:46mandredI'm confused, should I have to do anything special to get rockbox to boot on my clip+ just installed and it just boots to the OF.
20:46:14kugelfunman: codec buffer to iram patch: 1MB more audiobuffer
20:47:56funmanmandred: you should just follow the steps on SansaAMS wiki page
20:47:57kugelmandred: have you installe a bootloaer?
20:48:38mandredI followed those in the clip+ manual, thye seemed to be the same.
20:49:33mandredbootloader being the clippa.bin renamed from patched.bin?
20:49:55funmanclppa, not clIppa
20:50:31mandredAha, my mistake. Thank you.
20:51:15mandredAh actually doing the firmware upgrade now. Thanks.
20:53:03funmanif I remove set_cpu_frequency() call from init() -> correct value again
20:53:59kugelmind totally blown
20:54:27 Join saratoga [0] (~9803c6dd@gateway/web/freenode/x-bngjvwnqvufqbmxn)
20:55:48 Quit saratoga (Changing host)
20:55:48 Join saratoga [0] (~9803c6dd@rockbox/developer/saratoga)
20:56:59saratogadoes anyone know why the D2 doesn't seem to use IRAM?
20:57:31 Join christoffer [0] (
20:57:55funmanhead asplode => bed
20:57:55 Quit christoffer (Client Quit)
20:57:57 Quit funman (Quit: free(random());)
20:58:42kugelsaratoga: what do you think is the approximate maximum codec buffer needed? the standard seems to be 1MB
20:59:32 Join stripwax__ [0] (
20:59:33saratogain practice 300KB on the clipv1 seems to play almost everything
20:59:43saratogaits just AAC and Vorbis that can use more then that, and very few files do
21:00:05kugelare all codecs enabled on the clip?
21:00:19saratogavery old vorbis files can use more, but few exist, and very long AAC files due to our crummy MP4 parser
21:00:37saratogayes they're all enabled, except maybe AAC-HE (can't remember if I disabled it)
21:01:18kugelah, all mem related #ifdefs from codecs/SOURCES are removed, I didn't notice
21:01:56kugelwe have plenty iram (1MB) on as3525v2, and I'm going move the entire codec buffer into it
21:02:08saratogaSBR and PS support are disabled on targets <= 2MB of RAM
21:02:17kugelbut it might be desirable to move some cores stuff into it as well
21:02:51saratogaonce the AAC parser is fixed (cross fingers for GSOC 2010), they can be reenabled and the codec buffer shrunk to 512KB on all targetrs
21:02:57***Saving seen data "./dancer.seen"
21:02:58saratogaMP4 parser I mean
21:03:01kugelthe iram seems to be 50% faster when boosted, but the numbers come from the svn version of test_mem without your improvements
21:03:24saratogaplaying with test_mem the results were inconsistent with my changes, so I'm not really sure what was going on
21:03:40saratogait might have been a side effect of how pclk was set at the time (IIRC just 6 MHz unboosted)
21:03:57kugelhow inconsistent?
21:04:12ThomasAHfunman: (in case you'll read the log:) I'll try and report tomorrow
21:04:16saratogai sometimes got a speed up with LDM and sometimes not, which didn't make sense to me
21:04:26saratogasince LDM shouldn't really be faster on ARM9
21:04:42saratogaFWIW unrolling the load loop gave a 4% speedup
21:04:51saratogamostly just due to reduced loop overhead, gcc nonsense
21:04:58kugelI thought ldm is 2+n cycles on arm9 (and ldr always 2)?
21:05:12saratogaldr is a single cycle if its not used on the next clock IIRC
21:05:33saratogaand LDM is a single cycle per address for N>1 and N not used on the next cycle
21:05:52saratogaLDR 1 1S 1N Normal case, not loading PC.
21:05:59saratogaso 1 cycle for the normal case
21:06:13saratogaLDM n 1S+(n-1)I 1N+(n-1)S Loading n registers, n > 1, not loading the PC.
21:06:56kugelwell, the cycle numbers are really only meaningful if the ram isn't too slow aren't they?
21:07:05kugelotherwise, test_mem would be pretty useless
21:07:13saratogabut test_mem mostly just tests the cache
21:07:18saratogawhich should be 0 cycle
21:07:26kugelwell, that's intended
21:07:41saratogaso mostly what you see is how well pipelined the loads are
21:07:42 Quit dfkt (Read error: Connection reset by peer)
21:07:46kugelit shouldn't give the raw speed but what you can expect in the normal usecase
21:07:55 Join stripwax_ [0] (
21:08:02 Join dfkt [0] (dfkt@unaffiliated/dfkt)
21:08:24saratogait tests some combination of pipelining and how quickly the memory can handle cache line refills
21:08:31kugelI thought about disabling the caches but then the numbers aren't helpful anyway
21:08:43saratogaa random stride would give you memory access performance
21:09:03saratogacache lines are 32 bytes, so if you step 32 bytes at a time you'll get memory everytime
21:09:13saratoga(on almost all our arm tagets)
21:09:22 Quit stripwax (Ping timeout: 248 seconds)
21:10:20kugelI was inspecting why lcd update was so slow, and there sequential reads cound
21:10:40kugelbut I'm sure we could improve it further
21:12:01kugelsaratoga: we never had a report that a file can't be played on the clip did we?
21:13:05saratoganot that I know of
21:13:21saratogabut I think clip owners probably don't use many exotic files verses some other players
21:13:34saratogathere are certainly ones that do not play
21:13:46saratogaany AAC file over about 5-10 minutes should fail
21:13:56saratogabut i guess no one uses AAC anyway
21:14:11linuxstbsaratoga: The SoC page doesn't mention improviing the MP4 parser (but it's a good idea).
21:14:18saratogayes we can add that later
21:14:23saratogait should be quite simple I think
21:14:24 Quit merbanan (Read error: Operation timed out)
21:14:41saratogai had a hack that did it for all my files ages ago, but lear was skeptical it would work for weirder files
21:15:08saratogabasically i just downsampled the seek table and assumed that chunks were stored sequentially in the mp4 stream
21:15:27saratogait seemed to work but Mp4 is so needlessly complicated I have no idea if that was a valid assumption for all files
21:16:18saratogaand I've never managed to get a copy of the spec
21:16:28saratogai have like every mpeg document ever but not the mp4 one
21:16:36saratogabecause that is the only one the mpeg people decided to hold secret
21:16:53saratogabecause I guess making sure that no software supports MP4 correctly is in their best interest?
21:18:18CIA-5New commit by 03kugel (r25491): as3525v2: Move codec into iram freeing 1MB for the audio buffer and also a small decoding speedup (iram seems to be 50% faster than dram when boosted ...
21:20:38saratogai wonder if the IRAM speed up on amsv2 is due to actual changes in the IRAM, or just that the ARM9E has a low latency higher bandwidth tightly coupled memory interface built right into the CPU
21:21:01CIA-5New commit by 03kugel (r25492): Revert part of r25489 as it didn't fix the problem, that the CPU frequency debug screen shows the wrong frequency after boot, properly.
21:21:15ThomasAHfunman: r25490 + pastie/904095 + voice menus enabled freezes somewhere between using it for one second and one minute
21:23:25freddybRecording on Fuzev2 is much improved by shortening the delays in button-fuzev2.c
21:23:57freddybIt's gone a few minutes at 44100 now. (I also used gcc -O2, tho.)
21:24:37saratogayou used what?
21:24:48kugelfreddyb: but the delays probably need to be increased, the lcd is glitchy when boosted atm
21:25:52freddybShould the delays be based on the CPU HZ?
21:27:00kugelthey should be independant of the cpu ideally
21:27:58freddybSaratoga: Me?
21:28:32saratogayeah what did you compile with O2?
21:28:52freddybThe whole thing except the boot loader.
21:29:35freddybWell, I change GCC_OPTS in the builddir Makefile
21:32:40 Quit Eugenpaul (Quit: Eugenpaul)
21:38:03 Quit mandred (Quit: leaving)
21:41:49 Join kugel_ [0] (
21:42:05 Quit kugel (Disconnected by services)
21:42:09 Nick kugel_ is now known as kugel (
21:42:13 Quit kugel (Changing host)
21:42:13 Join kugel [0] (~kugel@rockbox/developer/kugel)
21:44:38freddybKugel: Sorry if this is a dumb question but can button reading be made interruptible by the audio?
21:44:55kugelwhat do you mean?
21:45:25freddybButtons are read in an interrupt that is higher than the audio stuff, right?
21:46:49 Quit TheSeven (Disconnected by services)
21:47:02 Join The_Seven [0] (~theseven@rockbox/developer/TheSeven)
21:47:12 Nick The_Seven is now known as TheSeven (~theseven@rockbox/developer/TheSeven)
21:47:26kugelfreddyb: no
21:47:59kugelall interrupts have the same priority; except the pcm callback which is an fiq and has higher prio. I don't know if that's used by recording
21:48:40saratogafreddyb: the individual makefiles specify the gcc level, i don't know if you should change those
21:49:08 Join JohannesSM64 [0] (
21:50:12 Join perfectdrug [0] (~marko@
21:51:35perfectdrugwodz: (logs) I finally added the SVG for the MPIO HD200 for the manual, the rest follows:
21:52:33 Join mitk [0] (
21:58:53CIA-5New commit by 03mcuelenaere (r25493): Add simple bootchart -> gnuplot shell script
22:00:44 Quit Strife89 (Quit: Going home.)
22:00:52 Join froggyman [0] (~me@unaffiliated/froggyman)
22:01:38 Quit n17ikh (Ping timeout: 245 seconds)
22:05:10 Quit perfectdrug (Quit: Leaving.)
22:06:48S_a_i_n_twith FS #11101 installed I get "/cygdrive/c/Cygwin/Rockbox_Source/patched/firmware/usb.c:60: warning: ‘reverse_usb_handling’ defined but not used" does this actually *mean* anything?
22:06:56S_a_i_n_t(the bootloader seems to work fine)
22:07:30S_a_i_n_tOh, when building a bootloader of course...*duh*
22:07:35gevaertswell of course. It means that reverse_usb_handling is defined but not used
22:07:55 Join n17ikh [0] (
22:07:57S_a_i_n_tdoes it mean anything *bad*?
22:09:12S_a_i_n_tI mean, I assume there are more than just this one option that are defined, but not implemented as default...but none of the other complain about the fact.
22:09:44 Quit stripwax__ (Read error: Connection reset by peer)
22:09:44gevaertsthis is what #ifdef is for
22:09:57S_a_i_n_tSo, its just shit code?
22:11:01S_a_i_n_tMisinterpreted you there I think...basically, *why* is it complaining?
22:11:26S_a_i_n_tIt still seems to work, but has a cry about building the bootloader.
22:13:09gevaertsIt's complaining because reverse_usb_handling is defined but not used. Seems clear enough to me...
22:13:38 Quit bmbl (Quit: Bye!)
22:16:18S_a_i_n_tgevaerts: but it *is* used, it just isn't enabled by default. As are a few other things I'd imagine, that don;t complain about it.
22:16:33gevaertsIt's *not* used by the bootloader
22:16:37 Join sierra2kilo [0] (
22:16:44gevaerts(on systems without bootloader usb)
22:17:00gevaertsWell, probably even on those
22:17:18sierra2kiloHey all. Got a quick question about the unstable firmware for the clipv2.
22:19:19sierra2kiloI've tried both the archive and current versions, and every time I get a checksum that's just slightly off. No idea why.
22:19:32 Quit yorick (Quit: Poef!)
22:19:49 Join domonoky [0] (~Domonoky@rockbox/developer/domonoky)
22:20:41 Quit mikroflops (Ping timeout: 265 seconds)
22:21:04 Join mikroflops [0] (
22:21:07gevaertsdomonoky: you know about the theme site, right? The admin bit seems to have some issues
22:21:21gevaertssierra2kilo: which checksum precisely?
22:21:52domonokygevaerts: yes, i know a bit about the themesite. what is the problem with the admin thing ?
22:22:15gevaertsdomonoky: I always seem to get all themes instead of only those for the player I want, and I can't hide a theme
22:22:32sierra2kilogevaerts: On this last install it wants 2A54079 and I got 2A5408D
22:22:42gevaertssierra2kilo: "it" being what?
22:23:38sierra2kiloWhen it's loading the firmware and it says "Checksum" and then "Sum".
22:24:33 Quit xiainx (Quit: Leaving)
22:25:10gevaertssierra2kilo: did you install the right firmware? I think this might be something you get if you e.g. install a clip firmware on a clipv2
22:25:19domonokyuh, something broke, it shouldnt show all themes. the "not able to change theme status" is known, if there are too many themes displayed, i suspect we just have too much data for the POST
22:25:53 Quit geertvdijk (Quit: ChatZilla 0.9.86 [Firefox 3.6.3/20100401080539])
22:25:55domonokyyou can hide the theme if you search for it first (so it only displays a small amout of themes)
22:27:50gevaertsThanks. That works
22:28:11sierra2kilogevaerts: I did, the clipv2 file.
22:29:29gevaertssierra2kilo: you have a clipv2 and not a clip+?
22:34:11gevaertsWhat does the bootloader print after "Model name:"?
22:40:14sierra2kiloWeird. Copied over the .rockbox folder manually and it worked. Crisis averted!
22:41:59sierra2kiloThanks for the help. Not sure why the rbutilqt method didn't work, but I'm just glad my FLAC is now lusciously gapless.
22:42:31 Join xiainx [0] (
22:42:36 Nick Ypsy is now known as YPSY (
22:45:20 Quit sierra2kilo (Quit: CGI:IRC (EOF))
22:46:57 Join Blue_Dude [0] (
22:50:58 Quit lpereira (Quit: Leaving.)
22:53:07CIA-5New commit by 03Blue_Dude (r25494): Restructure some bookmarking code, preparatory to adding version info to bookmarks. Saves some bin size as a bonus. No functional changes yet.
22:53:23stripwax_saratoga - my timings of the old rockbox mdct_arm.S versus stock tremor versus mdctexp-patched tremor was a bit bogus. I had said that old mdct_arm.S versus stock tremor had no improvement, but I'm pretty certain I had a broken build. rerunning now.
22:55:08 Join drostie [0] (
22:55:27 Quit mitk (Quit: Leaving)
22:57:46 Join b1uebrother [0] (~dom@
22:58:20 Quit b1uebrother (Changing host)
22:58:20 Join b1uebrother [0] (~dom@rockbox/developer/bluebrother)
23:00:08kugelfreddyb: maybe you can find out which delays are too long?
23:02:58***Saving seen data "./dancer.seen"
23:03:00 Quit Status ()
23:04:40 Quit S_a_i_n_t (Ping timeout: 265 seconds)
23:05:08 Join S_a_i_n_t [0] (S_a_i_n_t@
23:05:39 Quit b1uebrother (Quit: leaving)
23:05:54 Quit xiainx (Ping timeout: 276 seconds)
23:09:32 Quit Blue_Dude (Quit: ChatZilla 0.9.86 [Firefox 3.6.3/20100401080539])
23:11:21 Join xiainx [0] (
23:13:31 Quit mt (Ping timeout: 240 seconds)
23:14:36shaiHi :) When I downloaded the iPodUbuntu_Plus theme, and since I use Hebrew, all the menues are unreadable. I'm assuming that cuase of the coding. How can I fix this so I can actually use this theme?
23:15:46AlexPYou need to chose a font that has the glyphs you need and is the same size as that used by the theme
23:16:05shaiAlexP, how do I do this and check that?
23:16:11AlexPTry them out?
23:16:19AlexPThe size is in the name
23:16:24shaiWhen I apply the theme, I can't read the menues...
23:16:33AlexPBut not all fonts have all glyphs
23:17:08 Quit Chronon (Read error: Operation timed out)
23:17:37shaiAlexP, so if I can't read the menues, how can I change them?
23:17:45AlexPI think there is something about what fonts have what glyphs at">
23:18:03AlexPEnable voicing, and/or note down how many clicks it takes
23:18:36AlexPFor non-English voicing you'll need to create your own voice files though
23:18:46linuxstbOr use English, and only switch to hebrew after changing the font.
23:19:14S_a_i_n_tIsn't the only font for hebrew Unifont?
23:19:25AlexPI doubt it is the only, no
23:19:28linuxstbOr you can edit the font name manually in the iPodUbuntu_Plus.cfg file.
23:19:34linuxstb(from your PC)
23:19:46shaiWorked like a charm :) linuxstb
23:20:15 Quit liar (Remote host closed the connection)
23:20:28shaiI changed to English, then set the font 15-Adobe-Helvetica and then change the lang. back to Hebrew and I can see all the menues.
23:20:52AlexPS_a_i_n_t: As a note, please don't guess if you don't know, it only confuses things
23:21:43shaiThanks, guys... this is great now :)
23:21:54 Join dockimble [0] (~dockimble@
23:22:19 Nick fxb is now known as fxb__ (
23:26:46linuxstbAnyone know the answer to ?
23:26:56 Join panni__ [0] (
23:28:37AlexPlinuxstb: It has come up before - isn't it something to do with the current format for storing resume info would need to be rewritten to do that and nobody has?
23:29:53 Quit panni_ (Ping timeout: 264 seconds)
23:30:27linuxstbI guess the initial problem is that the resume is in nvram.bin (or real NVRAM)
23:31:09linuxstbLet's just stop people deleting files - it causes all sorts of problems....
23:31:36AlexPYou can only delete files if all resume info and playlists are wiped at the same time :)
23:31:51linuxstbSo let's only support MTP, where we can control that...
23:33:10*AlexP slaps linuxstb
23:34:36linuxstbBut just storing the filename would have other problems - e.g. if the playlist contained that file multiple times, or if the user deleted that file due to be resumed.
23:34:36 Quit evilnick_B (Quit: Page closed)
23:34:55 Nick fxb__ is now known as fxb (
23:35:29AlexPIndeed, which I think is why nobody has done this yet - it needs a little thinking about
23:36:20pixelmait's also that not storing filenames is what makes resume real quick
23:36:45AlexPI guess the playlist needs to be stored too, and if the file to be resumed has gone then it tries the next in the playlist etc
23:36:48 Quit jd (Read error: Connection reset by peer)
23:36:56linuxstbBut I think it's a valid bug - Rockbox _should_ resume to the track it was last playing, regardless of whether the fix is easy or not.
23:37:04AlexPoh yes, I agree there
23:37:05 Join jd [0] (
23:37:05 Quit jd (Changing host)
23:37:05 Join jd [0] (~jd@Wikipedia/HellDragon)
23:38:22*AlexP adds some of that to his reply
23:38:31AlexPshameless plagerism :)
23:39:10AlexPheh, cheers :)
23:41:51 Join Chronon [0] (
23:45:22 Join kwbr [0] (
23:46:50 Join webguest36 [0] (
23:47:07kugelthe google page has a separate field for the abstract, but in our template it comes after the information about me. should I cut out the abstract (with or without a reference at the point where the abstract is in our template?), or just copy it over and have it duplicated?
23:47:36 Part kwbr
23:47:36 Quit webguest36 (Client Quit)
23:47:37 Join Strife89 [0] (
23:48:00 Join kwbr [0] (
23:48:51 Quit jgarvey (Quit: Leaving)
23:48:55linuxstbkugel: I would say either duplicate it, or remove it from our template.
23:49:18kugelI'll duplicate then because we're asked to follow the template closly
23:49:25kwbrwould it be possible to synchronise the time of my rockbox device with my desktop computer?
23:49:39linuxstbkwbr: That depends on your device. What is it?
23:50:10kwbripod, video
23:51:13kwbrlinuxstb: I use
23:51:17linuxstbThen yes.
23:51:55kwbrlinuxstb: cool. How would I do that?
23:52:17linuxstbWhat OS do you use?
23:52:24kwbrlinuxstb: I am on linux
23:53:57gevaertskwbr: if you're on a semi-recent debian-like system, install libgpod-common, and you'll have the ipod-time-sync tool
23:56:19kwbrthanks. I'll try that
23:56:23Strife89Is there a comparable tool for Windows?
23:56:25amiconnsaratoga: It's not really surprising that ldm sometimes makes things faster on arm9, at least if the code in question resides in non-single-cycle, cacheable ram
23:57:46amiconnOh, also if it'S non-cacheable
23:57:46kugelshall I make it public?
23:57:49gevaertsStrife89: no idea. Maybe rbutil could be taught to to it, it already does raw scsi commands...
23:57:55kugelit = the application
23:58:29Strife89gevaerts: Definitely a nifty project to try.

Previous day | Next day