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

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

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

#rockbox log for 2005-08-15

00:11:46 Join ]RowaN[ [0] (
00:29:45 Join muesli- [0] (
00:30:06 Quit muesli- (Client Quit)
00:38:15 Part zezayer
00:46:56 Quit Moos (Read error: 110 (Connection timed out))
00:52:34***Saving seen data "./dancer.seen"
00:53:36 Quit hicks (Read error: 110 (Connection timed out))
00:54:34 Join XavierGr [0] (
01:11:24 Join pike| [0] (
01:20:20 Quit XavierGr ()
01:23:21 Quit pike| (Read error: 104 (Connection reset by peer))
01:25:41 Quit pike (Read error: 110 (Connection timed out))
01:39:32 Quit ender` (Read error: 113 (No route to host))
01:49:56 Join cYmen [0] (
02:01:43 Join pike [0] (
02:16:22 Join t0mas [0] (
02:16:28t0masgood night :)
02:16:58t0masdoes anybody know what the maximum value of a pcm_playback.c internal sample is?
02:17:20t0mas(read from the DMA memory)
02:24:03]RowaN[fast forward seeking no longer works in wavs (iriver ihp120) .. dunno if anyones aware of that
02:24:20]RowaN[its been like that for a few days since last week i think, in the daily builds
02:24:32]RowaN[or maybe it never worked and im losing my mind
02:25:11*t0mas needs Slasheri
02:25:29]RowaN[bedtime for Rowan
02:25:31t0masthe backlight dimming code has to be changed some way
02:25:40t0masbut I can't find out how
02:25:57 Quit ]RowaN[ ("zzz ahh rockbox zzzzz")
02:52:38***Saving seen data "./dancer.seen"
02:59:07t0masok... I have the right beat detection now
02:59:33t0mascan be tweaked a little for some music types
02:59:39BBubbeat detection?
02:59:43BBubfor the peak meter?
02:59:58t0masnow I only need a good backlight dimming trick
03:00:54 Quit cYmen (Read error: 110 (Connection timed out))
03:01:57t0masthe detection works for on/off backlight on the remote now
03:02:01t0masI have a flashing remote :D
03:04:22BBubsounds like fun
03:04:55t0masit is
03:05:01t0masand with dimming it would be even better...
03:05:07t0masdimming player screen on bass tones
03:05:18BBubbut dimming is not possible with the remore, right?
03:05:19t0masit will eat batteries
03:05:26t0masit isn't
03:13:01 Join [CJ]fLaRg3 [0] (
03:17:30 Quit Strath (Read error: 110 (Connection timed out))
03:19:29 Quit [CJ]fLaRg3 ("Leaving")
03:21:18t0masoptimization should be done tomorrow :P
03:21:31t0masit boosts 3x as much with beat detection (and output to the remote)
03:22:46BBubdou you know if the remote will be included before the feature freeze?
03:38:24 Quit Sucka ("a bird in the bush is worth two in your house")
03:58:14 Join Strath [0] (
04:05:46 Join QT [0] (i=as@madwifi/users/area51)
04:15:49 Quit QT_ (Read error: 110 (Connection timed out))
04:41:03 Quit dpassen1 (Read error: 104 (Connection reset by peer))
04:47:29 Quit edx (Read error: 110 (Connection timed out))
04:52:39***Saving seen data "./dancer.seen"
05:01:04 Quit Strath (Read error: 110 (Connection timed out))
05:02:20 Join solex [0] (
05:12:45 Quit solex_ (Read error: 110 (Connection timed out))
05:21:37 Quit ned12 (Ping timeout: 14400 seconds)
05:29:14 Join tvelocity [0] (
05:44:21 Quit tvelocity ("Leaving")
05:54:16 Quit crash_ (Read error: 110 (Connection timed out))
06:15:55 Join edx [0] (
06:25:35 Quit courtc (Read error: 104 (Connection reset by peer))
06:25:44 Join courtc [0] (
06:48:47 Join LinusN [0] (
06:52:40***Saving seen data "./dancer.seen"
07:20:08amiconnGood morning
07:30:08LinusNmorning amiconn
07:55:01LinusNamiconn: u there?
07:56:04LinusNthere is a bug in wps-display.c which was introduced by the wps aligning patch
07:56:45LinusNit uses lcd_putsxy() for non-scrolling lines, which doesn't reset the scrolling flag for that line (of course)
07:57:01LinusNso we need to make the line stop scrolling somehow
07:57:19LinusNi have (so far) two solutions to this problem
07:57:43LinusN1) introduce a new lcd_x function for stopping the scrolling on a selected line
07:58:35LinusN2) add a hack in wps-display.c that forces a scroller reset by calling lcd_puts_scroll() with an empty string, which will reset the scrolling
07:58:56LinusNdo you have any comment?
07:59:05LinusNanyone else?
08:01:05amiconnThat confirms my opinion that the wps alignment patch is a hack in its current form
08:01:51amiconnDoing it the proper way requires some more low-level work
08:03:21LinusNhow do you think it should be solved?
08:03:32amiconnI think the most elegant solution with the current design would be to add a parameter to lcd_puts_scroll() (or introduce lcd_puts_scroll_ex for that) to hand it the desired alignment
08:03:42amiconn...i.e. let the driver do the alignment
08:04:31amiconnThe long-term solution would be imho to switch to pixel-based alignment.
08:04:31LinusNhmmm, but how should non-scrolled align be handled then?
08:04:51amiconnThe alignment is done only if it doesn't scroll
08:05:25amiconn...and for non-scrolling lines, the problem with the scroll flag doesn't exist
08:05:30LinusNyes, but what if there is no %s in the wps line
08:05:50LinusNthe problem is with alternating sublines
08:06:07LinusNand only one subline scrolls
08:06:13amiconnThen how was that done before?
08:06:26amiconnI think the problem must be rather old then...
08:06:27LinusNit used lcd_puts()
08:06:32LinusNbefore the patch
08:06:37LinusNand that resets the scroller
08:07:22amiconnMaybe we could have lcd_puts_ex(), with 2 additional parameters
08:07:24 Join lostlogic_ [0] (
08:07:38amiconn(1) the alignment and (2) whether it should scroll long lines or not
08:08:09amiconnAll of this isn't clean, I know
08:08:46amiconnWith pixel-based alignment, the line-based text functions would be dropped completely
08:09:01amiconn...and there would be a function to put a scrolling text in a box instead
08:09:22amiconnOf course the number of scrolling boxes would have an upper limit
08:09:26LinusNi'm not sure that's what i would want
08:09:33amiconnWhy not?
08:09:50LinusNwell, that would hopefully be controlled by the wps
08:10:26LinusNthe current wps code combines the aligned strings to a full line if they don't fit
08:12:09LinusNalso, the wps can output several strings on the same line, with different alignment
08:13:29LinusNin fact, all we would need is a lcd_puts_ofs() with only an x offset
08:15:23amiconnI think the box layout is much more flexible
08:15:35amiconnLine-based has several problems
08:16:29amiconnHmm, btw, what combination feature do you mean?
08:16:44amiconnIirc the current wps code can only use one alignment per line??
08:20:45 Quit lostlogic (Read error: 110 (Connection timed out))
08:23:22LinusNah yes, i forgot that the multiple alignment patch is not committed
08:24:01LinusNi agree that the box layout is better, but there is no time to do that before the release
08:25:17amiconnIt isn't even high priority for me
08:25:23LinusNmaybe i should add the lcd_puts_scroll(0, i, "") hack, and do the box layout in the next version
08:25:34amiconnThere are multiple other things that need to be done before
08:26:09amiconnRockbox on iriver still doesn't support all features (even not all major features) of rockbox on archos
08:26:28LinusNand the database viewer bugs
08:27:49LinusNthe context menu has all kinds of weird behaviour, it can even hang
08:28:04LinusN(in the database mode)
08:28:52amiconnOn iriver, or in general?
08:29:04amiconnI must admit that I seldom use the database
08:29:18amiconn(Basically only to check whether my tags are correct)
08:29:55LinusNin general, i believe
08:35:44LinusNi went for the quick-hack solution
08:46:05 Quit hardeep ("BitchX: its all day strong, all day long")
08:46:19 Join Rockdude05 [0] (
08:47:11Rockdude05hi, are there any plans to eliminate the need for the hard disk to speed up prior to shutting down? what information is saved in this process?
08:52:44***Saving seen data "./dancer.seen"
08:56:28 Join einhirn_ [0] (
08:56:58 Quit einhirn ("Miranda IM! Smaller, Faster, Easier.")
08:59:00LinusNRockdude05: settings and resume information
08:59:42LinusNthe hard disk doesn't spin up if it doesn't have anything to save
08:59:57LinusNit also saves the runtime information database
09:00:03Rockdude05ok, so no plans/intentions on saving it any other times? or possibly a menu option.... i dunno.... whats on your mind :P
09:00:26LinusNit always saves it when the disk spins up for other reasons
09:00:28Rockdude05its just for when, as loads of people ahve said, it shuts down and has a heart failure
09:01:07LinusNi don't know about the runtime database though, i didn't write that part of the code
09:01:13Rockdude05oh by the way, im experiencing a bug with the latest daily... after i plug in the usb in rockbox, the backlight wont turn back on until i reset
09:01:26Rockdude05(when i unplug it)
09:01:49LinusNis this the bootloader usb mode?
09:02:07Rockdude05no, in rockbox
09:02:24Rockdude05il quickly make syre im right, 1 sec
09:03:20Rockdude05oops yep
09:03:29Rockdude05i guess you know of it then
09:03:52Rockdude05although it seems random, which is a pain for you i guess
09:04:06LinusNi'll have a look
09:04:46LinusNwell, it does come back if you let it timeout
09:05:02Rockdude05oh ok
09:05:31LinusNit seems like rockbox expects the backlight to be on when it starts
09:05:43Rockdude05whilst im here as wel.... i get loads of freezes when i change crossfade duration and it restarts playback. is this your 'department'? :P
09:06:02LinusNnot really, but i guess it will be :-)
09:11:34LinusNRockdude05: btw, the solution to the problem when it can't save the settings when the battery is low is not to eliminate the need to save the settings
09:11:48LinusNit must store them somewhere
09:13:19LinusNhowever, it should not try to save if the battery is too low, or at least not retry forever
09:13:46Rockdude05so, what are your 'hints' when i have a player going nuts from a 0% battery which i left on by accident, because reset simply resets in obviously, and although the bootloader says its low battery, it still freezes and the backlight is on, sometimes flickering
09:14:32 Join Coldtoast [0] (
09:15:20LinusNyou mean it freezes when you try to turn it off?
09:17:32Rockdude05i must film it some time, its quite entertaining....
09:18:05Rockdude05ok, low battery, so when the disk spins up it immediatly spins down after about 2 secs (i can hear it clearly as my disk sounds like a lawnmower anyway)
09:18:29Rockdude05and does that repeatedly until battery has lowered even more, so that the disk cannot spin at all
09:18:51Rockdude05then the backlight is on and stays on
09:19:04LinusNi repeat: you mean it freezes when you try to turn it off?
09:19:13 Join ender` [0] (
09:19:16Coldtoastmine does that. freezes
09:19:28ColdtoastI've only ever let it get that low once
09:19:33Rockdude05refuses to turn the hard disk off
09:19:46Rockdude05and bootloader wont turn off either
09:19:50Coldtoastand it froze before the HDD tried to spin up
09:19:57LinusNthe bootloader only warns
09:20:18Rockdude05ok, and sits there if its low battery?
09:20:31LinusNno, it tries to boot rockbox
09:20:50Rockdude05ooh, not good
09:21:34LinusNthis is because we don't yet know the exact battery level where there is no point in trying
09:22:12LinusNand we don't want to patronize the user by saying that the battery level is too low when it would be able to boot if it just tried
09:22:34LinusNbtw, i fixed the backlight issue
09:23:08Rockdude05i see. so, as ive read, the battery cannot go below 3 volts, for risk of damage or blowng up or something. but when its 1%, ive read in debug the battery is about 3.3 volts, ish. so at what point does the hdd cut out?
09:23:48Rockdude05and at what point do i worry after my hdd has fits for about 20 mins?:P
09:25:06Coldtoastyou'd have to be a moron to let it sit there doing it for 20mins tho
09:25:17Coldtoastthat's what the reset button is for
09:25:25Rockdude05oh but like i said, i cant really turn it off
09:25:30Coldtoastyes you can
09:25:33Coldtoastreset button
09:27:02Coldtoastmine's frozen before with a depleted batery while I was halfway thru the 45min walk home from the gym and I just looked around on the ground for a bit of wood and hit the reset button
09:28:07Rockdude05then the bootloader loads and sits there
09:28:35LinusNRockdude05: what happens if you reset again?
09:31:03LinusNRockdude05: when it sits there, does it display the low battery warning?
09:32:30Rockdude05yes, with backlight on
09:32:47Rockdude05i can keep resetting and it comes to the bootloader
09:33:25Rockdude05i guess the first line or whatever in the bootloader should be the battery... hardwares guna get damaged when this thing goes 'public' :P
09:34:45LinusNdo you know that for a fact?
09:35:04Rockdude05oh no no, just assumed, but im sure you know what i mean
09:36:45Rockdude05also, when the old iriver firmware starts it loads the file system, which takes ages. this is handy, though, as the disk is constantly spinning up for new folders, which surely drains battery if its stopping and starting. is there plans to maybe make a file which includes the file tree, or make it an option to load it at startup?
09:37:19Rockdude05stopping + starting in rockbox ^
09:37:42Coldtoastthe iriver FW starts/stops the HDD FAR mroe than Rockbox does
09:38:11Coldtoastthe iriver FW spins has to spin the HDD up to load the next track
09:38:41Coldtoastrockbox spins up, buffers a few tracks, spins down
09:38:47LinusNthere are no plans of caching the entire hard disk tree in memory
09:38:54Rockdude05wel yes but it isnt the point. every less spin is always a bonus in my eyes. it also helps when youre a 'flicker' like me and like to browse (i have 8500 songs)
09:39:01Rockdude05ok thats fine :)
09:39:10Rockdude05and does the 10000 file limit apply in rockbox?
09:39:47LinusNof course not
09:40:19Coldtoastit looks like Replaygain support has been REALLY popular
09:40:57 Join ashridah [0] (
09:41:13Slasheri think that file tree caching could be added for iriver targets as an option. It eats quite much memory (few megabytes) depending on how much folders and songs there are on the disk. But that shouldn't be a problem because iriver has a "huge" memory anyway
09:43:14LinusNSlasher: agreed, an option would be ok
09:43:21Rockdude05"huge" being....? and how much will a file tree be?
09:43:53Rockdude05and how much resources are left
09:43:54SlasherRockdude05: 32 MB total off-chip ram
09:44:00Coldtoasthey. I had that progress bar bug again 2 days ago
09:44:32ColdtoastI've had it 4 times now
09:44:33Rockdude05ooh not bad
09:44:42LinusNa file tree could probably be somewhere between 300kbytes and 1megabyte for your 8500 files
09:44:50SlasherRockdude05: i think that for most user the consumed amount of memory would be something between 200 and 500 kB
09:45:13LinusNit depends on how it is stored in memory
09:45:18Rockdude05oh wel thats pants
09:46:12Coldtoastexample: I'm playing a track that's 2:30. when the track ends and it switches to the next one, the progress bar doesn't reset. stays filled. And the time will either stop at the track length or it'll continue past it. So if track length was 2:30, when it starts playing the next track, time doesn't reset. keeps going
09:46:14LinusNColdtoast: when the progress bar bug appears, is the "total time" correct for the file that played?
09:46:42Rockdude05as you guys are pretty good at this stuff... my iriver has been dropped a few million times from waist height to concrete pavements and my hdd resembles a lawnmower. would a format fix this or is it worth gettin a new hdd? or shall i get a iaudio and hope for rockbox?
09:46:44Coldtoaston my WPS I have the time display as position/length. so I can get times like 3:02/2:30
09:47:15ColdtoastLinusN: the track length stays as the last track that was played when the progress bar didn't reset
09:47:16LinusNRockdude05: a format will not fix it, but you can replace the hd
09:47:49LinusNyes, but is the 2:30 the correct time for that track?
09:48:24LinusNdoes it always happen with that track?
09:48:31Coldtoastif I manually skip tracks, it resets and works. As a test, I skipped back to the track where the bug happened
09:48:34Coldtoastand it didn't do it
09:48:43Coldtoastit's a totally rendom bug :/
09:48:56LinusNwhat happens if you start playing with that track?
09:49:06Coldtoastit behaves normally
09:49:27Coldtoastif I let the track play again and switch, the progress bar and time all work and reset properly on track change
09:49:49SlasherColdtoast: is the crossfade enabled?
09:49:57LinusNi have seen the bug too, but only when the total track time calculation is wrong
09:50:29ColdtoastSlasher: I got that "second of silence" thing againt he other day, btw
09:50:53Coldtoastwhere in the middle of a track, the sound drops out for a very short time
09:51:20Coldtoastit's annoyingly random too so I'm never looking at the player when it does it
09:51:25LinusNsounds like a buffering problem
09:51:32Slashercould you reproduce it if you play excactly same set of files?
09:51:35LinusNare you shaking the player when it happens?
09:51:53Rockdude05and is it normal to have a stuttering crossfade?
09:51:54ColdtoastI'm walking tho
09:52:03LinusNColdtoast: could be that
09:52:15Coldtoastyeah. what I'll do it move the player
09:52:17Slasherah, you was walking?
09:52:24LinusNstuttering crossfade is not normal
09:52:27Slasherthen try setting the anti skip-buffer setting to 1-5 minutes
09:52:29Slasherit should help
09:52:33ColdtoastI suually have it hooked onto my belt too
09:52:52Coldtoastbut I'll hook it onto the strap of my backpack and let ou know if it happens again
09:53:00Slasherthe hdd is not accessible always when walking
09:53:10Coldtoastthat way it's at my shoulder where there's less movement
09:53:11LinusNColdtoast: walking with it in the belt is actually the worst case
09:53:20Coldtoastyeah. I know :)
09:53:51amiconnLinusN: It is correct that the backlight thread expects the backlight to be on initially
09:53:58ColdtoastI've had it while walking slowly tho. but I'll keep an eye on it and see if it happens with the player at a more stable position
09:54:20Rockdude05oh well when i crossfade sometimes the previous track reduces volume then all audio cuts out, then a snip of the previous track plays, then straight into full volume next track < if anyone understands
09:55:19LinusNamiconn: yes
09:55:42LinusNat least for iriver
09:57:20ColdtoastI'm actually surprised I really like replaygain support
09:57:35Coldtoastgapless means nothing to me but I'm liking replaygain a lot
09:57:38 Quit Nibbler ("life is like a rental car, you fuck it up, and give it back.")
09:57:55 Join Nibbler [0] (
09:58:16Rockdude05hmm yeah, handy when using it for lineout at work (replaygain) - i get customers having heart failure on a loud track
09:58:31LinusNamiconn: the backlight init calls backlight_on() right after initializing the i/o ports, but the new dimming code doesn't turn on the backlight if the pwm has reached the limit
09:59:57amiconnThat's because of the protection from extraneous calls...
10:03:32LinusNi committed a fix
10:04:29Rockdude05man arnt you fast
10:04:45amiconnHowever, I don't know whether this fix is the best solution
10:04:51LinusNperhaps not
10:05:09LinusNbut it works, and is pretty obvious in the code
10:05:13amiconnIt always does a hard switch-on now
10:05:31Rockdude05" Use the system font in the debug screens " < i thought it wasnt possible to have varying fonts in rockbox with the current system
10:05:58LinusNi could have added a hard switch-on in backlight dim, but i think you wouldn't like that either
10:06:03amiconnAfair the current bootloader doesn't use the backlight, so the backlight should fade in at boot
10:06:20amiconnPerhaps something can be done about this...
10:06:33LinusNthe boot loader turns it off in usb mode
10:06:43amiconnAh, yes
10:06:45LinusNbut leaves it on otherwise
10:06:53 Quit midkay (Read error: 104 (Connection reset by peer))
10:06:59amiconnIt should be possible to read the current state from the port, correct?
10:07:21amiconnOr is that pin one of those which have separate input/output pins
10:08:05LinusNwhy bother?
10:08:16LinusNturning it on immediately is the way to go imho
10:09:21Rockdude05or even have the bootloader say nothing if rockbox is going to load fine.... itll look pretty
10:09:22amiconnIt could fade in if fade in is enabled
10:09:35Rockdude05sorry ... *puts handbag down*
10:09:52LinusNamiconn: you're probably right, it's just me who doesn't care about such cosmetic details
10:10:39LinusNRockdude05: in fact, having a silent boot loader is not a bad idea
10:10:57LinusNmakes it less evident if you need to rma it
10:11:24Rockdude05even make a little picture saying "Loading..."
10:11:42LinusNit's still noticeable, since it takes longer to start the original
10:12:13Rockdude05whats noticable?
10:12:23LinusNthat the bootloader is there
10:12:31amiconnBtw, there's still that bug that you can switch the unit on with a really short push of either power button even if both lock switches are set on
10:12:46Rockdude05which is why maybe a loading would be good as it's there whilst everything cranks up
10:12:55amiconnIt seems we need some more sophisticated lock switch handling
10:12:56 Join midk [0] (
10:13:33LinusNamiconn: the first thing that happens in the bootloader is to check the Play buttons
10:14:29amiconnJust try it: Switch both lock switches to locked, then do a really short tap of either play button
10:14:36amiconn-> Then unit will power on
10:14:52amiconnI manage to do this everytime if I want to
10:14:59amiconnWe should stop if either lock switch is enabled if we can't detect which power button turned on the unit
10:15:22amiconn..but only stop on the appropriate lock switch if we were able to detect the used power button
10:15:31LinusNi'll fix that
10:16:06amiconnThat leaves to handle the exotic case when both power buttons are pressed
10:16:23amiconnI think in this case we should also stop on either lock switch
10:16:54LinusNbut the current check would handle that, wouldn't it?
10:17:06 Join webguest89 [0] (
10:20:41Rockdude05any inking when any wideness settings etc will be implemented?
10:20:54LinusNRockdude05: huh?
10:21:05Coldtoaststereo width
10:21:20ColdtoastI'm assuming
10:21:25Rockdude05oh eq etc. its been mentioned before. stereo wideness etc for oggs and stuff that sound condensed etc etc etc
10:21:36LinusNah yes
10:21:45LinusNnobody works on it
10:21:48Rockdude05its just good for car stereos etc
10:21:51Rockdude05oh ok
10:21:54Rockdude05not a good sign ;)
10:22:04LinusNwell, we can't do everything at once
10:22:17Rockdude05ah ye i know
10:22:42Coldtoastwhat was the conclusion on the volum elimitng? sill there be an option for that?
10:22:44Rockdude05*gets out the whip and tells LinusN to work harder*
10:23:14LinusNColdtoast: i would want an option
10:23:42LinusNon the other hand, i'm quite happy with the volume as it is, since i never boost the bass/treble
10:23:56Coldtoastme too :) just wondered if you had decided not to do it
10:24:00amiconnI wouldn't want another option
10:24:13amiconnRockbox already has a ton of them
10:24:22LinusNhow would you want to solve this issue then?
10:24:40 Quit thegeek_ (Read error: 110 (Connection timed out))
10:24:52amiconnI actually think that the iriver way of handling it is better in a number of cases, but it should indicate what it is doing
10:25:39amiconn(Like, the treble/bass adjustment should show the set (desired) value along with the actual (limited) value
10:25:49Rockdude05solution? ever heard of trubass? hahaha
10:26:10amiconnRockdude05: You can't get around of hardware limits
10:26:35LinusNRockdude05: we will implement a similar thing eventuslly, but that is a completely different issue
10:26:48Rockdude05no i was mentioning the crappy 'bass booster' in iriver firmware my making middle frequencies stupidly high
10:26:58Coldtoastthe way it is at the moment, the top, what, 25% of the volume level is useless as it doesn't actually affect the volume. Maybe make the volume relevent to what eh max volume actually is. So 100% is ACTUALLY 100% of what the current max volume is (depending on the EQ)
10:27:16Coldtoasterr, useless if you boosted the bass to max, that is
10:27:34LinusNthe your headphones are useless imho :-)
10:27:39amiconnColdtoast: You mean, scale down the whole range instead of just cap?
10:28:03amiconnThat still won't make it any louder...
10:28:08Coldtoastno it won't
10:28:32Coldtoastbut you were saying you don't want the player to "lie" werent you?
10:28:38amiconn...not that I would care. I never use anything even near 100% with my earphones
10:28:57LinusNi often set it to 100% when i have it in my car
10:28:57amiconn(and earphones being the only case where I do use non-flat eq settings)
10:28:58Rockdude05might i add.... chuck the standard headphones and spend a few pounds/dollars on some decent sony etc ones - 24db bass on iriver phones is as good as 16db or less on good h.phones
10:29:19ColdtoastI have ti at 100% when I'm at teh gym cos it's so loud there
10:29:20LinusNbecause my fm transmitter is volume-triggered
10:29:42ColdtoastI need some Stymotics or something
10:29:44amiconnRockdude05: I never go above 6..8 dB of bass with any phones
10:29:58amiconnLinusN: Yes, but with fm transmitter I guess you use flat eq?
10:30:06LinusNi always use flat eq
10:30:29ColdtoastI use flat EQ most of the time but only cos I want the extra volume :)
10:30:38amiconnI didn't find any phones that sound decent with flat eq at the volume levels I am listening
10:30:57amiconn..especially as iriver doesn't have loudness
10:33:43Rockdude058db bass? my god.... itll sound like your in a sewer ;) and 100% volume.... i recommend getting a hearing aid
10:34:08Coldtoastthat's silly to day
10:34:21Rockdude05oh you know im not serious
10:34:31Coldtoastyou need to bear in mind that no two models fo earphones are the same
10:35:06Coldtoast100% on some are as loud as 70% on another
10:35:20Rockdude05not, but im sayng if you wat good bass at high volume, the iriver obviously cant hack it, so get some good headphones with good driver unts - only hope to be honest
10:35:36Coldtoastthat's untrue too. heh
10:36:00Coldtoastyou can get REALLY good bass with the Iriver firmware sing SRS
10:36:15Coldtoasterm.. using
10:36:31Coldtoastso you can't say the unit is incapable of it
10:38:11Rockdude05SRS uses other frequencies and settings that arnt all very low ones
10:38:20Rockdude05you could say it fools your ears
10:39:16webguest89LinusN: The fixes you did to convbdf for 16 pixel wide characters... Don't seem to be working for me
10:39:31LinusNwhich font?
10:39:51webguest89Nobby provided me an updated convbdf (see
10:39:57webguest89I re-created some .fnt files using that program e.g. win_crox4h.bdf
10:40:10webguest89I still get the problem with the display of wide characters
10:40:16webguest89And don't get any conversion warnings about characters being too wide
10:41:39amiconnRockdude05: Just remind you... I'm staying *far* away from 100% volume, and *therefore* need 6..8 dB bass boost (and treble as well) to get decent sound
10:41:52amiconnOn iriver, I use 55..60% volume
10:42:30amiconn...with earphones
10:42:49amiconn..and I believe they are ok-ish ones (Sennheiser MX-450)
10:43:19Rockdude05and more bass=bad sound at 60%? lower=good?
10:43:40 Join DarkkOne [0] (
10:44:02DarkkOneAnyone around?
10:45:12amiconnRockdude05: More bass is just too much bass, and less is too little, at least for my taste
10:45:45amiconnBasically I need the bass and treble boost as a replacement for the non-existing loudness function
10:46:08LinusNwebguest89: which chars are wrong?
10:46:40LinusNah, i see it
10:48:07DarkkOneDoes anyone by chance know why an h120 wouldn't turn on at all? I understand a purse got dropped on top of it. The charging light shows when it's plugged in, but the HD light never flashes, and nothing displays on the screen... (I think he has rockbox installed. A panicked friend has just called me.)
10:48:07webguest89was that ah for me?
10:48:32webguest89lower case m
10:48:40webguest89upper case M
10:48:47LinusNi see it
10:48:49webguest89goes afk
10:49:01ashridahDarkkOne: does it turn on when the reset button is pressed, but not the power button?
10:49:33DarkkOneHe's searching for something to press reset with, we'll see.
10:49:43DarkkOneWhat are you thinkin'?
10:50:49ashridahnot much, but reset bypasses some of the circuitry the play button uses.
10:51:14*DarkkOne mutters.
10:51:46DarkkOneHe can't find *anything* (I suspect he's just lazy again.) But thanks for the idea. He seems to just want to wait until the next time I'm in town. =/
10:51:48 Quit midk (Read error: 110 (Connection timed out))
10:51:56ashridahbut, chances are it's screwed. the charging circuitry is relatively separate from the rest of the unit.
10:52:09ashridahi'd suggest refraining from dropping things on the unit would be a good move in future
10:52:15DarkkOneIt was his girlfriend.
10:52:23DarkkOneShe's got this anti-technology aura.
10:52:31ashridahhow heavy was the purse?
10:52:32Rockdude05haha i threw mine 4 metres onto a pavement and all i got was a loud unit
10:52:44DarkkOneI've forbidden her from touching any computer I own. Checking her email can necessitate a reformat.
10:52:45***Saving seen data "./dancer.seen"
10:53:01Rockdude05even kept playing..... bounced away. oh how i love metal
10:53:04DarkkOneHe doesn't know.
10:53:39DarkkOneHe says it doesn't respond *at all* when the play button is pressed. No lights flicker, no HD sounds, nothing.
10:54:20*ashridah shrugs
10:54:23ashridahget her to buy him a new one
10:54:34ashridahshe'll learn once she pays often enough
10:55:09DarkkOneYeah, he's pretty upset since H120s are hard to find.
10:55:22DarkkOneWell, outside of overpriced auctions on ebay.
10:57:02DarkkOneWell, thanks. Apparently he's going to go explain to her why she needs to look where she's putting things in the future.
10:57:04*DarkkOne sighs.
10:57:06ashridahhow old is his unit?
10:57:19DarkkOneHis warranty expired like, two months ago he said. Which's why he called me.
10:57:37DarkkOneHe wanted me to take it to Best Buy for them (where he bought it) because the local one where he is and I have history. Heh.
10:58:20 Join hshah [0] (
10:59:16DarkkOneAh well. Mine still works. :)
11:04:41LinusNwebguest89: the new convbdf.exe doesn't accept the win_crox4h font
11:04:55LinusNError: Too wide characters (>16 pixels)
11:06:02 Quit Coldtoast (Read error: 113 (No route to host))
11:08:54hshahbah LinusN... stop making all these changes,,, ur making me recompile every time :p
11:09:22 Quit Rockdude05 ("Chatzilla [Firefox 1.0.4/20050511]")
11:10:20 Join B4gder [0] (
11:17:51 Quit DarkkOne ("Chatzilla 0.9.68a [Firefox 1.0.6/20050716]")
11:20:17 Join crash_ [0] (
11:30:12webguest89is back
11:30:54webguest89LinusN: i didn't get a message that characters were too wide. Is it possible that I have an old version despite Nobby 's assurances?
11:38:52 Quit webguest89 ("CGI:IRC")
11:38:57 Join webguest89 [0] (
11:39:02hshahLinusN - in the near future, will it be possible to edit the FM Radio WPS?
11:41:57 Join midk [0] (
11:42:47hshahalso, LinusN, under the debug menu... how come my Hardware Info doesn't do anything... im assuming thats normal :p
11:46:45 Join Bger [0] (n=Bager@
11:47:26amiconnhshah: Yes it is. It only does something on archos
11:48:02amiconnThat's because the iriver hardware doesn't differ within the same model, at least we don't know of such cases yet
11:48:28amiconnOn archos, there is a bitmask that tells us about some differences
11:53:03hshahahh rite thanks :)
11:53:20hshahany idea about the radio wps while ur at it?
11:53:50 Join hicks [0] (
11:56:11 Join webguest22 [0] (
11:56:59webguest22The h300s differ within the same model us vs. int. versions
11:57:07 Quit webguest89 ("CGI:IRC")
11:57:09webguest22so it'll come in handy there again
11:58:51amiconnhshah: There were some thoughts about a 'while fm screen' similar to the while playing screen, but I doubt that it is high priority
12:00:45 Join Moos [0] (
12:00:54webguest22not much you can do, is there?
12:01:04 Join phaedrus961 [0] (
12:02:00hshahok thanks amiconn
12:16:10 Quit webguest22 ("CGI:IRC (Ping timeout)")
12:18:03 Join Kelem [0] (n=Kelem@
12:26:46Kelemhe's alive ! ;)
12:27:15Bgerthere're more alive, but most of them doesn't respond to "hi";)
12:27:24Bgerto a simple "hi" i mean
12:27:48Kelemhehe, no matter, it was just for being polite, i'm not offended
12:28:03*B4gder waves to prove his aliveness
12:29:42 Quit B4gder ("Lämnar")
12:36:27 Join Chamois [0] (
12:38:19 Join ender1 [0] (
12:45:45 Join Sucka [0] (
12:52:46***Saving seen data "./dancer.seen"
12:56:28 Join B4gder [0] (
13:00:18 Quit B4gder ("Lämnar")
13:02:39hshahhow do i make a new wiki page?
13:02:55Bgertype a name
13:03:09Bgerand search for it
13:03:19hshahu what?
13:03:37Bgerfor example
13:03:42BgerBlaBla is the name
13:03:58hshahand then...
13:04:16Bgerthere's link "Create"
13:05:40hshaherm... im stuck
13:05:47hshahwhere do i type the name in
13:07:07Bgerwhat's the WikiPage you want to create
13:08:08 Quit ender` (Read error: 110 (Connection timed out))
13:08:22BgerUsefoolTools ?
13:08:32Bgertc, it's not
13:08:43hshahSimplified Guide to Modifying Rockbox
13:09:05hshahsee the USB Logo forum post in Rockbox iRiver forum#
13:10:04Bgerbut the wiki page must be without spaces
13:10:17Bger <= see this
13:11:05 Nick ender1 is now known as ender` (
13:12:56hshahnow time to work on it
13:13:28Bgerhm, i don't think this is the right name ...
13:13:56Bgerhm, maybe it is :)
13:14:07hshahwhat should i call it then?
13:14:15hshahits the n00bs guide to compiling
13:15:16Bgerthen it's ok
13:19:23 Join B4gder [0] (
13:24:59 Join webguest98 [0] (
13:25:19webguest98can anyone help me with a query?
13:25:27BgerSQL ?
13:25:33webguest98lol no
13:25:41webguest98mySQL works fine for me atm
13:25:42Bgerhaha :)
13:25:58Bgerjust ask ;)
13:26:26webguest98seriously though, does anyone know if Rockbox on an iRiver 300 series would work okay? okay being not crashing right off the bat
13:26:49Bgerit is not working AT ALL on H3x0
13:26:54webguest98lol k thanks
13:27:04webguest98darn... I want playlist otg...
13:27:20Bgerand DON'T TRY to flash it with the bootloader for h1x0
13:27:35webguest98well, that's why I came here to check
13:27:40Bgerwebguest98 wait a little bit more
13:28:00webguest98good luck with the project, can't wait till you get a H300 version
13:28:08*Bger too
13:28:12Bgeri have h3x0
13:28:22 Quit webguest98 (Client Quit)
13:29:36Kelemtalking about h3xx, do you know if the lcd found a couple of weeks a go was the good one ?
13:30:02Kelemor is the bootloader still to be completed ?
13:30:37Bgernoone has tried it
13:30:42Bger(the LCD)
13:31:27Bgeri think we'll know for sure about the time when linus connects his bdm to his h320
13:31:29Kelemi'm not really concerned (i own a h120) but i think that would be great for rockbox
13:32:06Bgeriaudio X5 too :)
13:32:08Bgerand M3
13:32:40Kelemthe more hardware supported, the better it will be :)
13:34:27Bgermaybe H3x0 and X5 will be the first players, which are still selled (was it irregular...) when rockbox already runs on them
13:35:55Kelemi wish I know rockbox run on my new hardware if i had to change mine
13:36:58 Quit Nibbler (Read error: 104 (Connection reset by peer))
13:37:22 Join Nibbler [0] (
13:38:28 Part Kelem
13:42:41 Quit B4gder (Read error: 110 (Connection timed out))
13:53:25 Join Bugder [0] (
13:57:54Bugderwhy would you need winrar to compile Rockbox?
13:58:05*Bugder reads the new wiki page
13:58:32Bgerfor unziping the source ? :)
13:58:45Bgerbleeding edge source
13:58:56Bugderbut surely cygwin contains such a tool
13:59:22*Bger is going closer with GIMP
14:01:15*t0mas is still looking at his "beat flashing" remote
14:08:00 Join Bgr [0] (n=Bager@
14:13:33 Quit Bger (Nick collision from services.)
14:13:37 Nick Bgr is now known as Bger (n=Bager@
14:13:59 Quit Bugder ("CGI:IRC (Ping timeout)")
14:32:50hshahhow do u think my new guide is going?
14:33:48 Join muesli- [0] (
14:36:08ender`why winrar?
14:38:37hshahto extract the source :p
14:38:53hshahim trying to make it easy as possible
14:39:22hshahand plus im writing down what i do...
14:39:58amiconnWinrar isn't free
14:40:18hshahyes it is
14:40:26hshahthe trial version is anyway
14:40:56hshahthe same way u can get a free version of winzip... but it just asks u to register etc
14:41:11amiconnIt's not free. Even the trial is time limited
14:41:30amiconn...unlike winzip, but still winzip is shareware (but free for personal use)
14:41:54hshahhmm... dammit... they changed it... it used to be free
14:43:10midkamiconn, hmm? i've never had the trial of winrar run out on me, how recently was it changed?
14:43:33Bgerafaik WinRAR runs, it just shows a warning dialog box
14:43:36Bgerwhen you start it
14:43:43Moos7-zip is free, no?
14:43:50Bgeryep, it's free
14:43:55Bgerunzip is free too :P
14:44:14Bgeranyway, winrar is 40days trial
14:44:34Bgerafter this it will continue to run, but it's illegal to use it ;)
14:44:36midkBger, right, i've only ever got the dialog box...
14:44:42midkhaha.. wow, spooky. :)
14:46:31amiconnmidk: All winrar version above 2.50 I tried had this timeout
14:46:55Bgeramiconn: they continue to run
14:46:56amiconnI don't care anyway, since I use neither winrar nor winzip
14:47:08Bgeramiconn what do u use ? ;)
14:47:10midk2.50? hmm? I've been using versions past that for longer than 40/45 days... :)
14:47:42amiconnI windows I use Filzip. It's free (though not opensource :/ )
14:47:55BgerFilzip ?
14:48:00amiconnIt does uncompress rar and ace as well (not that I need that often)
14:48:21Bgernew rar too, i suppose (version 2.9 and later) ?
14:48:46hshahok - i have ruled out the use of winrar now
14:48:53hshahjust use bash commands
14:49:35midkhshah, i'd suggest going into detail on using CVS to checkout, and simply mention that they can use the program of their choice to unpack a daily build tarball, if they so desire..
14:49:43midkwhich sounds like something you might be doing. :)
14:49:52hshahgunzip FILE.tar.gz and then tar -xvf FILE.tar
14:51:01hshahim going into detail on how i do the compiling etc... so that any n00b who decides "ive never touched programming in my life, and it would be great if i even knew what it was, but what the heck, i still want to add a patch to rockbox and compile my code, but how the hell do i do it"
14:51:10hshahcan easily follow the steps and do it
14:51:31midkthat's good, but it sounds like you're having a little trouble with suggesting a program to actually get the source to compile.
14:51:43midki'd personally find CVS a lot easier, two commands and you've got all the latest source...
14:51:45hshahcygwin does it...
14:52:02hshahi use winrar to extract the tar.gz
14:52:21midki'm talking about the guide.
14:52:35hshahim confused...
14:52:41muesli-hours later..impressing how long you can discuss about such a nonsens :D
14:52:49***Saving seen data "./dancer.seen"
14:52:55midkhaha. i'm saying it seems like you're saying how to get the source from a tar file, right?
14:53:03midkwith a program to extract it.
14:53:30midkand i'm pointing out that cvs is a faster way to get source more easily than unpacking a tar file.
14:54:22hshahok... to tell u the truth i never really got the hang of CVS :p... so mebbe u could write me a n00bs guide on how to use CVS :p
14:54:32midkhaha... :)
14:55:36midkyou login to cvs in cygwin with "cvs login" minus quotes..
14:56:04midk[press enter when it asks for password after entering that].. then checkout with "cvs -z3 co rockbox" minus quotes.
14:56:16hshahim trying to make this guide into something i would understand... coz this is the first time ive done this sort of thing... so if i can explain how i did it... most n00bs should understand
14:56:17midkactually, use "cvs -z3 co rockbox-devel"..
14:56:28hshaherm... thats just confusing....
14:56:35hshahi think my way is easier :p
14:56:43midkyeah, that's cool.. i find cvs easier, but maybe it's a little trickier to learn. :)
14:56:54midkhaha... just two commands. :)
14:57:20 Quit Sucka ("a bird in the bush is worth two in your house")
14:57:21hshahmost noobs can download a file and type gunzip file.tar.gz and then tar -xvf file.tar to get the uncompressed code :p
14:57:25Slasheryou have to type the two long lines only once
14:57:37Slasherafter that cvs update will do the job in the future
14:57:50ender`no need to gunzip the .tar.gz file
14:57:58ender`just use tar zxvf whatever.tar.gz
14:58:05midkSlasher, true.. i never used update... i'd always just re-checkout if i needed. :)
14:58:07ender`(or tar jxvf whatever.tar.bz2)
14:58:15hshahok... thanks ender
14:58:25Slashermidk: oh, that's not very practical
14:58:37midkSlasher, i didn't do it often, just when i needed.
14:58:43Slashercvs update -dP (as necessary) will update everything fast
14:58:59midki'd work with source for quite a while before updating it if i didn't need to. :)
14:59:00 Quit Bger (Read error: 104 (Connection reset by peer))
14:59:28 Join Bger [0] (n=Bager@
14:59:39midki tend to work on something with the same source until i finish, then re-checkout and move the modified files back as long as they haven't been changed in the new checkout...
15:00:02Bgerthat's silly
15:00:03Slasherwhy bother that much hassle?
15:00:11Slashercvs update does everything for you
15:00:18midkhassle? i only did it once every half-week or so.
15:00:26Slasherit will merge your changes with the updated source files
15:00:28midki usually worked on plugins too...
15:00:56 Join BBub_ [0] (
15:01:00midkSlasher, i get that, if i was changing something that had been changed since the last update i did i'd probably give cvs update a shot...
15:01:08midka different story with plugins, i think. :)
15:01:34 Join Bugder [0] (
15:16:00BgerSlasher why did you "eat" your "y" ? ;)
15:16:33muesli-Bagder why did you eat your "a" ;)
15:16:46Bgeri'm not bagder :)
15:16:53muesli-a clone of him?
15:17:00ashridahgroupie! :)
15:17:28muesli-yeah, rbx groupie :D
15:18:02*ashridah ponders
15:18:15 Quit BBub (Read error: 110 (Connection timed out))
15:18:15 Nick BBub_ is now known as BBub (
15:18:16ashridahshould i be worried that the oracle error message reference PDF is 8.3 megs? :)
15:18:20 Join rooom_wrk [0] (
15:18:25 Quit rooom_wrk (Client Quit)
15:18:37*ashridah assumes it's a compendium of ALL oracle error codes, from everything
15:18:50SlasherBger: Hmm, how?
15:18:58Bgerdepends on what do you mean with "worried" ;)
15:19:24Slasherhmm :D
15:19:52ashridahBger: well. lots of error codes sort of hints that lots can go wrong
15:20:05Slasherah, my default nick is slasher but slasheri if the default is not available
15:20:09ashridah(not that it's my job to administrate the oracle setup, just connect to it and fire off sql queries
15:20:20Bgerashridah: on the other hand -> well documented
15:20:53Bgerou, yes, sorry, it was SlasherI
15:21:01SlasherIt was Slasheri :D
15:21:11Bgeri'm sorry
15:21:16SlasherBut Slashery sounds interesting
15:21:17Slasherhehe :D
15:21:19Bger(part of bulgarian joke)
15:21:34 Join Coldtoast [0] (
15:21:50Bgerbut i suppose #rockbox is not the right place to tell jokes ;)
15:22:05muesli-not really
15:22:10muesli-we hate jokes
15:22:11Bugderwe have too few bulgatian jokes in here
15:22:19muesli-will send you a howler otherwise
15:22:44muesli-am not kidding my mate :D
15:24:48Coldtoastinteresting... Thunderbird doesn't encrypt your email
15:25:21 Part LinusN
15:25:43Bgerhi, L
15:26:04Bgeranyone has any idea how to correct "reddish" picture in GIMP ?
15:26:16Coldtoastdelete the red channel :)
15:26:40Bgermaybe some other day
15:26:46Coldtoastdon't you have curves in Gimp?
15:26:52Bgeryep, i have
15:26:54Bgerbut ...
15:27:10Bgerthe white continues to be "reddish"
15:27:23Coldtoastare yo sure it's not your monitor?
15:27:33Bgeryes, i am :)
15:28:07ColdtoastI don't mean faulty :) you can adjust the temperature of your display
15:28:19Bgerhm, i think i got it
15:28:33Bgerno no
15:28:51Bgerit's just a picture taken with Konica film ...
15:28:57Bgeron konica paper. ..
15:29:18ColdtoastI was going to ask if it shoots RAW at all
15:29:23Coldtoastbut oh man.. film!?!!
15:29:37Coldtoasthow analogue!
15:29:48Bgerreally ? ;)
15:34:26 Quit ashridah ("Leaving")
15:34:47Coldtoasthaha. reading on this Linksys forum about a guy who's set up a CRON job to load a different firewall rule at 10:30pm every night so he could block his son's web access
15:35:16Coldtoastbut he's gotten annoyed his son gets around it by jumpin gon his neighbour's unsecured access
15:35:34Coldtoastthat's pretty funny
15:38:30Coldtoastand there's another who says his daughter uses EMule to download illegal music and he's worried he'll get busted. So he wants to know if he can restrcint her speed SO much she'll get frustrated and not bother doing it so he can get full speed with HIS EMule downloads of Seinfeld and Frasier. heh
15:39:02Coldtoastdon't ppl think about stuff before they post it? :)
15:39:14solexobviously not!
15:40:52solexmaybe he read all that buzz about illegal music and doesn't think it also applies to TV shows:)
15:41:13Coldtoastthat'd be my guess too. heh
15:45:09 Quit muesli- (Read error: 104 (Connection reset by peer))
15:45:22t0masjust like the guy about "chatting longdistance" on
15:46:14Coldtoastdidn't see that
15:46:30Coldtoastwhat was that about?
15:47:47 Nick lostlogic_ is now known as lostlogic (
15:48:32Coldtoastcan that POSSIBLY be real?
15:48:55Bgercool :)
15:49:05Coldtoastgotta love
15:50:18 Quit hicks (Remote closed the connection)
15:51:07hshahits obviously not real... and if it he... he was some retarded mofo
15:51:25hshah* if it is
15:51:29 Join hicks [0] (
15:57:28 Quit Bugder ("CGI:IRC")
15:57:37solexDamn, makes me want to lao while sitting in my classes...
15:57:51solex*laugh out loud
15:59:53SlasherHmm, does anybody know if the voice samples available from should work with iriver too once the api has been written?
16:00:05Slasherif those should work, i could try with them
16:00:27hshahhmm... why is my cygwin here different to the one at home...
16:00:39amiconnSlasher: I don't think so
16:00:50BBubi heard its in the progress
16:00:51Slasheramiconn: Hmm.. :/ what's wrong with them?
16:00:54amiconnThe voice files for archos contain pre-swapped mp3 clips
16:01:16Slasherthen i have to test with some other stuff
16:01:35amiconnYou know, the MAS mp3 codec wants the bits in the opposite order than the SH1 SPI sends them, so each and every byte has to be bitswapped
16:02:05amiconnThe voice files contain the clips pre-swapped in order to avoid swapping them every time they are loaded
16:02:11Slasherah, so bitswapping them again should make them work
16:02:21amiconnYou might have a look at the format, and think about a way to extend it
16:02:42amiconnWe'll probably need a byte indicating the clip format, i.e. codec used
16:02:59amiconn(and a special codec indication for mp3 pre-swapped for archos)
16:03:29Slasherbtw, what bitrate and sample frequency is used? i assume they are all mono samples
16:03:57BgerMPEG 2.5 ?
16:03:58amiconnIf we have an extended specification, I could build voice files for just about any codec where a command line encoder running under windows exists for
16:04:16Slasherhmm, sounds great
16:04:18amiconnThe clips in the current voice files are all MPEG2.5 layer 3 12 kHz mono
16:04:28Slasherah, ok
16:05:16amiconnI think we could start with non-swapped mp3 (as long as speex is not yet running on iriver)
16:05:25amiconnWe could use better quality on iriver as well
16:05:41amiconnThe 12 kHz is just to cut down file size as much as possible
16:05:59amiconnThe file has to fit entirely into RAM with the current design
16:06:16Slasherwith iriver that wont be a problem
16:06:23amiconnyes, exactly
16:06:43Slasherbut the bitrate etc. should be low enough that we could use voice ui even with q10 ogg files for example
16:07:07amiconnToo high a quality doesn't help speech that much
16:07:18amiconn...but I know that 16 kHz sounds way better than 12
16:08:04amiconnThere is one more limit in the current voice files - there are no frames with a bitrate >64 kbps (the clips are VBR)
16:08:38amiconnThis is because the MAS has a bug with MPEG2.5 - it doesn't play frames with bitrates >64kbps
16:09:38Slasherthe voice codec will be also slower if we use libmad because there is not enough iram space to put both stacks in iram
16:09:43amiconnAlso, the clips don't contain any tags or extra headers. No id3v1/id3v2/xing/lame header, just the raw frame
16:13:45Slasherhmm, has the current kernel any simple mechanism to prevent single thread from waking up? if not, i might have to implement one
16:15:53Slasherbecause when we hotswap the codec, also the old codec code memory will be replaced with the new codec (codecbuf[]) and we definately don't want the old codec thread to wake up when the code memory has a different code
16:16:00 Quit t0mas ("CGI:IRC")
16:16:14Bgerjust exit the thread ?
16:16:31Slasherthat's not possible, we have to keep the thread alive
16:16:33 Join t0mas [0] (
16:16:35Slasherwe only suspend it
16:16:45amiconnSlasher: I think we need a separate codec memory space
16:16:54t0masI just had to click a link
16:17:00Slasheramiconn: Hmm.. i don't think that's necessary =)
16:17:09*t0mas doesn't have to go to school for a few more weeks :D
16:17:20t0masa crane has fallen on my school :D
16:17:26amiconnSlasher: You want to decode interleaved instead of parallel?
16:17:37 Join XMaster-ShadowX [0] (
16:17:43Slasheryes, it's the only way to go
16:17:43t0masfor those how can read dutch:
16:17:51Slasherbecause we have shared IRAM also
16:18:10amiconnNot if we link the codec a second time
16:18:19Slasherhmm, yes..
16:18:33amiconnWe could even have the secondary codec not using iram at all
16:18:35Slasherbut i am not sure which one would be easier/better solution..
16:18:41Slasherboth are hard to implement
16:19:10Slasherwith the current solution i am trying, you could use exactly the same codecs for voice and audio
16:19:16amiconnI'm not sure whether it is a good idea to load the voice codec from disk
16:19:38amiconnIf we do, they should of course not be duplicated, but there is a disadvantage
16:20:05amiconnWith disk-based codecs, there is a lot more that can go wrong
16:20:27amiconn...and with the voice UI not working, it will be hard for a blind to figure out what it is
16:20:36Slasherhmm, true..
16:21:07amiconnOtoh, if we build the voice codec into the core, this will be a hard selection
16:21:08Slasherbut if you are not using voice ui, it always eats some ram if it's static
16:21:23amiconn..and if we want to change codec later, the voice won't work...
16:21:44amiconnSlasher: Yes, correct
16:22:16amiconn...but the voice UI will eat some RAM permanently anyway, even if disabled
16:22:35Slasherwhy it would if it's dynamic?
16:23:00amiconnIt will at least increase code size
16:23:09amiconnDo you think decoding interleaved will be fast enough even with longer voice sequences?
16:23:14Slashertrue, but that's quite minor increase..
16:23:52Slasheramiconn: if both codecs can do about 200% realtime it should be.. but i am going to find out =)
16:23:56amiconnThere is spell mode for directries and files names, and the queue can hold up to 64 entries...
16:24:25amiconnWhat about letting the codec thread doing the decoding for both audio thread and voice UI?
16:24:35amiconnThen there is no thread concurrency problem...
16:24:55amiconnAnother option would be using a mutex to access the codec slot
16:25:02Slasheri think that's not possible.. what if the audio codec loaded is ogg decoder for example and we have mp3 voice data?
16:25:17amiconnThen the thread needs to change codec
16:25:30Slasheryes.. it needs to unload the previous codec
16:25:38amiconnThe voice codec should be loaded to RAM if the voice UI is enabled
16:25:57Slasherthat could cause glitches to playback if the codec wasn't designed to be unloaded between decoding
16:26:06amiconnHmm, yes
16:26:34amiconnA codec would need to save state if it's not a stateless one like wav or mp3
16:26:58amiconnThat renders your idea of a shared slot pretty much unusable...
16:27:07SlasherHmm, yep. That might be one solutions also
16:27:10amiconn...regardless of the interlock mechanism
16:28:25amiconnHmm, all persistent data should be static, correct?
16:28:49Slasherthe implementation i am currently trying, would save codec IRAM state
16:29:12amiconnWe would have to make sure no callback (from codec to core) is running, then it could save the whole codec RAM into a buffer
16:29:25amiconnI'm not sure whether IRAM is enough.
16:29:35Slasheryes, it will do that
16:29:45amiconnThe codec might store some data in DRAM
16:30:07 Quit hicks (Remote closed the connection)
16:30:26amiconnThis could still be done within one thread, or you could go for the mutex
16:30:45Slasherbut about the callbacks.. do you mean that i couldn't initiate the codec swap in a callback from codec to core?
16:31:03amiconnI think this won't work
16:31:16amiconnThe callback would return to an address that's no longer correct...
16:31:47amiconnYou would need to mangle the stack, but that would be really dirty...
16:31:54Slasherbut we will suspend (sleep or something like that) the thread before it returns
16:32:29Slasherand the thread would remaing suspended before we swap the codec again and resume the thread
16:32:46amiconnHmm, that might work
16:33:03amiconnStill doesn't sound like a clean way though...
16:33:36Slasherbut it could be easier/cleaner than saving codec state in codec itself and reloading codecs
16:33:45amiconnI don't get why the swap should be initiated by the codec anyway
16:34:00amiconnDon't you think the core will know better when it's necessary to swap?
16:34:51Slasheryes of course.. but the callback would be good place to do the swap because that's the point when we have finished decoding one frame
16:35:04amiconnI think that it might work as follows:
16:35:27amiconn(a) codec thread receives a request to decode a bunch of voice frames
16:35:46amiconn(b) codec thread lets codec finish the current frame or similar
16:36:07amiconn(c) codec thread copies codec RAM+IRAM slot to memory buffer
16:36:20amiconn(d) codec thread copies voice codec into slot
16:36:39Slasherplease note that the codec thread is the codec itself, so it cannot control codec when it's running
16:36:41amiconn(e) codec thread lets voice codec decode the bunch of frames
16:36:45 Quit Bger (""In the other world, in paradise, the beauty of women surpassed even the beauty of Bulgarian women" Adaloloddin Mohammed Balhi")
16:37:04Slasheroh, you mean the audio thread i think..
16:37:29amiconnSlasher: Yes, the codec thread will be int the codec itself most of the time, but there must be some sort of periodic return
16:37:35amiconnNo, I mean the codec thread
16:38:20Slasherthe codec keeps running until there is nothing more to decode or we have to switch to an other codec
16:38:21amiconn(f) codec thread reverses the codec swap process and continues decoding audio frames
16:38:42amiconnHmm. How does the codec notice that there is no more data?
16:38:57Slashercallback returns null and size as 0
16:39:23Slasherthe codec requests more data from the core
16:40:38amiconnStrange design, imho
16:40:48amiconnBut I might be wrong...
16:40:57Slasherhmm, but it still works :)
16:41:09amiconnSo the codec thread basically is the codec itself, unless a codec swap is due
16:41:30Slasherbasically the idea would be same what you thought about
16:42:11Slasherand there is also audio thread controlling buffer (most importantly keeping the buffer filled)
16:43:12amiconnThen we would need a second thread for the secondary codec, and if we want to use the same memory slot, both threads must be interlocked that only one can be within the codec code at a time
16:43:35amiconn...and it needs to swap from a callback...
16:43:39Slasherthat's true.. that's why we have to sleep the other thread
16:43:56amiconnJust use a mutex
16:44:17Slasherah, mutex sounds fine :)
16:44:21amiconncodec_slot_mutex or somesuch
16:45:33amiconnYou will still need a mechanism to tell the codec thread that the callback should neither hand the codec more data, nor hand it NULL to exit, but enter the lock state
16:46:08amiconn...and then hand it more data or NULL after leaving the lock state
16:46:27SlasherHmm, i think the callback itselft could do that because it goes into the playback system and can determine what it has to do
16:46:36Slasherso we could just lock the callback
16:46:55amiconnYes, but you need a way to tell it to do that
16:47:04 Quit einhirn_ ("Miranda IM! Smaller, Faster, Easier.")
16:47:14SlasherHmm, that might be true
16:47:25amiconnI think it might work as follows
16:48:02amiconnThe voice thread should work the same way as the codec thread, just for the voice codec
16:48:48amiconnEach of the 2 hreads would lock the mutex before calling codec code, and the callback would unlock it, check some things, yield, then lock again
16:49:28amiconnThe yield() is crucial here, it allows the other thread to obtain the slot if it is waiting in mutex_lock() because it has got work o do
16:50:13Slasheryes, that should work
16:51:14amiconnAdditional yield()s within the codec are always possible, they won't change the lock state
16:52:53***Saving seen data "./dancer.seen"
16:52:55amiconnHmm, we'll need a method to track which codec is loaded in the slot
16:53:18Slasheryep, i have already done that
16:54:06amiconn...actually we need to track which thread used the slot last
16:54:18amiconnEven if they use the same codec, the state might be different
16:54:21Slasherthere is now current_codec -variable that would indicate the status as following: -1, only single audio codec loaded. 0, primary audio codec loaded. 1, secondary voice codec loaded
16:55:47amiconnWhy 3 states?
16:56:31Slasherif status != -1, it indicates that we will have to do a codec hotswap. Then we will update the status to indicate which codec is currently active
16:56:57 Quit Chamois (" HydraIRC -> <- The professional IRC Client")
16:57:16Slasherif status == -1, there are no dual codecs loaded and hotswap should not be tried
16:57:53amiconnIf each thread just compares whether the status isn't equal to its own codec loaded, then hotswap, we would only need 2 states
16:58:25SlasherAh, hmm.. yes
16:59:16amiconnThis would even work with more than 2 threads if we'll ever need that ;)
16:59:26Slasherhehe :)
17:07:00 Quit Coldtoast (Read error: 104 (Connection reset by peer))
17:07:06hshahphew - finished the basic instructions
17:09:05 Quit courtc (Read error: 113 (No route to host))
17:12:11 Join hicks [0] (
17:19:35 Quit hshah ("CGI:IRC")
17:33:58 Join einhirn [0] (
17:34:34 Join amiconn_ [0] (
17:51:33 Quit amiconn (Read error: 110 (Connection timed out))
17:51:33 Nick amiconn_ is now known as amiconn (
18:08:59 Quit XMaster-ShadowX (Read error: 113 (No route to host))
18:24:32 Quit pike ()
18:25:52 Join pike [0] (
18:26:11 Join thegeek [0] (
18:31:49 Join Lear [0] (
18:33:23 Join webguest05 [0] (
18:34:12webguest05Hi Lear, have you given the shuffle and 'Track' feature any more thought ? :)
18:34:44webguest05ReplayGain Track that is
18:34:59Learno, not really... should be easy enough for anyone who's interested to implement though. :)
18:35:35webguest05yep , I'm fishing for a customer, no joy yet
18:52:56***Saving seen data "./dancer.seen"
19:07:17 Part webguest05
19:08:51 Join matsl [0] (
19:32:40 Quit einhirn (Read error: 104 (Connection reset by peer))
19:34:27 Join hshah [0] (
19:34:47hshahhello all
19:35:19hshahany chance someone here with a fast internet connection quickly do me a favour?
19:35:36hshahits for a wiki page... so you would be helping Rockbox
19:38:09BBubwhat is it?
19:39:34hshahI need someone to download
19:39:45hshahand then install it following the instructions on here
19:40:01hshahthe setup.exe is only around 280kb
19:40:10hshahbut then the program downloads other files...
19:40:18hshahi need to know the size of the other files downloads
19:40:21hshah(total size)
19:41:01amiconnThis totally depends on which components of cygwin you select
19:41:27BBubit also depends on the versions
19:41:46amiconnSorry that I can't do it; I already have cygwin installed, a rather big installation including X11 etc
19:42:49hshahother than the default ones only patch has been added
19:42:57amiconnTotal size here is around 418 MB (that's excluding rockbox sources etc)
19:43:17hshahas i said above... it would need to installed following those instructions in the wiki
19:43:31hshahi did it on 56k but i can't remember how long it took or how much it downloaded
19:45:18hshahcan't either of u two download it to a different directory for me and then delete it afterwards?
19:46:11amiconnCygwin stores configuration data in the registry, so no, not that simple
19:47:02hshahoh rite...
19:47:15hshahanyone hiding here who doesn't use cygwin
19:47:52BBubim already downloading it ;)
19:48:00hshahoh rite... ok cool :)
19:48:02amiconnHmm, why do the instructions say to download both the devkit and cygwin?
19:48:15hshahthose who are hiding, can remain hidden
19:48:32amiconnAfaik the devkit should contain a complete (minimal) cygwin dev environment
19:48:58hshahi wanted the patching thing i think... thats why i got both
19:49:15 Join whistlerbrk [0] (
19:49:21whistlerbrkAnyone around?
19:49:45amiconnhshah: The devkit doesn't contain patch?
19:49:47BBubhshah: 13,6 mb
19:49:52hshahas long as its easy and as long as it works... its prefect for n00bs like me
19:50:01hshahi don't think it does...
19:50:10hshahthanks BBub
19:50:21whistlerbrkI wanted to help with testing and development, who should I talk too to get involved/
19:50:57amiconnhshah: That's strange; something to fix for Bluechip then...
19:51:35amiconnI mean, there is no point in downloading several megabytes (more than the devkit itself) just to get a missing utility
19:52:05hshahthere must be an addon
19:52:08hshahwell i think i have one...
19:52:12hshahbut i dunno how to install it
19:55:24whistlerbrkis there anyone I can talk to here about getting involved?
19:56:56whistlerbrkguess not
19:57:00 Quit whistlerbrk ()
20:01:16 Join courtc [0] (
20:01:24hshahok... i have removed the need for cygbin now
20:14:15 Quit hshah ("CGI:IRC")
20:24:20 Join LinusN [0] (
20:24:54 Join rooomish [0] (
20:25:01 Quit rooomish (Client Quit)
20:35:24 Join gursikh [0] (
20:35:27 Part LinusN
20:36:28 Part gursikh
20:39:59 Join Jozi|afk [0] (
20:40:24 Nick Jozi|afk is now known as Jozi (
20:52:59***Saving seen data "./dancer.seen"
20:56:16 Join hshah [0] (
20:58:14 Quit hshah (Client Quit)
21:14:39 Quit Jozi (Read error: 110 (Connection timed out))
21:18:22 Quit gromit` (Read error: 104 (Connection reset by peer))
21:18:24 Join gromit`` [0] (n=gromit`
21:23:47 Join [1]Moos [0] (
21:23:47 Quit Moos (Read error: 104 (Connection reset by peer))
21:23:51 Nick [1]Moos is now known as Moos (
21:49:47 Quit t0mas ("CGI:IRC")
21:49:54 Join t0mas [0] (
22:17:50 Join bagawk [0] (n=lee@unaffiliated/bagawk)
22:27:05 Quit BBub (Read error: 104 (Connection reset by peer))
22:35:24Slashert0mas: hmm, could you send your "light organ" patch?
22:35:40Slasheri could modify it to support the backlight dimming
22:40:33 Join Jozi|afk [0] (
22:40:56 Quit Jozi|afk (Client Quit)
22:53:00***Saving seen data "./dancer.seen"
22:54:44Mooswhat is this "light organ" patch?
22:58:26*bagawk looks
22:58:31bagawkthis sounds interesting
23:05:58bagawkMoos, where is that?
23:09:46 Join hardeep [0] (i=hardeeps@norge.freeshell.ORG)
23:21:15 Quit matsl (Remote closed the connection)
23:24:31 Quit Lear ("Chatzilla [Firefox 1.0.6/20050720]")
23:42:02 Join ashridah [0] (
23:48:09 Join tvelocity [0] (

Previous day | Next day