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 2009-05-19

07:32:10***Started Dancer V4.16
07:32:10***Connected to on port 6667
07:32:10***Logfile for #rockbox started
07:32:18***Server message 501: 'logbot :Unknown MODE flag'
07:32:18Mode"logbot :+i" by logbot
07:32:19 Join logbot [0] (n=bjst@gateway/web/cgi-irc/
07:32:19 Join _lifeless [0] (n=lifeless@
07:32:19 Join saratoga [0] (i=9803c6dd@rockbox/developer/saratoga)
07:32:19 Join BHSPitMonkey [0] (n=stephen@unaffiliated/bhspitmonkey)
07:32:19 Join gevaerts [0] (n=fg@rockbox/developer/gevaerts)
07:32:19 Join cool_walking_ [0] (i=cb3b81c3@gateway/web/ajax/
07:32:19 Join perrikwp [0] (i=18ac0c41@gateway/web/ajax/
07:32:19 Join SirFunk [0] (
07:32:19 Join bubsy [0] (i=Bubsy@unaffiliated/bubsy)
07:32:19 Join killan [0] (
07:32:19 Join pixelma [50] (n=pixelma@rockbox/staff/pixelma)
07:32:19 Join amiconn [50] (n=jens@rockbox/developer/amiconn)
07:32:19 Join CaptainKwel [0] (
07:32:19 Join timc [0] (n=aoeu@
07:32:19 Join parafin [0] (
07:32:19 Join martian67 [0] (n=martian6@about/linux/regular/martian67)
07:32:19 Join krazykit [0] (
07:32:19 Join Xerion [0] (
07:32:19 Join arohtar [0] (
07:32:19 Join GodEater [0] (n=yeahrigh@rockbox/staff/GodEater)
07:32:19 Join Seed [0] (
07:32:19 Join kachna|lappy [0] (
07:32:19 Join daurn [0] (n=daurnima@unaffiliated/daurnimator)
07:32:19 Join linuxstb [0] (n=linuxstb@rockbox/developer/linuxstb)
07:32:19 Join avacore^ [0] (
07:32:19 Join barrywardell [0] (n=barry@rockbox/developer/barrywardell)
07:32:19 Join efyx_ [0] (
07:32:19 Join Rob2222 [0] (
07:32:19 Join JdGordon [0] (n=jonno@rockbox/developer/JdGordon)
07:32:19 Join tvelocity[a] [0] (
07:32:19 Join kadoban [0] (
07:32:19 Join Ridayah [0] (
07:32:19 Join bagawk [0] (
07:32:19 Join KBH [0] (
07:32:19 Join Torne [0] (
07:32:19 Join HellDragon [0] (i=jd@Wikipedia/HellDragon)
07:32:19 Join DataGhost [0] (n=dataghos@unaffiliated/dataghost)
07:32:19 Join J-23 [0] (
07:32:19 Join R[a]ndom [0] (
07:32:19 Join alexbobp [0] (
07:32:19 Join Galois [0] (
07:32:19 Join markun [50] (n=markun@rockbox/developer/markun)
07:32:19 Join Lss [0] (
07:32:19 Join TheDJACR [0] (n=TDJACR@Wikipedia/Thedjatclubrock)
07:32:19 Join antil33t [0] (n=Mudkips@
07:32:19 Join Chex_ [0] (
07:32:19 Join rasher [50] (n=rasher@rockbox/developer/rasher)
07:32:19 Join jordan` [0] (i=gromit@
07:32:19 Join _Auron_ [0] (
07:32:19 Join evilnick_home [0] (
07:32:19 Join jmillikin [0] (
07:32:19 Join Tristan [0] (
07:32:19 Join FlynDice [0] (
07:32:19 Join lightbulbjim [0] (
07:32:19 Join courtc [0] (n=court@unaffiliated/courtc)
07:32:19 Join fyrestorm [0] (
07:32:19 Join jon-kha [0] (
07:32:19 Join dionoea [0] (n=dionoea@videolan/developer/dionoea)
07:32:19 Join suom1 [0] (
07:32:19 Join Slasheri [0] (i=miipekk@rockbox/developer/Slasheri)
07:32:19 Join tchan [0] (n=tchan@lunar-linux/developer/tchan)
07:32:19 Join spyder [0] (
07:32:19 Join n17ikh [0] (
07:32:19 Join Zarggg [0] (
07:32:19 Join fred_2 [0] (
07:32:19 Join blithe [0] (
07:32:19 Join Bagder [241] (n=daniel@rockbox/developer/bagder)
07:32:19 Join Hadaka [0] (
07:32:19 Join shadearg [0] (
07:32:19 Join agaffney [0] (n=agaffney@gentoo/developer/agaffney)
07:32:19 Join crwll [0] (
07:32:19 Join fuzzie [0] (
07:32:19 Join rwong [0] (
07:32:19 Join lostlogic [50] (n=lostlogi@rockbox/developer/lostlogic)
07:32:19 Join Kohlrabi [0] (
07:32:19 Join tmzt [0] (
07:32:19 Join tarbo [0] (n=me@unaffiliated/tarbo)
07:32:19 Join jhulst [0] (n=jhulst@unaffiliated/jhulst)
07:32:19 Join Bombe [0] (n=droden@freenet/developer/Bombe)
07:32:19 Join yosafbridge [0] (
07:32:19 Join planetbeing [0] (
07:32:19 Join CIA-38 [0] (
07:32:19 Join crashd [0] (
07:32:19 Join webmind [0] (
07:32:19 Join pabs [0] (
07:32:19 Join freqmod [0] (
07:32:19 Join fxb__ [0] (
07:32:19 Join trisiak [0] (
07:32:19 Join FOAD [0] (
07:32:19 Join wookey_ [0] (
07:32:19 Join Foxx [0] (
07:32:19 Join msi [0] (
07:32:19 Join BlakeJohnson86 [0] (
07:32:19 Join Bawitdaba [0] (
07:32:19 Join bittin```` [0] (
07:32:19 Join Unhelpful [0] (n=Militant@rockbox/developer/Unhelpful)
07:32:19 Join vedlith [0] (
07:32:19 Join andy` [0] (
07:32:19 Join ze [0] (i=ze@
07:32:19 Join SUSaiyan` [0] (
07:32:19 Join scorche [50] (n=scorche@rockbox/administrator/scorche)
07:32:19 Join AlexP [0] (n=alex@rockbox/staff/AlexP)
07:32:19 Join Tuplanolla [0] (n=jani@unaffiliated/tuplanolla)
07:32:19 Join soap [50] (n=soap@rockbox/staff/soap)
07:32:19 Join maraz [0] (
07:32:19 Join ufoman [0] (
07:32:19 Join goffa [0] (n=goffa@
07:32:19 Join flux [0] (
07:32:19 Join jfc [0] (
07:32:19 Join Beta2K [0] (
07:32:19 Join Kopfgeldjaeger [0] (
07:32:19 Join preglow [0] (i=thomj@rockbox/developer/preglow)
07:32:19 Join scorche|sh [50] (n=scorche@rockbox/administrator/scorche)
07:32:19 Join @ChanServ [0] (ChanServ@services.)
07:32:19 Join thegeek [0] (
07:32:19 Join liiwi [0] (
07:32:19 Join feisar-- [0] (
07:32:23CtcpVersion from freenode-connect!freenode@freenode/bot/connect
07:32:23***Server message 477: 'logbot #rockbox :[freenode-info] why register and identify? your IRC nick is how people know you.'
07:33:09 Nick fxb__ is now known as fxb (
07:33:32 Nick fxb is now known as fxb__ (
07:51:25 Join bzed [0] (
07:56:29 Join matsl [0] (
08:03:44 Join ender` [0] (
08:18:09 Quit kachna|lappy (Read error: 113 (No route to host))
08:18:59 Join webguest74 [0] (n=86478d2c@gateway/web/cgi-irc/
08:22:23 Quit webguest74 (Client Quit)
08:25:22 Join Zagor [242] (n=bjorn@rockbox/developer/Zagor)
08:35:45 Join B4gder [241] (n=daniel@rockbox/developer/bagder)
08:37:03 Join bertrik [0] (
08:38:27Mode"#rockbox +o Zagor " by ChanServ (ChanServ@services.)
08:38:32Topic"Please read before speaking: | Please direct offtopic/social chat to #rockbox-community" by Zagor (n=bjorn@rockbox/developer/Zagor)
08:38:36Mode"#rockbox -o Zagor " by ChanServ (ChanServ@services.)
08:41:28 Join BXCracer [0] (
08:44:54 Join Rob2223 [0] (
08:50:56linuxstbZagor: \o/ Was it simply a power outage - i.e. the server itself is fine?
08:51:29 Quit Rob2222 (Read error: 60 (Operation timed out))
08:51:41Zagoras far as I know, yes. the outage was short, but the machine didn't power up itself after for some reason.
08:54:59 Quit Bagder ("*plopp*")
08:57:03pixelmaseems like it wasn't up for the daily manuals build round yet
08:57:39 Join Bagder [241] (n=daniel@rockbox/developer/bagder)
08:57:47Zagorno, it came up just two hours ago
09:12:55 Quit Bagder ("*plopp*")
09:14:42 Join kugel [0] (n=kugel@rockbox/developer/kugel)
09:15:30 Join flydutch [0] (
09:15:30kugelpixelma: did you try if BUTTON_OFF|BUTTON_REL works?
09:16:00kugelI mean without BUTTON_OFF_PRE at all
09:16:22 Join kachna|lappy [0] (
09:16:33kugelthat _PRE shouldn't be needed
09:20:57 Join kugel_ [0] (
09:20:59 Quit kugel ("exit(0);")
09:21:25 Quit kugel_ (Client Quit)
09:21:31 Join kugel [0] (n=kugel@rockbox/developer/kugel)
09:21:45 Quit BHSPitMonkey ("Ex-Chat")
09:26:01 Quit bertrik (Read error: 60 (Operation timed out))
09:26:46 Join petur [50] (n=petur@rockbox/developer/petur)
09:30:16 Join Thundercloud [0] (
09:32:12***Saving seen data "./dancer.seen"
09:35:59 Quit kugel (Read error: 60 (Operation timed out))
09:39:51 Join Bagder [241] (n=daniel@rockbox/developer/bagder)
09:45:40 Join shodanX [0] (
09:50:28 Quit Thundercloud (Remote closed the connection)
10:03:18 Join fyre^OS [0] (
10:04:25 Join kugel [0] (n=kugel@rockbox/developer/kugel)
10:04:39 Quit kugel (Client Quit)
10:04:54 Join Dillizar [0] (n=Elive_us@
10:05:00 Join kugel [0] (n=kugel@rockbox/developer/kugel)
10:05:02 Quit cool_walking_ (" ajax IRC Client")
10:06:08Dillizarmy player is PTP if i put rockbox it will act like a normal USB? and will my linux recognize
10:06:27linuxstbDillizar: What player?
10:06:43 Join ze0 [0] (i=ze@
10:07:31Dillizari know its not supported but i have some version of ot :) thanks to Toffe
10:07:41 Quit markun (Remote closed the connection)
10:07:51linuxstbDillizar: According to this table, yes -
10:08:41linuxstb(the "USB" column is ticked, which means Rockbox controls the USB, providing a standard mass-storage interface).
10:09:48linuxstbMore info, including some install instructions are here -
10:10:15 Quit ze (Read error: 113 (No route to host))
10:10:15 Nick ze0 is now known as ze (i=ze@
10:10:34pixelmakugel: yes, it's needed to make a difference between the simple Off and Off+Menu. Otherwise you exit the plugin if you are slightlly quicker with the Off button in that combo...
10:11:21pixelmathis won't happen now anymore
10:11:27kugelpixelma: that shouldn't happen due to or'ing with |BUTTON_REL
10:12:20kugelthe button driver only returns X|BUTTON_REL if the button was released, not if it's hold down or used with another button
10:12:54 Join markun [50] (n=markun@rockbox/developer/markun)
10:15:07pixelmaI try it now but you could do it too. At least the accidental Off is fixed now - and you could have seen it in first place...
10:15:36pixelmaand the restart is in the manual
10:15:43kugelI can only test in the sim. I'm not sure if there was a hardware-specific reason the _PRE was added
10:16:03kugelyes, that's true, sorry about that
10:17:17 Join Zambezi [0] (
10:18:19 Join mt [0] (n=MTee@
10:18:26pixelmakugel: it's needed
10:18:59pixelmaotherwise if you hold the Off button slightly too long in that combo you also exit after the reastart
10:19:04pixelmaor restart
10:19:34kugelah, that makes sense, but it shouldn't behave like that :(
10:20:05 Quit fyrestorm (Read error: 110 (Connection timed out))
10:20:36pixelmawhy not?
10:21:23kugelbutton_rel shouldn't come in after a combo imo
10:27:27Dillizarhow long do i need to wait for a activation code to registrate?? i am waiting for 20min now
10:27:55linuxstbDillizar: I would expect it to come immediately. But what activation code are you talking about?
10:31:12*GodEater didn't think the wiki sent an activation code
10:31:27GodEaterI thought you just registered, and then came here to ask for write access
10:33:49GodEaterDillizar: MarjanTrajkovski ?
10:34:07Dillizaryes i need to resend it again
10:34:46GodEaterresend what ?
10:35:02markunDillizar: someone here needs to give you write permission, it's not automated
10:35:25GodEaterassuming MarjanTrajkovski is your wiki name, then you now have write access.
10:35:26Dillizarno its ok
10:35:40GodEaternow what ?
10:35:45*GodEater is horribly confused
10:35:58Dillizarfirst i was waiting for email
10:36:06GodEaterwhich it doesn't send you...
10:36:16Dillizarso i press the backspace button of my keyboard and send me back
10:36:29Dillizarand clicked again for a new mail
10:36:31Dillizarand i got it
10:38:20Dillizarso how do you install the rockbox if i dont have windows
10:40:45markunDillizar: it's probably in the manual, did you check it?
10:41:16Dillizaryes but i can read for utility tools smt like that
10:41:16GodEatermarkun: he doesn't have a supported device
10:41:34Dillizarso i cant use the tools
10:41:59BagderDillizar: do you have a question of some sorts?
10:42:19Dillizarwell how to install rock box on gogear
10:42:38Bagderyou join the dev people who work on it
10:42:39Dillizari have used this but doesnt help a lot
10:42:57markunDillizar: a new port is a lot of work:
10:43:11Bagdermarkun: that port is however already pretty far down the road
10:43:19markunah, sorry :)
10:43:23Bagdermi4 based PP
10:43:57GodEaterseems we have two gogear ports on the go
10:44:11Bagderright, I believe they are quite similar
10:44:29GodEaterbut have yet to hear which one Dillizar has :)
10:44:30Dillizari have compile the source and now i need to install it inside the gogear but because its ptp my lunux cant read it
10:44:46 Quit jmillikin (Remote closed the connection)
10:44:50Bagderptp? you transfer pictures to it?
10:45:03Bagderthat's handy
10:45:18Dillizari have a cable from the player to the camera
10:45:25Dillizarso i dont need a pc to put the pics
10:46:10Dillizarpicture taking mashine
10:46:34Bagderaaah, that's what a camera is. ok then, I'm happy I could help. byebye
10:46:40GodEaterDillizar: could you please speak in complete sentences ?
10:46:47GodEaternone of us know what you're asking
10:47:18Dillizarwell only windows xp can recognize my mp3
10:47:32*Bagder sighs
10:47:35Bagderyour mp3?
10:47:43Bagderas in your single mp3 song?
10:47:44GodEaterDillizar: first things first - WHICH GoGear do you have ?
10:48:45GodEaterok - so the instructions for "Recovery mode" from the wiki don't work ?
10:49:07Dillizarno cuz i cant access my mp3 player
10:49:21GodEaterthat's not the right answer
10:49:23Dillizarstupid MTP
10:49:37BagderDillizar: please read GodEater's question and answer that
10:50:21DillizarGodEater, i cant go to recovery mode i dont know why i am asking the channel of my linux but no one is there :(
10:50:38Bagdersyntax error
10:50:52GodEaterDillizar: installing rockbox on your player requires that you get recovery mode working. There is no other way.
10:51:04BagderDillizar: so you followed the steps in the wiki and those didn't work?
10:51:05GodEaterif you can't make it work, we can't help you
10:51:30DillizarBagder, well my grub is smt like -2 sec :)
10:52:03Bagderand please use complete words
10:52:07Dillizari need to press esc when the grub is booting so i can enter the recovery mode
10:52:19 Join advcomp2019 [0] (n=advcomp2@unaffiliated/advcomp2019)
10:52:25GodEateryour GoGear has grub on it ?
10:52:32*GodEater finds this unlikely
10:52:49Dillizaron my gogear?? wow i was thinking on my linux
10:52:53Dillizarlol sorry mates
10:53:00GodEaterDillizar: then you aren't reading the wiki page
10:53:10GodEaterDillizar: please go away and read it
10:53:11Dillizarnot right it seems
10:53:24Dillizark sorry need to read it again
10:53:48Bagderreally, you need to focus and pay attention
10:55:09 Quit B4gder (Connection timed out)
10:58:50amiconnkugel: button _REL always fires when a button is released, that's what it's for. If you need to distinguish short/long/combo, you need to check preconditions
10:59:59amiconnThis is in no way hardware specific, it's how the generic part of the button driver works. The (only) exception are buttons without release handling (wheel, archos remote)
11:01:19 Quit perrikwp (" ajax IRC Client")
11:01:32 Join perrikwp [0] (i=18ac0c41@gateway/web/ajax/
11:02:51kugelamiconn: I think, if you pressed A+B, and hold B slightly longer, it should give B|BUTTON_REL, IMO
11:03:42 Join nibbler__ [0] (
11:05:04amiconnIt fires both A|BUTTON_REL and then B|BUTTON_REL
11:05:57kugeleven worse :(
11:06:23kugelA|B|BUTTON_REL would make more sense imo
11:07:09Dillizarcan i install it in Service Mode? or i must in recovery mdoe
11:08:13amiconnkugel: Nope, it wouldn't, unless you release both buttons at *exactly* the same time (i.e. within a single tick)
11:08:31kugelor just within whatever the current release timeout is
11:10:00kugelor we enlarge that timeout for combos
11:10:05 Quit mt (Read error: 104 (Connection reset by peer))
11:19:55GodEaterDillizar: the document is quite clear - the install must be done in recovery mode.
11:20:15DillizarGodEater, i cant enter :(
11:20:24GodEaterthen you're out of luck I'm afraid
11:21:03Dillizartoffe82 has a gogear like mine i will wait for him to tell me how he went to recovery mode
11:21:51 Join Horscht [0] (n=Horscht@xbmc/user/horscht)
11:23:40markunDillizar: he's usually here every day
11:23:48Dillizaryes i know
11:23:52Dillizarbut i am gmt +1
11:24:01Dillizarand i dont know his gmt so
11:24:05Dillizari will just wait
11:24:13gevaertsgmt is the same everywhere
11:24:49markunDillizar: he's in the west of the USA I think
11:25:11Dillizarand i am east of europe :)
11:25:21Dillizari think we are 12h apart
11:25:47BagderUS and europe are never 12h apart
11:25:56Bagderunless you count hawaii perhaps
11:26:34Dillizar8 then :)
11:26:47Dillizarbut he is west doe
11:32:14***Saving seen data "./dancer.seen"
11:32:47*Dillizar is away: Screenshot of Elive:
11:33:00gevaertsDillizar: don't do that
11:33:17Dillizarit wasnt me is the xchat
11:33:24Dillizari will remove my away msg
11:34:20gevaertsthis is clearly stated in the guidelines for this channel
11:35:19Dillizaryes but this is the first time i used the /away command
11:35:31Dillizarits just from my os
11:36:54 Join Mathiasdm [0] (
11:54:04 Quit BXCracer ("Quit")
11:55:19 Join robin0800 [0] (
12:10:41 Quit _lifeless (Remote closed the connection)
12:10:57 Join _lifeless [0] (n=lifeless@
12:21:44 Join robin0800_ [0] (
12:25:08 Quit Mathiasdm ("Yuuw!")
12:32:11 Quit robin0800 (Read error: 110 (Connection timed out))
12:33:28 Join funman [0] (n=fun@rockbox/developer/funman)
12:37:11 Join shodanX_ [0] (
12:37:20 Quit shodanX_ (Client Quit)
12:38:33 Quit Ridayah ("The Rise and Fall of the Heavens themselves is dependant upon Humanity's belief and disbelief.")
12:39:16 Join Ridayah [0] (
12:40:27 Quit shodanX ("leaving")
12:40:33 Join shodanX [0] (
12:41:42funmanYesterday's irc log is absent?
12:42:11funmanOh I see mentions of a power outage in the first logs of today :/
12:43:18 Quit Horscht ("Verlassend")
12:48:17 Join robin0800 [0] (
12:48:39 Join kushalone [0] (n=kushal@
12:51:04 Nick fxb__ is now known as fxb (
12:53:20 Quit daurn (Read error: 110 (Connection timed out))
12:54:27 Quit robin0800_ (Read error: 110 (Connection timed out))
12:59:52 Join Grahack [0] (
13:00:12 Quit funman ("leaving")
13:02:19 Quit Grahack (Client Quit)
13:05:32 Quit kugel (Read error: 104 (Connection reset by peer))
13:07:57 Join BXCracer [0] (
13:09:46 Join daurnimator [0] (n=daurnima@unaffiliated/daurnimator)
13:10:34soapBagder, sorry for breaking the build table.
13:11:02soapI assume you my filesystem has /plenty/ of free room, the commit caught me "with my pants down".
13:17:57 Join Grahack [0] (
13:19:32 Quit Grahack (Client Quit)
13:20:32 Join Grahack [0] (
13:20:58 Part Grahack
13:32:18***Saving seen data "./dancer.seen"
13:37:38 Join Horscht [0] (n=Horscht@xbmc/user/horscht)
13:43:59 Join LambdaCalculus37 [0] (n=rmenes@rockbox/staff/LambdaCalculus37)
13:51:39 Join Horschti [0] (n=Horscht@xbmc/user/horscht)
13:52:49 Quit arohtar (Client Quit)
13:53:03 Join faemir [0] (
14:06:23 Join froggyman [0] (n=187b533e@gateway/web/cgi-irc/
14:07:19 Join Mathiasdm [0] (
14:23:15 Quit Horscht (Read error: 110 (Connection timed out))
14:23:47 Quit kushalone (Client Quit)
14:25:07 Quit froggyman ("CGI:IRC")
14:26:32 Quit Mathiasdm (Read error: 104 (Connection reset by peer))
14:26:33 Join mt [0] (n=MTee@
14:26:39 Join Mathiasdm2 [0] (
14:29:06 Join Mathiasdm [0] (
14:29:38mtI need a little help with ci->request_buffer(realsize, reqsize), My understanding is that it allocates <reqsize> of data from the file buffer to a pointer, as long as ( realsize > reqsize ). Correct ?
14:30:01 Quit Mathiasdm2 (Read error: 113 (No route to host))
14:31:05 Join renke [0] (
14:33:43mtAnd after calling request_buffer(), advance_buffer() has to be called to advance the pointer to the file buffer by the amount of data used from request_buffer() - which would be reqsize ?
14:34:09 Join SirFunk_ [0] (
14:34:40 Join {phoenix} [0] (
14:36:28 Quit Horschti ("Verlassend")
14:43:30 Quit Mathiasdm (Read error: 113 (No route to host))
14:43:54 Join Mathiasdm [0] (
14:47:05 Quit nibbler__ (Read error: 110 (Connection timed out))
14:48:29 Quit SirFunk (Read error: 110 (Connection timed out))
14:49:31 Quit Mathiasdm (Read error: 113 (No route to host))
14:49:41 Join Mathiasdm2 [0] (
14:51:21 Join dfkt [0] (i=dfkt@unaffiliated/dfkt)
14:57:45 Join firedix [0] (n=firedix@
14:58:58 Quit {phoenix} (Remote closed the connection)
15:00:40 Quit Dillizar ("I ♥ Elive")
15:09:21 Quit Mathiasdm2 (Read error: 104 (Connection reset by peer))
15:09:38 Join Mathiasdm [0] (
15:11:08 Join Horscht [0] (n=Horscht@xbmc/user/horscht)
15:14:24 Join itcheg [0] (i=62db4c46@gateway/web/ajax/
15:25:36linuxstbmt: If you look at flac.c as an example, it calls ci->request_buffer as follows - "buf = ci->request_buffer(&bytesleft, MAX_FRAMESIZE);"
15:26:35linuxstbMAX_FRAMESIZE is the maximum size of a FLAC frame (in bytes). The function returns a pointer to the data (buf) and sets bytesleft (an integer) to the number of bytes it has given you (either MAX_FRAMESIZE, or smaller at the end of the file).
15:27:20linuxstbYou then process some (or all) of that data, and then call ci->advance_buffer() with the number of bytes you have consumed. You then call request_buffer() again.
15:32:19***Saving seen data "./dancer.seen"
15:33:48 Join footwork [0] (n=4af941c9@gateway/web/cgi-irc/
15:34:32 Quit footwork (Client Quit)
15:38:45 Quit CaptainKwel (Remote closed the connection)
15:39:14 Join evilnick_7 [0] (i=0c140464@gateway/web/ajax/
15:39:22 Quit kachna|lappy (Read error: 110 (Connection timed out))
15:45:18 Quit itcheg (" ajax IRC Client")
15:47:26 Join Horschti [0] (n=Horscht@xbmc/user/horscht)
15:52:58 Join Strife89 [0] (n=michael@
15:55:01 Quit robin0800 (Remote closed the connection)
15:55:33 Quit SirFunk_ (Read error: 110 (Connection timed out))
15:57:29 Join SirFunk_ [0] (
16:01:05Horschtihow can I increase the plugin cache so I can keep listening to music while playing pacman?
16:02:13martian67thats not the issue
16:02:16LambdaCalculus37Horschti: Music playback with Pacbox works on 64MB targets.
16:02:29*Horschti retries...
16:02:30LambdaCalculus37My Gigabeast can play music and Pacbox at the same time.
16:02:58 Quit Mathiasdm ("Yuuw!")
16:03:59GodEaterLambdaCalculus37: it's not a memory issue
16:04:16GodEaterthe 80GB ipod 5.5 won't play music at the same time as pacbox
16:04:33HorschtiCPU too weak?
16:04:43GodEaterthe gigabeat F can do it
16:04:47GodEaterand I think that's a 32MB target
16:04:52LambdaCalculus37It is.
16:04:58LambdaCalculus37GodEater: Thanks for clearing that up.
16:05:04 Join jgarvey [0] (
16:05:05GodEaterany time :D
16:05:11linuxstbWasn't stripwax working on that though? The problem on ipods will also be that pacbox uses IRAM - and plugins and codecs cannot both use IRAM together.
16:05:18HorschtiIpod 5.5G is a 64MB target..
16:05:22 Quit Horscht (Read error: 110 (Connection timed out))
16:05:38GodEaterlinuxstb: no clue - that's not a project I was aware he was looking at
16:06:00linuxstbI _think- stripwax was looking at using the second core for pacbox, but now that codecs are using the second core, that won't help...
16:06:42 Join Dillizar [0] (n=Elive_us@
16:06:54GodEaterHorschti: yes, I know it's a 64MB target, as I was saying, it's not a memory issue
16:07:23Horschtiah, missed the line about the Gigabeat F being able to do it
16:08:08linuxstbThe final comment on that task sums up the current situation...
16:08:18*GodEater bows before linuxstb's encyclopaedic knowledge of what's in flyspray
16:10:24linuxstbMaybe a solution would be to make pacbox multi-threaded, with one thread on the main CPU, an one on the COP. That should hopefully make it play well with most codecs. But of course, someone who cares needs to do it...
16:12:11 Join itcheg [0] (i=62db4c46@gateway/web/ajax/
16:12:13GodEaterforgive me if this is a stupid question but :
16:12:17Horschtiok, no biggie. I thought it was a memory issue and easily solvable. I see now that it is not and that's ok. I am not going to ask you to "fix it", I'll just do either or...
16:12:37Horschtithank you for the information :)
16:12:42GodEaterwhen people say something like "Make X multi-threaded" - is it really that simple? I always thought you'd need to have a good idea of what needed a thread for itself each time.
16:13:14GodEaterI'm not sure for instance in pacbox what lends itself well to multi-threading
16:13:16linuxstbGodEater: What about "Make X multi-threaded" implies that it's simple. You're right, it's not.
16:13:39GodEaterhehe - I didn't mean it would be simple at all :)
16:13:57GodEaterIt's just on the surface I'm not sure I'd know where to start
16:13:59linuxstbIt would need benchmarking, but I think splitting the CPU emulation and the screen updates would be fairly even.
16:14:27linuxstbThat's quite easy to benchmark - just disable screen updates and see how fast it goes.
16:16:32GodEaterah ha
16:17:38linuxstbIn fact, that could be a nice way to up the framerate in general...
16:25:09 Join robin0800 [0] (
16:25:40 Quit jfc (Read error: 104 (Connection reset by peer))
16:26:03 Quit J-23 (Remote closed the connection)
16:26:07 Join J-23 [0] (
16:29:28 Join jfc [0] (
16:30:26 Quit jfc (Client Quit)
16:30:43 Join jfc [0] (
16:32:29 Join toffe82 [0] (n=chatzill@
16:34:28 Quit itcheg (" ajax IRC Client")
16:36:15 Join funman [0] (n=fun@rockbox/developer/funman)
16:37:33 Quit Strife89 ("Huzzah!")
16:43:30DBUGEnqueued KICK CIA-38
16:43:30CIA-38New commit by 03funman (r20986): Sansa Clip : ignore previous setting of CGU_DBOP (do not use ORR)
16:43:40CIA-38New commit by 03funman (r20987): FS #10219 (AMSSansa Debug Clocks) by Jack Halpin ...
16:44:55CtcpIgnored 1 channel CTCP requests in 0 seconds at the last flood
16:44:55*GodEater waits for the conversation to continue here
16:44:59*linuxstb spots some machine-gun committing...
16:45:23linuxstbOne commit quickly after another...
16:46:01funmanthanks to git all my commits are committed to svn with one command
16:47:23linuxstbHave any alternatives been suggested for the list of targets on the proposed new rockbox website, or are most people generally happy with it?
16:47:52*GodEater is not aware of any alternative proposals for it
16:48:57GodEaterdo you have an idea of how to redo it that sits well with the overall layout ?
16:49:01gevaertsthe problems I see with it are (a) it doesn't scale. Once we add two or three manufacturers the layout breaks (same problem if people want smaller windows), (b) it doesn't mention unsupported targets (sansa and ipod need those)
16:49:08*linuxstb repeats what he said in -community - that he would prefer a static list, and also for that list to indicate targets we don't support (e.g. the newer ipods and Sansas - with links to development status, if any), as well as ones we do.
16:50:24gevaertsGodEater: the *current* way doesn't sit well with the overall layout :)
16:50:41gevaertsunless you're happy with no new ports...
16:50:44GodEaterI'm waiting to see the proposed better solutino
16:50:52linuxstbGodEater: Not really... One idea that comes to mind is a table with manufacturers down the left, targets going across, and then colour-coding to say if they are supported (green?), in active development (amber?), or not supported (red?). But that wouldn't be a small table...
16:51:10GodEaterindeed not
16:51:15linuxstbI guess maybe we could simply move that info elsewhere, with a big "Supported devices" icon
16:51:17GodEaterI don't think there's a nice way to get them all on the front page
16:51:24GodEaterthat might work better
16:51:38GodEateror perhaps "Can I use Rockbox?"
16:52:13*GodEater would really like to see the wiki template changes made as soon as possible
16:52:17GodEaterwhile we're on the subject :)
16:52:26GodEaterthey're not very contraversial as I understand it
16:53:06linuxstbCan they be done independently of the new main website? (I've no idea what you're talking about...)
16:53:31GodEaterthere were images in the forum thread for the proposed wiki pages design
16:53:49GodEaterand I don't see why they couldn't be done independently
16:54:10linuxstbI assume they change the colour scheme though?
16:54:26GodEateronly a bit
16:54:30GodEaterit's still predominantly blue
16:54:35 Join rvvs89 [0] (n=ivo@pdpc/supporter/base/rvvs89)
16:54:58 Join itcheg [0] (i=41d59de2@gateway/web/ajax/
16:55:32pixelmahuh? The new proposal is almost white IIRC
16:55:47GodEaterI'm trying to find it now
16:55:49pixelmaI mean the background ;)
16:56:07GodEaterlooks blue to me
16:56:41GodEaterunless there was another one ?
16:56:46pixelmabecause I made it so
16:58:10***SPY: Authentication failed for Zagor
16:58:17***Server message 901: 'logbot logbot n=bjst rockbox/bot/logbot :You are now logged in. (id logbot, username n=bjst, hostname rockbox/bot/logbot)'
16:58:17Mode"logbot :+e" by services.
16:59:07pixelmahis proposal was this:
16:59:15GodEateryes it was - but yours was more popular was it not ?
16:59:21ZagorI plan on doing the wiki and other pages at the same time. I would have to split up the templates to do one without the other.
16:59:33GodEaterah ok
17:00:01gevaertsZagor: could we have all those things in svn somehow?
17:00:28gevaertsnot knowing whether you have the latest version is a bit annoying
17:00:36*GodEater thinks that's a stellar idea
17:00:47Zagoryeah I suppose we could simply branch the www module
17:01:02*gevaerts noticed that macku's latest zip isn't the same as
17:02:17Zagorgevaerts: no I changed quite a bit in the css, stripping out the line heights etc.
17:03:40 Quit Zagor ("Don't panic")
17:07:09pixelmathat .css is not easy to work with - many definitions to make sure it looks good in one size (fixed font sizes and whatnot) but not many things to take care of when someone does not use the browser at this size
17:07:40gevaertspixelma: I think I fixed most of that
17:09:19GodEaterhas anyone tried to track down a Gigabeat X image yet ?
17:09:43*toffe82 has a real one in his bag :0
17:09:55GodEaterthen you can take a nice picture of it for us!
17:11:22pixelmathe not completely tab-able menu is important IMO, had a try on changing that but wasn't completely successful yet (found whereabouts I need to change something, but somehow it didn't work yet)
17:11:25GodEaterthat looks a LOT like the S
17:11:38toffe82just missing the 2 buttons
17:11:44GodEaterso it is
17:12:12LambdaCalculus37GodEater: Tis a nice player indeed... if you can find one.
17:12:23GodEaterI never tried looking ;D
17:12:51toffe82LambdaCalculus37: I will ship the gogear's this afternoon
17:12:53GodEaterif those pictures are anything to go by though, the OF is a complete car crash
17:14:51LambdaCalculus37toffe82: Thanks! :)
17:15:17*LambdaCalculus37 will start building up the GoGear manual pages
17:15:27pixelmaGodEater: your car crash comparison reminds me of something... you are the first person (I think) who would call the wikipedia front page this way (which tries to look nice in a flexible way)
17:15:44LambdaCalculus37toffe82: So as I understand it, the install process isn't quite finalized for the GoGears, correct?
17:16:08GodEaterpixelma: I called the wikipedia front page a car crash ?
17:16:16*GodEater is clearly getting dementia
17:16:44toffe82LambdaCalculus37: there is not an easy install, you have to go in recovery to load the bootloader
17:16:51pixelmayou called all pages that try to be flexible a car crash, something like that
17:17:20GodEaterah ok
17:17:24GodEaterthat makes more sense
17:17:52GodEaterwell, I still stand by that with your browser below a certain size
17:18:35toffe82LambdaCalculus37: I check quickly yesterday and when you connect to the usb , it reboot by himself, perhaps a last version of rockbox fix the proble
17:20:02LambdaCalculus37toffe82: Okay, so since there's no finalized install method, I'll have to use the current method of install to describe the process.
17:20:09LambdaCalculus37When it's finalized, we can have it edited.
17:20:25gevaertsGodEater: the problems begin when "below a certain size" means "not exactly the same as the designer uses"
17:21:11toffe82LambdaCalculus37: the 1630 is the most usable, the 6330 needs tweaking on the touch pad, and the sa9200 doesn't have the touchpad
17:21:27GodEatergevaerts: I don't see any point try to design a page which works all the way down to 1x1
17:21:35GodEateryou have to call it a day somewhere
17:22:36LambdaCalculus37toffe82: How much tweaking on the touch pad?
17:22:58LambdaCalculus37toffe82: Also, I have a friend who also owns an HDD6330, so maybe she can help me test it out and provide feedback.
17:23:20toffe82for the sa9200, you have to find the where it is connected
17:23:43toffe82so disassemble of the code
17:25:30funmansaratoga: did you discuss with kugel why he disagrees that RBUtil support isn't mandatory for a SansaAMS release?
17:25:56toffe82LambdaCalculus37: I include the dock and remote for the 6330, it can be used also to charge the 3 , some connector
17:27:21Dillizartoffe82, :D
17:27:31toffe82LambdaCalculus37: and also the japanese training I promise you a long time ago ;)
17:27:47Lssquestion will an ipod fitted with a 240gig hdd be able to be rockboxed
17:28:04evilnick_7Lss: Yes
17:28:05Dillizartoffe82, how to you put your hdd6330 in to recovery mode
17:28:26Lssanything special that needs to be done?
17:28:49Lsskind of like how the 30, 60/80 gig 5g ipods are not exactly the same
17:29:11evilnick_7Lss: Make sure it's a supported target (so iPod Video is the most recent), and you *might* have to use a special build with support for the larger (block sizes?)
17:29:22Lsshumm ok
17:29:30Lssthe 240gig hdd isnt exactly that cheap
17:29:37Lssit would suck if i got it and it didnt work
17:29:42Horschtiindeed it is not
17:29:51Horschtiwell, it does work.
17:29:59Lssyou have seen one?
17:30:20Horschtino, but there are people reporting it to work.
17:30:36Horschtiyou need a custom build, that's all
17:30:52Lssthats awesome
17:30:56Lssnow to find one
17:30:57Horschtievilnick_7 was just emphasizing that you need an ipod that is "old enoiugh§"
17:31:30Horschtias in: don't go off buying a new ipod from the store, those are not supported. You need an Ipod Video, not an Ipod Classic
17:32:03Lssno i already have an 80gig 5.5g
17:32:19Lssrockboxed of course i cant stand using itunes
17:32:21***Saving seen data "./dancer.seen"
17:32:36 Quit petur ("work->...")
17:32:58Horschtiif you got the money... go an strengthen the economy by buying a 240GB HD :p
17:33:14Dillizartoffe82, ok in recovery mode i need just to unzip the rock box in to the player but what about the FWImage.ebn
17:33:56toffe82Dillizar: first is the bootloaderon the small partition, then unzip rockbox on the big partition
17:34:05 Quit matsl (Read error: 110 (Connection timed out))
17:36:33 Quit robin0800 (Connection timed out)
17:38:52gevaertsGodEater: of course not, but designing a site for 1024 pixels wide and breaking at 980 isn't ok
17:39:12 Join robin0800 [0] (
17:39:34LambdaCalculus37toffe82: I'll let you know when the package arrives. :)
17:39:38*LambdaCalculus37 has to go for now
17:39:46 Quit LambdaCalculus37 ("Fwump")
17:41:11Dillizartoffe82, the recovery mode doesnt work :(
17:41:57toffe82Dillizar: try again.. are you connected to your computer...
17:43:13 Quit robin0800 (Remote closed the connection)
17:43:16Dillizaryes toffe82 Disconnect the player from the computer and turn it off.
17:43:29DillizarHold down the Volume + button and continue to hold it as you reconnect the player to the computer.
17:43:38Dillizarbut doesnt reconnect
17:43:41toffe82you don't have to disconnect if I remeber
17:44:26 Join Grahack [0] (
17:44:26Dillizarbut then how can i tuen it off??
17:44:45toffe82Dillizar: first comment and link to the video too
17:46:31funmanFlynDice: ping
17:46:50FlynDicefunman: hi funman
17:47:16funmanFlynDice: so, what are you working on, AMS hero ? ;)
17:47:25Dillizartoffe82, so i need Philips Device Manager so i can go to recovery mode
17:47:50toffe82just follow the procedure and it should appear as mass storage
17:48:19funmanI did a bit of update/cleanup of SansaAMS & TargetStatus wiki pages, and now the only blocking issue is FS #10048 - where i'm a bit out of idea now.
17:48:27FlynDiceha ha... I was going to look at your latest sd boundery thing. I've finally got some time now! I see you liked the patch ;)
17:48:45funmanI want to play a bit with the pcm code, add some debug, but I fear that there is other problems than PCM
17:48:47 Join bmbl [0] (n=Miranda@unaffiliated/bmbl)
17:49:22funmansd bank boundary is simple, and after all it's only a problem for 8GB+ users (not me..!)
17:49:48Dillizartoffe82, yes but that is Service Mode
17:51:05FlynDiceI've got the 8 gb e280, but I'm on-call for the paying job right now, how difficult is reconfiguring to return to a normal state after teesting?
17:51:37Dillizarand i have only one partition
17:53:24toffe82Dillizar: I don't have so much time now , If you look the rockbox log and search for lowlight, you should find more information, I try to find out but need to work now..
17:53:57funmanFlynDice: as difficult as typing "mkfs /dev/sdb" (or formatting if you use windows)
17:57:15funmanFlynDice: I still find a bit strange that using an uncached buffer for SD DMA transfers works fine, while using *_dcache() functions in mmu-arm.S doesn't. I suppose that we were doing something wrong there, since these functions work fine for other targets (when using correctly?)
17:58:49funmanjhMikeS gave us some good advice, I don't know who else (experimented with caches on arm9tdmi) could have a look at our patch.
18:01:01FlynDiceYes, strange is a word that keeps entering my mind every time I experiment with something it seems. The thought occurred to me that perhaps enabling the mmu somehow changed the clock freqs which got me onto finding something to check them
18:01:58funmanI more think that it changed timing, and clock freqs need to be adjusted with more care
18:03:30FlynDiceThe other thing I wanted to check was why simply mirroring the DRAM uncached seems to negate the need for the *_dcache() coherency routines.
18:04:27 Quit Grahack (No route to host)
18:05:01 Join JdGordon| [0] (i=46012a8c@gateway/web/ajax/
18:05:09FlynDiceOr pehaps it's a side effect of the last ata_sd patch
18:05:40 Quit JdGordon| (Client Quit)
18:05:57funmanFlynDice: i told you that your patch has no effect !
18:06:47 Join Grahack [0] (
18:07:01 Join Strife89 [0] (n=michael@
18:07:10Dillizarwhat if i dont coppy the FWImage.ebn! will it still work :P
18:07:19Tornefunman: questions about ARM cache behaviour?
18:07:35 Join l403 [0] (n=l@
18:08:02*Torne might be able to help if you point him at the problem :)
18:08:05FlynDicefunman: Not my embarrassing effort, but yours just before that. There were multiple errors in what I was trying to do but the code seemed to work well because I don't think it ever made it to my code
18:10:06Tornewhat patch is this?
18:10:07funmanTorne: yes, I think we are not maintaining cache coherency correctly during DMA transfers, are you willing to read a patch ?
18:10:07l403guys. I was building gcc following the steps in the guide setting up binutils and stuff and the make of gcc failed
18:10:14Tornefunman: sure
18:10:19Tornefunman: i'm an embedded kernel developer for a living
18:10:34Tornefunman: not done a lot of rockbox dev but i know how arm's caches work ;)
18:10:44funman : ata_sd_as3525.c
18:10:52l403the config at the end of make failed. here's the log
18:11:37Tornefunman: lemme apply that to my tree and i'll have a look
18:11:48funmanFlynDice: there are logical errors in your patch, look at my post of may 13th (fs 10048)
18:11:53funmanTorne: do you own a Sansa AMS device ?
18:12:08Torneso i can't actually test anything
18:12:19funmanwe tested for you, we only need understanding ;)
18:12:49funmanstarting a dma transfer from an uncached memory location works fine.
18:12:56Torneso are you mapping it fully cached/buffered?
18:12:59funmanstarting a dma transfer from a cached memory location doesn't
18:13:08Torneand your coherency updates are not seeming to work
18:13:19funmanthat's exact
18:13:29Tornewhich direction does it go wrong in? or both?
18:14:21funmanwell, maybe not for out, i didn't test extensively. The filesystem corruption could come from wrong transfer when reading
18:17:16Torneheh. rockbox has yet another different set of terminology for cache ops
18:17:17Tornehooray :)
18:17:33 Join kachna|lappy [0] (
18:18:13funmanhum and the patch doesn't apply anymore, do you want an update?
18:18:18Torneit applies to that file
18:18:19Torneit's ok
18:19:01Torneok i'm suspicious that you are dumping the cachelines before *and* after DMA for the read
18:19:47funmanI added cache dumping after DMA after - just to try. But no cacheline should have been fetched during the transfer
18:19:48Tornethat should only be needed if there is a risk someone might read the buffer while the dma is in progress
18:21:00 Join robin0800 [0] (
18:21:43Tornebut, yeah, it looks like you are doing the right ops there :)
18:22:05Tornes'the same things any of our drivers do ;)
18:22:07funmanok .. i'm going to try again then.
18:22:21Tornehm, actually
18:22:29Tornei hope the functions drani the writebuffer
18:23:10 Join domonoky [0] (n=Domonoky@rockbox/developer/domonoky)
18:23:42Tornealso i assume that your buffer is cacheline aligned? :)
18:24:00funmanhum it isn't necessarily
18:24:03FlynDicefunman: Oh yes, there are embarassing errors in that patch! It appeared to work when I did it though. But when I removed that code and removed the cache coherency routines I could still boot rockbox. Before that, without the *_dcache routines I couldn't get beyond the sd_init
18:24:08Tornetry cleaning/invalidating the entire cache
18:24:11Torneit'll be slow
18:24:22Tornebut if that works and by-range doesn't then that points to the problem ;)
18:24:42saratogafunman: yeah kugel and i came to agree on his position and i updated the wiki to reflect it
18:24:58funmanFlynDice: try removing the diff from ata_sd_as3525.c (using the uncached buffer) and it won't work for sure
18:24:59Torne(also it may not actually be slower, large cache ops are slow as hell on ARM, over some threshold it's better to just kill the whole thing)
18:25:07saratogawe decided ams could be supported once the MMU, clock and write corruption problems are fixed
18:25:19saratogaand I guess the clip could be released without the write fix since its 4GB only
18:25:31funmansaratoga: hillshum added the manual as a requirement (and i do agree with him)
18:25:43funmanhum there's no 8GB Clipv1 ?
18:26:00saratoganot as far as I know
18:26:24Tornefunman: though having said that there's no dump_dcache(void);
18:26:35gevaertswell, the write fix should be ready and only needs testing, right?
18:26:58saratogareally we can put downloads up whenever the MMU issue is fixed even if we don't call it supported
18:27:04saratogagot to run
18:28:04Tornefunman: try if (write) clean_dcache(); else invalidate_dcache();
18:28:09Torneand nothing afterward.
18:28:21Tornethat will be a lot slower probably, but eliminates any alignment issues :)
18:29:28Tornethe functions in arm-mmu.S make some effort to handle unaligned addresses, it looks like, but some of the comments imply it doesn't handle every possible case
18:31:28 Join Casainho [0] (
18:31:41CasainhoHello :-)
18:31:50CasainhoNOR flash $2 for 32Mbit = 4Mbyte
18:32:16Casainhodoes anyone know if Rockbox bootloader + firmware + config files can be placed on a 4Mbytes?
18:32:30Tornerockbox itself is under a meg
18:32:42Tornecodecs/plugins/etc are usually read from the storage fs, not flash
18:32:53Torne(well, ok, that might also be flash)
18:33:10funmanhum I disabled write (to not break my filesystem) and invalidate_dcache() instead of dump_dcache_range() sure fixes read transfers
18:33:26Torneinvalidate is doing more work, since it's writing back dirty lines
18:33:33Tornebut that doesn't matter, because you're about to overwrite it anyway
18:33:54CasainhoTorne: but, all of it would be placed on 4Mbytes?
18:34:14gevaertsCasainho: 4MB is going to be a bit tight
18:34:28Casainhoso, 8mbytes ok?
18:34:40Tornewhat're you trying to actually do?
18:34:48Casainho"base board will cost around $50 USD and have just RAM and MMC Card Connector, possibly some NOR Flash for the basic software as it is quite cheap - $2 for 32Mbit = 4Mbyte."
18:35:02Tornea full rockbox for, say, ipod is 10mb
18:35:08Tornebut that's with all the codepages/languages
18:35:13CasainhoI am trying to choose 4Mbytes or 8Mbytes
18:35:26gevaerts8MB will hold a standard install, but if you want things like doom data or voices, which are also stored in .rockbox, that won't work
18:35:31Casainhoah, ok, so maybe 16Mbytes
18:36:02Torneif you're going to put a regular FAT filesystem on the nor flash to hold .rockbox then you'll need an FTL for it, too..
18:36:07linuxstbAnd then at least double it to try and predict the future...
18:36:14Tornewhich may or may not interfere with using it as an actual boot flash, depending how you do your FTL
18:36:26Torneflash translation layer
18:36:30Torneto make flash behave like a block device
18:36:51funman mcrrhs p15, 0, r1, r0, c6 @ Invalidate DCache range
18:36:55Casainhobecause memory card seems much cheaper than that memory..
18:37:09funmani do not remember seeing that in arm922tdmi reference manual
18:37:22gevaertsCasainho: you're much better off putting a bootloader and possibly the main rockbox binary in flash (in which case 4MB is enough) and use a "fixed" MMC/SD card for .rockbox
18:37:38funmanit even says that we can access cp15 "only with mcr/mrc"
18:37:42Torneyeah. only the rockbox bootloader actually needs to be in NOR flash
18:37:48Casainhogevaerts: ok, nice then. I want to save on money ;-)
18:37:53Tornemaybe the binary itself for boot speed
18:37:57Tornebut not the .rockbox folder
18:38:08Tornein fact it is considerable effort to do so since you need an FTL then
18:38:10funmanhum i'm not reading the correct implementation ..
18:38:12Torneas FAT is not flash-compatible.
18:38:14dionoeawouldn't codecs in flash also help a bit?
18:38:16gevaertsCasainho: can you put two MMC connectors on it? That would be about perfect I thingk
18:38:38Tornedionoea: not a lot..
18:38:45Tornedionoea: codec is only loaded when skipping between tracks that need different ones
18:38:48Casainhogevaerts: no, the board is already desingend and on prodution
18:39:14gevaertsCasainho: in that case you'll have something that's not very usable...
18:39:19Casainhoand I don't know yet if tthe MCU (LPC3130) can hold 2 memory cards...
18:39:23TorneCasainho: if you don't know what an FTL is then you are not in a position to be deciding this, I have to point out.. putting a FAT filesystem in Flash is not entirely trivial
18:39:39Torneand i don't think rockbox has any capacity to load rocks from anywhere other than FAT
18:39:39CasainhoTorne: ok, thanks
18:39:49 Quit robin0800 (Remote closed the connection)
18:40:07TorneMemory cards have their own FTL implemented in hardware
18:40:21Torneif you are using raw flash chips you would have to do it yourself in software which isn't something rockbox has afaik :)
18:40:26 Join robin0800 [0] (
18:40:40Tornealso lots of FTL algorithms are patented up the wazoo, msotly by samsung (though the linux folks tend to just Pretend Not To Notice)
18:41:22Torneanyway i need to head off for now
18:41:43Tornefunman: ping me at your leisure if you need any more help with cache stuff
18:42:11funmanTorne: which comment lead you to think the functions don't handle every possible unaligned case?
18:42:26funman(i'm reading dump_dcache_range for arm9)
18:42:49Tornedump_dcache_range says "will not do write back except for buffer ends not on a line boundary"
18:42:59Tornei'm nto entirely sure what whoever wrote that meant :)
18:43:05Torneit just generally makes me slightly suspicious
18:43:34funmanwell, i understand that we don't write back cachelines which are entirely in our buffer.
18:44:06funmanbut we do for cachelines which are half in our buffer, half somewhere else.
18:44:14Tornethat's.. interesting
18:44:22Tornei would generally assume that was broken :)
18:44:51Tornethat has the potential to break stuff if buffers are close to each other and not line-aligned.
18:45:10Tornethough, er, idaelly they wouldn't be ;)
18:45:52Torneyou probably want your buffer to be 32-byte aligned.
18:45:59Tornejust because it's nicer :)
18:46:03Torneanyway, gone now for rael :)
18:46:12funmanthanks for your help !
18:48:34 Join JdGordon| [0] (i=836b0070@gateway/web/ajax/
18:52:54FlynDicefunman: On a slightly differnt note, do you have an opinion on the synchronous vs asynchronous clocking schemes?
18:53:38funmanyes, I think we need to see what are the lowest frequencies we can use to minimize power usage
18:53:52 Quit Casainho ("ChatZilla 0.9.84 [Firefox 3.0.10/2009042316]")
18:54:40 Join Llorean [0] (
18:58:06 Quit Tuplanolla (Read error: 60 (Operation timed out))
19:03:57 Join miepchen^schla [0] (
19:04:50funmancan mcr/mrc update the cpsr?
19:06:06 Join Tuplanolla [0] (n=jani@unaffiliated/tuplanolla)
19:08:40funmanwhat should the *modified* virtual address look like : like the virtual address, with bits 4:0 unset? or something additional?
19:09:09 Quit flydutch ("/* empty */")
19:11:55funmanpage 2-6 (page 32 of pdf) of arm922t technical reference manual says : "ProcID translates the Virtual Address issued by arm9tdmi core to a MVA, seen by MMU & Instruction cache".
19:16:19 Join BryanJacobs [0] (
19:16:53funmanfrom google : "The VA is combined with the PID to form a MVA" : so I understand that the MVA may be different from VA only if FSCE is used
19:19:54gevaertsHorschti: one more thing to try : in usb_storage.c, near the bottom, change "tb.inquiry->DeviceTypeModifier = DEVICE_REMOVABLE" to "tb.inquiry->DeviceTypeModifier = 0;" and see if it makes a difference
19:20:02 Quit bubsy ("Mrrrrreow!")
19:20:45Horschtisometimes i think i should keep the virtual machine running :D
19:21:26*amiconn still doesn't understand why the website would need a redesign
19:22:15*gevaerts can think of some ways to make usb_storage.c itself even faster, but he doesn't want it to get a reputation similar to that of the playback code
19:22:35Horschtiwhat's the rep of the playback code?
19:23:03Horschtigevaerts, usb_storage.c in firmware/usbstack, correct?
19:23:28gevaertsHorschti: yes
19:24:32gevaertsHorschti: after that, be more careful about safely removing hardware :)
19:24:45Horschtii always remove safely
19:25:28Horschtiwait... that makes me think of something that could have manipulated the results all along... gonna check something...
19:26:24*Horschti slaps forehead
19:27:04funmanFlynDice: thanks to Torne I have some ideas on where to work.. see you later
19:28:08Horschtiof course, setting the policy to "optimize for performance" should increase my writing speeds... Obviously the OF identifies differently to the OS therefore setting the policy only applies to the OF...
19:28:41evilnick_7Horschti: Interesting point.
19:29:18Horschtiyes, ofc. OF shows up as "Apple Ipod USB Device"
19:29:41HorschtiRockbox shows up as the HD Name "Toshiba somethingtomethin"
19:30:22Horschtiwhy the hell didn't i check that... All the rsults and test I made were futile.
19:30:27Horschtisorry gevaerts
19:30:56FlynDicefunman: so long I'll post any results I get from the boundery patch
19:30:59Horschtii'll try with the speed policy first now...
19:31:12evilnick_7Doesn't the OF show up as "Apple Ipod USB Device" and then again as the type of disk inside the ipod?
19:31:29gevaertsnot as far as I remember
19:31:32Horschtino, it shows up as "Apple Ipod USB Device" in hardware
19:31:43evilnick_7Ah, okay.
19:31:44funmanFlynDice: don't forget to post your method as well! (patch + dd command)
19:32:22***Saving seen data "./dancer.seen"
19:32:55FlynDicesure thing.. hey i'm looking at the debugclocks patch and it's not "live info anymore. Was that on purpose?
19:33:08FlynDiceer "live"
19:33:35funmanhum what do you mean?
19:34:18FlynDicethe while loops just enclose the get_button calls and not the actual frequency calculations
19:34:38FlynDicethe buttons are reading much better though... ;)
19:35:27funmanoh .. yes, i wanted to make button reading realtime (too fast presses were ignored)
19:35:53FlynDiceyes that's the problem I was having too
19:36:05 Join matsl [0] (
19:36:30funmando we really need realtime data? there's no clocks changing on the fly except i2so (and i2si)
19:37:30Horschtigevaerts, enabling the speed policy profile resulted in slightly faster writing speeds of 7.8Mbyte/s and read speeds of 13.2
19:37:41Horschtiso I'll try the other thing you posted
19:38:06FlynDicewell I found it useful to see what was changing when and in what relationship. It gave me a better feel for what was going on
19:38:13gevaertswell, 6.3 vs 7.8 isn't "slightly" :)
19:38:28 Join bertrik [0] (
19:38:41FlynDiceI was bouncing between sync, async and fastbusses
19:38:43Horschtinot as much as I expected. I always thought there was a realy big difference between the profiles
19:39:03gevaertsI'd be interested in the speeds you'd get on a fully defragmented drive, but that's not really practically doable...
19:39:20Horschtii think it is pretty much defragmented...
19:39:36Horschtioo defrag mentions a fragmentation of only 3%
19:40:08funmanFlynDice: you mean with cpu boosting?
19:40:19funmanbertrik: hi! have some minutes to speak about i2c on sansa ams?
19:40:31FlynDiceyes you can tell what is being changed on the fly
19:40:59gevaertsHorschti: I get 14.2 read and 9.4 write on the raw device (30GB ipod). You're going to get slightly different speeds of course (80GB, right?), but I expect slower write than read at this point.
19:41:15Horschtiyes, 80GB
19:41:51Horschtii just compiled with the change you made above (and UDMA 4)... I'll test again
19:43:17funmanFlynDice: i can't think of a solution with live update and realtime button handling
19:43:30FlynDicefunman: I just modified it slightly and the "live" part and buttons work fine.
19:43:32funmanif you think live update is more important, then it should be changed
19:43:40funmancool !
19:44:28FlynDiceit'll take a few minutes to figure out the clip parts, you want me to post it on the forum or email it?
19:44:53funmanwhatever is the fastest - i really need to go
19:45:03Horschtigevaerts, "tb.inquiry->DeviceTypeModifier = 0;" did not have any effect on speeds
19:45:04FlynDicesee ya
19:45:22Horschti7.8 write, 13.3 read
19:46:04gevaertsok. I'm experimenting a bit here
19:46:06 Nick Horschti is now known as Horscht (n=Horscht@xbmc/user/horscht)
19:49:07 Quit funman ("long life vlc")
19:54:34gevaertshm, I can increase write speed by 1.3MB/s. About half of that comes from *reducing* the transfer buffer size, and half from not waiting for the disk I/O (which of course makes error reporting break, so I won't do it in an official build)
19:54:35bertrikFlynDice, my impression so far is that we've mostly been "tinkering" with all the clock and bus related settings
19:54:51bertrikWe just need to sit down and put some design into it I think
19:56:49bertrikAFAIK the datasheet is quite clear about the limits of the clocks and the conditions for the various bus modes. I haven't really looked into this myself, so probably it's easier said than done
19:58:29FlynDicebertrik: I agree with that assessment, that's why I tried to come up with something I could look at to see just what was going on with the clocks!
19:58:44 Join xnyhps [0] (n=xnyhps@2001:470:1f14:da:219:e3ff:fed7:c57c)
20:00:06FlynDiceThe thing is that some things that are not supposed to work do indeed work...
20:00:20 Quit crwll (
20:00:20 Quit bittin```` (
20:00:21gevaertsHorscht: for some extra speed, look for BUFFER_SIZE near the top of usb_storage.c and change it from 65536 to 32768 (i.e. halve it)
20:00:22NJoinbittin```` [0] (
20:00:50NJoincrwll [0] (
20:04:58 Quit Rob2223 (Read error: 104 (Connection reset by peer))
20:05:35 Join Rob2222 [0] (
20:06:20 Join antitrons [0] (n=Mudkips@
20:06:27 Quit antil33t (Nick collision from services.)
20:06:50 Join Lear [0] (i=chatzill@rockbox/developer/lear)
20:07:52 Join bubsy [0] (i=Bubsy@
20:11:03 Join juane414 [0] (n=chatzill@
20:12:00juane414Hey all just finished my new WPS... anyone willing to test it out and let me know what they think??
20:12:19Horschtyou know...
20:12:28evilnick_7Was this the one for the iPod Video?
20:12:30Horschtit would be helpfull if you mentioned what target ;)
20:12:43juane414yes ipod video
20:12:53*evilnick_7 spins the roulette wheel and wins
20:13:05juane414still haven't made the .cfg for it yet but the wps works
20:13:20Dillizarany one here has a gogear??
20:13:23juane414spent the last 3 days on it and i'd love some feedback
20:14:00Horschto could ttry it
20:14:07evilnick_7juane414: I'm at work now, but can test it on target in a few hours
20:14:08 Join dfkt_ [0] (
20:14:14Dillizarwhere can i find manual for installing rockbox on gogear
20:14:21juane414sounds cool
20:14:33Horschtjuane414, i can try it right now...
20:14:33juane414evilnick_7: i'll actually be in bed then...
20:14:50juane414Horscht: great how can i send the files to you?
20:14:55 Join fyrestorm [0] (
20:14:55 Quit perrikwp (Read error: 104 (Connection reset by peer))
20:15:13gevaertsjuane414: I'd like to see it as well :)
20:15:21Horschtupload it to some 1-click hister or something
20:15:26Dillizari need some help with manual install
20:15:31juane414know any good ones?
20:15:35Horschtrapidshare or similar
20:16:28gevaertsjuane414: can you dcc it to me? I hate downloading things from those 1-click things...
20:16:28amiconngevaerts: Smaller buffer is faster? /me puzzled
20:16:28gevaertsor even mail, if you prefer
20:16:46juane414oh email would work
20:16:56 Quit fyrestorm (Read error: 104 (Connection reset by peer))
20:16:58 Join perrikwp [0] (i=18ac0c41@gateway/web/ajax/
20:17:00 Join fyrestorm [0] (
20:17:08juane414can you guys just pm me your email address?
20:17:16 Join Thundercloud [0] (
20:17:34gevaertsamiconn: that's because of this double buffering scheme. A 64k buffer means read 64k from usb, write 64k, and so on. 32k means read 32k from usb, write it to disk while getting the next 32k, write 32k.
20:18:49juane414ok i'll send out the files in a minute
20:19:08juane414just to warn you guys i haven't tested any fonts so it might look goofy depending on what font you try
20:19:44saratogashould we have closed all the out standing feature requests?
20:20:01evilnick_7juane414: Which font did you design it with?
20:20:02AlexPDillizar: You can't find a manual as it isn't supported yet
20:20:13AlexPsaratoga: Yes, I think so - I was just about to have a look
20:20:17Horschtgevaerts, halfed the buffer, read: 8,06 MByte/s write: 8,52 MByte/s
20:20:21saratogai'll do it now
20:20:33Horschtstill UDMA 4
20:20:42gevaertsHorscht: ok. I'll probably end up using different buffer sizes for read and write
20:20:51saratogaoh theres no way to close multiple tasks at once
20:20:58 Quit dfkt (Nick collision from services.)
20:21:01 Nick dfkt_ is now known as dfkt (
20:21:11DillizarAlexP, but they already did and i have the rock box for hdd6330 i put it in to recovery mode but i have only one partition
20:21:20juane414the GNU Unifont is what I was using
20:21:24juane414seems to work well with ti
20:21:35Dillizarand i dont know where to put the FWimage.ebn
20:21:38AlexPDillizar: I was telling you that there is no manual, and no support for it yet
20:22:01evilnick_7juane414: Can you include that info on the email? I won't get chance to look at it for a while and will no doubt forget which font it was!
20:22:09gevaertsamiconn: we could still double-buffer with 64k transfers by lying about write completion, but I really don't want to do that
20:22:30DillizarAlexP, but toffe82 has rockbox on his hdd6330
20:22:42AlexPDillizar: And he is a developer
20:22:50juane414also one more warning... gmail isn't going to let me attach the bitmaps in a folder so they are all individual files, so you'll have to make a folder to put the bitmaps in
20:22:54 Quit perrikwp (" ajax IRC Client")
20:22:55AlexPDillizar: It has not been released, and is unsupported
20:23:05Dillizarhe gave it to me
20:23:06AlexPjuane414: zip it
20:23:07Horschtjuane414, zip it up :D
20:23:09saratogaDillizar: just wait for someone to answer you and stop spamming the channel
20:23:14juane414haha lol good point
20:23:16AlexPDillizar: None of this changes that it is unsupported
20:23:28 Join perrikwp [0] (i=18ac0c41@gateway/web/ajax/
20:23:35Dillizar:) saratoga spamming??
20:23:44 Join high-rez [0] (
20:24:26high-rezI'm trying to install rockbox onto my 5th generation 30g ipod... But the rockbox util won't detect the device (under os x)
20:24:53AlexPhigh-rez: It needs to be a winpod
20:25:02BryanJacobslinuxstb: you there?
20:25:10high-rezAlexP: Doh.
20:25:33AlexPhigh-rez: The easiest way to achieve this is to recover it with itunes on Windows, but there is a manual method outlined on the wiki where you can convert from a Mac and without losing data
20:25:54AlexPhigh-rez: But it is highly recommended to have a back up of everything before starting
20:25:54high-rezI don'ty mind losing data. I don't even really use the thing. :)
20:26:23AlexPhigh-rez: You can install fine to a winpod from OSX, and a winpod works fine with OSX, but Rockbox only supports FAT32, not HFS+
20:26:34AlexPhigh-rez: And just reformatting won't do it :)
20:26:35juane414okay guys files are sent
20:26:47juane414let me know what you think of it and any suggestions for improving it
20:26:53AlexPhigh-rez: If you have access to a Windows PC with itunes by far the easiest way is to use that to restore it
20:26:56high-rezAlexP: Ok i'll go look for this howto
20:27:07high-rezI have windows insalled under virtual box
20:27:12high-rezso it shouldn't be a problem
20:28:16AlexPhigh-rez: I assume that works on a Mac too, but I don't know
20:29:04amiconngevaerts: There is no way to delay the acknowledge?
20:30:10juane414hey guys it's possible that the album art isn't working either
20:30:15juane414i didn't really test that out
20:30:19juane414so if it's not let me know
20:30:41amiconnlinuxstb: Does the mac allow to format a partition as fat32 (i.e. on mac-partitioned hdd, not mbr)?
20:30:48gevaertsamiconn: I think we could tell the OS that we cache things, and implement the various cache flush methods. I'm not sure if it's worth it though
20:31:33 Join Strife1989 [0] (n=michael@
20:31:48 Quit fyre^OS (Read error: 110 (Connection timed out))
20:32:00Horschtjuane414, album art works.
20:32:02 Nick evilnick_7 is now known as How_unfortunate_ (i=0c140464@gateway/web/ajax/
20:32:27Horschtjust the playlist position seems wrong...
20:32:42BryanJacobsamiconn: you could always use mkfs.vfat if there's no native method
20:32:51BryanJacobsMacOS is a UNIX
20:32:52juane414playlist position?
20:33:00amiconngevaerts: I'm curious whether the OF(s) are doing this
20:33:20Horschtjuane indicating the current position of the track in the playlist.
20:33:40juane414ah gotcha
20:33:41Horschtit should say 2/9154, but instead it say 2/10
20:34:43juane414hmmm... one sec
20:35:42juane414ah that's not playlist position
20:35:50juane414thats %in
20:35:52juane414track number
20:36:02 Quit Strife89 (Read error: 110 (Connection timed out))
20:36:13Horschtyeah, that makes sense..
20:36:23juane414i intend on adjusting the info a bit
20:36:35juane414i'd like to add next track info
20:36:38juane414and a few others
20:36:38Horschtright... i should have skipped to the next track to notice that :D
20:37:08Horschtare the graphics final?
20:37:28juane414nothing is final
20:37:53HorschtI like the idea of the .wps, but the graphics could have a more "clean" feel to it.
20:37:53 Quit blithe ("Lost terminal")
20:38:04 Join blithe [0] (
20:38:28juane414yea i can try messing with colors and contrast to try and smooth it out
20:38:43juane414i used some sharply contrasting colors
20:39:21Horschtyeah, make it look a little bit more like a real dashboard
20:40:13juane414I can try :)
20:40:30 Nick How_unfortunate_ is now known as evilnick_7 (i=0c140464@gateway/web/ajax/
20:40:37juane414BTW... does album art need to be .bmp?
20:40:56AlexPor jpg
20:40:57Horschtno, it can now be .jpg
20:41:07juane414i can't seem to get it working...
20:41:16Horschtfolder.jpg works
20:41:26evilnick_7in the same folder as the music files
20:41:30Horschtdepending on your build number
20:41:41juane414okay i had cover.jpg
20:41:43Horschtit has only been implemented ~3 weeks ago
20:41:58gevaertscover.jpg should work
20:42:00evilnick_7cover.jpg *should* work too
20:42:28Horschtwell, how old is your rockbox build?
20:43:27juane414it works on the sim... but not on my ipod...
20:43:38Horschtso not a SVN build just the 3.2 release?
20:43:46Horschtyou have to update to a current build
20:43:47juane414yea the 3.2 release
20:44:24 Join tessarakt [0] (
20:44:49AlexPjuane414: 3.2 is a while old now - jpg album art was added after that
20:45:00juane414okay that must be the problem then
20:46:03juane414do i just copy-past the .rockbox folder to my iPod?
20:46:19Horschtyes, overwriting all files
20:47:07bertrikfunman, if you're reading the logs, yes I have some minutes. I experimented a bit with ams i2c yesterday and can reproduce the same behaviour as having high clocks. Can't fix it yet though (hoped to fix it by looking at interrupt status instead of the "busy" bit in the SR)
20:48:21AlexPBryanJacobs: If you have questions about stuff it is best to ask the channel - there are often other people about that can help
20:48:42juane414Horscht: any other suggestions you have for improving the theme?
20:48:50BryanJacobsAlexP: I was just going to tell linuxstb about some results from a conversation from yesterday
20:49:05BryanJacobsif anyone is interested, I discovered that WavPack blocks are routinely ~16k
20:49:14BryanJacobsand the correction blocks can be 40+k
20:49:38BryanJacobsmeaning that the current buffer space is insufficient to hold a "normal" WV+WVC block pair
20:49:44AlexPBryanJacobs: It is often useful to just share - you never know who has what knowledge :)
20:49:44GodEaterlinuxstb will likely read the from the logs :)
20:49:52BryanJacobsall righty then
20:49:58GodEaterbesides, we're ALL interested in the GSOC projects :)
20:50:03AlexPvery much
20:50:32BryanJacobswell, that's what I found out by writing a little wavpack file parser and trying it out on the "Bad Horse Chorus" encoded using -b256 -c
20:51:00BryanJacobsright now I'm seeing what the pattern of metadata blocks looks like
20:51:09CIA-38New commit by 03gevaerts (r20988): Use different read and write buffer sizes. Due to interaction between common transfer sizes used by most OSes (64k) and the double-buffering system we ...
20:51:21*gevaerts looks at Horscht for more testing :)
20:52:30BryanJacobswhat's the best way to convert uchar[3] into uint32_t architecture-independently?
20:52:36BryanJacobsI don't want to test for endianness
20:53:01*GodEater assumes amiconn will have a good idea about that
20:53:03BryanJacobsor, really, to fseek(... , SEEK_CUR) based on the amounts in a little-endian uchar[3]
20:53:47bertrikuchar[2] << 24 | uchar[1] << 16 | uchar[0] ?
20:53:59BryanJacobsthat's not arch independent, right?
20:54:09BryanJacobsI mean, what if the platform we're on is big endian
20:54:09gevaertssure it is
20:54:23bertrikwith different shifts ... 16, 8 and 0
20:54:54BryanJacobson little endian, should be [2] << 16 | [1] << 8 | [0]
20:55:05BryanJacobson big endian, should be [0] << 16 | [1] << 8 | [2]
20:55:23gevaertsthat depends on where that uchar[] comes from
20:55:32BryanJacobsI got it backwards
20:55:33LearBryanJacobs: metadata_common.c in Rockbox contains some different versions you could look at.
20:55:43BryanJacobsthe uchar[] is always little endian, no matter what
20:55:44*Horscht stares at gevaerts :)
20:55:48saratogaBryanJacobs: if blocks are that big in wavpack, i guess you'll need to split them
20:56:00BryanJacobsLear: will look
20:56:14saratogado you know how many time domain samples wavpack needs to decode one PCM sample?
20:56:15BryanJacobssaratoga: yeah, not looking forward to that - why is the buffer 32k?
20:56:26gevaertsHorscht: that last commit is the best I can do without overhauling usb_storage entirely :)
20:56:27BryanJacobssaratoga: sampling rate
20:56:41saratogaBryanJacobs: to save memory and because for most formats thats a lot of data
20:56:44BryanJacobsoh, you mean NEEDS
20:56:50Horschtjuane414, no, appart from the colors/graphics it's a nice theme. Allthough I prefer bigger album art myself (huge, 200x200)
20:57:02Horschtgevaerts, as in reverting to default?
20:57:07BryanJacobslet me check how many are necessary - I think it's theoretically up to the whole block
20:57:15Horschtoh wait
20:57:28Horschti should read what people write more carefully
20:57:31BryanJacobssaratoga: doesn't an iPod 5.5G have 64MB of RAM?
20:57:41juane414Yea there is not that much space for huge album art
20:57:42BryanJacobsisn't 32k a little ... spartan for that platform?
20:57:45saratogaBryanJacobs: since this is a time domain coder (I assume anyway) it probably looks at some number of past samples are used to try and predict the next sample, but its probably a relatively small number otherwise decoding would not be possible on CPUs as slow as we have
20:57:45 Quit Dillizar ("I ♥ Elive")
20:57:49juane414thanks for the feedback i appreciate it
20:58:14BryanJacobssaratoga: I need to look at the reference decoder b/c that's not in the docs
20:58:22Horschtjuane414, yeah. I understand that. Huge album art simply doesn't fit the concept.
20:58:37saratogaBryanJacobs: do you have a link handy to the docs?
20:58:47BryanJacobs <−− file format
20:58:47gevaertsjuane414: looks nice :) I'd like to see smaller intervals, but you're probably already near the image buffer limit I guess
20:59:25BryanJacobs <−− user docs
20:59:31gevaertsHorscht: what I committed should be as fast for read as before, and slightly faster on write than the last one you tested
20:59:39juane414gevaerts: not necessarily, I was planning on trying to make smaller intervals for at least the song progress dial
20:59:48Horschti'll give it a whirl, gevaerts
20:59:58juane414gevaerts: maybe not for the battery and volume though
21:00:04saratoga"it is possible to decode regular stereo (or mono) WavPack files without buffering an entire block, which allows the memory requirements to be reduced to only a few kilobytes ifdesired."
21:00:07saratogasounds good . . .
21:00:14 Quit xnyhps ("Leaving.")
21:00:39 Quit juane414 ("ChatZilla 0.9.84 [Firefox 3.0.10/2009042316]")
21:01:03BryanJacobs/me cannot seem to find docs on exactly HOW
21:01:03saratogais there a doc explaining the format it self and not just the bitstream?
21:01:29BryanJacobssaratoga: I'm searching but all I'm pulling up other than those two links is the library_use.txt file; how to use the library
21:01:37gevaertsJdGordon|: if you want to go to the limit, you could make short line images, and use more than one for each needle
21:01:50BryanJacobsI think the reference decoder is the only place to go for how decoding is actually performed
21:01:53gevaertshm, juane414 left :\
21:02:46saratogawikipedia has a very nice overview
21:04:03saratogait seems the author has designed this format with very low memory usage in mind, so I do not think you will have serious problems getting it to work
21:04:17BryanJacobsI just don't know the bounds on the requirements...
21:04:27BryanJacobsI'm sure it can be done, I'm just not yet sure how much must be pulled in
21:05:00BryanJacobsthat wiki article isn't technical enough
21:05:21BryanJacobs"An adaptive algorithm continuously determines the most efficient of the three to send based on the changing balance of the channels"
21:05:41BryanJacobsit seems to imply that there's a one-frame lookbehind, but then there's that line
21:06:04BryanJacobsthat could be unbounded state
21:06:18saratogathis will likely be of use for you:
21:06:19linuxstbBryanJacobs: Just to say that I'm not around, but will check the logs later...
21:06:29BryanJacobsthanks for sharing
21:08:58saratogaprediction seems to be quite simple needing only a few bytes of stored samples
21:10:44 Join {phoenix} [0] (
21:10:54BryanJacobsso then what we do is store the block header+metadata somewhere in the codec memory, and repeatedly buffer frames rathen than whole blocks?
21:12:25saratogafrom the sound of it you could just interleave the lossy and correction blocks in arbitrarily small blocks (say 256 bytes)
21:12:35 Join arohtar [0] (
21:12:51BryanJacobsso let's try that first, before tackling the thrash-seeking issue, eh?
21:12:59saratogai think the trick will be figureing out how to sync them, if the block size is very large it might be difficult to figure out which byte in the correction file goes with which byte in the lossy file without decoding it first
21:13:25BryanJacobsit's OK as long as we know when we've run out of one or the other
21:13:27saratogai think figuring out how to do this from a simple command line decoder program on your PC would be a great first step
21:13:37BryanJacobsthat's what I'm doing right now, actually
21:13:45BryanJacobsI've got the wavpack parsing for both WV and WVC
21:13:52saratogaits more complicated thent hat because if you get too far ahead of one then it may not be buffered in time
21:13:53gevaertshm, optimal write buffer size seems to vary between devices. Not entirely unexpected of course...
21:14:01BryanJacobsI was thinking next I'd try to get some sound
21:14:26high-rezSo... My iPod shows up as a disk drive when I plug it into windows, but its not really accessible...
21:14:44high-rezE.g. if I try to open the disk itself I get an error.
21:14:46BryanJacobsgevaerts: what? optimal values for parameters vary between different embedded devices? never!
21:14:51saratogawhile buffering you'll need to dynamically watch the position in both the correction and lossy files to keep blocks sorted in time (i.e. make sure you don't put a correction file block 10 minutes away from the lossy block it needs)
21:15:02 Part Grahack
21:15:12 Join bluebrother [0] (n=dom@rockbox/developer/bluebrother)
21:15:14evilnick_7high-rez: Have you set it to Disk-Mode in iTunes?
21:15:17gevaertsBryanJacobs: exactly! Why should they be different? :)
21:15:33high-rezevilnick_7: that may be what I'm missing... :)
21:15:55BryanJacobssaratoga: so, how about we at first fill the buffer 25/75 lossy/lossless, since that seems to be the approximate ratio
21:16:12BryanJacobsand then as one depletes we refill it
21:16:36BryanJacobsour time lag can't ever be too great because we only fill up to some preset high-water mark of readahead
21:16:52saratogathats a good first start but for large block sizes you'll eventually get more then 32KB apart and fail (though maybe large enough blocks for that to happen aren't used in practice)
21:16:58 Quit faemir (Read error: 104 (Connection reset by peer))
21:17:09BryanJacobshow would we get too far apart?
21:17:17high-rezevilnick_7: Just "enable disk use" right ?
21:17:22Horschtgevaerts, read: 13,3 MByte/s write: 8,85 MByte/s
21:17:33BryanJacobsimagine two distinct virtual buffers
21:17:36saratogayou don't know the bitrate of the correction blocks in advance, so if you assume a value eventually you'll be very far off
21:17:57Horschtso, as you projected, slight increase in write speed.
21:18:04BryanJacobsbut we know how many frames were actually unpacked
21:18:12BryanJacobsbecause they're put into the PCM buffer
21:18:16saratogaimagine trying to buffer a 10MB block and you get the bitrate wrong by 1%, by the end the correction file will be 100KB off and decoding will fail
21:18:28evilnick_7high-rez: It's been a *long* time since I did it but I think it's literally just a check-box when you first connect the iPod
21:18:31BryanJacobswhy do we need this "bitrate"?
21:18:39BryanJacobswe can read whenever we need to
21:18:48Horschtread speed stays the same. Since I have FS 9708 i.e. UDMA enabled, wouldn't the ipod also use less power during filetransfers?
21:19:13BryanJacobsevent: the codec has finished with what I read from disk (or at least one stream of it) action: read some more
21:19:47BryanJacobssee what I mean?
21:20:14BryanJacobssince we, as you pointed out, don't know how many bits we need per frame, don't even try to keep a constant flow from the disk
21:20:50BryanJacobsthe WV stream can be traditionally buffered, and the WVC stream can be a run-dry system
21:21:02saratogaBryanJacobs: you can't read whenever you want, it all has to be buffered before decoding begins
21:21:20BryanJacobsunderstood. Let me explain more clearly:
21:21:48BryanJacobssequence of events: 1) buffer is preliminarily filled 2) codec is invoked and decodes as much as possible
21:22:08BryanJacobs3a) codec runs out of WV frames first 4a) disk reads more WV frames
21:22:21BryanJacobs3b) codec runs out of WVC frames first 4b) disk reads more WVC frames
21:22:36BryanJacobswhy do we need to know the bitrate for this procedure?
21:23:13BryanJacobsas long as our buffers don't clobber eachother...
21:23:39saratogawhile that is one solution it will lead to reduced battery life
21:23:57saratogai was under the impression that the solution we wanted was to interleave the two files into a common buffer during load time
21:24:02 Quit Strife1989 ("Huzzah!")
21:24:19BryanJacobsdefine "interleave"
21:24:30BryanJacobsin order to get an AVI-style interleaving we need sync
21:24:49BryanJacobswhat we really want is to keep the amount of WV and WVC frames available approximately constant to reduce spinups
21:24:54gevaertsHorscht: I don't know about power. I'd guess most of the power goes to keeping the disk spinning anyway
21:24:57BryanJacobsthat could be done by an adaptive algorithm
21:25:10*gevaerts decides that he has done enough usb work for the day
21:25:20BryanJacobs"if we fall behind on WVC frames, next time buffer a greater WVC:WV ratio"
21:25:59Horschtgevaerts, just for the record, I cleaned the table a bit:
21:27:02Horschti removed all old results, which were not representative at all since they were not using the speed profile of windows anyways, so they were not a fair comparison to the OF results
21:28:52gevaertsHorscht: maybe it would be interesting to have pure svn there, as well as udma1 (as that's probably most likely to get in)
21:28:55 Join hd [0] (
21:29:15 Quit hd (Read error: 104 (Connection reset by peer))
21:30:16Horschtok, gonna do that
21:32:25***Saving seen data "./dancer.seen"
21:35:25saratogaBryanJacobs: I'm not sure, you should probably ask linuxstb since he knows the buffering code better
21:35:50BryanJacobseh, I'm going to watch Doctor Who now, I'll finish this tomorrow
21:35:55BryanJacobsttyl all
21:36:09 Part BryanJacobs
21:40:41 Join hd [0] (
21:41:11 Quit BXCracer (Remote closed the connection)
21:41:17 Join moredhel [0] (
21:41:30GodEateroooh, a feist fan
21:41:55 Join Grahack [0] (
21:44:00 Quit Grahack (Client Quit)
21:44:05 Quit arohtar (Read error: 104 (Connection reset by peer))
21:44:09 Join BXCracer [0] (
21:45:25 Quit HellDragon (Success)
21:45:37Leargevaerts: No big change in Rockbox install time with the latest USB change on my e200... :)
21:46:57gevaertsLear: unfortunately on e200 speed is mostly limited by the sd driver
21:47:00 Join Grahack [0] (
21:47:15 Quit Grahack (Client Quit)
21:47:26 Join Grahack [0] (
21:47:37 Join froggyman [0] (i=47ba0b80@gateway/web/ajax/
21:47:40 Part Seed
21:47:45Leargevaerts: Well, the previous change did make a big difference at least.
21:47:48gevaertsLear: actually, my latest commit may have made writes a bit slower on e200...
21:48:47saratogado faster classes of SD card write faster in rockbox on the sansa or is it limited by the controller?
21:49:15LearMy quick one-off test was a little faster, but by less than 1 percent. The install seem more limited by number of files than amount of data though.
21:49:19 Quit Grahack (Client Quit)
21:50:03JdGordon|gevaerts: ?
21:50:12bertriksaratoga, do the sansas use 1-bit or 4-bit mode?
21:50:28Horschtwell, that's a FAT32 issue... afaik, specifialy FAT32 is a lot slower in writing several small files than few big ones...
21:50:41gevaertsJdGordon|: I wanted to talk to someone else with a j, but he left...
21:50:56JdGordon|ah ok :)
21:51:02saratogai'm not sure
21:55:42 Join Grahack [0] (
21:56:25LearHorscht: OF is noticeably faster though; half the time compared to Rockbox.
21:59:49 Join Zagor [242] (
22:00:50gevaertsLear: try reading. My measurements say that we're a lot faster there :)
22:02:46high-rezThe install themes stuff is broken - at least when using a proxy... It requests "http://rbutilqt.php?" ....
22:07:50domonokyhigh-rez: seems to work fine for me, (at least without proxy) :-)
22:08:01domonokyon which OS is this ?
22:08:18Leargevaerts: Well, I rarely do that... :)
22:08:50high-rezdomonky: windows (which runs under virtual box on an os x host)
22:09:49 Join webguest83 [0] (n=ad2180dd@gateway/web/cgi-irc/
22:10:39domonokyhigh-rez: then please open a bug report, with as much details as you can..
22:10:49 Quit webguest83 (Client Quit)
22:10:58domonokyand why dont you use the mac version ?
22:11:44high-rezdomonky: I needed to firmat the device with fat32 first...
22:12:23domonokyhigh-rez: thats true, but you dont need rbutil for that :-)
22:12:54AlexPhigh-rez: Once you have a winpod, you can use the Mac native tools
22:15:59 Quit itcheg (" ajax IRC Client")
22:16:13 Join SirFunk__ [0] (
22:17:52high-rezdomonky: I'll validate with the mac client now...
22:18:30bluebrotherindeed, there is something broken with theme installation using a proxy
22:18:55bluebrother(linux, current svn)
22:19:11high-rezEverything worked on windows, except the themes.
22:19:49 Part Grahack
22:21:04 Join MuscleNerd [0] (
22:24:34high-rezDitto on OS X - the themes part is broken when using a proxy:
22:24:35high-rez1242764655.501 45 TCP_MISS/503 2150 GET http://rbutilqt.php? - DIRECT/rbutilqt.php?target=ipodvideo& text/html
22:24:46high-rez(that's from squid)
22:27:30 Join Mathiasdm [0] (
22:30:15pixelmabluebrother: about the manual - would you have an idea what's causing the insertion of blank lines that _sometimes_ happens after \opt or \nopt (something to check for at least)?
22:30:42Unhelpfulgevaerts: if the plugins alone don't work, i've put up a full test build + the three plugin versions here:
22:31:04 Quit SirFunk_ (Read error: 110 (Connection timed out))
22:34:09 Quit bubsy ("Mrrrrreow!")
22:34:23bluebrotherpixelma: well, I could imagine them causing double linefeeds, thus creating a new section in LaTeX. Have you tried adding trailing comment marks to each opt?
22:35:24Unhelpfuli assume putsxy clips correctly/safely? could the hang for the scaler benchmark on gevaerts' x5 be due to writing past the end of the LCD framebuffer?
22:35:30 Quit perrikwp (" ajax IRC Client")
22:35:52gevaertsit doesn't get near wrapping
22:36:31gevaertsit freezes after printing the very first line, which fits. Anyway, I'm now trying your build
22:37:20planetbeingDoes anyone have the iPod nano 2G SST flash dump here?
22:38:43 Quit Lear ("ChatZilla 0.9.84 [Firefox 3.5b5pre/20090517065612]")
22:39:08GodEaterlinuxstb might
22:39:27pixelmabluebrother: \opt and \nopt sometimes don't, sometimes do. I experienced this before (or additional spaces, especially with \nopt) and then the % trick helped but yesterday just reordering the list helped. What puzzles me the most is that the behaviour seems to be so unforseeable, at least I couldn't find a pattern...
22:39:30planetbeingAll right.
22:39:55Unhelpfulgevaerts: thanks a ton! so, my build hates your build, or something?
22:40:07gevaertsprobably :)
22:40:23planetbeingBTW, I just looked and the newer nano 5G firmware stuff have nearly the exact same format as the 8900 format, which we do have a vulnerability for.
22:40:25Bagderplanetbeing: I do
22:40:38Unhelpfulwould you mind loading an oversized bitmap into sliding_puzzle? that should confirm that the coldfire asm in the core scaler for that build is correct :)
22:40:38BagderI think, lemme check closer
22:40:41planetbeingNot sure if Apple fixed the implementation on the nano though.
22:40:53 Join arohtar [0] (
22:41:08planetbeingAlso, I'm 99% sure that those things aren't signature checked.
22:41:31 Quit JdGordon| (" ajax IRC Client")
22:41:38planetbeingSince where the offset for the RSA signature should be is a number that points exactly past the end of the firmware file. =P
22:42:24*pixelma wonders whether daily built manuals for today are available now
22:42:37Bagderpixelma: no, we didn't trigger any new ones
22:42:48planetbeingHuh, that thing has a Syscfg partition like on the iPhone.
22:43:20pixelmaI see, will wait for tomorrow then. I guess it's not that important
22:43:24 Quit moredhel (Read error: 104 (Connection reset by peer))
22:44:52gevaertsUnhelpful: "Failed to load bitmap"
22:44:59planetbeingBagder: thanks
22:47:04Unhelpfulgevaerts: weird. thanks for testing, i think i'll give that coldfire emu a go and see if i can slap together a bit of a test framework for the thing :)
22:47:31gevaertsUnhelpful: I'll try to load the same images on another player
22:48:17froggymanare there any guides to syncing source files (.c)?
22:49:07Unhelpfuli'd probably best build a complete install for lambacalculus as well, if he still wants to test that
22:50:26Lloreanfroggyman: There is no practical way to write a guide for it.
22:50:42 Quit bmbl ("Woah!")
22:50:43LloreanYou need to read the file, figure out what's wrong by understanding the source, and update it.
22:50:48gevaertsUnhelpful: they also fail on a new gigabeat f build
22:51:03*gevaerts guesses that he has a collection of "interesting" jpg files
22:51:13Unhelpfulwait, did you say jpg?
22:51:24gevaertsoh wait...
22:51:25AlexPoh dear, it appears that alex wallis took kugel's (bad) advice
22:51:26Unhelpfulsliding_puzzle isn't using load_image yet :)
22:51:39froggymanLlorean: i mean like adding a new plugin, that doesnt have a patch yet
22:52:25GodEaterAlexP: huh ?
22:52:42AlexPGodEater: dev mailing list
22:52:58gevaertsUnhelpful: it's your fault for not having touched the scaler for weeks so I instinctively think about jpg whenever you say something
22:53:45Unhelpfulyes, of course, that's it. ;)
22:53:56AlexPGodEater: yep
22:54:06GodEaterwhatever is in that email is causing firefox to crash when it displays it in gmail
22:54:17Lloreanfroggyman: It's still the same issue. You need to understand how it works, what has changed that prevents it from working, and fix it
22:55:46high-rezSo will iPod accessories work with rockbox? E.g. my car has an ipod adapter - will it work? :) P.S. sweet software :D
22:56:02AlexPSome, yes
22:56:13high-rezThat's just friggin neat.
22:56:14evilnick_7high-rez: You'd have to try it to be sure, there is limited support
22:56:17gevaertsUnhelpful: it looks broken. Most of the screen keeps its background colour, and there's a white vertical line on the right of the image display area
22:56:29AlexPhigh-rez: It depends on what commands the accessort tries to use
22:56:55high-rezWhat is the "proper" place for me to put my music ? E.g. should I just create a new directory under the devices root called mp3 or is there a preferred location?
22:57:20AlexPhigh-rez: Anywhere you damn well want
22:57:23evilnick_7high-rez: Anywhere you want, (but not in the .rockbox directory)
22:57:24Unhelpfulgevaerts: that sounds broken. at least we know the rework of the C math isn't a loss on coldfire. :)
22:57:34AlexPhigh-rez: Although the .rockbox directory isn't a good plan
22:57:40high-rezIt's all about freedom eh ? :D
22:57:46AlexPyup :)
22:58:10high-rezYou guys rock. Thanks for freeing my ipod from apples evil grip. :D
22:58:36AlexPno worries :)
23:04:44 Join Mathiasdm2 [0] (
23:04:46 Quit tessarakt ("Client exiting")
23:04:54 Quit firedix ("Ex-Chat")
23:06:38 Quit {phoenix} (Read error: 104 (Connection reset by peer))
23:07:26*Zagor is having second thoughts about the new web design :-(
23:08:29*bluebrother wonders what those second thoughs are
23:09:51Zagoras we have discussed before, I'd like to make the index page more user (non-dev) oriented. that means moving a lot of the content to the dev page, so the front page is cleaner. but the new design was not made for that, so it means I'd have to re-design parts of it. which I suck at...
23:10:24CIA-38New commit by 03bluebrother (r20989): Clean up and improve themes install window debug messages.
23:11:06bluebrotheras in new design not made for a dev page?
23:11:22 Join mcuelenaere [0] (n=mcuelena@rockbox/developer/mcuelenaere)
23:11:58Zagorthe new design was made to make room for basically everything the current frontpage has, and then some.
23:14:11bluebrotherso? Wouldn't it be sufficient to have a link to the dev page and use a different layout on the development pages?
23:14:22bluebrother(i.e. something more developer-friendly)?
23:15:13Zagoryes. but it means I have to design a new front page. or at least refactor it substantially.
23:15:57Zagor...when the whole point of the prolonged process with macku was to avoid that
23:16:12bluebrotherwell, I've made up something a while back using some layout someone posted on the forums
23:16:27bluebrothersomething I'd consider both user- and dev-friendly :)
23:16:32 Join _fml [0] (n=4fd3ccda@gateway/web/cgi-irc/
23:16:41bluebrotherin case you want to give it a look:
23:17:45ZagorI like the polished feel of the new design. and we have sort of reached consensus about it.
23:18:03*froggyman likes the design
23:18:15Zagorthat's why I'm having a "doh" feeling about having to change it before we even start using it
23:18:58*bluebrother doesn't like the new design. It distracts too much
23:19:29Zagorthe front page is cluttered, yes. that's my point.
23:20:09 Quit jfc (Remote closed the connection)
23:20:49bluebrotherhow can it be clean when it's cluttered?
23:20:52 Join jfc [0] (
23:21:39ZagorI didn't say it is clean. I said I want to make it cleaner by moving dev stuff to a separate page.
23:21:53 Quit evilnick_7 (" ajax IRC Client")
23:21:59pixelmaI think this full 3 columns layout is one of the main reasons it looks cluttered
23:22:04ZagorI don't think it is possible to make a non-cluttered combined dev+user page
23:22:35Zagorpixelma: I think it is a combination of that and the lack of a central focus point
23:22:57 Quit Mathiasdm (Read error: 110 (Connection timed out))
23:23:40pixelmaand my main grep currently is that the dropdown menu is not navigatable with tab (unless you turn off css)
23:24:09pixelmato be more precise - it isn't fully navigatable, you can reach the top level things
23:24:51Zagoryeah I imagine our blind users might be less than thrilled about that...
23:25:14*pixelma should have a look at it the cleaned up css, if it is available somewhere
23:26:19pixelmathanks, will try to look at it tomorrow
23:26:28Zagorit's my cleanup plus gevaerts' size-dynamics changes
23:26:47pixelmanot sure what will come out of it though...
23:27:24gevaertsI seem to remember the buttons to jump to two lines if you made it narrow enough with my changes
23:27:51pixelmathe download buttons?
23:28:35Zagornot anymore. I modified some % values.
23:28:50*bluebrother really hates websites that use that much of javascript
23:29:05gevaertsthat was actually on purpose :)
23:29:38Zagorgevaerts: haha, ok. since nothing else wraps I felt it strange that they would.
23:30:14gevaertsIt still looks good if they are aligned (which I did IIRC), and it gains you another 200 pixels or so in narrowness without breaking everything
23:30:49Zagorhmm no, I didn't change that. I changed values for the announcements/about/support columns.
23:32:28***Saving seen data "./dancer.seen"
23:33:36 Join perrikwp [0] (i=4aa794a0@gateway/web/ajax/
23:34:09Zagorthat looks good. I seem to have missed a few things in my merge.
23:34:28 Quit BXCracer (Remote closed the connection)
23:34:31gevaertsThe two main breakages on that when narrowing are the manufacturer menu and the announcement section
23:35:36gevaertsat least with firefox. I haven't tested anything else with it
23:36:04Zagorwhat breaks in the announcements? it looks fine to me?
23:36:24pixelmait vanishes behind the buttons
23:36:56 Join spyder_ [0] (
23:37:09 Part spyder_
23:38:03gevaertsthe mechanism it uses for making only one appear at a time assumes a fixed height
23:38:34 Join itcheg [0] (i=62db4c46@gateway/web/ajax/
23:39:05pixelmadid those buttons work in his version?
23:39:33Zagoryes, they work on too. only I mucked up the css so it doesn't look right.
23:39:59gevaertshm, I don't remember breaking them...
23:41:38amiconnTab navigation doesn't seem to work at all with the new design, neither in firefox nor in IE8, Opera 9 or SRware Iron
23:41:51pixelmaand why again is the font in the more blueish text part smaller than above (in the white)?
23:41:55Zagorgevaerts: I think you are missing a # on line 487?
23:42:22pixelmaamiconn: it does - but the frame is surpressed and the links are just unnamed anchors #
23:42:36pixelmaall of them, so you can't see where you at
23:42:44Zagorpixelma: because macku liked to set different sizes and spacing everywhere. that's one of the things I've been changing.
23:43:09Zagorthe links will be normal. they are just # now because it's a mockup
23:43:26pixelmaamiconn: except the dropdown menu
23:43:40gevaertsZagor: fixed
23:46:04 Quit renke ("leaving")
23:52:21 Quit matsl (Read error: 110 (Connection timed out))
23:56:11CIA-38New commit by 03alle (r20990): Add emdash, endash and ellipsis to 12-Nimbus, 13-Nimbus, 14-Nimbus and 19-Nimbus (FS #10213 with additions by me)
23:57:25pixelmathe new page will also look weird in browser that don't support transparent png. I know those become rare and you can't make it work in every browser... looks *nice* in Opera 6...

Previous day | Next day