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 2007-11-05

00:00:11lostlogiclinuxstb: yeah, I can't see why a seek would be a problem
00:00:28linuxstbIs there any difference between a seek_buffer and an advance_buffer?
00:00:36lostlogiclinuxstb: I was just going to look :-D
00:00:57lostlogiclinuxstb: from the beginning of a file no
00:01:04lostlogicseek => absolute, advance => relative
00:01:23roolkupixelma: I can only assume that the second (or both) are not called on HWCODEC - the code for registering the callbacks is the same
00:01:33 Quit bertrik ("bye")
00:01:34amiconnWhy is the metadata (before the audio) loaded at all as part of the audio?
00:01:47lostlogicamiconn: because the buffering api is stupid
00:01:51amiconnSeems like a waste to me.
00:02:07amiconnThe metadata reader already determined how much metadata is there
00:02:19lostlogicI suppose that TYPE_*_AUDIO types it could do special handling on but that seems to defeat the purpose of the bufferin gapi
00:02:26lostlogic'course it already does special handling for some frew things
00:02:32amiconnThis method will cause serious problems on low-mem targets
00:02:40linuxstbThe buffering api could just take account of first_frame_offset
00:02:50amiconnThe metadata can easily be larger than the whole buffer
00:03:08*amiconn remembers this weird track with 100 copies of the same jpeg attached
00:03:11lostlogicamiconn: yeah, I gathered from reading the id3 specs
00:03:15amiconn3MB of id3v2
00:03:15roolkupixelma: can you set a rating? This should be independent of the callback (IIRC)
00:03:33 Join cheesenuggett [0] (
00:03:36cheesenuggetthey guys
00:03:42cheesenuggetti need some help pretty badly
00:03:51lostlogicamiconn: yeah, that could work fine for leading tags but any ideas for trailing tags?
00:04:01linuxstbThe WMA metadata parser also sets first_frame_offset, not to skip an id3v2 tag, but to skip the ASF header (which includes ASF metadata). The WMA codec itself doesn't need it.
00:04:19amiconnTrailing tags can't be determined beforehand, if we don't want that wasteful seek
00:04:25cheesenuggettanyone down?
00:04:35Bagdercheesenuggett: just ask
00:04:41cheesenuggettok, i just wanted to make sure
00:05:04linuxstbamiconn: Is it really that wasteful, considering that Rockbox is about to read a few MB of data from the disk?
00:05:05pixelmaroolku: doesn't seem like it (or the keymap is broken on Ondio). I see the option in the WPS context menu but what I assumed would lead me to some kind of submenu (prssing "right") doesn't do anything.
00:05:07cheesenuggettso, i have a gigabeat f series, first attempt at installing the rockbox
00:05:11amiconnlostlogic: Why would it defeat the purpose of the buffering api? Buffering a file starting at a certain position >0 sounds like a pretty generic feature to mne
00:05:28cheesenuggetti just ran the rockbox utility complete install
00:05:29lostlogicamiconn: yeah, actually it already has that feature
00:05:30cheesenuggettit seemed fine
00:05:38cheesenuggettbut now my mp3 player wont work
00:05:40lostlogiccan be told to start from an offset on any file
00:05:52lostlogicI'm just tryign to work out the implications of using it to skip metadata
00:05:54amiconnlinuxstb: The seek happens in addition to the read.
00:05:55lostlogicI think it might work just as is
00:06:08cheesenuggetti can still plug in to the USB with the power cord
00:06:09roolkupixelma: there isn't a submenu - the number in the entry should increase when you press right
00:06:15amiconnAnd trailing tags are usually small
00:06:21cheesenuggettbut other wise it says it cant read
00:06:25pixelmaroolku: it stays at 0...
00:06:26lostlogicamiconn: yeah, but they are breaking gapless atm :)
00:06:27amiconnid3v1 is just 128 bytes irrc
00:06:34safetydan cheesenuggett, what do you mean wont work? Does it start at all, is there anything on the screen?
00:06:44roolkuhm, let me have a look at the code
00:06:59amiconnlostlogic: Why is that broken? It worked fine before MoB...
00:07:13cheesenuggettit says system error and shuts off
00:07:33lostlogicamiconn: prior to MoB the buffering thread would do the fiddling with teh tail of files after buffering them and adjust the file size to trim out the trailing tag
00:07:37safetydancheesenuggett: does it say that exactly or is their other text?
00:07:43lostlogicamiconn: that code was not moved into buffering.c because it really doesn't belong there
00:07:47amiconn...and I thought that it's afunction in the playback engine that skips trailing tags
00:07:55cheesenuggettit says exactly that and then an error code in the top left of a boc
00:08:21cheesenuggett0000020 is the error
00:08:30*amiconn thinks that trailing tags are a fundamental design mistake
00:08:42amiconnBut they exist, and we have to deal with them :/
00:08:45cheesenuggettit can still connect USB to my computer
00:09:18linuxstbcheesenuggett: And the USB connection is working - i.e. you can access the disk?
00:09:48 Join Thundercloud [0] (
00:10:00pixelmaroolku: many thanks :) Currently I let music playing for a while and will see if there's any runtime listed
00:10:02lostlogicamiconn: yeah, they are a design mistake, and it was in playback.c to skip them but it was part of the buffering process and buffering.c did not inherit the functionality, nor did playback relocate it to a new functional place
00:10:44linuxstbcheesenuggett: It sounds like the Toshiba original firmware is unhappy for some reason. If I was you, I would manually install the Rockbox bootloader - there should be instructions in the manual. (It's just downloading a file and copying it to your gigabeat)
00:10:54roolkupixelma: I have a suspicion that audio_current_track() might not work as expected on HWCODEC - that would explain the rating not increasing
00:11:12cheesenuggetti have that file
00:11:24cheesenuggettso i just extract it to the root folder correct?
00:12:10linuxstbcheesenuggett: The file is FWIMG01.DAT, which goes inside the GBSYSTEM/FWIMG/ folder
00:12:29pixelmaroolku: I'm not sure if it could help you but the logscrobbler support (which also didn't work in the beginning) was fixed a while ago
00:12:31cheesenuggetti dont know that folder
00:12:34cheesenuggetti havent seen it
00:12:44roolkupixelma: you could try the following experiment - export your modifications, and with a text editor change a few rating entries (remember the tracknames)
00:12:54linuxstbcheesenuggett: It's marked as hidden/system - you need to enable viewing of those files in explorer (assuming you're using Windows).
00:13:25roolkupixelma: then import the modified modifications and see if they are reflected in the WPS and or set rating menu
00:13:26 Join webguest40 [0] (i=51684caf@gateway/web/cgi-irc/
00:13:36cheesenuggetti formatted it a while ago, did that delete it?
00:13:47linuxstbYes, that would be the cause of your problem...
00:13:57cheesenuggetthow can i go about getting it back?
00:14:15pixelmaroolku: going to experiment
00:15:45cheesenuggettif it's even possible
00:15:59linuxstbcheesenuggett: I'm not sure... But if you're installing Rockbox, you can use the dummy version of that folder available here -;
00:16:30cheesenuggettso i just extract that to the root folder
00:17:33pixelmaroolku: the database_changelog.txt contains 2 lines; the first reads "## Changelog version 1" and the second is empty...
00:17:39 Join dandin1 [0] (
00:17:53cheesenuggettstill doesnt work :(
00:17:53linuxstbcheesenuggett: Yes. And then add the Rockbox bootloader (FWIMG01.DAT)
00:18:04Buschelbye, have a good night
00:18:19 Quit Buschel ()
00:18:33cheesenuggettwhere can i find that?
00:19:06barrywardellamiconn: wrt GPO32, the comment in lcd-color-nano.c:108-109 suggests that GPO32 also has an enable register. is that accurate?
00:19:29pixelmaroolku: I played 3 or 4 complete songs after reenabling the gather runtime data feature.
00:19:34roolkupixelma: hm, it seem to only export entries that actually contain modifications - let me check
00:19:45amiconnbarrywardell: Hmm, possible
00:19:48cheesenuggettthank you very much for your help
00:20:12*amiconn wonders why it's called 'D01' there
00:20:47cheesenuggettgrrrrr same thing
00:21:09linuxstbcheesenuggett: You put FWIMG01.DAT in GBSYSTEM/FWIMG/ ?
00:22:37barrywardellamiconn: svn blames linuxstb, but I suspect it was just copied from ipodlinux
00:23:04linuxstbMost of my code is copied from somewhere...
00:23:29 Part tihoc4n
00:23:34cheesenuggetti did a google search and someone on a forum had the same problem, what he did was use the gigabeat formatting utility and it came back
00:23:57linuxstbcheesenuggett: Then you could try that. I've run out of ideas - I've never had that problem with my gigabeat.
00:24:11cheesenuggettdang it says access denied
00:24:15cheesenuggettthis is pretty awful
00:24:31linuxstbWas it working before you tried to install Rockbox?
00:25:14Calebthis is gay
00:25:19cheesenuggettis it bricked?
00:25:21Calebchose a theme and now i can see shit lol
00:25:57linuxstbcheesenuggett: If it turns on, it's not bricked... What is saying "access denied" ?
00:26:13cheesenuggettthe formatting utility was
00:26:15cheesenuggettnow it isnt
00:26:25cheesenuggettthanks again, and also for being so patient
00:26:31cheesenuggettim not quite computer savvy
00:26:46roolkupixelma: I am going to fake you an exported file...
00:27:15linuxstbcheesenuggett: no problem.
00:27:26 Quit ompaul (Client Quit)
00:27:44pixelmaCaleb: reset your settings or open config.cfg in a text editor on your computer and change one of the values for foreground or background colour
00:28:18linuxstbbarrywardell: What do you think about making the "sansapatcher -bl" installation the default? I thought it was a good idea until I realised it means you lose recovery mode....
00:28:36barrywardellhmmm. D01 enable = 0x21000200, while D01 = 0x20000000
00:29:00barrywardelllinuxstb: I like the idea. With a couple of conditions
00:29:19barrywardell1) Bootloader USB mode would be nice (when Zagor has it ready)
00:29:56barrywardell2) A bootloader logo would also be nice (the reason for 7977)
00:30:17karashatanot fun
00:30:27 Quit n17ikh|Lappy ()
00:31:09 Quit Zagor ("bed time")
00:31:24*karashata hopes everyone disregards his message, his client swapped windows on him for some dumb reason
00:32:31linuxstbbarrywardell: Does that patch require the bitmaps in bootloader/bitmaps/ or does it use the ones in apps/ ?
00:32:52barrywardellit requires them in bootloader/bitmaps
00:32:58 Quit davina (Remote closed the connection)
00:34:13roolkupixelma: there is an example file at - if you change the filename entries to match two tracks on your player you can then import it and check if the ratings are used for those two tracks
00:35:12roolkupixelma: also do you know what fixed the scrobbling ? I can only find commits relating so SWCODEC
00:35:45 Join hcs [0] (n=agashlin@rockbox/contributor/hcs)
00:36:28cheesenuggettthis is quite frustrating
00:36:58cheesenuggetthow interesting
00:37:05cheesenuggettit started the initial rockbox dialogue
00:37:15cheesenuggettbut it couldnt find the firmware
00:37:39pixelmaroolku: let me see if I can find it, I just remember that it was a patch in the tracker. I'll also try the modifications test shortly.
00:38:09 Part webguest40
00:38:39 Join n17ikh|Lappy [0] (
00:39:13 Quit donutman25 (Read error: 104 (Connection reset by peer))
00:39:43cheesenuggettit works!
00:40:07cheesenuggettlast time: linuxstb: thank you
00:40:10linuxstbcheesenuggett: I'm glad it's working.
00:40:23cheesenuggetti was afraid i just screwed it over for good
00:40:36linuxstbcheesenuggett: So you just needed to run the gigabeat formatting utility to restore it?
00:40:53cheesenuggettthat may come in handy for many a noob to come
00:41:23cheesenuggettbecause the gigabeat formatting software reinstalls all missing hidden files
00:41:38cheesenuggettwhen i formatted outside of the gigabeat one it deleted everything
00:41:45cheesenuggettis my guess*
00:41:54linuxstbYes, sounds right.
00:42:18linuxstbDid you format it before or after running Rockbox Utility?
00:42:18cheesenuggettrockbox sounds pretty terrific, im excited
00:42:26roolkupixelma: thanks - I'll have a read
00:42:37cheesenuggettbecause it installed incorrectly
00:42:40linuxstbcheesenuggett: So there's still the mystery of why Rockbox Utility broke your gigabeat...
00:42:51cheesenuggettim sure it was my fault
00:43:15linuxstbDo you click on the "safely remove hardware" icon before disconnecting your gigabeat?
00:43:27 Join JdGordon [0] (
00:43:29cheesenuggettno i didnt, that doesnt pop up for my gigabeat
00:43:49cheesenuggettbut it does for other USB items
00:44:23TTThomascheesenuggett: you should still be able to right click the drive and select 'eject' or 'remove'
00:44:41cheesenuggettthat wasnt actually an option
00:44:45cheesenuggettmaybe i didnt see it.
00:45:06TTThomaswhich os are you loading it from?
00:46:25TTThomas:-/ i dunno. it always shows up in mine.
00:46:54pixelmaroolku: what do I have to place for the genre tag if isn't tagged, <untagged> or ""?
00:46:58linuxstbbarrywardell: Your patch appears to deal with remote bitmaps in the Makefile, but show_logo() isn't displaying a logo on the remote....
00:49:22barrywardelllinuxstb: I didn't give much consideration to the remotes since I don't have a target that has one. the remote stuff in the makefile was just copied from the apps version without too much thought.
00:49:28 Quit zicho (Remote closed the connection)
00:49:35barrywardelldo we want the logo shown on the remote?
00:50:45barrywardellor maybe I should just remove that stuff from the Makefile
00:51:14roolkupixelma: don't worry about the other tags - they will be ignored anyway, just the filename entry needs to be correct
00:52:17 Part cheesenuggett
00:52:54linuxstbbarrywardell: I don't know... I don't use a remote (although I have one for my H140), so have no view.
00:54:33*linuxstb realises the H140 remote lcd is same width/height/depth as his Logik DAX player, so the Rockbox and USB logos are taken care off ;)
00:54:58pixelmasame as the iFP then... :)
00:55:25 Quit Maha` (Remote closed the connection)
00:55:29roolkupixelma: the issue that is addressed by the first patch also exists for the two callbacks used with the database
00:55:51JdGordonanyonw want to reply to my email on the dev ml?
00:56:08roolkupixelma: first patch on the FS entry you gave me
00:56:17linuxstbpixelma: Even better...
00:56:46barrywardelllinuxstb: I don't think it's worth the extra bin size, but maybe others have a different opinion. There's no harm in having the option in the Makefile though
00:57:06linuxstbpixelma: Hmm, it seems the iFP port is using the Archos bitmaps...
00:57:09 Join jumpatrain [0] (i=tabac@gateway/gpg-tor/key-0xB9002659)
00:57:21 Quit Calcipher (Read error: 104 (Connection reset by peer))
00:57:28pixelmalinuxstb: it shouldn't
00:58:31pixelmaat least I remember drawing new graphics for the Archos ones that take the unusual aspect ratio into account (rockboxlogo, usblogo)
00:58:46linuxstbpixelma: Yes, that's what I mean. Assuming the iFP doesn't have the same aspect ratio...
00:59:21linuxstbBut anyway, when I fix them for my device, the iFP will automatically follow.
01:00:15pixelmaan own usblogo is there, don't see the rockboxlogo though
01:01:49pixelmacompare usblogo.100x20x1.bmp and usblogo.104x27x1.bmp ;)
01:01:58linuxstbAh yes, I missed those...
01:02:53 Quit Thundercloud (Remote closed the connection)
01:03:09 Quit ender` (" Smoking is one of the leading causes of statistics.")
01:03:27pixelmahmm... the Archos rockboxlogo was first with standard aspect ratio, you should be able to get it back
01:03:50linuxstbI can just copy the h1x0 remote logo
01:03:56linuxstb(it's probably the same...)
01:04:35pixelmaah yes didn't think of that
01:05:34linuxstbThat's a limitation of the bitmap build system - we can't use the same bitmap for the main screen on one target, and the remote on another without copying it...
01:06:03pixelmanot the same that was used on Archos (to big) but it will fit better anyways
01:07:03*pixelma was at typo classes for too long ;)
01:08:43roolkupixelma: I don't have high hopes my last commit will change anything for you, but give it a try after your import experiment please
01:08:53 Quit JETC- (Read error: 110 (Connection timed out))
01:10:50*lostlogic tries to get the early tags off of the buffer withou tbreaking anything
01:11:10hcslostlogic: omgitbrokesomething
01:12:52pixelmaroolku: alright. The import experiment showed nothing for the two tracks I edited (if I didn't do something stupid), I'll update my build and try with your latest commit. And thanks again for looking into it :)
01:14:53***Saving seen data "./dancer.seen"
01:15:12lostlogichcs: hachu
01:16:40roolkupixelma: could you please export the changes again, to see if the import has worked?
01:18:04 Join zicho [0] (
01:18:18pixelmahmm... it didn't, the txt is empty again
01:18:50roolkunot a good sign
01:19:36amiconnlinuxstb: Iirc the iriver remote logo is not the same as the old archos one, but one properly adapted to 128 pixels screen width
01:20:05 Nick fxb is now known as fxb__ (
01:20:53pixelmaI already said err... mistyped that ;)
01:21:05roolkupixelma: and your filenames were okay you think? leading slash, directories - no umlauts ;)
01:21:39lostlogicI need people who have mp3s and shorten files (and flacs) that they would recognize any breakage early in the song with to test this:
01:22:24pixelmaroolku: just wanted to ask you about the umlauts.... problem is: I have one album on the internal memory of the Ondio (album name has umlauts), wanted to take this because I was unsure about the MMC paths
01:24:55 Join Angryman [0] (i=50e0a118@gateway/web/cgi-irc/
01:24:59roolkupixelma: you would need to save the file in UTF-8
01:25:59amiconnFilenames are UCS-2 on VFAT - always
01:25:59Angrymanshort advice needed on how to get the adc_scan properly working in the plugin API... Anyone ready here to helpme a bit ,-)
01:26:25roolkuamiconn: no, we are talking about the txt import file for the database
01:26:43pixelmaroolku: I'll try again
01:27:02amiconnMaybe there's a bom problem then?
01:27:34roolkupossibly, can you save utf-8 without bom?
01:27:49*roolku checks for bom skipping
01:28:07amiconnNotepad++ can
01:28:28pixelmait wasn't utf-8 before because I didn't pay attention and copy and pasted it from the pastebin (I think firefox did not detect it as unicode...)
01:30:43roolkuthe original text was ascii only, so it didn't matter
01:31:33roolkubut the file reader doesn't skip the BOM, so better don't have one.
01:31:46Angrymanpixelma or amiconn /// I wont disturb you furthermore... If you have time (and the will) my problem is posted in the forums... Any advice would be appreciated... N8 all :-)
01:33:07pixelmaok, import/export worked now but the rating is still at zero and can't set it (that's with the old build). I'll update now but that'll take a while... :|
01:33:43 Quit Angryman ("CGI:IRC")
01:33:45lostlogicnvm the patch I linked, it doesn't work as well as I thought.
01:33:50 Quit lee-qid (Read error: 104 (Connection reset by peer))
01:34:37linuxstblostlogic: Is that patch also not buffering the tags?
01:35:42JdGordonpixelma: is the bitrate needed in the rec screen statusbar?
01:36:10lostlogiclostlogic: that's all it does is not buffer them and change mpa and shorten to seek, but it seems like mpa at least seeks earlier than that point for some reason
01:36:12roolkupixelma: did you try a database query as well? It seems odd that the imported ratings are not shown in the menu (WPS I could understand)
01:36:21 Join Angryman [0] (i=50e0a118@gateway/web/cgi-irc/
01:36:22lostlogicwhich results in having to get more of the file from disk
01:36:34 Join vitaminx [0] (
01:36:40roolkupixelma: in the exported file all the artist names etc should be filled in correctly
01:36:54 Join Thundercloud [0] (
01:37:13pixelmaroolku: hmmno, I didn't. That's the first thing I'll look at when extracting the new build is done
01:37:49vitaminxsry, anybody knows if there is *some* linux support for the kenwood hd20ga7 mp3-player anywhere?
01:38:09roolkuI need to sleep now - I'll read the log in the morning
01:38:41 Quit Angryman (Client Quit)
01:38:47 Quit roolku ()
01:39:41pixelmaJdGordon: is it currently? And I think if it is settable it should be on the screen somewhere - I only know that on the Archos you can set the average quality (the MAS records to vbr files) and it is there. But I'm not a big taper or so, you should probably ask petur
01:40:35linuxstbvitaminx: No. But I think the hd20ga7 is very similar to the Gigabeat F, so you might be able to port Rockbox to it relatively easily...
01:40:58JdGordonpixelma: ok, i thought you were a taper...
01:41:16pixelmaah, and on Archos the quality is shown in the statusbar.
01:41:37JdGordonthe freq and bitrate are both shown and both are going to be annoying :p
01:41:58linuxstbJdGordon: I remember that being discussed in the past - recorders are very keen on packing as much info as possible into the recording screen, and will probably shout loudly if you remove features...
01:42:23JdGordonyeah :'(
01:42:28pixelmaroolku: (for the logs) the rating is now shown in the WPS context menu and I can set it! :D
01:43:03linuxstbJdGordon: Is the same function used to display the status bar in the recording screen as in the rest of Rockbox?
01:43:29JdGordonI might split it though so the rec SB doesnt change
01:43:58*JdGordon not sure how best to do this :p... and has exams he best study for
01:45:34pixelmaJdGordon: the so-called "REP" comes with a much more full packed screen. I only saw a screenshot once and I can only remember thinking I would likely be lost there...
01:46:39amiconnJdGordon: Why do you want to split the statusbar drawing?
01:48:34pixelmaif I read his comments correctly it has to do with his customising statusbar proposal
01:48:54pixelma(in the dev mailing list)
01:50:56*amiconn is afraid of half baked solutions and bloatz
01:54:00pixelmaroolku: and it gathers runtime data too! :)
01:54:16 Quit courtc (Read error: 113 (No route to host))
01:58:50JdGordonamiconn: dont worry... not split indefinatly.. just untill i ahve a better idea how to do the rec part
01:58:59JdGordonit would al be together for commit
02:00:08*amiconn thinks that fixing the rec screen should be more important than a customisable status bar
02:00:48JdGordonyeah.. probably.. but /me never uses that screen so meh :p
02:00:54pixelmaroolku: the two tracks that were edited show different numbers as if it worked with the trick before too (looking at "most played") but now it works for the rest too
02:01:28kugelHey, now that you've changed the e200 keymap, isn't it a good point to change the keymap of the text editor/keyboard screen to?
02:01:40kugelLike in my patch
02:03:03 Quit cooz ()
02:03:45pixelmathat patch doesn't look right (the changes to the plugin itself I mean). But I have no e200 and can't comment on the usefullness of the different button assignments.
02:04:20 Quit ptw419 (Read error: 110 (Connection timed out))
02:04:29*amiconn did fix a lot of things he doesn't use in the past...
02:04:41 Join tkooda [0] (
02:05:01kugelyea, you are right. I mainly care about the keyboard screen changes
02:06:21kugelI can't say if the current state is more consistent with other targets, but my keymap for the keyboard screen is more natural and intuitive IMHO
02:07:20tkoodacan someone please recomend what used digitial audio player I should purchase? -I'll only be using it in my car for ~2-6 hour trips. ~4GB shoud suffice, car/usb power, mp3/aac/vorbis, nothing fancy. no nice interfae required. usb(2?) to load the audio.
02:08:01tkooda(for use with rockbox)
02:09:42tkoodanobody has a recomendation of what I should get to run rockbox on?
02:09:59kugelI'd say sansa
02:10:08tkoodawhy sansa?
02:10:24tkooda..what whast series?
02:10:34kugelEvery rockbox target will do the job, but sansa is one of the cheapest and it has a nice scroll wheel
02:10:58 Quit Thundercloud (Remote closed the connection)
02:11:24 Join emeraldd [0] (
02:11:28linuxstbkugel: Not quite all of them - the Archos devices won't do AAC or Vorbis...
02:11:37pixelmaI can only point to the
02:11:53tkoodacan the sansa e200 handle aac/vorbis/flac?
02:11:59tkoodathnx pix
02:12:53kugelAAC plays fine
02:13:04TTThomasdoes flac murder battery life on flash based storage? or just hdd based?
02:13:06pixelmait's missing the c200 but that has only a maximum of 2GB internal memory and the microSD support isn't ready yet
02:13:08kugelI don't use vorbis or flac
02:13:22tkoodawhat are the basic limitations that some players can do over others? (as far as the rockbox software is concerned? -some can't do flac/aac?)
02:13:38kugelbut I recall a table which stated how well codecs on which targeds do
02:13:40tkoodacan any sansa do microsd?
02:14:07linuxstbTTThomas: I wouldn't say FLAC murders battery life. It's very efficient CPU-wise, so that helps compensate for the more frequent disk accesses.
02:14:15tkoodaso, if I get a small sansa (~1-2GB), I'd be able to add a ~8GB(?) microSD later?
02:15:12TTThomaslinuxstb: i had a couple flac albums on my gigabeat. didnt have good luck with the battery life
02:15:19linuxstbI would also suggest the Gigabeat F20 - it's also cheap, and has a fast CPU, so can play some very CPU intensive codecs (e.g. HE-AAC, APE) that other targets can't handle.
02:15:22pixelmatkooda: with the e200 you could already do that, it will hopefully come later on the c200
02:15:37linuxstbTTThomas: I only use FLAC on my gigabeat, and have never run out of power...
02:15:42kugeltkooda: yea, it can eat up to 32BiB
02:15:57TTThomaswhich model?
02:16:12linuxstbAn F20 upgraded with an 80GB disk
02:16:17tkoodapixelma, aah.. thanks.. -so anone got any reason not to get a sansa e200, and a 4-8GB microSD?
02:16:30tkooda..for playing mp3/aac/flac/vorbis in a car
02:16:43tkooda(don't care about the interface or size)
02:16:46tkooda(just cost)
02:16:49TTThomasmine is an f40. any idea if they are very different?
02:16:51kugelisn't a 8GiB Sansa cheaper than 2GiB Sansa + 4-8GiB microSD?
02:17:11 Quit spiorf (Remote closed the connection)
02:17:16 Join advcomp2019_ [0] (n=advcomp2@unaffiliated/advcomp2019)
02:17:24 Quit advcomp2019 (Nick collision from services.)
02:17:32 Nick advcomp2019_ is now known as advcomp2019 (n=advcomp2@unaffiliated/advcomp2019)
02:17:46linuxstbtkooda: I always think you can never have enough disk space, so for in-car use, I would definitely go for a hard-disk device (20GB or more), even if you only have small requirements now...
02:18:28amiconnWhere did the MajorChanges link go? :>
02:19:15tkoodalinuxstb, makes sense. -can you recomend a HD player?
02:19:22midgeyit recently disappeared from the front page...
02:19:49linuxstbtkooda: I already have - the Gigabeat F.
02:20:19tkoodaa quick search for ~4GB microSD's found one for ~$35 (and I thought I saw an 8GB for not too much recently)
02:20:37linuxstbtkooda: You can only buy them used, but an F40 (40GB) can probably be found for around $100
02:21:07 Part pixelma
02:22:45linuxstbtkooda: Looking at at the moment, there are lots of F40 auctions ending in the next couple of hours (first one in 8 minutes...)
02:23:53tkoodais the 20GB model still called "F40"?
02:24:32linuxstbNo, 20GB is F20
02:25:55tkoodaare there any codec/other limitations of the F40?
02:26:17tkoodaany worry about skipping if in a car?
02:26:24lostlogicgoing to the recording screen breaks playback on my ipv, known?
02:26:30TTThomasi've never had it skip
02:27:12linuxstbtkooda: No. The Gigabeat F has the fastest CPU of all current Rockbox targets, so works very well.
02:27:28tkoodaaah, good to know
02:27:39linuxstblostlogic: Yes.
02:27:50tkoodathink I'll find an F20 for much cheaper?
02:28:13krazykittkooda, no, probably not
02:28:18lostlogicat least it's not a new bug
02:28:29lostlogicwtf why is some client setting the filechunk to 262k!?
02:28:30TTThomastkooda: probably not, od had the f40 at an outrageous price a couple monthes ago, so they are all over the second hand market
02:28:34lostlogicwhy is that even configurable?
02:28:46tkoodaTTThomas, "od"?
02:28:50TTThomasoffice depot
02:29:19 Quit vitaminx ()
02:30:06 Quit jhulst ("Konversation terminated!")
02:30:08lostlogicok, new question: why do all codecs set the filebuf buffering chunk size to random things?
02:30:13lostlogic(not random I'm sure)
02:30:38tkoodaTTThomas, nabbed that F40 for $68
02:30:42lostlogicbut what's the reason behind this setting? it only applies to how much data the buffer reads from disk at a time, ntohing to do with how much the codec can read at a time.
02:30:51linuxstbtkooda: That was quick...
02:30:56linuxstb(and a bargain)
02:30:56tkoodathanks for the advice guys
02:30:59TTThomashope your happy with it
02:31:03TTThomasi am
02:31:29TTThomasadded bonus. fastest cpu = good performance on videos
02:33:18 Join saratoga [0] (i=9803c6dd@gateway/web/cgi-irc/
02:33:28TTThomasanyway back to tv
02:35:00tkoodawhat's that max video resolution/codec/bitrate the gigabeat can play (i.e. if I were to get the RCA AV cables for it?)
02:35:53saratogaregarding flashing the sansa bootloader, another problem is that it currently ignores the hold key
02:36:07saratogaso if you hit power with hold on, it will still bootup
02:36:08 Join martin__ [0] (
02:39:28 Quit zicho (Read error: 110 (Connection timed out))
02:41:24tkoodaanyone else here got a F40?
02:43:48tkoodahow much cpu/ram is in the F40?
02:43:55tkoodapardon the pm thomas. :P
02:43:59TTThomas i'm not sure. i know the unit it self is limited to 320x240. I dont know if RCA output is even functioning in rockbox.
02:44:11TTThomasits alright, i'm not registered so i couldnt respong
02:44:12tkoodaaah. I'll hold off on 'em then
02:44:36saratogaCPU is an ARM9 core at about 300MHz, theres 32MB of ram
02:45:51TTThomasmy mp3 player has a faster proc than my parents computer
02:46:22lostlogicah _hah_ found the breakage, I think that not buffering leading tags will work now... I think.
02:48:49TTThomastkooda: when the player comes. I recommend making a full backup of the file system on it. having one saved my butt during a couple of botched upgrades
02:50:02tkoodaaah.. thanks for the tip. will be sure to do.
02:50:05TTThomaseverytime i have a working build, i back it up. i locked mine up twice with my last upgrade
02:50:56tkoodawent from joining the chan, to having a $68 F40 shipping in less than 15min, heh
02:50:59 Join jhulst [0] (n=jhulst@unaffiliated/jhulst)
02:51:10tkoodafigured you guys'd know what's decent for rockbox
02:51:19tkoodabeen meaning to for ages
02:52:14tkoodais car charger any kinda standard/common format that I could use some other one I might have, or is it some funky/proprietary plug or something?
02:52:39tkoodai.e. "it's interchangable with nokia charger plugs", etc?
02:52:51TTThomasit can be charged from miniusb or direct power plug
02:53:00TTThomasor from the dock that comes with it
02:53:13tkoodaso a car->miniusb would work?
02:53:23krazykitshould, yes.
02:53:25TTThomasfairly certain
02:53:50TTThomasprovided you have charging from usb turned on in the rockbox settings
02:55:39tkoodaall the F40 car chargers I can find look like some crappy "kit" (generic w/the end for F40).. and for $20. -I think I'd rather get a generic car->usb->miniusb
02:56:08tkoodais usb and miniusb same (5.5? or is mini 3.x?) voltage?
02:56:54lostlogicAnyone who has lots of codecs with lots of tags, please test the commit I just committed. My confidence is high, but ya never know what I might have broken.
02:58:55jhMikeShow do you set things up in the sim to know where a segfault occurred? (if that's in the wiki, I'm glad to just read that)
02:58:59tkooda <−− think this "Motorola Mini USB Car Charger" for only $2.78 will work for my F40?
03:01:43 Quit martin__ (Remote closed the connection)
03:09:08safetydanjhMikeS: are you talking about compiling with symbols enabled? or just how to use gdb?
03:09:43jhMikeSsafetydan: to get the traces. I thought symbols were already in it.
03:10:04 Quit cynic ("Time to mosey.")
03:12:08safetydanjhMikeS: yeah it does compile with -g by default. To get the traces you do something like "gdb ./rockboxui" and then type "run" and do whatever you need to do to get it to crash
03:12:16safetydanthe debugger should stop at the point of the segfault
03:12:25safetydantyping "bt" will give you a backtrace
03:12:27lostlogicrebuffer_handle is quite severely broken for the backwards case.
03:12:29lostlogicquite severely
03:12:42jhMikeSsafetydan: ah. thanks. shoudn't take too long for do the crash thing.
03:12:47safetydanyou can also use "p <variable name>" to print the value of a variable
03:13:07midgeylostlogic: was your last commit an attempt to fix fs 8069?
03:14:56***Saving seen data "./dancer.seen"
03:17:48 Quit kugel ("ChatZilla [Firefox]")
03:18:01*JdGordon wants to close 8090..
03:18:48lostlogicmidgey: no.
03:18:59lostlogicmidgey: that will require some more work at the end of the files.
03:19:02tkoodaanyone ever try a car->miniusb charger that was origionally designed for a cellphone with their F40? (they might have an IC that prevents overcharging the batt, but will it still work with the F40?)
03:19:58midgeylostlogic: good, because i didnt want to disappoint and tell you it still occurs :)
03:20:23midgeylostlogic: mp3, nsf, spc, and ogg seem to load metadata fine here
03:20:42lostlogicmidgey: thanks for the update :)
03:20:43midgeyi tried adx, but i dont think i have those tagged (they still played)
03:21:01lostlogicstill played and didn't have wierd gaps or weird buffer behavior is what I was hoping fo rwith that commit ;)
03:21:56midgeyi didnt notice anything, but i didnt try and stress it too hard
03:22:30lostlogicMy brain is still spinning on how to fix gapless, but nothing's popped up yet
03:22:36lostlogicso I'm fixing other bugs in the meantime
03:22:47midgeyfair enough
03:28:23jhMikeSheh...that got me right to the problem :)
03:29:34jhMikeSat least mpegplayer isn't using 100% CPU now during playback
03:34:10*jhMikeS notices the sim audio is really loud now :\ ??
03:37:56 Quit joshin (Read error: 110 (Connection timed out))
03:41:27 Join criznach [0] (
03:43:42 Quit jumpatrain (Remote closed the connection)
03:44:24jhMikeShey, I've got 5-threads going here and two streams and code that is starting to dwarf core playback (in lines, but still rigorously object oriented). It can't be that bad there. ;)
03:45:20lostlogicnico might get a beating (or at least a teasing) for this one.
03:45:26 Join jumpatrain [0] (i=tabac@gateway/gpg-tor/key-0xB9002659)
03:46:02 Quit saratoga ("CGI:IRC (EOF)")
03:46:10lostlogicrebuffer_handle virtually never could work for a backward seek. in most cases such a use will cut off the end of the playing track
03:46:29*jhMikeS will volunteer :p. /me needs some frustration outlets from RealLife (TM).
03:48:39jhMikeSwhy would it not? doesn't playback just require that? -buffer windows *cough*-
03:49:16lostlogicwell rebuffer_handle is only called with a codec tries to seek off of the buffered data
03:50:09lostlogicso in normal use you'll never run into this bug... but when resuming a song (which sometimes makes a codec seek a bit) it can mean that the tail of the song is trimmed by as much as a full frame
03:50:14jhMikeSso if the left edge of buffered data is after the left edge of the seek, it's not (all) there
03:50:33 Quit animeloe ("This computer has gone to sleep")
03:52:25 Join animeloe [0] (n=animeloe@unaffiliated/animeloe)
03:55:36 Join PoisonEye [0] (
03:59:20 Part PoisonEye
03:59:58lostlogicI had to give up on fixing that for now, it got too ugly
04:00:45safetydanInteresting little blog post on atomic locks for armv5 and lower arches
04:04:47jhMikeSsafetydan: we already have that but a better version that allows locking on one variable as both the lock and value
04:05:34jhMikeSit only needs to give up one of it's values as the busy value
04:07:19 Join eigma [0] (n=cat@
04:11:15 Join gb40 [0] (i=63f388f0@gateway/web/cgi-irc/
04:11:16jhMikeSan earier DC version had those routines, but they didn't make it in since they weren't so useful. I guess semaphores implement atomic increment though.
04:11:19 Join evanstapler [0] (
04:11:26 Quit gb40 (Client Quit)
04:11:55 Join gbf40 [0] (i=63f388f0@gateway/web/cgi-irc/
04:12:31evanstaplerDoes anybody mind helping me with a problem I have...?
04:13:16safetydanjhMikeS: well it was an interesting article for me as it gave a bit of an overview of different ways to do it.
04:13:26safetydanevanstapler: it's easier if you just state what your problem is
04:13:58evanstaplerOkay. When I run freedoom, the backlight shuts off on my 5G iPod 30GB and will not turn on until I restart the iPod.
04:14:32jhMikeSsafetydan: see kernel.c for a more efficient way (on PP where swp(b) isn't broken) in the semaphore_* functions.
04:14:38gbf40heyy, can anybody help me out with a bricked Gigabeat F40?
04:14:43evanstaplerI've searched the foruma nd didn't find any info, so I assumed maybe it was an isolated case...
04:15:07TTThomasgbf40: have you tried the dissasmbly guide yet?
04:15:58 Join webguest78 [0] (i=4a469ecd@gateway/web/cgi-irc/
04:16:20safetydanevanstapler: sounds like a bug. You should report it in the tracker (flyspray)
04:16:56gbf40disassembly guide?
04:17:18TTThomasyeah. there is a guide for getting back into the file system of a bricked f40
04:17:27evanstaplerSafetydan: Alright, thank you for the help!
04:17:44TTThomassaved my butt the other day. hold on.
04:19:14evanstaplerSafetydan: How would I go about reporting the bug, I don't see the option on the bugs page.
04:20:08gbf40Very interesting
04:20:17gbf40The error code I have on my 'beat is an unusual one
04:20:49gbf40System Error 00020003.. unit tries to load, hard drive spins up and down a few times, then the hard drive stops harshly and the error message loads up..
04:21:09gbf40Lets see if taking the poor thing apart helps any.. haha!
04:21:29 Join davidfg4 [0] (n=david@
04:21:31TTThomashave you changed your build lately?
04:21:55webguest78i tried to use someones theme on rockbox and the menu lettering got all messed up. For various reasons, i need to know which one means copy and which one means paste. I connect connect my sansa to the computer so i have to replace the rockbox folder over the old one using the file manger in rockbox. I am using a micro sd card which hold the other rockbox folder. This is the order it goes in under the menu
04:22:46eigmajhMikeS: I have a feeling I'm going to ask you about what you were just talking about: do you have any references/keywords/leads on learning about inter-processor synchronization? (for between the m:robe's ARM and its DSP)
04:24:22lostlogic <−− and I'm bitter because it's a pain and I can't fix it tonight because I have other stuff to do.
04:26:25safetydanevanstapler: have you signed up for an account?
04:28:05krazykitwebguest78, check the manual for how to reset the settings when you turn it on
04:28:43evanstapler safetydan: Assuming it's not the same as a forum account, no.
04:28:52evanstaplerSafety dan: I guess I can start there...
04:30:23safetydanevanstapler: you'll need to sign up for a flyspray account which is unfortunately not the same as the forum
04:30:42evanstaplersafetydan: Gotcha, confirming the account now :)
04:31:02jhMikeSeigma: I think that's different in a big way for that. Sort of an RPC type deal.
04:31:07webguest78krazykit: i'm looking now but can't find which link it is under
04:31:10zesafetydan: is it justified that your nick always reminds me of the comic "red meat"?
04:31:41krazykitwebguest78, ...or you could download the PDF and use the search feature in your PDF reader.
04:32:12davidfg4I am having trouble with tools/
04:32:15davidfg4It says "ROCKBOXDEV: patch is required for this script to work. Please install and re-run the script."
04:32:18eigmajhMikeS: I meant in terms of synchronization primitives −− I've only heard of the mailbox paradigm so far (it's well described in one of Cadenux's BSPs) but I was hoping you had some more references about this stuff.
04:32:27davidfg4I can't find this patch
04:33:28davidfg4Where is it?
04:33:28 Quit gbf40 ("CGI:IRC (EOF)")
04:33:32 Quit jhulst ("Konversation terminated!")
04:34:00jhMikeSeigma: no papers. Those need to communicate in different address spaces on two separate OSs. I'd love to make libmpeg2 look the same to the threads but be a hardware decoder underneath...seems doable.
04:34:19 Join cynicz [0] (
04:34:23 Part cynicz ("Leaving")
04:34:39eigmajhMikeS: ah, sorry, I didn't mean anything related to libmpeg2 or any particular codec.. just in general
04:35:41jhMikeSstill no reference material. I suppose I'd just learn how the DSP processor works and figure it out from there. The stuff for shared memory doesn't apply here iirc.
04:35:58eigmaokay, thanks
04:41:13davidfg4Anyone else have this same problem?
04:41:14jhMikeSI think the status is just signaled by the DSP or something after sending commands
04:41:35krazykitdavidfg4, no, it should download all the patches it needs by itself
04:42:46 Quit evanstapler ()
04:42:57 Quit criznach ("ChatZilla [Firefox]")
04:44:54davidfg4I did a fresh cvs checkout on my Ubuntu 7.10 computer and got the error before I even got any prompts.
04:45:49 Quit webguest78 ("CGI:IRC")
04:47:55 Join courtc [0] (
04:48:07 Join Mouser_X [0] (i=cf9bb003@gateway/web/cgi-irc/
04:49:47Mouser_Xlostlogic: I read in the logs that midgey tested your patch/commit (I forget which), and said something about ADX. ADX files don't support tags.
04:49:55Mouser_X(I just thought you'd like to know.)
04:50:06 Quit Bjoern-Erik ("Reconnecting")
04:50:16midgeyi had a nagging suspicion about that
04:50:18 Join Bjoern-Erik [0] (
04:51:50 Quit dandin1 ()
04:51:52Mouser_Xmidgey: I really wish they did, but it's a limitation of their original format.
04:53:05Mouser_XADX files are (usually) ripped from a game ISO. In making in_cube (which Rockbox's ADX support comes from), HCS wanted to keep the files in their original forms. Thus, no tag/metadata support.
04:54:15midgeyright, all my adx files come directly from my dreamcast games
04:54:35*Mouser_X has an ADX encoder on his PC.
04:54:56Mouser_XI've been strongly considering transcoding some other formats into ADX for use in Rockbox.
04:54:58safetydanze: I've no idea.
04:55:50 Join TiMiD[FD] [0] (n=TiMiD[FD@
04:55:56 Join RoC_MasterMind [0] (
04:58:07TiMiD[FD]recently I started to get illegal instruction errors with the newest builds
04:58:12TiMiD[FD]is that a known bug ?
04:58:22 Quit karashata (Read error: 110 (Connection timed out))
04:58:37TiMiD[FD]maybe a callback pointer that gets uninitialized
04:58:45 Quit homielowe (Connection reset by peer)
04:58:46TiMiD[FD]in the new buffering engine
05:00:01Mouser_XWhat target are you using? iPod, Gigabeat, etc?
05:00:25Mouser_XAlso, is it the *very* newest build? That can make a difference.
05:01:56lostlogicMouser_X: thanks
05:04:50Mouser_Xlostlogic: Glad to help where possible (though, that may be limited in some areas...)
05:05:10 Quit advcomp2019 (Nick collision from services.)
05:05:38lostlogicMouser_X: good, go fix 8092 fo rme, because I'm bitter about it.
05:05:59Mouser_XThis would be one of the areas that I'm limited in...
05:06:19 Join advcomp2019 [0] (n=advcomp2@unaffiliated/advcomp2019)
05:06:33 Quit miepchen^schlaf (Read error: 110 (Connection timed out))
05:07:00 Join miepchen^schlaf [0] (
05:07:15 Join joshin [0] (n=joshin@VDSL-130-13-11-116.PHNX.QWEST.NET)
05:07:47Mouser_XAlso, I have an odd (but unimportant) bug to mention: In Rockbox, charging while plugged into USB is disabled. However, on all the computers I've plugged my Gigabeat into (through USB of course), it's charged itself. Strangely, the only exception to this is my PC, which is currently offline and in need of a new power supply.
05:09:30 Join Jon-Kha [0] (
05:13:02TiMiD[FD]Mouser_X: I'm using a h140 with yesterday's build
05:13:23TiMiD[FD]but the last week's one (after the buffering commits) did that as well
05:13:23 Quit z35 ("Leaving")
05:13:25 Join homielowe [0] (
05:13:35 Join z35 [0] (
05:14:00Mouser_XTiMiD[FD]: Try the newest builds.
05:14:10TiMiD[FD]I'm at work I can't :)
05:14:18TiMiD[FD]besides it's not easily reproductible
05:14:23Mouser_XSorry, I misread. However, trying a newest build would still be a good idea.
05:14:33 Part tkooda ("thanks guys!")
05:14:46TiMiD[FD]it only happens sometimes when doing a lot of skipping
05:14:58***Saving seen data "./dancer.seen"
05:15:10Mouser_XWhat's your Rockbox version?
05:15:26TiMiD[FD]hmm I don't really know
05:15:31 Join bb [0] (n=bb@unaffiliated/bb)
05:15:33Mouser_Xlostlogic did a lot of work on the buffer stuff yesterday. Perhaps you got stuck somewhere on there.
05:15:45TiMiD[FD]I compiled the svn yesterday around 9am
05:16:06TiMiD[FD]which means something like sunday 0am GMT
05:16:19TiMiD[FD]yep I saw the svn logs
05:16:29Mouser_XRockbox will tell you the version number if you check system>rockbox info (from the main menu)
05:16:38TiMiD[FD]I know I'm a dev ...
05:16:50TiMiD[FD]but I'm at work
05:16:56*Mouser_X only shows up here when he's bored.
05:17:06TiMiD[FD]just wanted to know if that happened to others
05:17:11lostlogicTiMiD[FD]: what's the bug you're having?
05:17:11TiMiD[FD]same here :)
05:17:38TiMiD[FD]freeze with illegal instruction
05:18:09lostlogicooh! I haven't seen an illegal instruction in quite a while :-P
05:18:23TiMiD[FD]last time I saw one it was my fault :)
05:18:38TiMiD[FD]unintialized callback pointer ...
05:19:14TiMiD[FD]maybe it's already fixed in svn I don't know
05:19:23TiMiD[FD]I'll try to update today
05:19:37TiMiD[FD]when I'll go back home
05:20:07TiMiD[FD]lostlogic: are you one of the buffering guys ?
05:20:40lostlogicTiMiD[FD]: Nico_P wrote it, I'm the old timer who was beating playback into shape in 2006
05:20:55lostlogicI just came back to bug-fight because it's fun.
05:21:01lostlogic(even if I'm bitter about a particular bug right now)
05:21:24Mouser_X8092 if you want to check it out TiMiD[FD].
05:21:57TiMiD[FD]sounds fun to fix bugs ...
05:21:57 Join Llorean [0] (n=llorean@
05:22:04TiMiD[FD]I'll make some for you
05:22:19TiMiD[FD]Mouser_X: I check the tracker, will get me smth to read !
05:22:33 Quit eigma ()
05:24:51TiMiD[FD]if I remember, it only happned in cases when I went previous by some tracks and then went t next tracks
05:25:12TiMiD[FD]that buffering code is getting pretty complicated
05:25:23TiMiD[FD]looks like a malloc
05:25:35lostlogichehe, in some ways it is one
05:25:50lostlogicat leaset it's not sitting around adding complexity to playback.c any more
05:25:59TiMiD[FD]I don't know if it's possible to have a track broken out in small chunks
05:26:19TiMiD[FD]the most important is to have a clean API
05:26:24lostlogicTiMiD[FD]: not really −− codecs have widely varying needs for contiguous reeds
05:26:29TiMiD[FD]then what goes behind works or not ...
05:26:35lostlogicwe use the 'guardbuffer' to guaratee at least 32k contiguous read
05:27:05 Quit bb_ (Read error: 110 (Connection timed out))
05:27:13TiMiD[FD]that's why it's so complex
05:28:06TiMiD[FD]with tracks broken up, it would maybe be easier (but I'm not really accurate in that area)
05:28:06lostlogicand relates to why 8092 is so goddamned frustrating
05:29:11TiMiD[FD]because when you go backwards the buffer needs to remain contiguous
05:29:45lostlogicyeah −− the old code just used a complete rebuffer in that situ, but even when I tried to do that I somehow failed, which is when I got bitter and threw the code at flyspray
05:30:07TiMiD[FD]maybe a "packet based" approach would be better
05:30:17TiMiD[FD]tracks would be splitted in small packets
05:31:01TiMiD[FD]and codecs would have to require next packet after they finish reading the information of one packet
05:31:03lostlogictoo hard to give codecs good memory without constant memcpy
05:31:24TiMiD[FD]codecs could read directly from the bufer data
05:31:35TiMiD[FD]looks somehow like a software mmu
05:31:38lostlogicno, because they need each frame to be contiguous
05:32:03lostlogicso if the data is packeted they would need to copy the data they need to a temporary buffer each time they hit a packet boudary
05:32:04emeralddyou'd need a paging system
05:32:26lostlogicand this is why buffering is a 'plain' ring system with no fragmentation allowed ;)
05:32:35TiMiD[FD]ok good to know
05:32:56TiMiD[FD]is there a priority handling in the buffer ?
05:32:59lostlogicwith the plain ring, this only happens every 30m of data
05:33:04lostlogicTiMiD[FD]: nope, just a ring
05:33:11TiMiD[FD]like "this track is more important than this one" ?
05:33:23TiMiD[FD]hmm nevermind
05:34:03TiMiD[FD]would a code that "steals" the track before the data currenlty being read be difficult to implement ?
05:34:40lostlogicTiMiD[FD]: unfortunately yeah −− because before it could do that (in the current world) it would hafta tell the playback system that it was doing so
05:34:53lostlogichmm... or maybe it wouldn't... hmmmmmm
05:34:58lostlogicnow that I'm thinking about it...
05:35:18lostlogicgah, but it would need to handle cases wher eit needs to fill up more than the immediately prior track's data space for the seek
05:35:27lostlogicso I think it's simpler just to force a full rebuffer... hmmmmm
05:35:38TiMiD[FD]so the playback system manages a list of the buffered tracks too ?
05:35:44 Quit RoC_MasterMind ("Leaving")
05:36:13TiMiD[FD]just my thought on this frustrating situation :)
05:36:26lostlogicyeah, I appreciate the thought
05:36:38TiMiD[FD]I'm a java programmer so I'm not very used to that kind of situations
05:36:43lostlogicbut yeah, there is some duplication of logic between the two both managing track lists
05:37:01emeralddJava has it's own kinds of memory issues :)
05:37:03lostlogicbut basically the playback system knows about an audio track, which generally includes between 2 and 3 buffer handles
05:37:16*lostlogic is a java programmer for his day job too
05:37:30lostlogicas we add things like album art, a track will know about more handles
05:37:35TiMiD[FD]hehe same here :)
05:37:40lostlogicthe buffer doesn't know about tracks, it only k nows about types of handles
05:37:50lostlogicit's an acceptable separation of logic, IMO
05:38:08lostlogicbut I do wish that there was a way for teh buffering system to "demand" the freeing of a specific track
05:38:29TiMiD[FD]that would add complexity
05:38:48emeralddYou could add a lock to the handles,
05:38:50TiMiD[FD]i'm sure there is a simple solution
05:39:02emeralddthat would let you see which one is currently being used at that moment
05:39:33TiMiD[FD]what happens when a codec reaches the end of the buffer ?
05:39:48TiMiD[FD]I mean can this situation happen ?
05:39:48lostlogicemeraldd: oh, I know which one is being used −− the one that I am holding a fscking seek request on is tha active one :-D
05:40:25TiMiD[FD]or does the playback engine requires the buffer filling when it detects the track's buffer is getting low
05:40:36lostlogicTiMiD[FD]: ie the end of the ring? that's what the guardbuffer is for −− for the current request, that wraps the buffer, we copy up-to 32k of data onto the guard buffer to give the codec a contiguous section of data satisfying its request
05:40:48lostlogicoh, sorry, answered wrong question.
05:41:04TiMiD[FD]still good to know
05:41:11lostlogicthere are two mechanisms that might trigger as a codec runs out of buffer.
05:41:41lostlogicthe one that's supposed to wrok is that the codec should have set a threshold with the buffering api that says "get more data if I run this low" at that threshold the buffering thread wakes up and starts filling the buffer
05:42:23lostlogicif that doesn't work (which is one of the bugs I've killed lately) then when the codec tries to request data that goes beyond the end of the buffer, the buffer then starts to fill
05:43:00TiMiD[FD]the other one is when the buffer has to loop ?
05:43:02 Join hurr [0] (
05:43:25lostlogicthe "wrong answer" I gave above, yes.
05:43:43TiMiD[FD]so basically every codec can work with 32kb of data
05:44:16TiMiD[FD]I mean a full frame could be fitted in 32kb for all supported codecs
05:44:26lostlogicall of 'em so far, other than ADX, which I believe hcs if fixing (because it's not a technical limitation, just a coding choice that makes it erquest more)
05:44:31lostlogicTiMiD[FD]: yeah
05:45:01lostlogicthere are also a couple that require their full file, for which I added a TYPE_AUDIO_ATOMIC to buffering which forces those (small < 512k) files to be buffered without wrapping)
05:45:11lostlogicthose are NSF and SPC
05:45:17lostlogicand SID maybe too
05:45:30Mouser_XMOD as well, if it ever gets supported.
05:45:37TiMiD[FD]so packet based approach could work in those cases
05:46:02TiMiD[FD]if packets are >=32kb
05:46:05 Join karashata [0] (
05:46:18TiMiD[FD]or if we make big packets for those special codecs
05:46:26lostlogicTiMiD[FD]: no, because the buffering code has no way to know where to put the packet boundaries
05:46:37lostlogicso it would still have to copy data around based on what the codec actually requests
05:46:38TiMiD[FD]uh ?
05:47:18lostlogicfor a VBR codec for instance, frame sizes will vary considerably within a single track
05:47:32lostlogicso buffering has no way to know where to place the boundaries of packets so that they align with frame boundaries
05:47:53TiMiD[FD]isn't that the same in the case of the special ring end ?
05:48:44TiMiD[FD]forget it
05:49:05TiMiD[FD]in that case we know the start boundary
05:49:37TiMiD[FD]so apart doing a parser buffering engine the packet approach cannot work
05:50:18lostlogicI mean _large_ packets might be practical (ie > 1mb)
05:50:53lostlogicbut then the smaller things that the buffering api handles would need to be coallesced tos hare packets
05:51:13TiMiD[FD]and having 1mb packets may be bad when memory starts to fragment
05:51:18TiMiD[FD]I mean unefficient
05:51:22emeralddis queue_send blocking?
05:53:04JdGordonits only swcodec though...
05:55:00TiMiD[FD]i wonder how do they do that on the original iriver fw
05:55:09TiMiD[FD]so far the buffering seemed pretty efficient
05:57:16lostlogicyeah, I've often wondered how the OFs do it
05:57:23lostlogicbut not enough to try and RE it out
05:57:41TiMiD[FD]nah that would be unsane to read in assembly
05:59:39TiMiD[FD]complete rebuffer vs track stealing .. .I wonder if it's worth the complexity of track stealing
06:00:34TiMiD[FD]in cases when the buffer is already almost empty a complete rebuffer is probably the right thing to do
06:01:15lostlogicTiMiD[FD]: the fact is that a seek backwards off of the buffer is a fairly rare case, so I believe a full rebuffer is acceptable
06:01:26TiMiD[FD]when it's full ... I wonder if it's worth the battery spare
06:02:13TiMiD[FD]I indeed almost never use it
06:02:57lostlogicthe other option that occured to me and I dismissed would be to rebuffer the track in-place, _knowing_ that it will not be able to fit entirely on the space available for it and then do something tricky when it reaches the end of the buffered data to get the rest (another temporary handle elsewhere on the buffer?)
06:03:20TiMiD[FD]not a very clean solution
06:03:28TiMiD[FD]anyway since you have to spin the disk
06:03:51TiMiD[FD]complete rebuffer makes sense at long as it doensn't takes ages to complete
06:04:10TiMiD[FD]maybe 3 or 4 seconds on 32b units
06:05:31lostlogiccomplete rebuffer on the ipod video takes a bit longer
06:05:53lostlogicbut you're right, any reasonable solution gets everything over with in a single disk spin
06:07:25TiMiD[FD]one thing that I would like would be the playback engine to request the buffering of next folder's tracks
06:07:38TiMiD[FD]I don't think it would be difficult to implement
06:07:42lostlogicyeah, you are definitely not the only one
06:08:19jhMikeSthe playback engine shouldn't care about folders (I guess I repeat that one alot :P)
06:08:21lostlogicunfortunately, the directory flow code is a hack on top of playback, so it creates a whole new playlist and loads that... in that situ, the playback/buffering api doesn't have the info on what to buffer until it's too late
06:08:47TiMiD[FD]I was there when that code was created :)
06:08:57TiMiD[FD]and I thought it was very dirty
06:09:18lostlogicso really it's a matter of fixing the directory flow mode code, not of fixing playback or buffering :-P
06:09:30lostlogic(that's my way of saying "I'm never going to look at that"
06:09:35TiMiD[FD]or the plaback's flow code
06:09:38jhMikeSplayback should go through the playlists via a similar interface and playback from UI controlled through that
06:09:46TiMiD[FD]hehe neither do I intend to :)
06:09:51lostlogicjhMikeS: hrm?
06:10:35jhMikeSplaylists should be the play/pause/stop interface when using them...the playlists should control playback through that.
06:11:12TiMiD[FD]jhMikeS: no next/previous track ?
06:11:16jhMikeSpush the track lists to the playback engine through that
06:11:40TiMiD[FD]would be nice to abstract the playlist from the playback engine
06:12:14jhMikeSof course next/previous...but that's really a playlist concern. if playback has the track ready, just the same as skipping now.
06:12:33TiMiD[FD]the playback engine would only request "givme the track next/previous to this one"
06:12:55jhMikeSplayback engine doesn't request, it's just told
06:13:39TiMiD[FD]so buffering would be done by the playlist engine ?
06:13:44jhMikeSit can be asked if trackX is there of couse...basically make it a dumb thing that just is given a track list to buffer and play
06:13:52jhMikeSyes, it would
06:14:28jhMikeSI already partially did the's sitting somewhere in a .diff on my HD :)
06:14:49 Join midgey_ [0] (
06:14:57jhMikeSanything could be tracklist provider then, a plugin, whatever
06:15:15TiMiD[FD]a nice thing
06:15:49 Quit midgey_ (Client Quit)
06:16:23TiMiD[FD]althought in the case of previous/next folder it would have to handle some kind of endless playlist ...
06:16:33TiMiD[FD]since we're not gonna add the whole disk
06:17:19jhMikeSthe playlist would worry about the directories
06:17:40TiMiD[FD]special entries for directories ?
06:17:59TiMiD[FD](wouldn't be very convenient)
06:18:38jhMikeSno, just as a boundary when building the tracklist for the engine
06:19:15TiMiD[FD]so in case of normal playback the user wouldn't be able to see which will be the next track when reaching a directory's boundary
06:20:31TiMiD[FD]hmm would require some handling in that case by the playlist engine
06:20:59jhMikeSwhy not? if the playlist can give the next track whereever and when ever. I suppose a placing of the metadata alone on the buffer would be enough...but no "to buffer" tasks to perform.
06:22:33TiMiD[FD]so who would do the buffering ?
06:22:37 Join ddalton [0] (
06:22:44jhMikeS"what to play" really stikes me as playlist. "how to play" is engine.
06:22:55ddaltonHow do I adjust the voice file volume when building it?
06:23:00ddaltonI want it to be on 2.1
06:23:07ddaltonso I can hear it over music
06:23:10jhMikeSTiMiD[FD}: the buffering would not change from now
06:23:56TiMiD[FD]so only buffering of the tracks in the current playlist
06:24:29TiMiD[FD]on the iriver it was handled pretty simply
06:25:10TiMiD[FD]at the start the device scans all files in the disk and gives them numbers (breadth first)
06:25:13jhMikeSnot sure what you mean. The playlist would have whatever boundaries and loops that the playback settings demand I suppose.
06:25:20TiMiD[FD]then the next track to play is the next number...
06:26:14 Quit desowin ("use linux")
06:28:23 Quit lazka_ (Remote closed the connection)
06:28:42TiMiD[FD]wait and see :)
06:29:08TiMiD[FD]i won't touch to any piece of code in that area
06:30:04jhMikeSlol...I'll just jump head first into anything even if I have no idea what I'm getting into. :)
06:30:21 Join stevenm [0] (
06:30:28TiMiD[FD]I would if I had time :)
06:30:46 Quit midgey (Read error: 110 (Connection timed out))
06:30:50jhMikeSeven if that means rewriting it just so I understand how it works
06:30:54TiMiD[FD]but nowadays I only touch to some plugins when my day becomes boring and when i need to fake working
06:31:32TiMiD[FD]btw it would be nice if devs stopped producing 2-3k lines files
06:31:39jhMikeSrockbox isn't fake working...that's the real stuff ;)
06:32:01TiMiD[FD]bah it's a cute thing
06:32:08jhMikeSdon't look in mpegplayer when I'm done then
06:32:41TiMiD[FD]I mean when a file grows that big maybe thinking about splitting functionnalities wouldn't be bad
06:32:42jhMikeSstream_mgr.c is hardly close to done and it's 1439 lines :p
06:33:43emeralddhow does the buffering code deal with a file that is larger than the available memery ( Say 128M file on 32M player)?
06:33:47jhMikeSvideo players are nasty complicated things no matter how nicely abstrated and object oriented
06:34:32jhMikeSemeraldd: core buffering?
06:34:58TiMiD[FD]working on more than 600 lines files really becomes difficult
06:35:31TiMiD[FD]I think that 400-500 lines is a reasonable upper limit
06:35:39jhMikeSthe video, audio, disk buffer threads are < 600 lines
06:36:23TiMiD[FD]I'm taking a look at your monstruous file
06:37:07jhMikeSemeraldd: it's just FIFO with the oldest data being whatever is opposite the direction of the buffer movement. oldest is discarded to make room for new.
06:37:37TiMiD[FD]but for example the buffering code could have been splitted in 3 files (lists, buffer space management, buffering api)
06:38:56emeralddjhMikeS: so you would have a single handle with what would fit of the file in memory . . . .
06:38:58jhMikeStoo much splitting can actually make things hard to follow. even if the file is large, clean functional grouping seems to be more helpful.
06:40:10jhMikeSemeraldd: just the section of the file buffered so far. if a seek or playback needs data from outside that window, the above conditions come into play.
06:40:12TiMiD[FD]when you can identify functional groups, splitting makes sense
06:40:27TiMiD[FD]that makes navigation easier
06:40:39TiMiD[FD]instead of using your mouse's wheel :)
06:41:15jhMikeSsure it does. only thing is stream_mgr.c is really one functional group in this case. mpegplayer's been split into many files and it is pretty easy to follow.
06:42:27 Quit emeraldd (Remote closed the connection)
06:43:05TiMiD[FD]hmm yeah in some specific cases it makes sense to keep functions grouped in one file
06:43:18TiMiD[FD]but in the case of the buffering api it would have been nice to split it
06:43:35 Quit stevenm ("Connection reset by beer")
06:43:38jhMikeSthere is a bit of code that could probably be split off...the plans are to make it easy to support different containers/formats/stream.
06:43:50lostlogicTiMiD[FD]: buffering is kinda good the way it is −− use of statics to specify calling restrictions should not be too easily discounted
06:45:14jhMikeSthis sort of coding makes me like C++ quite a bit...but objects can be logically created in C and that's where I'm going...a strict heirarchical architecture.
06:45:19 Join hannesd_ [0] (
06:46:55*lostlogic mutters profanity about the term architecture :-P
06:47:22jhMikeSgotta better term?
06:48:25lostlogicah, structure, yes.
06:48:36lostlogicarchitecture causes nightmares of overengineered crap :)
06:48:58TiMiD[FD]architecture is my daily work
06:48:59jhMikeStechtural archistructure
06:50:05TiMiD[FD]with classes like AbstractDatabaseFactoryManagerCache I'm sure my company will need me forever :)
06:50:07*jhMikeS is glad that painting crap is his...rockbox programming is better than watching the paint dry
06:50:55lostlogicjhMikeS: rockbox is playtime for me, a tiem to get away from overengineered crap and into code that is elegant and purposeful
06:51:16TiMiD[FD]hmm I wouldn't say all parts of code are elegant
06:51:21jhMikeSyeah :)
06:51:28lostlogiclaugh, of course not
06:51:39lostlogicbut I'm free to work on the parts I want to and try to make them as elegant as I can
06:51:44TiMiD[FD]though it got way better since when I came here
06:51:48jhMikeSthe spaghetti code is definitely underengineered
06:52:58jhMikeSI think the target tree helps keep things in line a bit better
06:53:23lostlogicyeah, that was Linus' brilliant presentation at devcon '06
06:54:53jhMikeSthere ended being too much duplication sometimes. I killed that in the PCM sections (a full peaks calculator for each driver just annoyed me).
06:55:11lostlogicyeah, that was an expected initial side effect
06:55:25TiMiD[FD]or the graphical API ...
06:55:55TiMiD[FD]with every target redefining the scrolling thread ...
06:56:02jhMikeSthe graphical API is definitely a bad case
06:56:36jhMikeSbut it doesn't now. I put that in scroll_engine.c.
06:56:51 Join Siku [0] (
07:00:20 Quit DogBoy (Read error: 104 (Connection reset by peer))
07:00:32 Quit hannesd (Read error: 110 (Connection timed out))
07:00:33 Nick hannesd_ is now known as hannesd (
07:00:56TiMiD[FD]I think the fonts related functions are still remaining in target trees
07:01:00 Join DogBoy [0] (
07:01:53jhMikeSsome of the implementation could just be code inlined into the master driver functions
07:05:01jhMikeSis there some rule I don't know about for doing '#include "<file>.c"' in another .c ?? (just wondering if that would raise eyebrows)
07:05:35lostlogicjhMikeS: doesn't that cause the functions to end up compiled multiple times?
07:05:47ddaltonjhMikeS: its not good to include c files if that is your question
07:06:01ddaltonI would have thought so
07:06:04jhMikeSwhy? it would be the static inlines that get inlined into the master display driver
07:06:25jhMikeSyou wouldn't add it to SOURCES
07:06:30JdGordonjhMikeS: yeah, there has been some raised eyebrows over that in the past
07:06:55lostlogicjhMikeS: ah, then why not just put them in a .h?
07:06:55JdGordonsame as not putting code in .h'
07:07:52jhMikeSthat's why I don't want it in an .h if it's really basically a code file...what I do want is private inlining to avoid more function call overhead where it's expensive.
07:08:30lostlogicI doubt anyone would beat you with it since you haev a good reason
07:09:33 Join salp [0] (
07:09:47jhMikeSI guess itty bitty utility inlines in the .h isn't bad (since that's done commonly) where mostly it's still declarations
07:10:59salpHi, I'd like to ask a basic question...I just installed Rockbox on my Iriver h10 20gb and want to add music, but when I plug it into my computer it doesn't go into the mode it used to that allowed my computer to recognize it...what do I do to access it through my computer?
07:10:59jhMikeSI'd have a good reason to do this for thread.c as well since there's alot of CPU selected code in there.
07:12:12JdGordoninlines in the .h is probably nicer than including .c
07:12:14safetydansalp: it should appear as another drive in explorer
07:12:48salpIt isn't appearing as another drive, unfortunately
07:13:08 Part toffe82
07:13:34salpbefore, when it was connected everything was locked on out on the iriver itself, but now I can use it as normal
07:13:50jhMikeSJdGordon: even for 100s of line of just code? what about inline/noninline combo stuff? all inlines in the .h and non-inlines in a separate...say...thread-pp.c? would make following it more difficult instead of just including all that in the .c file.
07:14:16JdGordonwhy cant they added to sOURCES?
07:14:40jhMikeSthe .c must be inluded if inlines are to be inlined. they'll be function calls otherwise.
07:15:00***Saving seen data "./dancer.seen"
07:15:43jhMikeSsince inlines also must be static, they won't be seen externally
07:16:26JdGordonyou'd probably be safe to do it for very special cases... but im sure there will be lots of complaints if it starts being used where it really shouldnt be
07:17:40jhMikeSShould there be a disclaimer? :) I only want to do it where it really makes sense otherwise I agree the usual pattern should be used.
07:18:02JdGordona disclaimer wouldnt hurt...
07:18:16JdGordondunno if its really needed though :p
07:19:13jhMikeSjust a warning about being caned in Singapore
07:19:25 Nick fxb__ is now known as fxb (
07:20:04 Quit Mouser_X ("CGI:IRC (EOF)")
07:21:21 Join Mouser_X [0] (i=cf9bb003@gateway/web/cgi-irc/
07:23:02 Quit Mouser_X (Client Quit)
07:24:49salpdo I need to hold any special buttons while connected to my computer to put my iriver into a mode where it can be recognized by my computer as a drive?
07:25:20 Quit karashata ("Leaving.")
07:25:22lostlogicsalp: I'm not sure on that target, have you read the manual?
07:25:40 Join Mouser_X [0] (i=cf9bb003@gateway/web/cgi-irc/
07:25:47 Join karashata [0] (
07:27:23salpyeah, I have and have seen no information about having to do anything special to connect with the computer...I'm thinking I may have a problem here
07:27:31 Join webguest89 [0] (i=ddd7585a@gateway/web/cgi-irc/
07:28:12lostlogicsalp: generally rockbox puts the player into the correct mode to interoperate with the computer automatically... at least it does on the two targets that I've owned
07:29:09Lloreanlostlogic: The iRiver H10 probably doesn't, being portalplayer
07:29:24lostlogicLlorean: but even on ipod it automatically reboots to OF for it
07:29:41LloreanBut not on the Sansa e200/c200 for some reason.
07:29:49LloreanAnd I *think* I recall it not doing it on the H10, though I'm not sure.
07:29:49lostlogicsalp: for portalplayer targets (ie ipod and h10) we reboot to the original firmware for usb
07:30:01lostlogicsalp: so you might need to manually reboot into th eof to use usb mode
07:30:42*Llorean goes back into hiding.
07:31:17salpso to manually reboot I'll need to stick a paperclip in?
07:31:46lostlogicsalp: no, just shut down
07:31:51karashataRockbox on the H10 will reboot into the OF for USB mode if the select button is held down while it's booting
07:31:52salpand thanks, I appreciate the help
07:32:02lostlogicand do what karashata says :)
07:32:22salpit worked! Thanks a million!
07:32:49salpone more quick question...I can't seem to find my FM tuner in Rockbox...where is it?
07:33:02karashataThe FM tuner isn't working for the H10 yet
07:33:22salpah hah
07:33:30karashatait uses the same chip as the other iriver players but accessing it is different because of the different processor
07:33:38salpoh, that I'm connected, I don't see my files
07:33:44salponly .rockbox and text
07:34:05karashatamake sure you have "view hidden files and folders" checked in the folder layout options
07:34:42karashataor whatever it's called, you're using XP right?
07:34:45salpthere it is...thanks a lot karashata
07:35:06lostlogicaw man, I was getting all set to panic with salp there :)
07:35:22karashataI recommend you uncheck the "hidden" folder property on the music folder so you can see it easier
07:35:33salpwhere can I do that?
07:35:48 Part safetydan
07:35:53Mouser_XRight-click the music folder.
07:36:09Mouser_XCheck properties (sorry about the enter, I forgot this part.)
07:36:10karashataright click the folder, go to properties (should be near the bottom), then in the general tab, uncheck the box labelled "hidden"
07:37:18salpgreat, it worked...and I can finally see how much space I've used up! That was one of my biggest complaints with the original firmware
07:37:45salpI'm in the process of replacing all my .wma files now
07:37:52karashatahmm... I've always been able to see how much space was used up, even with the OF
07:38:19salpreally? not'd never let me see
07:38:27karashatanot on the player itself, mind you, but WMP kept tabs on what space was left, and if the device was in UMS mode, Explorer showed it as well
07:38:45Mouser_Xsalp: Rockbox can play WMA. Just pointing it out, if replacing those proves to be a problem.
07:39:01salpit can? I thought I read that it can't on the iriver
07:39:19Mouser_XOh, sorry. I don't know.
07:39:29salpoh, maybe I never checked on the UMS mode
07:39:55karashataI don't have any WMA files so I can't test to see if they work on the H10 or not
07:39:56 Quit webguest89 ("CGI:IRC (EOF)")
07:39:58 Nick fxb is now known as fxb__ (
07:40:01Mouser_XIt can on software decoders. I don't know if the iriver is a hardware, or a software, decoding target.
07:40:07karashatathough it would be easy enough to find out
07:40:55salpwell, they used to work, but they're rights protected and they don't work anymore
07:41:08salpit's a real pain too, because they're things that I burned from my own cds years back
07:41:38karashataRockbox doesn't do DRM stuff
07:42:23salpNot a big deal...I can just download the albums again
07:47:17karashatahmm, well, lossless WMA doesn't work, apparently
07:50:15 Join vnl [0] (n=nitin@
07:50:46 Part vnl
07:51:16karashatakay, CBR WMA does work, but lossless VBR WMA doesn't
07:52:19Lloreanlossless WMA isn't the same codec.
07:52:23LloreanNot all "WMA" is one codec
07:52:59karashataahhh, okay
07:58:52 Join Stalwart [0] (
07:59:16Stalwartmaybe someone is interested - i just found WooT is offering sansa e250 2gb for just 30$
07:59:31scorchewe know
07:59:52Stalwartnice offer
07:59:55scorchebut we have just been talking about it in the channel where talk of that sort belongs ;) (rockbox-community)
08:00:17Stalwartdidn't visit this chan for a while :D
08:04:02 Quit nanok (Read error: 110 (Connection timed out))
08:10:55jhMikeSStalwart: how's shipping charges at that place?
08:11:59scorchejhMikeS: 5$
08:12:05StalwartCaleb, has experience in wooting
08:12:06 Quit TiMiD[FD] ("leaving") are in -community...sorry for not replying there
08:12:16*jhMikeS thinks he might woot this one
08:14:58 Join jamesshuang [0] (i=44219091@gateway/web/cgi-irc/
08:15:08 Join crwll [0] (
08:15:41jamesshuanghey everyone
08:15:46jamesshuangjust a quick verification question
08:15:54jamesshuangdoes rockbox work on the sansa e250?
08:16:21scorcheyes, and we know why you are asking :)
08:16:27jamesshuanglol, woot deal eh
08:16:39jamesshuangthat's pretty bloody cheap for a rockbox-able player
08:17:18jamesshuangI already have a gigabeat, but I guess it's nice to have a solid state one for jogging and stuff...
08:17:21jhMikeSwooted :D
08:17:27 Join dagrom [0] (i=43b9132b@gateway/web/cgi-irc/
08:17:40Mouser_XIndeed. Nice stuff and price all around.
08:18:19jhMikeSIt's worth $30 just to not have anyone asking to borrow my dev device all the time :p
08:18:20scorchejhMikeS: you got one?
08:18:37 Join Rob222241 [0] (
08:18:51scorche"Thank you for your purchase." too
08:18:59*scorche blushes shamefully
08:19:17*scorche pokes lostlogic to get one for testing ;)
08:20:05dagromwow i just found out about rockbox tonight and my nano has never been happier - thanks all you smart people for making such awesome stuff - i love you all - *goes looking for the manual now*
08:20:05 Quit crwl (Read error: 110 (Connection timed out))
08:20:30scorcheit isnt hard to find :)
08:20:42dagromhehe yeah there are so many
08:21:09dagromi just did the manual unzip and copy i was so happy it was so easy and everything just worked
08:22:18jamesshuangyay... wooted for me :-D that's an incredible deal
08:25:10lostlogicjhMikeS: wait, does your commit mean that the same thread locking the same mutex ttwice and unlocking it twice is safe?
08:25:29jhMikeSlostlogic: yep, the current owner isn't blocked
08:26:20 Join GodEater_ [0] (n=bryan@rockbox/staff/GodEater)
08:27:12 Join artedlc [0] (n=artedlc@
08:27:41 Part jamesshuang
08:27:42lostlogicjhMikeS: when is it unlocked for other entries, after the last unlock on the return path, I hope/assume?
08:28:57jhMikeSyou must call mutex_unlock for every granted call to mutex_lock (balanced calls)
08:28:57 Quit BigBambi (Read error: 113 (No route to host))
08:29:21lostlogicok, good
08:29:24lostlogicI'll keep that in mind.
08:30:11jhMikeSit reduced the number of locks needed. the adc driver on e200 was doing this as part of the DC commit (to have the code running for awhile)
08:32:21hurrwhat's the probability of bricking a device by loading rockbox onto it?
08:32:30lostlogichurr: vanishingly small
08:32:35ddaltonwhat player.
08:32:43hurririver h320
08:32:46ddaltonnot really any. Just make sure you use the right bootloader
08:32:55jhMikeSmutex previously weren't owned objects so semaphores needed to be introduced to have the functionality that codec swapping required. but they're also useful in the DC SPC codec to make the output queue block the output when full and the input when empty.
08:33:08ddaltonnone unless you get a screwed up bootloader
08:33:23ddaltonjust get the one from the site and you'll be fine
08:33:32hurrah, ok, good, thanks. it was the primary reason behind my hesitation to try rockbox
08:33:42 Join ender` [0] (
08:34:02ddaltonyou could probably get it replaced it is hard for them to tell what you did.
08:34:03lostlogicjhMikeS: yep yep, I getcha
08:34:20lostlogicand now I sleep
08:34:42Mouser_Xlostlogic: Did you see the Sansas? Want one for testing? :P
08:35:07dagromthanks again everyone - i just read through the list of credits on my nano to see if i knew anyone but it turns out i don't know any smart people - you are all incredible and keep up the great work (and is there really someone with the last name Teh? that's awesome)
08:35:21*jhMikeS will refrain from expressing the reasons for everything he codes...could run for volumes :p
08:35:26scorcheMouser_X: i already poked him, but he ignored ;)
08:36:03Mouser_Xscorche: I saw that. Since I didn't see a reply, I figured another nudge wouldn't hurt.
08:36:06 Join LinusN [0] (i=linus@rockbox/developer/LinusN)
08:36:13 Quit Rob2222 (Read error: 110 (Connection timed out))
08:36:31lostlogicokokok, I'm gettig one
08:36:31 Quit dagrom ("CGI:IRC (EOF)")
08:37:02Mouser_XHeh. Awesome.
08:37:07lostlogicclick this stupidly large button
08:37:21jhMikeSthey only allow one purchase/woot?
08:37:35Stalwarti guess
08:37:48Mouser_XI think you can buy up to 3. Maybe it has to be 3 on the same order?
08:37:57 Quit artedlc ()
08:37:58lostlogicnot that I really have much use for a 2 gig player, but you're right, good for testing
08:38:00*jhMikeS can get around any restrictions here
08:38:03lostlogicwhy did my ipod just stop playing, do I have a bug?
08:38:07scorchejhMikeS: hrm?...1-3 per person per woot
08:38:48jhMikeShmmm....shoulda done 3 :) but if I wooted once, can I woot two more times?
08:39:10scorchei dont know
08:39:24lostlogicunfucking believable. there's still a bug that made my ipod freeze after playing for hours. whatever, sleep time.
08:40:55*jhMikeS just goes and reads the terms there to avoid OT chatter
08:42:44hcsok, who decided to remap the sansa buttons... the down button for the context menu was nice (not having to hold something...)
08:43:04jhMikeSaha...1 order/accout-day. np anyway.
08:43:32Mouser_XjhMikeS: Dang...
08:43:53*scorche directs hcs to the forum post, the tracker entry, the mailing list post, and Llorean
08:43:55karashatahcs: Llorean commited that change
08:44:40hcsscorche, karashata: thanks, I could have found out if I wans't just whining
08:44:46jhMikeSthe remapping mutated into something more pleasant than the initial patch I tried it seems
08:45:05karashatahcs: doesn't bug me any, I tend to be quick at finding things anyway
08:45:34 Join viv [0] (n=nitin@
08:46:01 Quit viv (Remote closed the connection)
08:46:25jhMikeSlostlogic: btw...if I start babbling...just tell me to stfu...I prefer that to hints...and the anger helps me code better too. :p
08:47:53amiconnjhMikeS: What do you think should we do with pp5024.h (re my question yesterday)?
08:49:09jhMikeSamiconn: do those multiple GPIO bitmasks apply to pp502x in general? I figure it's not much worse than having mcf5250.h to change some defines.
08:49:30amiconnDid you have a look at pp5024.h in current svn?
08:49:40amiconnIt's empty apart from including pp5020.h
08:49:49ddaltonis it possible to beep when fast forwarding/rewinding? or is it like the paused thing.
08:49:53ddaltonsame problem.
08:50:04amiconnThe gpio ints are grouped on the others as well, and there is GPIO2_IRQ as well
08:50:21amiconnFound that when adding microsd detection to c200
08:51:11jhMikeSamiconn: not for some time. what about #defines in there for the SD interface? Is that pp502x general?
08:51:29jhMikeSsome of that stuff should get out of the source files
08:52:28amiconnOne would think it is though, as the c200 is PP5022
08:53:25jhMikeSPP5022 has an AS3514 too though or it's external there?
08:53:36amiconnexternal afaik
08:57:01jhMikeSwell, heck it seems pp502x.h would be sufficient since they obviously stopped "innovating" before tanking.
08:58:31amiconnI still need to find out why microsd cards don't work on c200. The detection isr definitely works now
08:58:56amiconnTested it by putting some buttonlight manipulation into it
08:59:34amiconnI think the card selection is wrong - I'd like to ask low_light where he found that, or whether it's just guesswork based on e200
09:03:02jhMikeShe said from the emu which is also where the e200's came from
09:03:39jhMikeSmaybe just try the e200 selection?idk
09:05:28jhMikeSI guess that wouldn't work since the main card works with the current one
09:07:45amiconnBtw, there *are* a few differences between pp5020, pp5022 and pp5024. One thing where they differ is the clock setup (pp5020 different from 5022+5024), another is the ata controller (all different)
09:08:22amiconnThe latter is not yet researched further, but an ata reg dump looks quite different between 5020 and 5022
09:08:40amiconn(and 5024 has no ata controller as far as we know)
09:09:55jhMikeShave you tried a longer timeout for the debouncer? ata will disable the card if a few retries doesn't result in a successful init. retailos actually waits several seconds (that is by the spec for worst case)
09:11:27amiconnIt doesn't work even with the card plugged at boot
09:12:11 Join Zagor [0] (
09:12:27jhMikeSGPIOL bit 3 is high when plugged? seems wierd.
09:13:26amiconnyes, why?
09:14:15 Join jmspeex_ [0] (
09:14:34GodEater_does anyone know if there's a windows ssh client which implements the network interface tunnel (e.g. tun0) equivalent to the *nix one ?
09:14:40jhMikeSI dunno...perhaps ask low_light for the c200 modded emulator
09:15:04***Saving seen data "./dancer.seen"
09:15:45hcsGodEater_: not that I've heard of, best I know is normal tunnelling + socksap
09:16:25GodEater_hcs: not good enough :(
09:16:32hcsGodEater_: then again, I'm no expect
09:17:58 Quit Mouser_X ("CGI:IRC 0.5.9 (2006/06/06)")
09:20:27 Join petur [0] (n=petur@rockbox/developer/petur)
09:20:53 Quit jmspeex (Read error: 110 (Connection timed out))
09:24:00jhMikeShaha...resume restarts playback at the start of the file even though it shows the correct resume time in the WPS
09:24:27 Quit jumpatrain (Remote closed the connection)
09:25:34jhMikeSseeking once brings it back to the correct point
09:26:26 Join jumpatrain [0] (i=tabac@gateway/gpg-tor/key-0xB9002659)
09:28:35linuxstbamiconn: Your HDD power-off patch increased runtime (with approx 200kbps VBR mp3 files) from 6h 9m to 6h 50m on my ipod color.
09:29:48hcslinuxstb: I got 7h 20m with a similar setup, unfortunately I don't have the "before" time handy.
09:30:10hcsI seem to recall only around 6 hours, though
09:33:40 Join corevette [0] (
09:33:52 Part corevette
09:43:25 Nick parafin|away is now known as parafin (
09:47:06 Quit karashata ("Leaving.")
09:47:45 Quit markun (Read error: 145 (Connection timed out))
09:54:16 Join davina [0] (
10:01:29 Join pixelma [0] (i=pixelma@rockbox/staff/pixelma)
10:01:38linuxstbhcs: I'll charge again and do another test...
10:01:58linuxstbhcs: Have you done any tests in the Apple firmware?
10:05:49 Join FOAD_ [0] (
10:15:55 Quit atsea-39 ("using sirc version 2.211")
10:16:36 Quit Siku (Read error: 145 (Connection timed out))
10:17:57 Join sasukee [0] (n=SK@unaffiliated/sasukee)
10:18:20 Quit sasukee (Client Quit)
10:18:48 Quit FOAD (Read error: 110 (Connection timed out))
10:18:49 Nick FOAD_ is now known as FOAD (
10:22:33 Part pixelma
10:22:38 Join CaptainSquid [0] (
10:25:32 Join pondlife [0] (n=Steve@rockbox/developer/pondlife)
10:28:24amiconnlinuxstb: Hmm, so about +10%. Somehow I expected more, but still an improvement :/
10:28:38*amiconn started a runtime test on mini g2 this morning
10:29:32 Quit nnakihfdn (Read error: 110 (Connection timed out))
10:38:06*barrywardell remembers his H10 getting about 10% improvement when disk poweroff was enabled too
10:40:58barrywardellnow that I think about it, I'm not even sure if we have the right gpio for the small H10. Might be worthwhile for someone who owns one to look into it.
10:41:47linuxstbamiconn: I'm planning on double-checking those figures, so I'll let you know for sure in a few days. But even 10% is a nice improvement.
10:42:27 Join roolku [0] (
10:42:30amiconnbarrywardell: I can verify that tonight
10:43:16*amiconn thinks we should split CPUPortPinAssignment per architecture, and add the missing PP bunch
10:43:17barrywardellgreat, if we have it wrong, maybe bit 24 of GPO32 would be a good place to look
10:44:41amiconnTesting that is fairly simple: Just put the port switch somewhere in the debug menu, set a long hdd spindown timeout, and then enter that debug item with the disk spinning. If it was the correct pin -> click
10:45:06barrywardellthat's exactly how I found it on the big H10
10:46:05amiconnI just used that as a verification on ipod. Found the pins from disassembling the rom
10:47:19barrywardellI pretty much just happened across it. the click is hard to miss
10:47:27 Nick parafin is now known as parafin|away (
10:48:14amiconnNot that hard on the mini (had to put the mini right next to my ear)
10:51:18 Join mhamlin [0] (
10:55:49 Part mhamlin
10:58:08 Quit barrywardell ()
11:00:01salpI've got a quick question about Rockboy...for an iriver h10 do I have to install anything special or is it ready to play those games with the normal rockbox firmware? Also, I've read that it runs slowly, but it is so slow that it's not worth trying?
11:00:58GodEater_salp: no - you don't need to install anything special. No idea on your other question though.
11:02:07salpThanks GodEater
11:03:20GodEater_salp: just remember. For some reason people always seem to think that Rockboy is launched from somewhere in the menus. It itsn't. Just copy your rom files to somewhere on the player, and then select them in the filebrowser to play them.
11:03:22 Join atsea- [0] (i=atsea-@gateway/tor/x-61a8263ef1d3743d)
11:04:53salpgreat, thanks for the info. I'm excited to give it a shot. I need to download a chess game that's a little easier than the one that comes pre-installed...that thing is tough!
11:05:17GodEater_<−− not a fantastic chess player either
11:05:31 Join cheesenuggett [0] (
11:05:57cheesenuggetthey guys i was wondering if you possibly know of a good mp3 organizer that will automatically sort my mp3s
11:06:09cheesenuggettbecause i have about 4000 and it's sort of a mess
11:07:06rvvs89Amarok is pretty good
11:07:34GodEater_as is winamp
11:07:49cheesenuggetti want something that will sort the mp3s into folders
11:08:00Bagdermv !
11:08:05cheesenuggettlike do all the artist > album name
11:08:16rvvs89iTunes? :P
11:08:32cheesenuggettdoes itunes do that?
11:08:53salpit's not too easy to find .gb roms...everything I'm finding is either .gbc or .gba
11:08:53cheesenuggettwhy have i hated itunes all my life
11:09:13salpme too cheese...I can't stand itunes
11:09:13cheesenuggettbecause the end goal would be to be able to move all my folders cleanly into my rockbox
11:09:17rvvs89Because it's a monopolistic program?
11:09:44cheesenuggetti just started using rockbox
11:09:53cheesenuggetti moved all my music into a music folder
11:09:57cheesenuggettand it was a mess
11:10:02cheesenuggettall the tracks in the same folder
11:10:35cheesenuggettso itunes can organize all of them into folders that i can drag onto my mp3 player?
11:10:47rvvs89Again, yes
11:10:51cheesenuggettalright sorry
11:10:54cheesenuggettim wayyyy tired
11:10:58cheesenuggettthanks for the help yo
11:11:00 Join lee-qid [0] (
11:11:07rvvs89I'm sure there's a free program that does it too
11:11:16cheesenuggettyeah ive been looking around
11:11:25cheesenuggetti found a great one but couldnt find a damn keygen
11:11:34rvvs89That's not what I meant by free...
11:11:54pondlifecheesenuggett: Use the Rockbox database?
11:12:04cheesenuggettwhat do you mean?
11:12:11pondlifeTo access your music
11:12:18rvvs89Yeah, rather than the file browser
11:12:24pondlifeThe Godfather is a good organiser too
11:12:34GodEater_or easytag
11:12:42*rvvs89 away
11:13:19cheesenuggettwoah the database works that way
11:13:24cheesenuggetti deleted all my music
11:13:43GodEater_cheesenuggett: please try not to use "enter" in place of ","
11:13:53cheesenuggettalright dude, sorry
11:14:50cheesenuggettthis is the second time today you guys helped a little ol' noob out, thanks!
11:15:05***Saving seen data "./dancer.seen"
11:16:02 Part Stalwart ("WeeChat 0.2.6")
11:23:24 Join barrywardell [0] (
11:37:25cheesenuggettso how do i use the godfather organiser?
11:38:11 Nick Soap is now known as Soap_ (n=Soap@rockbox/staff/soap)
11:43:13 Join pnakicofd [0] (i=0@
11:48:29 Part salp
11:49:22 Join Thundercloud [0] (
11:56:06 Quit cheesenuggett ()
12:04:38preglowlinuxstb: i would have expected more impressive new runtime figures from almost all ipods :/
12:04:52preglowboy, do i wonder where the hell all that current goes
12:05:58linuxstbAs the test was running, I was wondering if my WPS (icatcher) is helping consume power - there's a scrolling line which was scrolling throughout the text (next track info).
12:06:14preglowprobably is
12:12:19 Quit RaRe` (Read error: 110 (Connection timed out))
12:13:09 Quit lespea (Read error: 104 (Connection reset by peer))
12:13:16 Join Rob2222 [0] (
12:14:08 Join RaRe [0] (
12:14:17 Quit ddalton ()
12:24:04 Join n1s [0] (
12:24:58 Join spiorf [0] (
12:26:15 Join Nico_P [0] (n=nicolas@rockbox/developer/NicoP)
12:27:29JdGordonrasher: ping?
12:30:51 Quit Rob222241 (Read error: 110 (Connection timed out))
12:31:22 Quit maddler ("connection reset by beer!")
12:31:47 Join maddler [0] (
12:38:43 Nick fxb__ is now known as fxb (
12:39:41 Join mrkiko [0] (
12:40:25mrkikoHi all!
12:41:14mrkikoThe H 300 port seems out of control competely from this morning; pressind quickly enough to skip from a track to the next will cause a crash! :)
12:41:32mrkikoBut my reader still works fine on the original firmware - so that's not an hardware problem...
12:45:27mrkikoIn addition, fs#8003 can be triggered easily now, after a successful recording...
12:45:41preglow-O4 does _not_ exist, right? tons of people seem to think it does
12:47:13n1spreglow: ?
12:53:43preglowseems like a simple question to me...
12:53:46n1swhat is "-04"
12:53:50preglowthe gcc flag
12:54:01n1sah, nope don't think so
12:54:47n1sO3 usually makes code slower in my experience so why would anyone want O4? :P
12:55:26jumijozeto be the ultimate ricer :)
12:56:46amiconnpreglow: The current goes into those pp functional units we don't use because we don't know about them, and which we don't disable because of that same reason
12:57:17amiconnOn PP5002 it was a very noticeable effect. With all DEV_EN bits set, the ipod got warm to the touch
12:59:22amiconnOn the outside that is - with both the hdd and the battery between the pcb and the metal back cover
13:00:40mrkikoDoes someone know how many advantages would have a DB approach for a player ?
13:00:45mrkikoi.e.: iPod style.
13:01:30 Join perrikwp__ [0] (n=chatzill@
13:02:22*amiconn thinks it has only disadvantages
13:02:33*linuxstb doesn't really understand the question
13:02:51*mrkiko agrees with amiconn
13:03:04 Quit lee-qid ("aufwiederbyebientotsayonara")
13:03:33mrkikoI asked: in your opinion, what advantages would have a DB-style approach into an mp3 player ? Why in your opinion Apple choosen to use itunesDB scheme instead of raw file storage?
13:04:11n1smrkiko: this is OT for this cannel
13:04:59mrkikon1s: ... really? I was thinking why not to adopt this solution for rockbox...
13:05:21mrkikobut I can't see many advantages...
13:05:52n1smrkiko: then I don't understand your question ;)
13:06:11mrkikon1s: sorry, but I should go...
13:06:17amiconnAn optional DB makes some sense, depending on usage cases. A mandatory db is just silly
13:06:19mrkikon1s: at later
13:06:33mrkikoamiconn: right
13:06:34 Quit mrkiko ("leaving")
13:07:28 Join agm3nt [0] (
13:15:07***Saving seen data "./dancer.seen"
13:16:54n1sDoes anyone have any ideas for a looong string to put at the top of the scroll setting menus? I don't see the point of having one for speed and one for step size
13:17:28n1sand it should be longer than "Scroll Step Size Setting Example Text" because that doesn't scroll everywhere
13:17:45 Join Sedgewick [0] (
13:19:16 Quit perrikwp (Read error: 110 (Connection timed out))
13:19:48 Join Arathis [0] (
13:22:17 Join perrikwp___ [0] (n=chatzill@
13:22:22 Nick perrikwp___ is now known as perrikwp (n=chatzill@
13:23:55linuxstbCan we set any kind of margin on scrolling lines, to ensure the text fits?
13:24:08linuxstb(or rather, doesn't fit...)
13:25:24linuxstbThe m:robe's screen for example will need very long text if we are going to guarantee scrolling with any font...
13:25:28n1spadding with spaces works but is kind of ugly
13:25:57linuxstbMaybe we should just leave that until we have viewports, so we can fix the scrolling margins.
13:27:07n1syep, makign it scroll that way on a 640 pix wide display (mrobe landscape) and 5 pix font is a _lot_ of spaces :D
13:27:27JdGordonI tihnk adding a special case for that is needed
13:28:02linuxstbWhat kind of special case?
13:28:35JdGordonfudge the title in the setting screen
13:30:43JdGordonI really dont think there is a clean way to do that though
13:31:46JdGordon... except if we add a rediculous amount of spaces in english.lang depending on the target?
13:32:52JdGordon... or maybe not
13:33:47linuxstbCouldn't the string just be manipulated at runtime to pad with spaces, or maybe repeat the string multiple times, based on LCD_WIDTH and current font width?
13:33:54JdGordonnope.. its gonna have to be messy
13:34:08JdGordonyeah, thats the way it will have to be
13:35:20amiconnn1s: The text could be padded with itself until it's long enough
13:35:47amiconnJust keep in mind that there is a size limit for scrolling text
13:36:08n1samiconn: yes, true, would still be really long for the worst case thought (640 pixel display and 5 pixel font)
13:36:14JdGordonwhich is actually possible to exceed without scrolling!
13:36:27amiconnI doubt that
13:36:42linuxstbWhich reminds me of the second part of this bug report -
13:36:46amiconnThe maximum length is MAX_PATH + something lcd size dependent iirc
13:36:49JdGordonhmm.. ok maybe not
13:37:03JdGordon640/5 = 128.. well under the max title length
13:37:36linuxstbIn the example wma file given with that bug report, the comment field is around 650 characters, which seems to cause problems with the "view track info" screen...
13:38:30JdGordoni bet it does! the list can only take lines up to MAX_PATH
13:38:37JdGordonmassive buffer overflow there
13:39:24linuxstbBut where should it be checked?
13:39:42linuxstbShould the metadata parsers limit tags to certain lengths?
13:39:59JdGordonyes, I dont tihnk the list code checks the line length at all
13:40:23amiconnlcd_puts_scroll just cuts whatever exceeds its limit iirc
13:41:26 Quit perrikwp__ (Read error: 110 (Connection timed out))
13:41:33linuxstbWhy does the list need to check lengths? Surely it doesn't copy strings?
13:42:29 Join perrikwp___ [0] (n=chatzill@
13:42:52JdGordonit doesnt.. but the buffer which it passes to the callbacks are MAX_PATH long and not usually \0'ed to make sure they fit correctly
13:44:15*preglow wishes some hero would pop up and implement viewports
13:46:41amiconnJdGordon: If they aren't \0 ed that's clearly a bug
13:47:42JdGordonumm... I should have said, they assume snprintf() makes sure they are \0'd... im nt sure if it does or not (another project i was in had a bug where it wasnt if it fit exactly)
13:51:13Calebrockbox is sweet
14:00:37 Quit perrikwp (Read error: 113 (No route to host))
14:04:10*JdGordon is king!
14:04:28JdGordonthis also means we can deprecate a string or two
14:04:59JdGordonalthough "Scroll Speed Scroll Speed Scroll Speed ..." for the title might look bad
14:06:02n1show 'bout "Scroll Setting Example Text" or somesuch?
14:06:14JdGordonthats what it displays atm
14:06:53JdGordonthat code pads it with the title.. would padding be spaces be better? im not so sure
14:09:14amiconnPadding with text is better
14:09:24amiconnWhat about "Scroll settings example text..."
14:09:28JdGordonhmm... padding with space is simpler code.. but it doesnt look so good unless we check the new string lenght
14:10:26 Part LinusN
14:11:23JdGordonanything in that paste look bad?
14:11:28JdGordonI tinhk ill commit and goto bed
14:12:34pondlifeAnyone else finding playback resuming totally broken today?
14:13:06pondlifeI can happily get stupid times displayed too... -0:-18 et al
14:13:45JdGordonn1s: is there a FS task to go with the commit
14:13:50pondlifelostlogic: You seen this?
14:14:01n1sJdGordon: 7288
14:16:06barrywardellIs there a list of which DEV_EN,DEV_EN2,DEV_EN3 bits are set on different targets? maybe that would help to figure out what they all do.
14:16:25barrywardellthere are some differences between what my H10 has and what my Sansa has
14:18:10 Quit Thundercloud (Remote closed the connection)
14:18:20barrywardellH10 has DEV_EN = 0xC240197f, DEV_EN2 = 0x00002020, DEV_EN3 = 0x0007003F
14:19:03barrywardelle280 has DEV_EN = 0xC440597F, DEV_EN2 = 0x00000000, DEV_EN3 = 0x0007003f
14:22:08GodEater_I can only see DEV_EN on my 5.5G ipod, it's set to : 0XC2C1197F
14:24:16barrywardellamiconn added the other two to the debug screen on saturday
14:24:28GodEater_I'll update then
14:24:58amiconnDEV_EN3 is still to be verified
14:25:26amiconnIt might be something different, although a routine in the ipod roms suggests that it is DEV_EN3
14:26:19JdGordonbarrywardell: if your interested... my DEV_EN2 = 0x00002000
14:26:22 Part Caleb ("WeeChat 0.2.6")
14:26:33JdGordonis that missing 2 the radio?
14:26:56linuxstbipod Color is 0xc2c3197f, 00002000, 0007003f
14:27:22amiconnI doubt it. DEV_ENx are most probably just clock gates, and the radio is external
14:28:47linuxstbpondlife: Which file format (your resume problem) ?
14:29:02*GodEater_ wishes someone would just commit 7738, he's fed up of having to sync it
14:30:04pondlifeOnly tried the sim so far
14:30:06linuxstbpondlife: Can you check another filetype, e.g vorbis? I expect it's the id3v2 stripping...
14:30:27JdGordonGodEater_: you need to nag jhMikeS about that one
14:30:48GodEater_JdGordon: I have. He's very stubborn ;)
14:31:08pondlifelinuxstb: It was rev 15465 that broke it
14:31:30JdGordonstubborn because he cant be bothered? or stubborn because its not done properly?
14:31:50GodEater_JdGordon: I've no idea. I'm not sure what is still wrong with it
14:32:00GodEater_it appears to work fine to me
14:33:02n1sGodEater_: does it still add settings and are they needed?
14:33:33GodEater_yes and I don't know
14:33:59*GodEater_ doesn't see DEV_EN2 and 3
14:34:14n1sIMHO we should only add settings if they are really needed not because we can...
14:34:46JdGordonneed to... want to... same thing :p
14:34:50*JdGordon going to bed
14:34:53 Quit JdGordon ("Konversation terminated!")
14:34:57 Join zicho [0] (
14:35:36amiconnGodEater: Does it also work for 1st..3rd gen and mini 1st gen?
14:36:05amiconn(those have different wheel drivers than 4th gen and newer)
14:36:21GodEater_amiconn: good question :(
14:36:33*GodEater_ resigns himself to sync'ing manually for a lot longer
14:36:43barrywardellJdGordon: the missing 2 is the ADC
14:39:07GodEater_well that's bloody weird.
14:39:32GodEater_current build is offering the DEV_EN2 and 3 - but my own build, done with a freshly sync'd svn (albeit via git) doesn't
14:39:59 Quit maddler ("connection reset by beer!")
14:39:59GodEater_DEV_EN2 on 5.5G is 0x00000000, and DEV_EN3 is 0x0007003F
14:40:00 Join japc [0] (n=japc@
14:40:15 Join maddler [0] (
14:41:58pondlifelostlogic: Seems like seek_buffer isn't doing enough to replace advance_buffer. Reverting;r2=15465 fixes resuming for me
14:43:59 Quit japc (Remote closed the connection)
14:46:06 Join darkapostrophe [0] (
14:52:24 Quit zicho (Remote closed the connection)
14:54:06 Quit spiorf (Remote closed the connection)
14:56:07 Join JETC- [0] (
14:56:21 Join mitchie [0] (
14:56:32 Nick mitchie is now known as msmitchie (
14:56:45msmitchiehallo all
14:57:46 Join nicktastique [0] (n=nick@unaffiliated/nicktastic)
14:58:51msmitchieMy current build doesn't show the bitmap oics which the originalSansa software does...
14:59:04msmitchiepics I mean
14:59:58 Quit Dark_Apostrophe (Connection timed out)
14:59:58 Nick darkapostrophe is now known as Dark_Apostrophe (
15:00:10linuxstbWhat format are those pictures? And can you find them in the file browser?
15:01:28msmitchiebmp and couldn't unless I change the file setting to show all
15:01:52 Quit delYsid (Read error: 104 (Connection reset by peer))
15:02:41linuxstbRockbox doesn't have a bmp viewer, the best you can do (I think) is open them with rockpaint - which should happen when you select the file in the file browser.
15:03:35msmitchieis tht another plugin?
15:03:40linuxstbpondlife: Have you tried removing the request_buffer() call, and just calling advance_buffer() instead of seek_buffer() ?
15:03:58linuxstbmsmitchie: It's a viewer plugin that should be installed by default.
15:04:17linuxstbWhat happens when you select a bmp file?
15:05:30 Join zicho [0] (
15:05:32msmitchieNothing, not even the pics tht came w the player itself
15:05:43*barrywardell created
15:06:03preglowlinuxstb: have you checked if ffmpeg svn has done any neat stuff to their flac.c?
15:06:12msmitchieI couldn't even view those I mean but they are there. Mine own folder isn't even shown
15:06:29linuxstbpreglow: No, I haven't looked at it since I first ported it.
15:06:46preglowlinuxstb: it's seen quite a bit of work
15:08:40linuxstbpreglow: Yes, I'm just looking at the commit log...
15:09:24preglowon the other hand, it works quite well as it is, but some of that stuff might make it faster
15:09:31preglowfaster still...
15:11:15linuxstbCommit 10586 looks interesting - claiming around 25% speedup
15:11:47linuxstbBut yes, we should pay attention to them.
15:12:28 Join Siku [0] (
15:12:56 Join desowin [0] (
15:13:53linuxstbGiven the dream of FLAC-on-SH, we can't make it too fast...
15:15:08***Saving seen data "./dancer.seen"
15:16:54*barrywardell wonders how dan_a came up with the comments in firmware/drivers/audio/as3514.c:142
15:16:56 Quit barrywardell ()
15:17:16 Join barrywardell [0] (
15:18:04 Join barrywardell_ [0] (
15:21:43preglowlinuxstb: well, as long as none of it breaks any asm it should be straight commitable
15:22:39 Join markun [0] (n=markun@rockbox/developer/markun)
15:25:23msmitchiegtg. Thx linuxstb
15:25:28 Quit msmitchie ("using sirc version 2.211+KSIRC/1.3.12")
15:25:38 Join Chronon_ [0] (
15:26:56 Quit desowin ("use linux")
15:27:59 Quit Chronon (Read error: 110 (Connection timed out))
15:28:33 Part kkurbjun
15:28:45barrywardellah, they were copied from the wmcodec code which was copied from ipodlinux
15:30:49 Join desowin [0] (
15:30:51 Join karashata [0] (
15:33:20 Quit zicho (Remote closed the connection)
15:41:55 Join zicho [0] (
15:42:51GodEater_Llorean ?
15:45:20 Join Chronon [0] (
15:46:48lostlogicpondlife: resuming's brokenness is most likely a result of FS8092 and the non-application of PRESEEK in the existing code.
15:47:37lostlogicpondlife: the read->advance method would force the whole file to end up buffered even when resuming would only require part of it and therebo obviate the need for a reverse seek and also the need for conf_preseek
15:47:56 Quit Zagor ("Client exiting")
15:50:21 Quit Chronon_ (Read error: 110 (Connection timed out))
15:50:38lostlogicpondlife: what filetype are you having resume trouble with?
15:53:37 Join MajorC [0] (
15:53:54davidfg4I fixed the problem I was having yesterday!
15:54:13davidfg4tools/ had said "ROCKBOXDEV: patch is required for this script to work. Please install and re-run the script."
15:54:27 Join vitaminx [0] (
15:54:39pondlifelostlogic: Sorry, was away - it's with mp3
15:54:41davidfg4what it meant was that I needed to install the program called "patch"
15:54:50 Quit vitaminx (Client Quit)
15:55:02pondlifeJust run the sim, press stop during the track, then press play to resume
15:55:07 Join vitaminx [0] (
15:56:00 Quit vitaminx (Client Quit)
15:56:16lostlogicpondlife: I don't (and likely never will) have a sim :-P
15:57:55lostlogicpondlife: but yeah, it's like I said −− an additional symptom of rebuffer_handle being broken −− that read, advance sequence would just force its way around
15:58:12pondlifeOK, will leave with you
15:58:19 Join Thundercloud [0] (
15:58:34pondlifePerhaps you could reverf that mod if it's going to take a while though..
15:58:37lostlogicno tsaying I can necessarily fix it though because 8092 ate my brain twice already
15:58:58lostlogicpondlife: hmm, I prefer to expose the symptoms of stupid bad that needs fixing, but maybe if it doesn't get fixed fairly soon
15:59:13pondlifeYes, this is a royal pain though
15:59:29lostlogicpondlife: interestingly resume is not one of the symptoms I get on target ;)
15:59:49pondlifeI'll try on H340 then
16:00:20pondlifeI can get either a segfault or the wrong audio played on sim
16:00:30pondlifeBut it never works ok :)
16:05:45 Quit zicho (Remote closed the connection)
16:06:12 Quit CaptainSquid ("Miranda IM!")
16:07:25 Join Windows-Server [0] (
16:07:46 Quit DataGhost (Nick collision from services.)
16:07:49 Join [Sur`DataGhost] [0] (
16:08:07 Part pondlife ("Gone")
16:10:01 Quit desowin ("use linux")
16:10:46 Quit Thundercloud (Remote closed the connection)
16:12:42 Quit Sedgewick (Connection timed out)
16:13:12 Join desowin [0] (
16:18:24 Join Crash91 [0] (n=evil91@
16:20:00 Join advcomp2019_ [0] (n=advcomp2@unaffiliated/advcomp2019)
16:22:06 Quit Windows-Server ("Money can't buy happiness but it can provide a better class of enemy.")
16:22:08 Join zicho [0] (
16:23:04 Join Windows-Server [0] (
16:23:55 Nick Windows-Server is now known as Sedgewick (
16:25:44 Join martin_ [0] (
16:26:16BagderGodEater: your server back to sanity?
16:26:38Bagderbuild server that is
16:32:14 Join pondlife [0] (n=Steve@rockbox/developer/pondlife)
16:33:58 Part pondlife ("Gone")
16:35:28 Part Crash91
16:36:11 Quit jumpatrain ("Leaving")
16:37:22 Quit advcomp2019 (Read error: 110 (Connection timed out))
16:39:54 Quit martin_ ("*.net *.split")
16:45:57 Join blufox_ [0] (
16:46:10blufox_hello, lol
16:47:01blufox_I don't have an Ipod, but I was thinking about getting a 5G
16:47:23blufox_I was just wondering if anyone had used the cover flow theme for rockbox ipod 5G
16:47:58*Bagder has no ipod and uses only bundled themes...
16:48:16blufox_I see
16:48:51blufox_by bundled, do you mean the included theme, and can I use the regular ipod theme if I install rockbox?
16:49:07Bagder"regular ipod theme" ?
16:49:15BagderI mean included in the rockbox zip
16:49:24 Join web-taz [0] (
16:49:42blufox_the default theme that ipods have with the regular firmware, or does it change when you put rockbox on
16:49:56Bagderrockbox is a complete replacement
16:50:01Bagderit uses nothing from the original
16:50:42web-tazi have got a problem converting videos for rockbox
16:51:10 Quit hurr ("NO")
16:51:39web-tazi encode via batch and vlc; and it always worked fine; but i wanted to watch some anime episodes on my gigabeat f30, and the files couldnt be opened
16:51:53web-tazalso at the pc only the sound was played
16:52:23web-tazi realized a problem with MKVs which habe several audiolayers, und subtitles
16:52:46web-tazbut these animes, just are xvid and mp3
16:52:56web-tazso i cannot imagine why it doesnt work
16:53:19karashataweb-taz: did you convert them to MPEG-1 or MPEG-2 format before trying to play them in Rockbox?
16:53:54 Join Thundercloud [0] (
16:54:07karashataactually, thinking a sec, that's a silly question... they wouldn't show up otherwise..
16:54:13Bagdertime to go make dinner
16:54:50karashatahowever, Rockbox can't do anything with Xvid compressed video AFAIK, so that's likely why it's not showing the video portion of the clips
16:55:43web-tazi used the vlc batch file to convert it to mpeg
16:56:14web-tazwith other files this worked very well
16:57:02karashatahmm, I don't use VLC myself so I can't help too much with that...
16:57:17karashatado you use Linux or Windows?
16:58:27web-tazi think it must be something with the animes
16:58:42karashataare you converting them from MKV files directly?
16:58:45web-tazbecause other files (i'm just trying another)
16:59:02web-tazi gave up converting mkvs
16:59:18karashataah, kay...
16:59:20web-tazbecause i must delete audio layers
16:59:33web-tazwhat I'm converting is avi
16:59:37web-tazxvid / mp3
17:00:13karashatadoes the avi file play back with the audio and the video both working in the VLC player?
17:03:03 Join toffe82 [0] (
17:05:26markunweb-taz: maybe you can host a small sample video for us to see what's wrong?
17:05:56web-tazsample from the not working converted?
17:06:04web-tazplease wait a minute...
17:06:58 Quit hcs ("Leaving.")
17:07:11web-tazoh fuck
17:07:22web-tazi have got so many tools to convert videos
17:07:28web-tazbut how do i cut it?
17:07:44markunnot sure :)
17:08:32markunI use mencoder and when I press ^C it just stops converting, but maybe vlc throws the video away
17:09:18web-tazi think so
17:09:32web-tazabort in windows mostly means -> return to start
17:11:56web-taznot goot
17:12:07web-taztmpgenc sais: illegal mpeg streams
17:12:13web-tazwhen i try to cut it
17:13:27 Join japc [0] (n=japc@
17:14:02lostlogicbarrywardell: there's red and it's not mine
17:14:45web-tazperhaps i should try to encode with mencoder
17:15:10***Saving seen data "./dancer.seen"
17:15:51barrywardell_lostlogic: yeah, I 've already commited the fix ;)
17:18:06 Quit japc (Remote closed the connection)
17:18:08 Join Arathis2 [0] (
17:22:11 Join mf0102 [0] (n=michi@
17:22:12 Quit Thundercloud (Read error: 104 (Connection reset by peer))
17:23:10lostlogicbarrywardell_: :) I'll be sure to commit some red of my own soon.
17:25:49 Join rzr\ [0] (
17:26:32rzr\Good evening everyone.
17:26:45rzr\Any Sansa experts here? :P
17:28:13rzr\I've been having a problem with playback using rockbox, it's being very unstable and will occasionally freeze entirely, resulting in me having to remove the battery to reboot the player. Anyone having a similar issue?
17:28:26rzr\Sansa e270 btw
17:28:52tuplanollai have, e280
17:29:15tuplanollayou can press the menu button for 15sec to shutdown
17:29:40tuplanollait freezes when i play wma files
17:29:42rzr\that long, ok
17:30:07rzr\Well, mine seems to freeze randomly. Playback will just stop, and I can't start any other file until I've rebooted the player entirely.
17:30:38tuplanollayou can select new track from the db but it just won't play it
17:31:01rzr\yeah, or browse the files directly and start one, but it just won't play
17:31:34n1srzr\: when did you last update?
17:31:35rzr\It's incredibly annoying, I'm forced to reboot the player several times a day, at work
17:31:38rzr\two days ago
17:31:46rzr\the problem was there before that too though
17:32:14n1srzr\: lostlogic whacked a couple bugs affecting sansas yesterday so you might want to try a newer build
17:32:55rzr\I can try
17:33:06 Join webtaz [0] (
17:33:07tuplanollaimma trying too
17:33:32amiconnbarrywardell: I think that 0x70000084 is actually an enable, so we should add a define for it as well
17:33:56 Join Thundercloud [0] (
17:34:00*amiconn named GPO32_VAL clever enough so the suggested name would be GPO32_ENABLE ;)
17:34:22amiconnOf course we could verify whether it's really an enable
17:35:22rzr\Such a weird bug though, it was the sole reason I updated on friday.
17:35:29barrywardellamiconn: have you verified it?
17:35:42amiconnno, not yet
17:36:06 Quit Arathis (Read error: 110 (Connection timed out))
17:36:17barrywardellI'll try it on my sansa
17:36:22amiconnShould be easy to verify on a G5; when disabling GPO32 bit 30, the HDD should be either always on or always off
17:42:31tuplanollarzr\: yeah, works for me now
17:42:48rzr\I've deleted my old rockbox folder
17:42:52rzr\and installed the new one
17:43:12rzr\so hopefully it'll work continuously now :P
17:44:14 Join nanok [0] (n=nanok@
17:44:27 Nick Arathis2 is now known as Arathis (
17:48:47rzr\hmm, ReplayGain in APEv2 tags are not supported? I just read in the manual that they need to be ID3..
17:48:56 Quit web-taz (Read error: 110 (Connection timed out))
17:49:01rzr\I thought replaygain was APEv2 only?
17:49:26n1srzr\: it depends on which app you use to write them
17:49:55 Join web-taz [0] (
17:50:09 Join webguest78 [0] (i=40db7e82@gateway/web/cgi-irc/
17:50:22rzr\I use foobar for all my music
17:50:42rzr\I make sure to tag my files using both ID3v2 and APEv2
17:51:40n1ssome apps, (i don't know which) are able to write RG info in id3v2 tags, and those tags are what rockbox reads
17:51:45n1s(for mp3)
17:52:52rzr\thats weird then cause replaygain seems to work on my player. I guess foobar writes it to both tagsets then.
17:53:51 Quit barrywardell_ ()
17:55:21rzr\it seems to play fine now. Time will show. :P
17:55:36 Join japc [0] (n=japc@
17:57:02 Part web-taz
17:57:57 Join japc_ [0] (n=japc@
17:58:52 Quit japc_ (Remote closed the connection)
17:59:10 Join japc_ [0] (n=japc@
18:01:29 Join scorche|w [0] (n=42c007b2@rockbox/administrator/scorche)
18:01:41 Join barrywardell_ [0] (
18:03:42 Join pondlife [0] (n=Steve@rockbox/developer/pondlife)
18:04:33pondlifelostlogic: I'm sure you know, but I've just retested and resume is no better since earlier :/
18:04:54 Join Domonoky [0] (
18:06:27 Quit petur ("*plop*")
18:07:09 Quit Siku (Nick collision from services.)
18:07:13 Nick [Sur`DataGhost] is now known as DataGhost (
18:07:15 Join Siku [0] (
18:07:36 Quit webtaz (Read error: 110 (Connection timed out))
18:07:50barrywardellamiconn: yep, it's an enable. the sd_card_mux doesn't work on sansa with the bit disabled
18:09:20 Join BigBambi [0] (n=Alex@rockbox/staff/BigBambi)
18:11:38 Join linuxstb_ [0] (n=linuxstb@rockbox/developer/linuxstb)
18:14:11 Quit japc (Connection timed out)
18:14:14 Join spiorf [0] (
18:16:34 Quit Siku ()
18:19:59Nico_Plostlogic: I'm finally reading your recent commits
18:22:52*BigBambi feels a bit sorry for the unfortunate Nano people
18:23:45 Quit linuxstb (Read error: 113 (No route to host))
18:26:10lostlogicpondlife: aww, I thought what I just committed might have made it better actually.
18:26:15lostlogicNico_P: fun fun
18:26:24Nico_Plostlogic: so much things to read :)
18:26:27 Quit desowin (Read error: 113 (No route to host))
18:26:36pondlifelostlogic: I hoped so too... :/
18:26:58lostlogicI have one queued up which removes conf_watermark because that shouldn't be configurable and appears to have been a hack
18:26:59Nico_Plostlogic: I'm not sure I agree with what you said about having to buffer the whole file then seek for resuming
18:27:11lostlogicNico_P: hrm?
18:27:13lostlogicnot for resume
18:27:21lostlogicor which statement
18:27:33lostlogicoh that's right, I"m supposed to be beating you up.
18:27:38lostlogicfor 8092
18:27:44Nico_Pjust a sec... not sure I understood right too
18:27:49pondlifeThe file should be buffered when I resume after stopping...
18:28:01Nico_P" the read->advance method would force the whole file to end up buffered even when resuming would only require part of it and therebo obviate the need for a reverse seek and also the need for conf_preseek"
18:28:02lostlogicpondlife: huh? stop clears buffer completely
18:28:20 Quit Chronon ("Client exited")
18:28:28lostlogicpondlife: stopping invalidates all tracks
18:28:36Nico_Plostlogic: FS #8092 is why partial tracks are gotten rid of
18:28:38pondlifeAh, ok. But does it have to?
18:28:53Nico_Pbut yeah there are other problems
18:28:55lostlogicpondlife: would be needlessly complex to not do so, imo
18:29:09lostlogicNico_P: partial tracks being gotten rid of does nothign to help 8092
18:29:13Nico_Ppondlife: same as playlist skipping ;)
18:29:20pondlifeBe nice to avoid an un-needed spinup in both cases
18:29:23lostlogicrebuffer_handle cannot work on the backward case.
18:29:34pondlifeThrowing away good data ;p
18:29:39n1spondlife: ever heard of pause? .P
18:29:45lostlogicn1s: beat me to it!!
18:29:50pondlifePause sucks
18:29:50Nico_Plostlogic: yeah but FS #8092 was showing up in a bad way when partial tracks weren't gotten rid of ;)
18:29:54pondlifeVoice doesn't work
18:29:58lostlogicNico_P: ah
18:30:31lostlogicNico_P: so what do you think about doing away with conf_chunksize
18:30:34Nico_Plostlogic: so about read->advance... I'm not sure I see what you mean...
18:30:39lostlogic(I misspoke earlier, conf_watermark is good)
18:30:53pondlifePause and stop should be unified really... and the buffer/playlist could be preserved in both cases.
18:31:02Nico_Plostlogic: then we hardcode the chunksize?
18:31:14lostlogicNico_P: so if a codec calls for a raed from the beginning of a file that is only partially buffered as its first action... it forces the whole file tob e buffered.
18:31:32lostlogicNico_P: yes, absolutely, chunksize is a decision of the buffering code and is nobody else' business
18:31:42Nico_PI quite agree
18:31:46pondlifeDon't those "per-codec" settings make little sense when buffering? I mean, as globals at least..
18:32:19lostlogicNico_P: the other thing that the read, advance method might have been doing now that I think of it is just skipping first_frame_offset worth of good data that should have been played
18:32:24Nico_Plostlogic: I can readd the ata idle callback if you're not working on it... I just need to get acquainted with your changes first
18:32:28pondlifeThey only refer to the current codec using the data, not other buffering, no?
18:32:29lostlogicand now that I think of it, that's why changing it broke pondlife
18:32:43lostlogicNico_P: yeah, I'm not working on that.
18:32:49lostlogicpondlife: exactly.
18:32:58pondlifeIMHO they should be playback, not buffering (or if they must be buffering, per handle)
18:33:03lostlogicpondlife: so if it were to be used it would hafta be done differently and in its current state it's just wrong
18:33:27pondlifeOr even better see if it can all be scrapped and KISSed better ;)
18:33:46Nico_Plostlogic: if you get rid of chunksize, we only have watermark left, right?
18:34:04lostlogicNico_P: yeah, my patch also makes it an explicit call not a conf switch call
18:34:32lostlogic<−− lostlogic not logical :-D
18:34:45pondlifeWatermark could be per-handle if needed....
18:34:54pondlifeThat sounds a sensible thing.
18:35:00Nico_Ppondlife: agreeed. Maybe we could even get rid of it
18:35:24pondlifeWhich codecs use extreme watermarks? FLAC?
18:35:37lostlogicpondlife: I like the per-codec watermark personally
18:36:03pondlifeYep, but it should be per buffer handle... Video may want a higher value...
18:36:33pondlifeI guess it depends on the data rate and how variable it might be.
18:36:34lostlogicpondlife: it is in effect per-buffer handle
18:36:40lostlogicbecause it's configured by the codec as it starts playing a handle
18:37:08pondlifeSo not going to accidentally affect another handle that's filling at the same-ish time?
18:37:26pondlifee.g. playing FLAC while buffering a SID
18:37:32pondlifeor whatever
18:37:47lostlogicpondlife: nothing to do with filling
18:37:52lostlogicpondlife: only to do with emptying
18:37:54 Quit barrywardell_ ()
18:38:00pondlifeOK, that makes sense.
18:38:04lostlogicso the currently _reading_ codec sets it
18:38:16pondlifei.e. buffer -> codec
18:38:20Nico_Plostlogic: so we'd use the base_handle value?
18:38:41lostlogicNico_P: yeah, I mean if we were to add a varialbe for watermark to each handle
18:38:42*pondlife was confused and thought it related more to the trigger point for filling.
18:38:42Nico_Pto know which handle is affected by the setting, I mean
18:38:46lostlogicbut I still don't think that's necessary
18:38:58Nico_Plostlogic: why?
18:39:04lostlogicpondlife: it is the trigger poitn for filling but it's only relevent to the currently playing handle
18:39:16lostlogicNico_P: because it's only relevant to the playing handle
18:39:20lostlogicnto to all handles
18:39:28zichois it possible to have shuffled playback when listening to a dynamic playlist?
18:39:31Nico_Phmm ok I thought it was set for each buffered handle in turn
18:39:32lostlogicand it's just as easy to have the codec say "I'm playing a handle of this watermark"
18:39:33pondlifeWhat if you have 2 consumers using buffering at the same time?
18:39:39 Join barrywardell_ [0] (
18:39:47pondlifeIn the future, I mean
18:39:48lostlogicpondlife: buffering would explode because buf_ridx wuold break.
18:39:58lostlogicpondlife: so we'll have bigger problems than the watermark.
18:40:04Nico_Plostlogic: that's not true
18:40:14Nico_Punless it changed
18:40:26lostlogicNico_P: hmm, your'e right, buf_ridx is only cahnged when first_handle is closed or shrunk
18:40:37 Quit perrikwp___ (Read error: 104 (Connection reset by peer))
18:40:39Nico_Pyes. the reading pointer is maintained per-handle
18:40:46 Quit karashata (Read error: 110 (Connection timed out))
18:40:55Nico_Pso multiple users isn't a problem
18:41:00lostlogicpondlife: ok, yes, then it would be better to have some kind of per-handle setting but the logic for knowing when data is low would get quite complex I think
18:41:04n1szicho: yes, turn on shuffle
18:41:24pondlifelostlogic: I guess you're low when any active handle is low
18:41:29pondlifei.e OR them all.
18:42:05*pondlife eats the last biscuit
18:42:12lostlogicpondlife: so you'd want per-handle checking too *shudder* not today.
18:42:21pondlifeNope, not today :)
18:42:51pondlifeLet's just get it bugless first :p
18:42:53Nico_Plostlogic: in r15466 you added a queue_send of Q_FILL_BUFFER *before* the queue_send of Q_BUFFER_HANDLE in buf_request_buffer_handle... wouldn't *after* make more sense?
18:43:11*Nico_P hoped the code would be bugless on his return from vacation :p
18:43:32pondlifeHah, it's not too bad
18:43:38zichon1s: It just hops to the next track on the last, in order.
18:44:01pondlifeCertainly seems that some ancient playback bugs may have vanished
18:44:34n1szicho: yes, turning on shuffle will reorder the playlist, it always plays in order, htere is also a "reshuffle" option in the wps context menu
18:44:41 Quit barrywardell_ ()
18:44:43lostlogicNico_P: yeah, that's a little broken, I just realized I had it fixed proper-like in a pending fix, but I'm trying to separate my commits into good atomic units before proceeding
18:44:56Nico_Plostlogic: I use got for that ;)
18:44:58zichon1s: oh.
18:44:59Nico_Por git
18:45:18lostlogicNico_P: yeah, hehe
18:45:27pondlifeget git
18:45:36lostlogicbut the start_fill _must_ go before the buffer_handle to start making space for the handle to buffer
18:45:44lostlogicotherwise it'll just be bumping against data already ont he buffer
18:46:29lostlogichaah, doesn't exist.. fill_buffer
18:46:42rzr\context menu button on the sansa seems to be changed or non.functional in the newest build
18:46:57lostlogicthe ponti is that a BUFFER_HANDLE without first _at least_ shrinking handles is pointless
18:47:08lostlogicbecause the buffer was most likely filled completely 30 megs ago and hasn't been emptied at alls inced
18:47:14lostlogicall since
18:47:33Nico_Pyeah but I think it makes more sense to give prio to the handle that needs it
18:47:57Nico_Pso maybe add a shrink_buffer call when Q_BUFFER_HANDLE is received
18:48:14Nico_Pbut I'm still reading through the changed code ;)
18:48:50lostlogicNico_P: exactly what the pending patch does :)
18:49:15lostlogicNico_P: turns out we need two cases for buffer_handle 1) a case that should initiate a new fill and 2) a case that is called during a fill by a new bufopen or suchlike
18:49:22lostlogicso I'm separating them out
18:49:23 Join DerPapst [0] (
18:49:25 Join karashata [0] (
18:50:26n1srzr\: yes it changed to long select
18:53:49lostlogicNico_P: so now I need you (or me) to figure out how to correctly trigger a complete rebuffer on a reverse seek off-buffer.
18:54:32lostlogicpondlife: are you sure that resume is still broken? because if its still broken then the resume you're dealing with is not being handled well even for the known bugs.
18:54:50pondlifeI'll try again if you like
18:55:11pondlifelostlogic: Which OS are you using?
18:55:34Nico_Plostlogic: I just tried and it's broken
18:55:45lostlogicpondlife: broken on target too?
18:55:49Nico_PI'll try to find out why
18:55:51lostlogicpondlife: cuz it definitely works on target for me
18:56:00pondlifeNot sure, but the sim is knacked
18:56:07lostlogicfsck the sim :-P
18:56:16pondlifeHah, you should use it more often
18:56:34lostlogicpondlife: I develop on a leased server in a data center, copying builds to one of several computers that I am actually typing through, so I have no way to build or run sims for my terminals.
18:56:43pondlifeNormally the sim works great and target is fscked
18:56:48nanokuhmm.. sorry to spoil the party, and all
18:56:56lostlogicnanok: what'd I break?
18:57:18nanoki notice the "context menu" on the sansa e200 target, on the current build, acts like "back" in the menus
18:57:21nanoklostlogic: :)
18:57:38n1snanok: the keymap changed a little
18:57:45nanoki see. i was about to ask if it;s normal. i take it it _is_ indeed normal
18:57:52nanokaahm, i see
18:58:05nanokn1s: any link to the changes?
18:58:15 Quit idnar (Nick collision from services.)
18:58:17 Join idnar_ [0] (i=mithrand@unaffiliated/idnar)
18:58:41*pondlife reboots his H340...
18:58:42n1snanok: it's in the svn log, try the "last 4 weeks" link on the bottom of the frontpage
18:59:01lostlogic*laugh* would be on the front page if I wasn't committing like 123 times a day
18:59:10lostlogicI should ease up on the rate of change.
18:59:15pondlifeYes, Sunday was an epic :)
18:59:25pondlifeIt's broken on target too
18:59:36lostlogicpondlife: *frown* why it works for me?
18:59:41barrywardellI'm not convinced I like that change. The power button does have "menu" written under it. It seems the obvious choice to bring up the menu
18:59:56n1slostlogic: at least many small commits are better than one huge that basically rewrites several files :)
19:00:17lostlogicn1s: yep, I'm trying to keep my commits to single functional ideas
19:00:37lostlogicpondlife: yeah, just booted up into a resume on an mp3 and it worked like a charm... *frown*
19:00:44pondlifelostlogic: Recipe is simple. Play a track halfway through. Press STOP to go to browser (not pause). Press PLAY to resume. WPS progress continues, but audio starts from the start
19:00:50pondlifeNot a boot-up resume, a STOP/PLAY one
19:01:05Nico_Plostlogic: sill broken on sim
19:01:40pondlifeSame thing happens on both... Do you know what the value passed in for resume position is?
19:01:56pondlifeIt seems to be resuming after the ID3 data, not at the last played position
19:02:04lostlogicpondlife: ahh, ok, I get it now... yes, resets playing position to beginning even though wps is where it thought it should be
19:02:08 Nick advcomp2019_ is now known as advcomp2019 (n=advcomp2@unaffiliated/advcomp2019)
19:02:15lostlogicpondlife: ok, this I can work on.
19:02:24pondlifeIf you let it play past end of track, things go horrible
19:02:31*nanok is safe from this, he didn;t figure out yet how to _stop_ not pause the sansa
19:02:49 Join Frazz [0] (
19:02:53pondlifeUse voice, then you'll learn not to pause :)
19:03:09lostlogicpondlife: codec bug.
19:03:17 Join bertrik [0] (
19:03:18pondlifeUsed to work :)
19:03:35lostlogicpondlife: the codec doesn't opbey the offset so the old way was actually skipping the first_frame_offset first bytes at a resume
19:03:37n1shaha lovely we now get even more gcc/server specific warnings... barrywardell's server spits out warnings in libmad/synth.c that no other server cares about (sims)
19:03:40lostlogicbut that wasn't very noticible most of the time
19:03:59lostlogicwill fix in codec.
19:04:18pondlifeWill test at some point
19:04:19Nico_Phaha I love when new code reveals old bugs
19:04:52Nico_Pamiconn: did I say something wrong?
19:04:53amiconnHDD poweroff didn't increase runtime noticeably on mini G2 :|
19:05:42pondlifeMaybe MoB is decreasing runtime in a proportional manner ;/
19:05:44amiconnProbably these microdrives do autosleep much like their solid-state twins
19:05:52n1sshould we perhaps introduce required versions of gcc/binutils for our build servers for their native arch so these things at least get consistent?
19:06:18pondlifen1s: Yes. Make the buggy gccs illegal.
19:06:31amiconnn1s: I don't think that's possible, with the mix of architectures and distributions
19:06:33barrywardelln1s: probably because it's running gcc 4.1.1? maybe I should update it?
19:06:51n1spondlife: that wont work, we couldn't use gcc at all then ;P
19:07:02pondlife4.2.1 and 4.2.3 seem to give more warnings than 4.1.1...
19:07:07pondlifen1s: True :)
19:07:33n1syep 4.2 has anoter bug that spits out most of the warnings on the build table...
19:08:05pondlifeCan't the oggmalloc warning be fixed in code?
19:08:31pondlifeI don't have the ability to build 64 bit sims.
19:08:31n1sbarrywardell: 4.1.2 seems ok...
19:08:47nanok15 minutes lostlogic Create two separate ways for a handle buffer request to be sent. One is when th...
19:08:51pondlifeOr 4.1.3
19:08:52nanoklostlogic: you bastard...
19:08:58barrywardelln1s: I can upgrade to that
19:09:32amiconnn1s: The warnings in cfft.c are in fact very odd: No matter how you try to cast, the warnings stay - *and* they occur consistently with all gcc 4.2.x
19:09:44pondlifeThat's a gcc bug, no?
19:09:48amiconnEven across architectures (x86 and x86_64)
19:09:48nanoklostlogic: if you keep it up, the repo server will have to use lvm snapshots to backup the svn and/or build the nightly builds
19:09:58 Join ilgufo [0] (
19:10:19amiconnpondlife: That's the question...
19:10:28lostlogicwhy am I bastard? I 'm having parsing difficulty
19:10:36n1samiconn: yes I have seen your comments on it, and those warnings are not there in the trsts with gcc 4.3 I have made but those were only m68k-elf builds
19:10:48pondlifeHmm, has the build hung, or is my patience wearing out, or my internal clock wrong?
19:11:18pondlifeAha, one of the latter 2
19:11:38pondlifelostlogic: Nice deltas btw
19:11:47lostlogicpondlife: my favorite hobby.
19:12:30pondlifeI shaved some a few bytes off playback.c earlier, but I'll wait until you're done before I change anything...
19:12:49lostlogicdoes shorten support resume?
19:12:58pondlifeYes, AFAIK
19:13:02pondlifeIt should
19:13:38pondlifeAh, no
19:13:38nanoklostlogic: uhm, it's supposed to be kind of a compliment :)
19:13:53pondlifeAt least according to the wiki - it doesn't have the seek box ticked
19:14:11pondlife"Decoder from ffmpeg project in SVN. No seeking support."
19:14:34lostlogicyeah,I dn't see it in the codec either :)
19:14:43lostlogicglad I'm not negatively hallucinating
19:15:12***Saving seen data "./dancer.seen"
19:15:28pondlifelostlogic: Mind if I commit a tiny playback hack?
19:15:35pondlifeNo logic change
19:15:55pondlifeJust an attempt to shave 4 bytes off
19:16:28lostlogicpondlife: I'm not in playback.c
19:16:41pondlifeSo your mod is just a codec fix
19:16:52 Join Bagder_ [0] (
19:16:53lostlogicfix for mp3 resume committed untested (because I'm that confident that it's not worse)
19:17:05*Nico_P will test
19:17:25*pondlife too
19:17:40*lostlogic too
19:18:05Nico_Plostlogic: the validity of a null HID causes a bunch of buclose(0) calls on playback stop
19:18:10lostlogicI hate windows. it takes lik 4x as long toe xtract a build from windows as from linux
19:18:13Nico_Pall the null track info structs...
19:18:28lostlogicNico_P: I know, I've been trying to decide if I care though ;)
19:18:51Nico_Pit's annoying with logf enabled and it feels inefficient
19:18:53pondlifeWhy do we start at 0, not 1?
19:19:00Nico_Ppondlife: we used to start at 1
19:19:06pondlifeI know
19:19:10pondlifePOSIX thing, is it?>
19:19:27nanokuhm.. i seem to be incapabale of finding the damn changelog. however, i cannot, for the life of me, figure out what might have replaced the "down" button in the function of "context menu"?
19:19:31*nanok is puzzled
19:19:35 Join bluey [0] (
19:19:54pondlifenanok: It's LONG SELECT on most targets
19:20:02nanokpondlife: AAAAAHM
19:20:10lostlogicugh, now there seems to be a problem where it resumes to the right spot but then doesn't continue playing until full buffer fill is done
19:20:12nanokpondlife: thanks, sorry
19:20:34pondlifeSo, you fixed one bug, but that revealed another, again ;)
19:21:16pondlifeThere's a glitch on resume now.
19:21:28pondlifeA bit of old PCM?
19:21:41pondlifeOr ID3 as audio?
19:21:52lostlogicpondlife: hmm, I haven't heard that, but maybe
19:22:11pondlifeFairly obvious in quiet passages
19:22:58pondlifeMight be voice related... that's MP3 of course.
19:24:42pondlifePerhaps it's my playback mod... will try clean
19:25:18lostlogicthere's definitely a problem with mp3 startup now.
19:25:24lostlogictrying to work out what it is.
19:25:56 Part blufox_ ("Leaving")
19:26:04lostlogicNico_P: btw, that 'problem' only happens on first lpayback start, so it's really no big deal (the bufclose(0))
19:26:14lostlogicNico_P: could be solved by a few LOC in audio_init
19:26:25lostlogiclike I said, just deciding if I care
19:28:12nanoki've noticed that the discharge curve from about 30 percent on, on the sansa (as reported by the battery bench) is very steap, more so than the rest
19:28:56nanokcould this be addressed in an easy way, is there allready a way to provide a.. uhm.. function for the way the battery dischrges on a particular target?
19:30:11lostlogicpondlife: ugh, now I can't reproduce the problem I was having with resuming... how is yours?
19:30:22 Join mirak [0] (
19:30:26pondlifeJust rebuilding clean
19:31:25 Quit webguest78 ("CGI:IRC (EOF)")
19:33:03 Quit Bagder (Read error: 110 (Connection timed out))
19:33:07 Join Alonea [0] (
19:33:25AloneaI think there is something wrong with mpegplayer. The video stopped moving after about 1.5-2 minutes, but sound continues. lemme try another video. Dithering and start menu are turned off.
19:35:05pondlifelostlogic: It's glitching slightly
19:36:34bertriknanok: yesterday my sansa wouldn't boot at 3.4V, while rockbox still played happily just before the reboot
19:36:36 Join cheesenuggett [0] (
19:36:50bertrikso the low battery threshold may be a little too low on the sansa e200 series
19:37:01Aloneaok, looks like its just that one. probably got corrupted somehow.
19:37:26pondlifelostlogic: Seems ok now I've enabled crossfade again.. maybe I'm just used to crossfade ;)
19:37:43lostlogicpondlife: glitching -> playing bad data as it resumes still?
19:38:03lostlogicpondlife: grr, why is the buffer fill process not rapidly satisfying the data needed by a resuming mp3?
19:38:12nanokbertrik: aham, that might be. i know usually liion are supposed to be quite flat, untill they get to the bottom, after which they become steap
19:38:24 Part cheesenuggett
19:38:56pondlifelostlogic: Hmm, the buffering should easily be able to beat the playback in a race
19:39:15Aloneaok, thats odd...another file, stops at same place.
19:39:23 Quit karashata ("Leaving.")
19:39:33nanokbertrik: who refused to boot? the bootloader, saying "battery is too low"?
19:39:39bertriknanok: yes
19:39:41lostlogicpondlife: yeah, I know, I'm going to get a good solid logf of the problem and stufy it.
19:39:43 Join karashata [0] (
19:39:44lostlogicstudy even
19:39:50nanokbertrik: or did it just refuse to boot alltogether..
19:39:51pondlifeGood luck.
19:39:54pondlifeI gotta go
19:40:01 Quit pondlife ("Read error: 110 (Connection slimed out)")
19:40:01lostlogicsee ya!
19:40:08nanokbertrik: ah, okay, than it's not _that_ bad
19:40:15bertrikI got the "battery is too low" message.
19:40:47nanokbertrik: but i do agree, it seems the last "20%" or so only means about 10 minutes of playing or something like that
19:41:05 Join ambiotic [0] (
19:41:21nanoki will run with the battery bench from now on, see if we can make out what would be best to shave off, without loosing a significant part of battery lifetime
19:41:55bertrikI wonder if it would be possible to predict battery level by counting time and estimated current instead of using voltage alone
19:42:15lostlogicbertrik: some of the older targets do that
19:42:22lostlogicI think
19:42:25lostlogicat least for charging
19:42:26bertrikor maybe combine them in a kalman filter, but I think we would need more information for that
19:42:34nanokbertrik: you mean like checking the steepness of the curve, localy, or the discharge rate, so to speak?
19:42:53Aloneaok, so 3/4 files I have tried so far stop at the same place. the 3 files that failed are full screen ratio and are about 45 minutes long. The one that didn't fail is widescreen and is about 5-6 minutes.
19:43:06bertrikestimate the discharge rate from processor usage, LCD brightness, etc
19:43:18Aloneahas anyone else had any problems playing video?
19:44:31 Join przemhb [0] (
19:44:46nanokbertrik: that sounds like a lot of work, and a bit indirect. isn't there a way to check the actual consumption (the battery/power circuitry should have this built in, so one doesn;t have to "guess it" based on what one consumes, but read it directly at the source, as it actually is)?
19:45:05 Join Lear [0] (i=chatzill@rockbox/developer/lear)
19:45:44bertrikI wish there was
19:46:09 Join petur [0] (n=petur@rockbox/developer/petur)
19:46:32bertrikthe battery itself has three terminals, but I guess that's just +, - and thermistor
19:47:34 Quit japc_ (Remote closed the connection)
19:47:42 Join merbanan [0] (n=banan@
19:49:12nanokbertrik: hmm, i see..
19:49:25n1sAlonea: I would guess on something happening when it tries to rebuffer
19:50:32nanoki will have to look at the battery bench results, but i have a hunch a "healthy" battery should have quite a consistent trend, untill it gets almost empty (at which point the steepness changes fast and we should shutdown)
19:50:55nanokmaybe this would be easyer to implement, combined with some voltage threshold, just to be safe?
19:51:13Aloneagonna see how file plays on computer. anyway, LOTR and everything else seems to be playing fine, but my ally mcbeal stops as soon as the opening song starts (intro video then it goes to the opening credits)
19:51:18 Part agm3nt
19:52:13n1sAlonea: it the stopping ones are from the same source it's probably something about them that mpegplayer don't like :-9
19:52:48nanokso basically, the second derivative should be the thing to watch for (the "delta_delta_voltage") −− is that the correct english word?
19:53:43bertrikthe second derivative is very sensitive to noise i think
19:54:06Alonean1s: ok, thats odd...I guess it is the file. WMP played up until the opening credits, skipped over them and started right after them. winamp was playing the video upside down...O.o
19:54:26*lostlogic high fives Nico_P
19:54:38n1sAlonea: seems safe to say the file is bad then ;)
19:54:41Nico_Plostlogic: ?
19:54:46nanokbertrik: hm, might be, but we don't need it for fractions of a second, "we don;t have to be that neat"
19:54:55lostlogicNico_P: MAX_TRACK vs MAX_TRACKS, I'm a little retarded ;)
19:55:08nanokif we do a rough calculation for every few seconds, with a decently safe threshold, it should be enough
19:55:09Nico_Plostlogic: I think it was my bug
19:55:13Alonean1s: i guess so, though its odd its only the ally files are like this. and they used to work just fine.
19:56:01 Join bluey_ [0] (
19:56:02nanokbertrik: i guess the neat term for this would be some sort of "dampening"
19:56:27Nico_Plostlogic: yep, annotate confirms
19:56:36lostlogicNico_P: hehe :)
19:57:09Nico_PI'm not sure how bad that bug was
19:57:24bertriknanon: I haven't really looked into it, but how about a simple piecewise linear approximation?
19:57:38lostlogicNico_P: me either :-\
19:57:47nanokbertrik: that even might work :)
19:59:36nanokbertrik: i will use my player with batt bench from now on, so there is something to start with. it will be worthless for benchmarks, as i cannot yet find the heart to give up usage of it and use the non-rockbox one i still have (not for long though). it's like trying to get from a car back to crawling, although crawling does get you there also, eventually..
19:59:42Nico_Plostlogic: the tracks *should* have been cleared by audio_clear_track_entries in most cases, so I think it wasn't too bad
20:00:02lostlogicNico_P: yeah, that's right, I commented ont he likely not-needed-ness of the release_tracks
20:00:43Nico_Pyeah the two need to be unified... audio_release_tracks was some kind of a security measure because I wasn't sure I fully understood audio_clear_track_entries
20:00:54lostlogicNico_P: any new thoughts on how to kill stupid trailing tags?
20:01:13lostlogic(that and the rebuffer_handle) are things I don't have ideas for fixing well
20:01:19Nico_Pdo they still break gapless playback.
20:01:33bertriknanok: by the way, what voltage do you see just after the OF charged the battery?
20:01:40Nico_Pmaybe get_metadata could somehow tell us how much of the file is really needed
20:02:05lostlogicNico_P: yeah, they do still
20:02:22bertriklast time I saw only slightly less than 4.0V, while LiIon should be able to go upto 4.2V
20:02:33lostlogicNico_P: yeah, I was thinking that but would hafta pass an additional parameter to bufopen and get_metadata would hafta seek+read at end of file, so it's an ugly solution.
20:02:46nanokbertrik: hm, i didn't check that yet
20:03:01bertrikI'd like to work on battery charging for the sansa e200
20:03:04nanokit goes down very quick after booting up, so i would have to do it fast to see it
20:03:13 Join jsor [0] (
20:04:08amiconnbertrik: There are 2 different LiIon systems. Most go up to 4.2V, some only up to 4.1V.
20:04:47bertrikamiconn: any idea which one is in the sansa, or how to find that out?
20:05:36lostlogicI think I'm done breaking things for now...
20:05:48bertrikI read that charging up to 4.1V is safer than 4.2V with only small impact on charged capacity
20:06:18lostlogicNico_P: If you disagree iwth my last commit (or it causes trouble) revert it... I don't like the idea of starting buffering a file that is > chunk on open, but I'm a bit of two minds.
20:06:40nanokbertrik: yes, and overcharging liion can be.. catastrophic :)
20:06:53Nico_Plostlogic: r15486?
20:07:10bertrikso, any opinion on a safe voltage level and safe max charging current?
20:07:19nanokbertrik: i might try to get a multimeter and actually measure the battery after charging, without booting
20:08:24nanokbertrik: i am no expert in liion, but i seem to remeber liion is not charged at constant current, but rather in pulses
20:08:30amiconnlostlogic: The code police has something against your handle init (variable declaration mixed with code)
20:08:40nanokit is quite a tricky business, charging liion, afaik..
20:08:57lostlogicamiconn: where which?
20:09:22amiconnThe 'int i;'
20:09:39bertrikI'm using this as a reference:
20:09:39*lostlogic didn't do it!
20:09:46*Nico_P confesses
20:10:25Nico_Pamiconn: isn't that accepted?
20:10:34Nico_PIMHO it's also much more readable
20:10:53 Quit bluey (Read error: 110 (Connection timed out))
20:11:03lostlogicNico_P: coding standards are no variable declaratiosn mixed with code... I noticed quite a few in buffering.c and started moving some of them out to the beginnings of functiosn as I was working.
20:11:05Nico_PIf I could I would've written "for(int i = 0; i < MAX_TRACK; i++)"
20:11:16*amiconn disagrees regarding readability
20:12:49lostlogic(I personally don't care one way or the other and used to argue in favor of mixed a couple years ago)
20:13:02Nico_Plostlogic: isn't this a somewhat obsolete standard?
20:13:30amiconnHmm. It's no longer in CONTRIBUTING
20:13:42Nico_Pamiconn: yeah, I just looked
20:13:51*amiconn wonders why
20:14:02Nico_Plostlogic: I also just got an audio "glitch" like pondlife reported on playing an OGG after having played an MP3 which finished
20:14:03lostlogicamiconn: I seem to recall it being brought up a couple of months ago
20:14:25Nico_Pthe last half sec of the MP3 played before the OGG started
20:14:33 Join jhulst [0] (n=jhulst@unaffiliated/jhulst)
20:14:40amiconnI know that mixed declarations and code are allowed in C99, but I still don't like it
20:14:50lostlogicNico_P: in continuous playback or after a stop?
20:15:05 Quit ilgufo (Remote closed the connection)
20:15:10Nico_Plostlogic: the playlist finished on the MP3, then I started a new one on the OGG
20:15:11amiconnBtw, the rule (that seems to be missing now) was not set by me, but I agreed with it 100%...
20:15:44lostlogicNico_P: odd, what target are you? gigabeat? I am not seeing this one
20:15:45amiconnbarrywardell: Around?
20:15:51 Join darkless [0] (
20:15:52Nico_Plostlogic: yeah, gigabeat
20:16:48 Nick idnar_ is now known as idnar (i=mithrand@unaffiliated/idnar)
20:16:55Nico_Plostlogic: it seems to be reproduceable... make a one file playlist with an MP3... let it finish (or seek forward to just before the end) and then start a new playlist
20:17:14 Quit Lear ("ChatZilla [Firefox]")
20:17:22lostlogicNico_P: hmm. I have no idea what I wuold have changed that would cause that
20:17:54Nico_Pjust tried with an mp3 as new playlist and it did the same
20:18:03*Nico_P could bisect using git
20:18:08bertrikwhat do you mean by glitch?
20:18:41Nico_Pbertrik: last split seconds of the previous track being played before the new one starts
20:22:14barrywardellamiconn: yes
20:23:17amiconnbarrywardell: What happened when you disabled the card select GPO in GPO32_ENABLE? Was it permanently set to internal or to microsd?
20:24:10 Join ilgufo [0] (
20:25:09barrywardellamiconn: I didn't check that much. Just the sdcard never appeared when it was inserted, so I guess it was permanently internal
20:25:48barrywardellyou're wondering if disabling sets it to zero or disables changes to the value?
20:25:55amiconnI'm asking because microsd doesn't work yet on c200
20:26:12amiconnMaybe the selector isn't enabled
20:26:19amiconnBut I just spotted sth in the driver
20:26:38*amiconn can't test; has to wait until pixelma returns
20:29:26amiconnsvn blames jhMikeS
20:32:01 Join lazka [0] (
20:32:04*barrywardell wonders what the problem is
20:32:44amiconnBoth ata_read_sectors() and ata_write_sectors() check for card presence directly instead of using the appropriate function
20:32:47nanokbertrik: reading that link. very nice. one question pops up: can we read the thermistor, on the sansa or any of the liion targets, or is it only used by the internal battery circuitry?
20:33:01amiconnBut both GPIO pin and polarity are different on c200...
20:33:24bertriknanok: i think we can read it from the 10-bit ADC in the AS3514
20:33:41bertrikbut it only shows a meaningful value when the charger is actually connected
20:34:00nanokbertrik: i;ll take that as "i think so, yes". the internal of the sansa are way beyond me at this time ;)
20:34:01bertriki can see the value go down if i warm the sansa up in my hands
20:34:03barrywardellah, so it detects, but can't read/write it
20:35:04 Join J3TC- [0] (
20:35:15 Join saratoga [0] (i=9803c6dd@gateway/web/cgi-irc/
20:35:31nanokbertrik: it would be interesting for charging checks (temperature is critical for liion, allthough not as much as i thought, nor in the way i thought it to be, according to that link), and to try and calibrate (and "live-calibrate" ) the battery gauge accordingly
20:35:54nanokdepending on temperature (that's a bit niftyer than it needs to be, i guess, at least for now)
20:36:18saratogaamiconn: are all the DEV_EN 1+2 bit writeable?
20:36:29bertriki think the battery charging logic already has high battery temperature detection built-in
20:36:44DerPapstThe DEV_EN value on my 5.5G doesn't match the one on the wiki but it matches with the DEV_EN value for the iPod color listed on the wiki page :S
20:36:53saratogai've got my sansa hooked up to a anmeter and was debating just flipping each one in sequence and measuring the current draw and if the device works
20:37:24amiconnLiIon charging circuits always do the critical things in hardware. Software controlled LiIon charging would be dangerous
20:37:29lostlogichmm... we still need to figure out how to draw 500mA from USB. the iPod video still discharges slowly when backlight is on and playing while USB'd
20:37:45bertriknanok: I don't think the battery is any real use except as an emergency charge shutdown signal
20:37:52amiconnsaratoga: All which are '1' should be disableable.
20:38:13saratogais it a safe assumption that "0" is always off?
20:38:45amiconnThat said, some not that criticall things are sometimes software controlable, but with a hardware safety fallback
20:38:50barrywardelllostlogic: that will happen once Zagor commits the USB stuff. the controller needs to request 500mA
20:38:52amiconnsaratoga: yes, '0' is off
20:39:24amiconnsaratoga: That test would have to be done on each different PP502x target individually though
20:39:49amiconnAlso compare what ipl already knows about DEV_EN
20:40:11lostlogicbarrywardell: ooh, there's progress happening in the area, taht's very nice to hear!
20:40:31amiconn (device enable, that's DEV_EN1 only)
20:40:50DerPapstIs the DEV_EN the same on every model or can there be different ones for the same model?
20:41:36nanokbarrywardell: !!! reeeaaaally good news
20:42:05*nanok is really pissed about not being able to at least charge the sansa while hooked on, and having to go to the OF
20:42:15amiconnDerPapst: DEV_EN is the same for all PP502x, but not all targets need the same functional units
20:42:20 Join bertrik_ [0] (
20:42:30nanokwhen i see that backlight on all the time, it pisses me off.. :/
20:42:34amiconnE.g the ipods need the opto controller for the wheel, *except* the mini g1
20:42:42amiconnOther PP502x targets probably don't need it
20:43:28amiconnsaratoga: I *think* that it might be possible to disable the ata controller while not in use, without having to reinit at re-enable
20:43:44saratogaamiconn: well i've only got a Sansa to play with so my hacking is somewhat limited
20:43:46amiconnThat is, if the DEV_EN bits just control clock gates as I presume
20:44:02saratogastill, there a half dozen on bits for the Sansa that we don't have a label for
20:44:20saratogai see the USB bit is on, might that be turned off since we have no stack anyway?
20:44:35amiconnThere are at least 64 DEV_EN bits. Perhaps we should make a list which bits are set on which targets?
20:44:48saratogaamiconn: theres one on the wiki page
20:44:54saratogabarry made it this morning
20:45:03amiconnNo, we need the usb controller for usb detection
20:45:16saratogawhat good does that do?
20:45:29DerPapstamiconn: ah ok. that explains why my DEV_EN differs from the one in the rockbox wiki
20:45:44saratogaat least on the sansa USB insert just freezes the player until you remove the USB plug
20:45:47barrywardellamiconn: I saw in the rockbox comments that DEV_EN 0x2 controls device clocking enable
20:45:52 Join The-Compiler [0] (
20:46:15amiconnsaratoga: Then something is wrong. It should reboot into the OF for USB access
20:46:38The-CompilerSomeone who speaks German here? I'd like to do an interview for my podcast about rockbox (
20:46:44amiconnIt is known that this doesn't always work. It's build dependent; the next build might work
20:46:54saratogai don't think that was ever implemented on the sansa
20:47:02bertrik_amiconn: that's a nice idea
20:47:13scorche|wbarrywardell: oh!...i should probably attach my results of my testing to that page...
20:47:25 Quit karashata ("Leaving.")
20:47:30amiconnbarrywardell: Thanks, will add values for the devices I have access to
20:47:34saratogaah got to run, will think about this more
20:47:50 Join karashata [0] (
20:48:21amiconnbarrywardell: Bit 26 being LCD is probably a bit too generic; there are 2 different LCD bridges in the PP502x
20:48:36 Join Lear [0] (i=chatzill@rockbox/developer/lear)
20:48:53scorche|wamiconn: would you like the results of that test as well? (just went through and tried every single bit and saw what broke)
20:49:17barrywardellyeah, I just took that stuff from pp5020.h. the ipodlinux page suggests there's a lcd bit in DEV_EN2 as well
20:49:53 Quit rzr\ ("Generic quit message...")
20:52:01 Part jsor
20:53:21amiconnOkay, Mini G2 and small H10 added.
20:53:22 Join darkapostrophe [0] (
20:53:38amiconnG5.5 is identical to the one already in the list
20:53:38 Join scorche` [0] (i=Blah@rockbox/administrator/scorche)
20:54:33barrywardellDEV_EN3 doesn't look like it varies at all
20:55:14 Quit karashata ("Leaving.")
20:55:57 Join karashata [0] (
20:56:21Nico_Plostlogic: using the ata idle callbacks, should I add a new queue event or do you think Q_START_FILL is OK?
20:56:53DerPapstamiconn: shall i add my G5.5? Dev_EN is different (the same as ipod color in that list)
20:56:59 Quit bertrik (Read error: 110 (Connection timed out))
20:57:47lostlogicNico_P: Q_START_FILL would need to be modified to not call buffer_handle if is < 0
20:57:51lostlogic(when cast to an int)
20:58:11lostlogicNico_P: actually...
20:58:27lostlogicata_idle-callback won't work in tht way because as soon as it returns ata shuts down the disk.
20:58:31lostlogicamiconn: back me up there?
20:58:32Nico_Poh yeah it doesn't call fill_buffer anymore
20:58:35lostlogicI think we'll need a new way
20:58:57 Quit atsea- (Read error: 104 (Connection reset by peer))
20:58:59amiconnbarrywardell: Yeah, it *might* be something different. Here's what makes me think 0x60006044 is DEV_EN3:
20:59:01lostlogicNico_P: but it could set filling to true which would cause a call to fill_buffer, or it oculd explicitly call it in this case
20:59:31amiconnA selector, a jump table, and lots of stubs fiddling with either 0x6000600c, 0x60006010, or 0x60006044
20:59:37amiconnThis is from the G5.5 ROM
20:59:41Nico_Plostlogic: have your commits solved the sansa problems?
20:59:50lostlogicNico_P: 'far as I know yeah.
20:59:57lostlogicbertrik_: how's playback/buffering for you?
21:00:11 Quit karashata (Client Quit)
21:00:16bertrik_lostlogic: I need to install todays SVN yet
21:00:42 Join karashata [0] (
21:00:48Nico_Plostlogic: "BUF_USED < buffer_len" is probably true always
21:02:19bertrik_hmmm, playback dropped out while i was looking at the buffering debug screen
21:02:36bertrik_pcm was struggling to keep full and eventually touched left side
21:03:45 Quit Alonea (Read error: 110 (Connection timed out))
21:04:12barrywardellamiconn: I see. It could still be a DEV_EN, but with the same things enabled for all targets.
21:04:46lostlogicNico_P: that's a good thing :)
21:05:11 Join ompaul [0] (n=ompaul@gnewsense/friend/ompaul)
21:05:23lostlogicNico_P: the concensus was to do some filling any time the queue is idle and the disk is on
21:05:24Nico_Plostlogic: I mean even if the buffer is as good as full, the condition probably won't be true
21:05:34lostlogicNico_P: I know
21:05:54Nico_Pthen it's useless! :p
21:06:40 Quit Dark_Apostrophe (Connection timed out)
21:06:40 Nick darkapostrophe is now known as Dark_Apostrophe (
21:06:41 Join TMM [0] (
21:06:50 Join linuxstb [0] (n=chatzill@rockbox/developer/linuxstb)
21:06:55lostlogicNico_P: maybe should not enter that block if(filling)
21:07:45lostlogicor maybe that should not be where it is and should only be explicitly triggered by ata going idle and not actually set the filling state but just do a single round of filling
21:07:49lostlogicthat might make more sense
21:08:21lostlogicstill ata_idle_callback is not an OK way to do a buffer fill for the raeson I mentioned earlier... we'd need to register for and respond to a sys message instead
21:08:26lostlogicamiconn: lil' help?
21:08:47 Quit scorche (Connection timed out)
21:08:47 Nick scorche` is now known as scorche (i=Blah@rockbox/administrator/scorche)
21:09:14Nico_Pyeah that's why I originally got rid of the idle callback but I was told polling ata_disk_is_active was wrong
21:09:26lostlogicNico_P: yeah, I know, which is why I #if 0'd it
21:09:38lostlogicwhich is why we need a new mechanism, which amiconn has decribed but which does not currently exist
21:09:57Nico_Poh, where is the description?
21:09:59*Nico_P didn't catch up on the logs
21:10:13lostlogicamiconn: so you speak of sys messages and registering/resopnding to them as an alternative to ata_idle_callback
21:10:31amiconnIt would replace the callback
21:10:34lostlogicunfortunately I have _absolutely_ no idea where we would start to implement that, could you by chance give a more detailed description of this idea?
21:11:22amiconnWell, we currently have sys messages, which are broadcast to all public event queues
21:12:05amiconnAnd right now we have 2 special sys messages, which *must* be replied by *all* threads with a public queues - usb connect and usb disconnect
21:12:05Nico_Pso I can drop using the callback?
21:12:51amiconnThat means there are several threads which need to reply to those messages just because they have public queues, even though they might not have to do anything special in the usb case
21:12:54lostlogicNico_P: if we use the callback, the ata thread would hafta be _blocked_ for the duration of the fill, not acceptable, imo
21:13:39amiconnAnd still, the usb handling is a bit messy, because the ata thread would need special handling, and it doesn't
21:14:28nanokhmmm.. have been running todays build for a litle while now
21:14:32amiconnIf we detect usb connection, *first* all threads which have something to flush have to do that, *then* ata would have to disable the disk
21:14:50amiconnCurrently, this disabling is done by the usb thread - somewhat messy
21:14:58nanoki don't seem to get that bug with the playback stopping, but it *seems* to have been replaced by skipping instead of stopping
21:15:10amiconnThe same goes for shutdown, which is a complete mess atm because it's all done in the ui thread
21:15:14***Saving seen data "./dancer.seen"
21:15:23nanokis this consistent to anybody elses observations?
21:15:24Nico_Pnanok: "today's build" isn't good enough... a revision is needed
21:15:38nanokNico_P: good point, sorry
21:15:41amiconnThis won't work for suspend, because suspend has to make sure all threads are ready for suspend *without* killing the threads
21:15:50*nanok forgets timezones are.
21:16:23lostlogicnanok: update, will probably go away
21:16:25amiconnSo my idea was to introduce a mechanism to register for certain system events (those which could require synchronised action - usb, shutdown/suspend)
21:16:58amiconnAnd if we have that, we could add an event for catching a 'disk spinning' opportunity
21:17:01 Quit ilgufo ("So Long, and Thanks For All the Fish -")
21:17:29 Quit The-Compiler ("Verlassend")
21:18:06bertrik_nanok: I just got a pcm buffer underrun with SVN 15486
21:18:15lostlogicamiconn: I definitely like the idea of not automagically registering queues for system events... removes the confusing distinction between public and private, right?
21:18:21 Nick bertrik_ is now known as bertrik (
21:18:30bertrikhard to reproduce though
21:18:34lostlogicbertrik: more infoz about how and when?
21:18:43amiconnNo it won't
21:18:56lostlogicamiconn: hmm... I must not understand the difference entirely.
21:19:11amiconnThere are still some simple sys event which don't require synchronised action, e.g. charger plugged/unplugged etc
21:19:24lostlogicbertrik: I had a problem with resuming an mp3 froma playlist where the next track is another codec, but I can't reproduce any more with the latest changes.
21:19:31lostlogicamiconn: ahhh, indeed.
21:20:26bertriklostlogic: I think I never tested that, all my albums are always just one codec
21:20:37lostlogicbertrik: damn
21:20:59 Join pixelma [0] (i=pixelma@rockbox/staff/pixelma)
21:20:59lostlogicas always, a logf with LOGFQUEUE and LOGFENABLE from playback and buffering is the big help :)
21:21:14amiconnPublic queues are those initied via queue_init(..., true). Only those are receiving boradcast events
21:21:41lostlogicamiconn: *nod* ok, so ... we need someone to implement this...
21:21:46*lostlogic not it
21:22:19bertriklostlogic: it seems the pcm bar has a hard time filling up because it has to compete with the usefl filling up, even at 80 MHz
21:22:25 Quit MajorC (Read error: 104 (Connection reset by peer))
21:22:44lostlogicbertrik: ah, I was afraid of that when I removed the yield_codec loop.
21:22:54Nico_Pyou removed it?
21:23:22lostlogicNico_P: it was one of the things freezing sansa
21:23:25Nico_Pbut it's replaced with somthing similar, isn't it?
21:23:25lostlogicas I was chasing gremlins
21:23:38Nico_Poh, do you know how?
21:23:39lostlogicyeah, but not a loop, just a standard length sleep
21:23:45bertrikhard to reproduce a complete underrun though, most of the time pcm just hovers at 50% for the 15 sec it takes the usefl bar to fill up
21:23:55lostlogicbertrik: yeah, that's what it _should_ do
21:24:25lostlogicNico_P: yes, because BUF_USED would never fall low even though usefl was low, the yield_codec loop would never exit and both pcm and compressed buffers would go to zero
21:24:47Nico_Plostlogic: so why not poll usefl in the loop?
21:26:57nanoklostlogic: aahm, okay
21:27:07*nanok getting today's today's build
21:27:28Nico_Pnanok: don't use daily builds
21:27:34 Quit amiconn (Nick collision from services.)
21:27:34Nico_Puse "current builds" instead
21:27:38nanokNico_P: current build i mean
21:27:40 Join amiconn_ [0] (n=jens@rockbox/developer/amiconn)
21:27:47nanokNico_P: i allways mix the terms up
21:28:00nanokaren;t the current ones daily?
21:28:02 Nick amiconn_ is now known as amiconn (n=jens@rockbox/developer/amiconn)
21:28:05lostlogicNico_P: could do that, didn't seem necessary, but maybe is... I hate calling update_data_counters, even though I know it's not that slow, it feels slow
21:28:20nanokit was my understanding they are done veery day at a certain hour..
21:28:29bertriknanok: i think they are near instant
21:28:39nanokah, darn
21:28:49nanoki suddenly feel overwhelmed..
21:28:52bertriksvn commit triggers a new build i guess
21:29:22Learnanok: daily builds are like that, every day a certain time. Current build is after each commit.
21:29:24DerPapst100 points to bertrik :)
21:29:35nanokaaahm, i see
21:29:41nanokokay, now i understand
21:30:09nanokbtw, what kind of computing power is behind it, to afford this? i mean, i know there's not so much code
21:30:18nanokbut there are a few targets, though..
21:30:32amiconnThe current builds are distributed
21:30:38 Quit J3TC- (Read error: 110 (Connection timed out))
21:30:40DerPapstby a buildserver farm
21:30:51nanokso as long lostlogic is not sleeping, i won;t even really know to what version i am updating, untill i can check it on the player..
21:31:03amiconnBuilt by several servers around the world
21:31:07Nico_Pnanok: now you see why "today's build" isn't worth much when there are ~10 commits a day ;)
21:31:10*amiconn runs one of those build machines
21:31:15nanokamiconn: ahm, distcc?
21:31:32Nico_Pnanok: not distcc, a custom script
21:31:44amiconnnanok: If you look here , the far right column shows how many build servers did the respective build round
21:31:49nanokNico_P: yup, i was under the obviously wrong impression that the current build is scheduled daily
21:31:51Nico_Pnanok: you can check the SVN rev in the info screen
21:31:52nanoknow it is clear
21:31:59amiconnnanok: Not distcc. Wouldn't help us
21:32:15 Quit mf0102 ("Verlassend")
21:32:16amiconndistcc is good for slow builds and fast networks. WANs aren't fast
21:32:44Nico_Plostlogic: well it seems that the loop was necessary... I agree on it feeling slow but we must remember there are only a couple hundred handles max ;)
21:32:48nanokamiconn: ahm, okay. so is this openmosix style? or beowulfish (that would be hard i think over wan)
21:32:55amiconn(I mean the numbers which are links as you might have figured)
21:33:25amiconnnanok: It's a perl script on the master that hands out builds to the various machines via ssh
21:34:20nanokamiconn: aahm, that kind :)
21:34:29nanokNico_P: thanks ;)
21:37:33Nico_Pbertrik: could you test a patch for me?
21:37:39nanokso i guess this is almost an openmosix-style approach
21:38:00Nico_Pbertrik: cool :) I'll have it soon
21:39:41nanoki am suddenly sad i decided to go for a thinkpad instead of a nice and chunky desktop pc..
21:40:27nanokthe problem with these thinkpads is that they tend to be addictive: once you used one, you can;t really use anything else and not be unhappy :-P
21:40:42nanokkind of like rockbox, and opensource
21:41:16 Join FreshPrince [0] (
21:41:18FreshPrincehi ppl
21:41:34FreshPrincewill rockbox work fine on my ipod video? :/
21:41:42 Quit karashata ("Leaving.")
21:41:50linuxstbYes, as long as it's an ipod video, and not an ipod classic.
21:41:59 Join karashata [0] (
21:42:02FreshPrincenope i have only an ipod video :)
21:42:09nanoknow.. i have 15486 in front of me (so to speak)
21:42:25nanokis somebody planning to obsolete that in say... the next five minutes?
21:42:29nanoklostlogic: ?
21:42:37 Join atsea- [0] (i=atsea-@gateway/tor/x-d366677b201f2e74)
21:42:40FreshPrincelinuxstb, @ the moment the ipod works also fine in my car.. i can listen the music in my car with ipod.. would that work too if i load rockbox on the ipod?
21:43:14linuxstbIt depends how you connect your ipod to the car. See
21:43:24 Quit Frazz ("Leaving")
21:43:54FreshPrincelinuxstb, i connect it with the normal connection below..
21:44:07FreshPrinceand not the headset connection..
21:44:39 Quit ambiotic ("pub")
21:44:44nanoklinuxstb: btw, i noticed i was not able to connect my sansa to my car "cassette adapter" (i did not try with the OF, though). i assumed the break in playback was caused by to low impedance on the cassettes circuitry
21:44:53nanokcan this be something rockbox related also?
21:45:20FreshPrincelinuxstb, is it also possible to delete rockbox from ipod and get the original firmware from ipod back?
21:45:42DerPapstFreshPrince: yes
21:45:45DerPapstread the manual
21:45:46 Quit karashata (Client Quit)
21:45:58bertrikNico_P: building ...
21:45:59FreshPrinceDerPapst, ok thx
21:46:09Nico_Pnanok: I might make r15486 soon if bertrik OK's it
21:46:20bertrikwhat should I test about it?
21:46:30 Join karashata [0] (
21:46:36bertrikoh, the yield stuff
21:46:36linuxstbFreshPrince: You can dual-boot between Rockbox and the original Apple firmware.
21:46:42nanokNico_P: okay, i;ll wait :)
21:46:45FreshPrincelinuxstb, ok thx
21:47:27nanok15487 you mean
21:47:46LloreanGodEater, GodEater_: You asked for me, quite some time ago?
21:47:49bertrikupdate_data_counters is used before it is defined
21:49:15Nico_Poh sorry about that
21:50:49 Quit barrywardell (Remote closed the connection)
21:51:10bertrikI don't think it's harmful
21:51:51 Quit karashata ("Leaving.")
21:52:02Nico_Pbertrik: check the PCM underruns, making sure the previous problem doesn't reappear (the one lostlogic fixed with his commit spree ;))
21:52:17bertrikby the way, i now see the usefl only starting to increase for the first time when pcm has reached at least 50%
21:52:36FreshPrincecheck this out:
21:53:32*DerPapst build something similar for his ipod too
21:53:39DerPapstbut with firewire :)
21:54:07nanokyeah, i need one of those for my sansa
21:54:36nanokbut, unfortunately, it would need a cable, the bllody plug for the sansa is totaly non-standard :(
21:54:47 Join J3TC- [0] (
21:54:52DerPapstsame for the ipod
21:55:13nanokFreshPrince: it's a must have. how can i go trekking for a week without that?
21:55:13DerPapstbut you can use the usb cable for your sansa and a febale usb thingy
21:55:17bertrikthat the one thing i don't like about the sansa
21:55:38*pixelma agrees 110% :/
21:55:52DerPapstbut hey.. it's got a/v out :D
21:56:12nanokDerPapst: yes, that's what i would probably do, but i hate having to have acable there. it could be made really compact with a custom, really short cable
21:56:22Nico_Pbertrik: what you said about pcm and 50%, is it with or without the patch?
21:56:42bertrikNico_P: with the patch
21:56:46*DerPapst is already used to nonstandart plugs
21:59:24bertrikwhen skipping backwards, I typically see pcm increasing to 50%, then PCM hovers at 50% while the usefl and real bars fill up
22:00:46 Quit sd (Nick collision from services.)
22:00:50 Join sd__ [0] (n=sd@
22:00:58Nico_Pbertrik: is 50% more than what it would drop to before the patch?
22:02:08bertrikbefore the patch, the usefl and real bars would already fill up before pcm was 50%, causing pcm to underrun in some cases
22:02:38Nico_Pso it's an improvement?
22:02:43bertrikyes, I think so
22:02:58FreshPrincenanok, ^^
22:03:06FreshPrincehmm.. i need some help.. i'm a n00b :(
22:03:07Nico_PI'll commit then
22:03:21FreshPrincei dont uinderstand the installation for rockbox −−
22:03:35FreshPrincei only need the rockbox utility to install it on my ipod right?
22:04:01bertrikwait, i'll try to run out the usefl without skipping
22:04:11bertrikok, that went just fine
22:04:12FreshPrincelinuxstb, hm.. ok :(
22:04:32linuxstbFreshPrince: Why the sad face?
22:04:43 Join karashata [0] (
22:05:06 Join eigma [0] (n=cat@
22:05:12FreshPrincelinuxstb, cause i have no idea howto do that.. i only read the manual but i dont understand it.. :(
22:05:25DerPapstc'mon... drag'n'drop isn't that hard :)
22:05:26FreshPrincelinuxstb, the rockboc utility. is this one the right link? ->
22:05:38FreshPrinceDerPapst, yay.. do u speak german? :P
22:05:50*DerPapst hides :P
22:05:54FreshPrincehöhö :)
22:06:01*FreshPrince speaks german
22:06:13FreshPrincewell i have to know if this one is the right one ->
22:06:33DerPapsti think so
22:06:33FreshPrinceis this rockbox utility?
22:06:42FreshPrincehm ok. i'm gonna download it.
22:06:43 Quit karashata (Read error: 104 (Connection reset by peer))
22:07:08DerPapst<−− food
22:07:58FreshPrinceok already downloaded.. now i'm gonna connect and mount my ipod video
22:08:21 Join Domonoky_ [0] (
22:08:21 Join Buschel [0] (
22:08:31 Quit bluey_ ("This computer has gone to sleep")
22:08:57 Join ilgufo [0] (
22:11:35FreshPrincelinuxstb, oh u are a delevoper :) hehe
22:12:01 Nick Bagder_ is now known as Bagder (
22:12:04Nico_Pbertrik, nanok: committed
22:12:13FreshPrincelinuxstb, ok i already unpack the rbutilqt.. now i get a dir rbutilgt/ do i have to move this to /mnt/ipod ?
22:12:16 Join Frazz [0] (
22:12:19 Quit Frazz (Client Quit)
22:13:42linuxstbNo, just run it. It's Linux app.
22:13:58FreshPrinceoh ok
22:14:02linuxstb^_a_ Linux app
22:14:19Domonoky_and dont use "complete installation" :-) use the other buttons.. :-)
22:14:54DerPapst"complete installation" isn't complete yet? :P
22:15:00bertrikNico_P: thanks. the buffering debug menu helps a lot with issues like this
22:15:15FreshPrincelinuxstb, ok i start it.. but now i get an error.. "Your configuration is invalid. This is most likely due to a new installation of Rockbox Utility or a change device path. The configuration dialog will now open to allow you correcting the problem"
22:15:16Domonoky_version 1.0.2 has a broken complete install ..
22:15:29Nico_Pbertrik: indeed. I'm really glad I added it... it was GodEater's idea
22:15:57DerPapstah ok
22:16:00Domonoky_FreshPrince: thats not an error, you have to set the configuration in rbutil.. like setting your player type..
22:16:02 Join Alonea [0] (
22:16:07 Quit Alonea (Client Quit)
22:16:15DerPapsttime to release a fixed one then eh? ;)
22:16:26 Quit J3TC- (Read error: 110 (Connection timed out))
22:16:31FreshPrinceDomonoky, that means in section "Select your device in the filesystem" what do i have to type there? /mnt/ipod?
22:16:32pixelmaor setting the language ;)
22:16:45FreshPrincepixelma, hehe yes :)
22:17:01Domonoky_FreshPrince. first try the "autodetect" button in the configuration :-)
22:17:08FreshPrincei did
22:17:15FreshPrinceipod video 5th gen, 30GB :D
22:17:19Domonoky_if it doesnt work. you can change the settings manually ..
22:17:32FreshPrinceok found it , i think i have to klick on ok now..
22:17:38 Quit Lear ("ChatZilla [Firefox]")
22:18:15FreshPrincenow it works :D hehe
22:18:22FreshPrincei restart the app.. :)
22:18:35Domonoky_if someone could make a linux release of rbutil, i could build a version 1.0.3 for windows..
22:19:19FreshPrinceDomonoky, if i select the whole installation.. it will install for me alll the things.. with bootloader etc?
22:19:27FreshPrinceor do i have to install after the installation the bootloader?
22:19:48Domonoky_yes, but this is broken, dont use it..
22:19:50FreshPrincefirst: installl whole installation, second: install the bootloader?
22:19:55FreshPrincewhats broken?
22:19:58FreshPrincethe bootloader?
22:20:10Domonoky_the install options on the first tab..
22:20:22Domonoky_use the seperate install options on the second tab..
22:20:23FreshPrincewhat should i use then?
22:20:36FreshPrinceinstall bootloader then install rockbox?
22:20:44FreshPrinceok thanks :)
22:20:45Domonoky_the extras..
22:21:25Domonoky_the buttons on the QuickStart tab, only combine those seperate install options..
22:21:37FreshPrinceah ok ^^
22:22:08FreshPrincewow.. this installation app is realy cool :D
22:22:20 Quit petur ("switching")
22:23:13 Quit nicktastique (Read error: 104 (Connection reset by peer))
22:23:27 Join petur [0] (
22:24:04FreshPrinceDomonoky, the installation of rockbox.. would that delete my original firmware from apple?
22:24:31Domonoky_no, you can dual boot... take a look at the rockbox manual..
22:26:06 Quit Domonoky (Read error: 110 (Connection timed out))
22:26:12n1shaha, the power management chip in the gigabeat s actually has both a stereo DAC and a limited recording ADC, both are of course not wired anywhere so can't be used...
22:31:03 Quit ilgufo (Read error: 110 (Connection timed out))
22:31:52 Join kugel [0] (i=kugel@unaffiliated/kugel)
22:32:08 Join mhamlin [0] (
22:32:58FreshPrincedamn it..
22:33:03FreshPrincei startet the bubble game..
22:33:07FreshPrincebut i can't exit it lol
22:34:38FreshPrinceok now it works
22:34:38Domonoky_FreshPrince: take look at the really nice manual :-)
22:37:12 Join criznach [0] (
22:37:42Domonoky_could someone with access to the download server move this file to it ? : domonoky/"> ?
22:38:16Domonoky_and could also someone make a linux and a mac build of this release ?? please :-)
22:38:41 Part mhamlin
22:38:53bertrikso what if.... we rewrote it in java? :P
22:39:24 Quit HellDragon (Connection timed out)
22:39:39*Domonoky_ doesnt thin that sansa- and ipodpatcher would work good in java ..
22:40:30ender`how about perl? :)
22:40:36Domonoky_but how about :-)
22:41:30lostlogicNico_P: did that original patch end up getting committed?
22:41:34lostlogicNico_P: if so, it's wrong.
22:41:41*lostlogic reads real diff
22:42:22lostlogicNico_P: that doesn't yield at all between chunks on a non-packet audio type.
22:42:36lostlogicNico_P: it still needs to do the yield, but not the sleep, imo
22:42:56Nico_Pyeah, true... I'll fix
22:43:15lostlogicNico_P: cool, probalby not noticeable on any target other than sansa...
22:43:17lostlogiclike most of this crap :-D
22:44:23FreshPrincei can't see some tracks..
22:44:28FreshPrincei only can see. <Untagged>
22:44:30FreshPrincewhy that? :(
22:44:42DerPapstbecause you files aren't tagged?
22:45:59DerPapstadd a "r" where it fits :P
22:46:22Nico_Plostlogic: done
22:46:46FreshPrinceDerPapst, i dont understand..
22:47:21DerPapstdo all of your mp3 or whatever you'r using have tags? e.g. artist, album, title and such
22:48:09DerPapstwell.. i'm not a rockbox expert... should work
22:48:39*DerPapst directs this question to others who know mote about that
22:48:47n1sFreshPrince: what format are your files in?
22:49:11FreshPrincemp3, wmv
22:49:21DerPapstwmv or wma?
22:49:27FreshPrincewma sry ^^
22:49:35DerPapstactually shouldn't matter i guess
22:49:35jhMikeSlostlogic: I don't know if you caught something in the logs that I observed. On e200, if I resume playback starts from the beginning of the file but the WPS shows the correct time for the resume.
22:50:46n1sFreshPrince: do the tags show up in wps and metadata viewer in wps context menu?
22:51:16lostlogicjhMikeS: already fixed, pondlife noticed first
22:51:33lostlogicjhMikeS: was a bug in mp3
22:51:35 Join stevenm [0] (
22:51:53stevenmhello people
22:51:55 Join safetydan [0] (n=safetyda@rockbox/developer/safetydan)
22:52:22 Quit merbanan (Remote closed the connection)
22:52:57linuxstbFreshPrince: What kind of tags do your files have? Are the tracks showing as "untagged" mp3 or wma?
22:53:43FreshPrincen1s, ehm.. dont know what u mean :(
22:53:57FreshPrincelinuxstb, shpwing as untagged.
22:54:03bertriknanok: are you familiar with thermistors?
22:54:07FreshPrincethe first 50 tracks are untagged
22:54:12FreshPrincethe others are fine..
22:54:38n1sFreshPrince: if you play a file that shows as untagged do the tags show in the wps?
22:54:48stevenmn1s, if you haven't seen, we can seek backwards now!
22:54:54 Join JdGordon [0] (
22:54:58 Quit Domonoky_ (Read error: 104 (Connection reset by peer))
22:55:04stevenmjust inefficiently
22:55:05n1sstevenm: yes I saw, very nice :-)
22:55:11stevenmthanks :)
22:55:32stevenmPITA to seek midi because of instrument changes and notes on for a long time
22:55:56n1sstevenm: I have been thinking about trying to turn it into a codec once most of the buffering bugs are worked out
22:56:10stevenmthat would be sweet
22:56:30stevenmis it just going to kick everything else out of the buffer?
22:57:36n1sstevenm: the idea is that we can have all the patches and stuff on the buffer together with regular streaming tracks or at least we should be able to come up with a way to do that :-)
22:58:06FreshPrincen1s, whats the wps?
22:58:21n1sFreshPrince: the While Playing Screen
22:58:33stevenmn1s, I don't exactly know the codec framework, but don't we only get X bytes?
22:58:34FreshPrinceno its not showing.
22:59:25n1sstevenm: with the recent changes to buffering code the buffering has become much more flexible and that should be possible AFAIU
22:59:38linuxstbFreshPrince: What kind of tags do your files have? Do you have any mp3 files with APE tags for example?
23:00:01 Join HellDragon [0] (i=jd@unaffiliated/helldragon)
23:00:12FreshPrinceAPE tags. :S
23:00:23FreshPrince<... understands nothing sry ^^
23:00:32stevenmn1s, oh, that would be nice. I guess we don't have to worry about "sharing" the patch set among multiple files yet, but it would be nice.
23:00:57stevenmn1s, I can think of at least a few things that would be shared.. piano definitely (loaded by default) and some of the drum patches.. guitar perhaps
23:01:16stevenmn1s, I haven't actually bothered to find out just how much ram an average midi file actually requires
23:01:18n1sstevenm: that would be nice but also a bit more complicated...
23:01:25stevenmn1s, no, definitely
23:01:36stevenmit would be insanely hard :)
23:02:25 Quit ompaul (Client Quit)
23:02:26 Join qweru [0] (
23:02:56 Join DarkDaemon [0] (
23:03:00FreshPrincelinuxstb, those are not playable.. :(
23:03:02 Join bertrik_ [0] (
23:03:15stevenmn1s, so we are good with seeking, I guess.. (minus some optimizations I can think of). We also have playing time. And I can think of a few things to put under WPS (text inside the files
23:03:17DarkDaemonhow can i put flac albums onto my 5.5g ipod :S
23:03:39linuxstbDarkDaemon: Just copy them...
23:03:49n1sstevenm: I was thinking about the synthVoice loop and all the getSample calls (or whatever it's a macro) in there, do you think there would be a way to just calculate cp_temp (former cpShifted) and acces memory just once?
23:03:57linuxstbTo your ipod...
23:04:02 Quit XavierGr (Nick collision from services.)
23:04:03 Join XavierGr_ [0] (
23:04:18linuxstbYour ipod is a hard disk, Rockbox lets you browse that disk and play files.
23:04:22DarkDaemonok so when im using rockbox would they appear under database?
23:04:44linuxstbYes, they would also appear in the database (assuming they're tagged)
23:04:51 Nick XavierGr_ is now known as XavierGr (
23:04:56stevenmn1s, I think under most conditions, that variable is accessed twice- once to read and once to write
23:04:58DarkDaemonthank you
23:05:34stevenmn1s, the getSample thing - it pulls out two samples and does a little bit of interpolation. I wonder if we can get away with just getting one sample.. I do remember it sounding all hissy without it
23:05:41n1sstevenm: I'm not worried about that varable but the getSample calls as that buffer is in slow ram
23:06:00 Quit DarkDaemon (Client Quit)
23:06:07stevenmn1s, I think getSample should be called only twice also.. let me check
23:06:21 Join webguest27 [0] (i=4541cfaf@gateway/web/cgi-irc/
23:07:13webguest27:) hiya all
23:07:24stevenmn1s, yeah, there are 5 calls, but most of the time only two of them actually run. the others handle cases when playback hits the edge of a loop
23:07:57n1sstevenm: so those other calls are rather infrequent?
23:08:05stevenmn1s, yes
23:08:27 Join SMB-MacMan [0] (
23:08:28n1salso I wondered why some of the calls for s2 are (cp_temp >> FRACTSIZE)+1 and some (cp_temp >> FRACTSIZE)
23:08:31SMB-MacManhey guys
23:08:48SMB-MacManjust wondering if someone would be able to help me with rockbox?
23:09:00n1sas you probably can tell i'm not too familiar with the dirty details
23:09:08stevenmn1s, to explain.. GUS patches can have a loop region. so once you hit the edge, you either rewind to the start of the loop, or reverse direction. the other calls happen if the last sample retrieved was outside of the loop region
23:09:16linuxstbSMB-MacMan: Just ask your question, if someone can help, they will.
23:09:58stevenmn1s, how is the slow ram accessed? how wide is the bus?
23:10:00webguest27random question - i just ordered a sandisc e250 from today's woot ( and found out i could install rockbox on it... but how large is a standard installation? the player is only 2 GB and i'm looking into SD cards also
23:10:11SMB-MacManI'm using Rockbox on a Toshiba Gigabeat F60, and I just wanted to know what format video clips have to be in, in order to play, i've tried VCD MPEG and it goes to a black screen for a sec then back to the menu?
23:10:34stevenmn1s, because most of the time, the two getSample calls access two consecutive 16-bit locations
23:10:37krazykitwebguest27, not enough to matter.
23:10:47stevenmn1s, wait, maybe not... durr
23:10:49webguest27badass, thats what i figured.
23:10:55webguest27very excited. gotta love woot.
23:10:59n1sstevenm: ah, ok I think I get it. slow ram is 32 bit but _much_ slower than fast iram
23:11:06webguest27thanks guys
23:11:30linuxstbSMB-MacMan: VCD mpeg should work, as long as it's the source .mpg file, and not the .bin file. See
23:11:35n1sstevenm: I played with doing those two reads as one 32 bit too but never got it working alright...
23:11:38 Quit jhulst (Connection reset by peer)
23:11:47SMB-MacManthanks Linuxstb
23:11:52stevenmn1s, Hmmm... I have to read into it more, but maybe we can get away with caching one of them
23:12:13 Join japc [0] (
23:13:04n1sstevenm: as we are not doing a lot of calculations slow ram access is what I would guess on is killing performance
23:13:15pixelmasomeone ever heard of/seen a microSD to MMC (or _thin_ SD) adapter?
23:13:39stevenmn1s, do we know if a 32-bit read takes as much time as a 16 bit read?
23:13:55SMB-MacManpixelma: I've seen them on ebay
23:14:06JdGordonpixelma: isnt mmc and sd the same size?
23:14:24SMB-MacManJD: MicroSD is smaller than regular SD :)
23:14:31 Quit atsea- (Read error: 104 (Connection reset by peer))
23:14:46pixelmausual MMC is thinner than a usual SD - or as thick as a thin SD...
23:14:47n1sstevenm: I'm not sure but I think it's a little slower for coldfire but one 32 bit read is faster than 2 16
23:14:56stevenmn1s, aah, ok
23:15:18SMB-MacManJD: Do you have your mini SD card already?
23:15:19***Saving seen data "./dancer.seen"
23:15:24 Quit webguest27 ("CGI:IRC (EOF)")
23:15:27pixelmaSMB-MacMan: do you happen to have a link, I can only find the thicker onses?
23:15:40 Join Calcipher [0] (
23:16:20SMB-MacManpixelma: I havent found any yet, only thing I have found is this:
23:16:31pixelmaJdGordon: I can't put the standard adapter that came with the microSD into my Ondio...
23:16:32*SMB-MacMan gets booted from channel for spam hahahah
23:16:48 Join guest01 [0] (i=c909d5b1@gateway/web/cgi-irc/
23:16:50 Quit bertrik (Read error: 110 (Connection timed out))
23:17:05JdGordonpixelma: ah, ok
23:17:10pixelmaSMB-MacMan: nah, that's only standard thickness too
23:17:16 Quit guest01 (Client Quit)
23:17:26stevenmn1s, I'd have to look into this then.. can we easily do a 32-bit read on that data? It is an array of 16 bit values
23:17:33stevenmmaybe have to endian swap?
23:17:44stevenmwait, no swapping.. but shifting
23:17:56SMB-MacMannot sure where to get the thinner ones sorry mate
23:17:56 Nick bertrik_ is now known as bertrik (
23:18:19stevenmn1s, also, what about all the variables used in synthVoice? Do we know if all of them live in registers?
23:18:29pixelmastill thanks
23:18:31n1sstevenm: I think it would be worth trying at least
23:18:55eigmaJdGordon: omg I fixed my m:robe!! :D
23:18:57stevenmn1s, I'll try to do it right now. beats the paging project :)
23:18:58n1sstevenm: no, I have not disassembled it so I don' t know
23:19:28JdGordoneigma: woohoo :D
23:19:34JdGordonwhat was wrong?
23:20:27eigmaJdGordon: turns out the commands I was sending over SPI were affecting the LCD in a persistent manner.. even disconnecting the battery didn't erase the changes I made
23:20:52eigmaJdGordon: I looked through the OF and it looked like some values the LCD was returning was causing the OF to skip some initialization commands
23:21:10eigmaJdGordon: I sent those commands manually myself, and voila, it's back to normal for good −− and now the OF sends those commands on every bootup
23:21:22JdGordongreat :)
23:21:34eigmaI still have no idea what chip it is that caused this, or what the commands mean :(
23:22:19JdGordonbut at least you know how to fix it if it happens again :p
23:22:50eigmayeah −− we have to be really careful around that part of the code (backlight/SPI) because it would be pretty scary if this happened on users' devices
23:24:41stevenmn1s, I got rid of the second call to getSample entirely
23:25:06n1sstevenm: great, and it still works?
23:25:18stevenmn1s, so, s = s2; s1 isn't ever retrieved, and there is no interpolation
23:25:48stevenmn1s, well.. as I expected, it still plays... but some instruments will sound a little weird because they are being up/downsampled without any interpolation at all
23:26:18stevenmn1s, I plugged in USB during playback (instant 22k -> 44k) and there are still some buffer misses. not as many as before, but still are
23:26:46n1sstevenm: does it affect sound as much when in 44kHz ?
23:27:00n1sdropping the interpolation I mean?
23:27:47stevenmn1s, probably wouldn't so much... honestly I don't remember the difference all that much (I also havne't really slept either). try it for youself and see what the sound difference is
23:28:21n1scould you pastebin the code, my brain doesn't work :-/
23:28:57stevenmn1s, just comment out s1=getSample((cp_temp >> FRACTSIZE), wf);
23:29:09stevenmn1s, then comment out the line directly after and write instead s = s2;
23:30:53stevenmn1s, it probably isn't a good idea to get rid of the interpolation entirely- it adds a high pitched hiss to instruments. ie monkey island sounds kinda ugly
23:31:05stevenmwe did away with the Ghetto Lowpass Filter by adding the whole s1/s2 thing
23:31:10n1saha, I thought you changed it to read 32 bits too, building now
23:31:38stevenmn1s, nah. this is just the best, best, best case we will have, as it also gets rid of some math that does need to be left in
23:31:41 Join MajorC [0] (
23:32:09 Quit Buschel ()
23:33:06 Part FreshPrince ("Leaving!! ;)")
23:34:40n1sstevenm: yep sounds a lot worse, the idea with reading both samples as 32 bit was just to see if it would be any faster but still keeping the interpolation in
23:35:20stevenmn1s, a better test I guess would be to keep the interpolation there, but instead of writing s1 = getSample, just set s1 = s2
23:35:44stevenmit wouldn't do much "interpolating" but it would give a better idea of how much speedup we'd get from 32bit reads
23:37:05 Join atsea- [0] (i=atsea-@gateway/tor/x-658e13f7717de55c)
23:37:09 Quit mirak (Remote closed the connection)
23:37:22 Join Guile` [0] (n=Guile@
23:40:57 Quit Nico_P (Remote closed the connection)
23:41:05 Quit DerPapst ("So Long And Thanks For All The Fish!")
23:41:40 Join hardeep [0] (
23:42:58 Quit linuxstb ("ChatZilla [Firefox]")
23:44:22 Join ddalton [0] (
23:44:28dionoeaSuppose that I have tracks A and B in the buffer. I'm playing A and skip to B. If i go back to A, why does it need to rebuffer it?
23:45:28ddaltonn1s: around?
23:45:46n1sddalton: yes, for a few mins
23:46:00dionoeaor, the otherway around: why does it unbuffer A when skipping to B? Couldn't it just mark the buffer as reclaimable but keep it until someone actually needs to reclaim it?
23:46:01ddaltonn1s: are you familiar with voice building?
23:46:14n1sddalton: a little
23:46:32ddaltonn1s: ok do you know how to make the clips record at scale 2.1?
23:46:40ddaltonor volume...
23:47:28n1sddalton: no but that would be a setting in your tts engine or something you need to use a post processing tool on the wav for if the tts engine can't do it
23:48:04ddaltonn1s: but in the other makevoices vb script we could modify it to do that. But now with the configure script I can't work out how.
23:48:36n1sddalton: which engine are you building with?
23:49:03ddaltonn1s: sapi5. (espeak)
23:49:28ddaltonn1s: it says enter sapi options. Should I type something here?
23:49:35ddaltonor press enter for defaults
23:49:38n1sddalton: I never used that and never built voices on windows so i can't help you with that
23:49:41ddalton0.6 is the default...
23:49:48ddaltonok thanks anyway.
23:50:24 Part przemhb
23:50:35ddaltonn1s: what about -vbr
23:50:38 Quit japc (Read error: 104 (Connection reset by peer))
23:50:57ddaltonwhat should I type here? I don't think it has anything to do with sapi...
23:52:31 Quit bertrik ("bye")
23:53:57 Quit Guile`` (Read error: 110 (Connection timed out))
23:57:05 Join Guile`` [0] (n=Guile@
23:57:34 Join Soap [0] (n=Soap@rockbox/staff/soap)

Previous day | Next day