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

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

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

#rockbox log for 2010-06-20

00:00:11pixelmabluebrother: this wish had been expressed a few times, also at DevCon IIRC :)
00:03:37Jerompamaury: "/* idem but for i=0 */" idem is a french word
00:03:48 Quit M3DLG (Quit: RAGE QUIT)
00:04:08pamauryno it's a latin one
00:05:03pamauryBut yeah I got the idea, I'll change that
00:06:38***Saving seen data "./dancer.seen"
00:09:07saratogai saw some old posts suggesting that volume was a little low on amsv1 targets too compared to the OF
00:09:11saratogahas anyone else noticed this?
00:12:20bertriksaratoga, I find it hard to tell/compare, but I hardly ever go louder than -20 dB on my clipv1
00:13:17bertrikin other words, not anywhere near too low for me
00:13:25 Join AsusFreak [0] (
00:14:09AzureSkyAnother problem, and it was working last night, its telling me arm-elf-eabi-gcc command not found....
00:14:37saratogathat means you either didn't build the compiler, or did not add it to your path
00:14:57AzureSkyit was working last night.........stupid cygwin......
00:15:07JeromAzureSky: launch
00:15:08saratogai'm guessing you exported it to the path?
00:15:13Jeromand build eabi
00:15:35saratogaand then closed cygwin
00:19:14 Part Jerom
00:20:18 Join Jerom [0] (~Jerom@
00:21:54AsusFreakHi JdGordon, you remember me - the guy who tried to implement a pictureflow integration with the Cowon D2. You made it possible to use the toucharea "hotkey" tag for the D2. It worked like a charm for both 3x3 grid mode and absolute point mode. But now something seems to be broken. 3x3 grid mode works like a charme, absolute point mode does not at all. Clicking on my album art (which is defined as "hotkey" area will lead to nothing. May you check the cod
00:22:28AsusFreakTo test this you may use the actual simulator with my AF_Flow theme..
00:45:40AsusFreakAnd now with further testing: the ffwd tag seems to fail as well in absolute point mode. Hmmm. Can anybody check this please?
00:46:26 Join Gabe_G23 [0] (~gabe@bzflag/player/GabrielG)
00:47:16 Nick CGL is now known as CGL_emo (~CGL@
00:51:44 Nick bimbel is now known as bmbl (~Miranda@unaffiliated/bmbl)
00:52:43 Quit bertrik (Quit: De groeten)
00:57:59 Quit bmbl (Quit: Bye!)
01:09:10 Quit Jerom (Remote host closed the connection)
01:12:51 Quit whydoubt (Ping timeout: 265 seconds)
01:12:55 Join whydoubt [0] (
01:13:36 Quit bluebrother (Disconnected by services)
01:13:40 Join bluebroth3r [0] (~dom@rockbox/developer/bluebrother)
01:18:23 Quit ender` (Quit: Religion is regarded by the common people as true, by the wise as false, and by the rulers as useful. -- Seneca)
01:23:17 Quit DataGhost (Ping timeout: 276 seconds)
01:29:51 Quit AsusFreak (Quit: CGI:IRC (EOF))
01:35:12saratogathe flash memory in my clip+ strikes me as pretty slow compared to other sansas, even in the OF
01:46:00AzureSkymaximizing profit by using cheaper parts, nothing surprising there really
02:06:39***Saving seen data "./dancer.seen"
02:12:26 Quit GeekShadow (Quit: The cake is a lie !)
02:13:57 Join BHSPitMini [0] (
02:22:24 Quit pamaury (Remote host closed the connection)
02:42:05dfktsaratoga, formatted it with the official SD formatting tool?
02:44:35 Join Strife1989 [0] (
02:47:51 Quit Strife89 (Ping timeout: 248 seconds)
02:52:18AzureSkyi was told to use the internal formatter rather then the sdformat tool, but i have done both on my players
02:54:17AzureSkywell its confirmed, 26937 is what broke microsd support for v2 fuze
03:08:14 Quit BHSPitMini (Ping timeout: 264 seconds)
03:31:34 Quit krazykit (Ping timeout: 265 seconds)
03:34:49 Nick Strife1989 is now known as Strife89 (
03:34:58JdGordonwtf? I thought putting functions into a .a would stop name collisions with the main binary?
03:35:11JdGordonstupid skip_whitepace() being in both
03:35:46saratogaits just formatted however sandisk did it
03:36:44AzureSkytry that
03:39:22 Join funman [0] (~fun@rockbox/developer/funman)
03:41:54 Quit Luca_S (Quit: CGI:IRC (EOF))
03:43:06dfkti got immense speed gains by using the formatter - not only with sd cards and the like, but also CF and sony memorystick
03:45:51 Quit steve|m (Ping timeout: 276 seconds)
03:46:17funmanFWIW, cowond2 has been checked to work with eabi but noone committed the change yet
03:47:01 Join WilliamC [0] (~WilliamC@
03:47:40AzureSkyfunman, if you havent seen, its confermed that its 37 that breaks sd card support 36 works, and changing the clock didnt work :(
03:47:47WilliamCSo, uh, how do I report a bug in a daily build?
03:48:32funmanAzureSky: ok
03:49:33WilliamCThemes no longer work on my Sansa e200
03:49:53saratogaupdate them
03:49:53WilliamCWell, they don't work when playing music
03:50:37WilliamCDoubt that will hjelp
03:50:56saratogawhy do you doubt that?
03:51:16WilliamCAll my themes worked until just a few days ago.
03:51:24WilliamCIncluding the default.
03:51:25AzureSkyyou know they changed the theme format dont you
03:51:41saratogaso what you're saying is that because something was up to date in teh past, it can't be out of date in the future?
03:51:52WilliamCUp to date 3 days ago
03:52:11funmansaratoga: trying flac @44.1kHz 16 bits (dunno about the quality) on clipv2
03:52:29saratogafunman: i just used the test_file flac and apes
03:52:45saratogai also noticed one or two glitches playing mp3s, but i'm not sure if I was imaging things
03:52:55saratogathe APE ones were really obvious though since the decoder did not catch the error
03:53:03saratogaand just played loud static
03:53:03funmanthe same thing happened to me for mp3 today
03:53:07 Join krazykit [0] (~kkit@
03:53:43funmanbtw which target did you use?
03:53:47WilliamCI install the daily build on a daily basis.
03:53:55funmanit might be different: fuzev1/e200v2 have codec buffer in DRAM and all the others in IRAM
03:54:37funmanconsidering the µSD bug on fuzev2 and the problems I had with PLLA@384MHz on clip+ i think we should revert the PLL change for AMSv2 to rule out problems with SD
03:56:11funmanfor SD i tested with test_disk on fuzev1/fuzev2, with µSD and internal, and with write&verify and speed test
03:56:57funmanlast time i had tried such a change it crashed quite fast but i supposed since then we had removed the last writing bugs from the drivers
03:57:31WilliamCAlso, I have it playing from my folder on my Sansa E200 but it doesn't seem to give me much variety.
03:57:47saratogafunman: test_disk gave no errors
03:57:52saratogaso i'm not sure what to make of it
03:58:05WilliamCIs there to randomize it more?
03:58:13funmanflac plays fine so far here on clipv2, i'll just revert the PLL change and let's see if we still see bugs
03:58:15WilliamCIs there a way to randomize it more
03:58:38saratogaAFAIK its all ready completely random so you can't get more random then that
03:58:58WilliamCSo, why isn't it playing some songs?
03:59:08funmanpress next a random number of times, at random intervals
04:00:12saratogaWilliamC: random playback doesn't mean it'll play every song ...
04:00:19saratogathats actually sort of the opposite of random :)
04:00:41 Join steve|m [0] (
04:00:41gevaertssaratoga: shuffle does play everything...
04:00:48gevaertsJust in a random order
04:01:07CIA-8New commit by funman (r26979): Revert r26937 (as3525v2: use 248MHz PLL) ...
04:01:13saratogayeah but theres no promise it'll ever play every song :)
04:01:26gevaertsyes there is
04:01:31gevaertsunless you stop it
04:01:34saratogaoh you mean shuffling a playlist?
04:01:57gevaertsIs there another random option?
04:02:02funmanIIUC playback settings -> shuffle does just that
04:02:20saratogarepeat shuffle?
04:02:25 Quit kramer3d_ (Ping timeout: 240 seconds)
04:02:45saratogaor does that just shuffle the current folder if i select it from the quickscreen
04:03:05CIA-8r26979 build result: All green
04:03:05saratogai assume not, since theres both shuffle and repeat shuffle
04:03:47funmanTheSeven: i think i have posted a patch which aligned fat stack buffers some time ago, but a simple cache strategy might be worth
04:03:51JdGordonrepeat shuffle will still shuffle the whole playlist before playing any song twice
04:04:04saratogawhy does the clip+ quick screen show "show files" twice?
04:04:04JdGordonunless that song was in the playlist twice, in which case it wont be played a 3rd time
04:04:21WilliamCMan, some of these themes suck
04:04:26saratogaso whats the difference between shuffle and repeat shuffle?
04:04:26JdGordonbecause if an item is empty it uses the opposite item
04:04:28WilliamCWTF, Screamiing Oprah
04:04:29funmansaratoga: same on clipv1 (no idea why)
04:04:52AzureSkya friend of mine asked me to ask if theres any chance of ading .pla playlist support(the playlists you can make using rclick in windows explorer or with alot of tools for devices like the fuze)
04:04:57JdGordonsaratoga: shuffle will shuffle the current playlist and stop, shuffle repeat will reshuffle at playlist end
04:05:07JdGordonit is redundant and iirc there is a patch to reove it
04:05:08saratogaah that makes sense
04:05:14funmanAzureSky: current build should detect µSD fine on fuzev2
04:05:26AzureSkyok i will giveit a test :D
04:05:28saratogaAzureSky: no plans for it
04:05:34JdGordonAzureSky: there is a patch to convert pla t m3u
04:06:23AzureSkyah didnt see one, it run on the device itself?
04:06:41***Saving seen data "./dancer.seen"
04:06:45AzureSkyi just use createM3U then edit the ones that are for multi folders
04:07:21WilliamCsaratoga, you're wrong
04:07:24AzureSkyone note, m3u/m3u8 dont work if you ad both sd card and internal memory items last i tested, but pla file do work
04:08:00WilliamCI have a bug to report
04:08:39funmansaratoga: btw about embedded art i think ucchan has posted a patch on flyspray
04:08:57saratogayeah but it wasn't suitable
04:09:09WilliamCI'll just purge the entire .rockbox folder.
04:09:57 Quit funman (Quit: free(random());)
04:11:07 Quit obo (Ping timeout: 265 seconds)
04:11:48saratogaah heres the playlist converter:
04:12:40AzureSkybut do the coverted playlists work on combine internal and external memory tracks?
04:13:21AzureSkyand yes, the new build does the trick for sd support :D
04:13:25 Quit dfkt (Quit: -= SysReset 2.53=- Ph'nglui mglw'nafh Cthulhu R'lyeh wgah'nagl fhtagn.)
04:14:13 Join Barahir_ [0] (
04:14:22 Join webguest02 [0] (
04:15:22 Quit webguest02 (Client Quit)
04:17:36 Quit Barahir (Ping timeout: 265 seconds)
04:19:47 Quit togetic (Ping timeout: 260 seconds)
04:20:27WilliamCHow do I save my dynamic playlist?
04:20:52WilliamCEverything I do either moves the cursor or adds/deletes a letter
04:25:15 Quit binaryhermit (Read error: Connection reset by peer)
04:25:26 Quit efyx (Remote host closed the connection)
04:25:31 Join binaryhermit [0] (
04:26:36WilliamCAlso, saratoga, wrong wrong wrong wrong wrong!
04:26:52 Join kramer3d [0] (~kramer@unaffiliated/kramer3d)
04:28:06WilliamCI guess I'll just have to wait for most of my favorite themes to get updated.
04:28:47saratogaWilliamC: ?
04:29:10WilliamCThe themes are still broken, actually all the themes but the default are broken.
04:29:23saratogawhich themes are these?
04:30:15mc2739WilliamC: there is an updater for the skin syntax change.
04:30:39WilliamCblack and blue glass, cabbie plus, cabbie v2, Vision, Sentinel
04:31:38WilliamCcabbie v2 works.
04:31:41 Join kramer3d_ [0] (~kramer@unaffiliated/kramer3d)
04:32:16saratogaWilliamC: where/when did you get them from?
04:32:25WilliamCThe utility
04:32:37WilliamCInstallation utility
04:33:44 Join togetic [0] (~togetic@unaffiliated/ibuffy)
04:34:19 Quit pixelma (Disconnected by services)
04:34:19 Join pixelma_ [0] (quassel@rockbox/staff/pixelma)
04:34:24 Quit amiconn (Disconnected by services)
04:34:26 Join amiconn_ [0] (quassel@rockbox/developer/amiconn)
04:34:27WilliamCNone of the themes I have installed were updated recently.
04:34:35 Nick pixelma_ is now known as pixelma (quassel@rockbox/staff/pixelma)
04:34:44 Nick amiconn_ is now known as amiconn (quassel@rockbox/developer/amiconn)
04:35:00 Join DerPapst1 [0] (
04:35:15saratogaWilliamC: other half of my question :)
04:35:34 Quit kramer3d (Ping timeout: 258 seconds)
04:35:44WilliamCCabbies V2 works but I don't remember installing that, it must have come with the build
04:37:13 Quit DerPapst (Ping timeout: 265 seconds)
04:37:43WilliamCHow do you save playlists?
04:37:53AzureSkyyou could try a fresh install of rockbox.
04:38:02WilliamCAzureSky, I just did
04:38:05mc2739WilliamC: You can fix them with the skinupdater - a windows version can be downloaded here -
04:38:43saratogaor probably just download the updated versions
04:38:57WilliamCThey don't exist though
04:39:10saratogahave you checked?
04:39:16saratogaIIRC almost all were updated
04:39:30saratogadid you check poorly?
04:39:56WilliamCNext you're going to be asking me if it's plugged in.
04:40:09saratogaAFAIK all themes on that site should be updated
04:40:13saratogadid you check any of them?
04:40:36WilliamCYES I DID GOD DAMN IT
04:40:59saratogawhich one? I'll try it
04:41:04saratogahave to run in a moment though
04:41:05binaryhermitrockbox utility seems to download the old, fubar versions of the themes rather than the new fixed version
04:41:14binaryhermitin my experience
04:41:21 Join Piki [0] (~Piki@unaffiliated/piki)
04:41:28WilliamCThe website says use the utility though
04:41:44binaryhermitif you download the themes manually you get the fixed versions
04:41:46saratogai only have a couple minutes, so if you want help, hurry up
04:41:51AzureSkyi use the svn version
04:41:57binaryhermitagain, in my experience
04:42:14AzureSkydosnt support skins for fuzev2 so i have to pretend its a v1 :P
04:42:41WilliamCI cannot find the skins I'm using
04:43:20WilliamCI am confused.
04:43:40saratogado they show up in rbutil?
04:44:26saratogaso which are they?
04:44:33WilliamCThe vision ones
04:44:33mc2739WilliamC: maybe try the latest svn version of Rockbox Utility -
04:44:45WilliamCSand, Wave and Wood
04:44:49saratogacabbie plus?
04:45:49saratogawell Wave and Wood doesn't appear to have come from the theme site, so no surprise its not updated :)
04:46:25WilliamCI'll use that update utility I guess
04:46:44saratogawhich ones were you saying didn't update on the theme site?
04:47:08WilliamCThe ones I'm complaining about aren't listed
04:48:00WilliamCActually Black n Blue Glass is listed
04:48:04WilliamCLEt me see
04:48:37WilliamCThat works
04:48:50WilliamCSo, the utility is giving me old themes?
04:49:05saratogacabbie plus is, and it works too from the theme site, but rbutil installs the old one
04:49:11saratogalet me see if a newer rbutil fixes that
04:49:43WilliamCSorry, I just get angry when I feel that people are treating me like an idiot.
04:50:31WilliamCI think I developed that from unsatisfactory tech support calls where they try to avoid doing any actual support.
04:51:09saratogayeah the SVN installer works fine
04:51:15WilliamCok, thanks
04:51:20saratogabut the current release still installs themes for the old syntax
04:51:30Pikihello, just did the manual install of Rockbox on a Fuze v2 (I know, it's an unstable port, but was working fine when i first installed it). after turning on the Fuze, Rockbox loaded normally like it did on my Fuze v1. i plugged in the v2 without shutting it off, and the screen went blank. i tried a soft reset (holding the power button in the on/off position for up to a minute). i also tried putting the power button in the hold position and plugging in while
04:51:32Pikiholding the submenu/playlist button, and tried the same thing again except that i held in the center/select button
04:51:46Pikisorry for the long post, but i was hoping for a bit of help
04:51:47WilliamCsaratoga, how do I save the dynamic playlist?
04:52:17saratogaWilliamC: no idea
04:52:44WilliamCShame there is no enter key and the select key just enters in new letters.
04:53:11saratogaPiki: if holding power for a long time doesn't work, i think you have to let the battery run out
04:53:38WilliamCHolding the power button normally works.
04:54:07WilliamCFirst time I had a kernel panic, I followed suit.
04:54:10AzureSkyim not worried if i brick the fuze, i paid for 2 year geeksquad warr that will replace it no questions asked :P
04:54:32WilliamCAzureSky, those warranties are a bunch of BS, btw.
04:54:34Pikisaratoga: any guesses as to how long a fully charged batttery will take to discharge? or will i just have to guess?
04:54:37saratogaif those keys are wrong, post a flyspray ticket and someone will probably fix it
04:54:57AzureSkyfunny, i have used it many times willy and its worked eveyr time on every device i have bought at bb
04:54:58saratogadepends what it was doing when it crashed, but the battery lasts about 20 hours fully charged playing music, so thats a rough estimate
04:55:11WilliamCAzureSky, if you want to get them to actually fix or replace anything, you'll get the urge to ram the phone up their rectum eventually.
04:55:14AzureSkyPiki, up to 3 or so days
04:55:44PikiAzureSky: for a blank screen? wow, i thought it would be longer
04:55:50AzureSkywilly, u go into store, they take the player give you a card with the players value when you bought it on it, you go pick ur replacement
04:56:28AzureSkyor u can do a strait swap in store, i have done this many times, never had a problem
04:56:30WilliamCAzureSky, oh, well, when I had an extended warranty they had me deal with a call center first, they tried to find reasons to not help me.
04:56:40AzureSkyPike thats what the Wiki says for time
04:56:47WilliamCThis was CircuitCity
04:56:57AzureSkyan bestbuy isnt like that
04:57:09AzureSkythey honestly just dont give a crap
04:57:22PikiAzureSky: not to be picky, but i'm not based on a street, i'm based on a "Piki" person
04:57:36 Quit TheSeven (Ping timeout: 264 seconds)
04:58:04AzureSkyan option that works well is to get a squaretrade warr on stuff, they replace it even if u smash it with a hammer :P
04:58:10saratogarockbox-community is a better place to chat about off topic things
04:59:52 Quit saratoga (Quit: Page closed)
05:00:46 Join HANSi [0] (~CLD@
05:02:52 Join TheSeven [0] (~TheSeven@rockbox/developer/TheSeven)
05:02:58 Quit saratoga_ (Quit: Page closed)
05:10:03 Part WilliamC ("Leaving")
05:15:34 Quit HANSi (Read error: Connection reset by peer)
05:20:28 Nick CGL_emo is now known as CGL_party (~CGL@
05:23:40 Part Piki ("Konversation terminated!")
05:29:35 Quit t0rc (Remote host closed the connection)
05:42:00 Join HANSi [0] (~CLD@
05:53:07 Quit anewuser (Ping timeout: 260 seconds)
06:06:45***Saving seen data "./dancer.seen"
06:07:55 Join kramer3d [0] (~kramer@unaffiliated/kramer3d)
06:08:20 Nick kramer3d is now known as destiny (~kramer@unaffiliated/kramer3d)
06:09:28 Nick destiny is now known as Azaroth (~kramer@unaffiliated/kramer3d)
06:12:09 Part HANSi ("Leaving")
06:19:41 Join JesusFreak316 [0] (
06:20:10 Quit JesusFreak316 (Client Quit)
06:37:50 Quit panni_ (Read error: Connection reset by peer)
06:49:05 Join hebz0rl [0] (
06:49:31 Quit Azaroth (Quit: Leaving)
06:55:58S_a_i_n_tsaratoga: There are still quite a few themes missing, and a few pass checkwps, but have invalid syntax.
06:56:25S_a_i_n_tUpdating the themesite didn't really go that well IMO
06:59:16 Quit mc2739 (Ping timeout: 276 seconds)
07:00:08S_a_i_n_tThere's 4 or 5 missing from Nano 1/2G alone, I really should look which ones they are, but I haven't had much luck trying to get one theme updated, let alone several.
07:00:32 Join mc2739 [0] (~mc2739@rockbox/developer/mc2739)
07:00:32 Join BHSPitMini [0] (
07:01:11AzureSkyseems most of the themes for my fuze use 24hr day on the playing screen, but every other screen they use the setting i choose(12hr)
07:01:51S_a_i_n_tI ened up updating mine myself, which is ok for me but I don't want to go behind an authors back and re-submit a theme.
07:02:00S_a_i_n_tAzureSky: You can edit that easily.
07:02:23AzureSkycan u give me a hand only 2 themes i use both have the issue
07:02:29S_a_i_n_tIt just means the theme forces 24hour instead of using a 12/24 hour conditional (which is poor form IMO)
07:02:48S_a_i_n_tThe manual can give you a hand...
07:02:53AzureSkyi can deal with 24 but i just dont want to :P
07:04:36S_a_i_n_tFind the clock line in the WPS and replace it with "%?cf<%cH:%cM|%cI:%cM%cP>"
07:05:13S_a_i_n_tthat line (I'm pretty sure I got it right...) will respect the user setting for 12/24 hour
07:06:13S_a_i_n_tIt will also pad the hour/minute with a 0 so it doesn't mess up alignment.
07:06:33AzureSky:) nice :D
07:11:41AzureSkyworking on it
07:15:45 Join Horschti [0] (~Horscht2@xbmc/user/horscht)
07:18:26 Quit Horscht (Ping timeout: 240 seconds)
07:20:57S_a_i_n_tAzureSky: If you really get stuck, pastebin the wps and I'll fix it up.
07:21:07S_a_i_n_tJust, at least have a go first ;)
07:21:29S_a_i_n_targh... S_a_i_n_t forgot to close the bold section.
07:26:34AzureSkygot 1 other that i need clock help on, edited like you showed and it didnt work, removed clock from the playing screen :P
07:27:05CIA-8New commit by mc2739 (r26980): FS #11418 by Peter Lecky: Slovak language update
07:28:47CIA-8r26980 build result: All green
07:32:28 Quit storm` (Ping timeout: 260 seconds)
07:35:32 Join storm` [0] (
07:35:44*S_a_i_n_t sees no reason why this code should not work.
07:36:12AzureSkywill test now
07:36:15S_a_i_n_tunless you included the ""s when you editied it?
07:36:22 Quit binaryhermit (Ping timeout: 264 seconds)
07:36:51AzureSkynot that i know of but i just cleaned out a stack of themes i dont like, i try and keep things light
07:40:41 Join binaryhermit [0] (
07:41:04AzureSkywaiting for refresh.....stupid fuze refresh of db every time i edit a file :P
07:42:21AzureSkyworked great :D
07:42:32AzureSkyi see what i did wrong to :)
07:42:55AzureSkyi suck at editing files that look like gibberish :P
07:45:08AzureSkyi gotta get a buddy to remote in again and get my cygwin to work again, so i can compile latist build with saratoga's volume patch
07:45:32S_a_i_n_twhat's wrong with your CygWin?
07:45:41S_a_i_n_tit can be a PITA, but is easy to setup.
07:46:52AzureSkyit was working last night now it keeps saying its missing the arm-elf-eabi-ld
07:47:06S_a_i_n_tJust delete the Gygwin folder completely, reinstall according to the wiki, don't download the toolchains though, checkout the rockbox source, run, done.
07:47:34S_a_i_n_tOh, Hmmm...ignore that then, that's overkill for that fix.
07:47:47AzureSkyno clue how to run that to be honest im not at all a programer
07:48:07S_a_i_n_tremoving the toolchain and re-running should fix that.
07:48:13AzureSkyi use to be able to script for sphere UO server emulator a bit thats it :P
07:48:33AzureSkyhow do i run the
07:48:43S_a_i_n_tbut, even that is probably overkill compared to what is probably needed to fix it.
07:48:51S_a_i_n_tone sec.
07:49:28S_a_i_n_tjust cd to "tools" then "./"
07:49:29AzureSkyi SUCK at this stuff, if i had an acct like my buddy with citrix easy support i would let you remote in and do it :P
07:51:26AzureSkyits doing its thing now
07:51:31AzureSkyhope this works :)
07:52:10AzureSkyi thought i was done using commandline all the time when i moved to windows 2000 years ago :P
07:52:11S_a_i_n_tIt should do, you told it to build arm-eabi by inputting "e" I take it?
07:52:45AzureSkydid the one that listed sansa and e as well
07:53:23AzureSkyone nice thing about having a 6 core, even when doing stuff like compiling i never endup being bogged down :D
07:53:47S_a_i_n_t*cough* /topic *cough*
08:06:47***Saving seen data "./dancer.seen"
08:08:40AzureSkyok, quick question, how do i go about using satatoga's volume patch?
08:30:44AzureSkyok, 1 more what is the command to update the source i already have to latist rather then redownload the whole thing again?
08:30:55S_a_i_n_t"svn update"
08:31:09S_a_i_n_tseems to simple no?
08:31:34AzureSkyyeah that is to simple
08:31:45 Quit leavittx (Read error: Connection timed out)
08:32:02 Join mitk [0] (
08:32:07S_a_i_n_t"svn −−help" will tell you other possibly commands.
08:32:16 Join leavittx [0] (~leavittx@
08:32:19S_a_i_n_tadding "−−help" to almost anythng works.
08:32:51AzureSkydo i need to add a revision number or url to the "svn update" command?
08:33:33AzureSkynm its working :D
08:33:44S_a_i_n_t"svn update" will always pull the latest revision.
08:33:58S_a_i_n_teven "svn up" wil work.
08:34:42 Join ast[1] [0] (
08:35:43 Quit ast[1] (Client Quit)
08:37:30*S_a_i_n_t points AzureSky to:
08:37:48S_a_i_n_tit answers all the basic "How do I...?" development questions.
08:50:53 Join Highlander [0] (
09:05:55 Quit kramer3d_ (Ping timeout: 260 seconds)
09:26:28 Join bmbl [0] (~Miranda@unaffiliated/bmbl)
09:28:40CIA-8New commit by jdgordon (r26981): fix the viewport tag [arams
09:30:23CIA-8r26981 build result: All green
09:38:47 Join Kitr88 [0] (~Kitar_st@
09:42:25 Quit Kitar|st (Ping timeout: 276 seconds)
09:43:15 Quit Kitr88 (Ping timeout: 248 seconds)
09:43:43 Join flydutch [0] (
09:48:43 Join Kitar|st [0] (
09:50:42 Join stoffel [0] (
09:57:14 Join Rob2223 [0] (
10:00:37*JdGordon spots a rather bad bug in the legacy skin parser :/
10:00:47 Quit Rob2222 (Ping timeout: 265 seconds)
10:06:49***Saving seen data "./dancer.seen"
10:09:23JdGordonI assume noone noticed that sharing fonts between skins doesnt actually work?
10:09:48 Join petur [0] (~petur@rockbox/developer/petur)
10:09:49JdGordonthey will work, but the font will be loaded twice!
10:11:29 Quit TheSeven (Ping timeout: 272 seconds)
10:13:07 Quit binaryhermit (Quit: Leaving)
10:13:53pixelmasharing fonts between skins - what do you mean with that?
10:17:00JdGordonif you load the same font in more than one skin, the font should only load once
10:17:13JdGordonI just noticed that is almost certainly not happening
10:31:56S_a_i_n_tI thought I noticed that by looking at the skin buffer...but I wasn't sure, nor could I confirm it.
10:41:44 Join hamish_ [0] (~hamish@
10:41:54 Quit hamish_ (Client Quit)
10:44:13 Join ender` [0] (
11:03:23 Quit mc2739 (Read error: Operation timed out)
11:05:03 Join mc2739 [0] (~mc2739@rockbox/developer/mc2739)
11:09:55 Join bertrik [0] (~bertrik@rockbox/developer/bertrik)
11:37:01 Join ender1 [0] (
11:37:19 Quit ender` (Disconnected by services)
11:37:23 Nick ender1 is now known as ender` (
11:39:57*S_a_i_n_t wonders where to start to try and make "make voice" smart enough to do "nothing to be done for `all`" instead of cimpiling a voicefile even if no changes were made...
11:50:24 Join Jerom [0] (~Jerom@
11:51:30 Join andromedea [0] (
11:58:35andromedeaI read through the links in the topic and can't help myself. I'm trying to install rockbox on my sansa e200 and keep getting "Bootloader mismatch! Aborting"
11:59:39bertrikandromedea, is it an e200 or and e200v2?
12:00:15bertrikI think you get this message when you installed the correct bootloader but the wrong rockbox main firmware
12:01:21bertrikWhat rockbox version exactly are you trying to install? I could have a try with the same version on my e200.
12:04:06andromedeaHere's the trick: it's potentially an R. I bought it refurbed, and it said "rhapsody" on it, but performed no rhapsody functions.
12:05:05bertrikSorry, I'm not familiar with the installation procedure for the rhapsody versions
12:06:51***Saving seen data "./dancer.seen"
12:18:21pixelmaandromedea: did it just say "Rhapsody" on its back plate or so? Sandisk is known to assemble those together in different ways... What were the USB mode settings in the original firmware called?
12:27:51andromedeait just says rhapsody on the back plate. ... and now it won't even turn on.
12:29:50andromedeausb setting were msc/mtp
12:30:09bluebroth3rwhat is producing this "Aborting!" message? Rockbox Utility?
12:30:48bertrikoh sorry, then I gave the wrong answer to your first question, I thought it was the bootloader complaining
12:31:23bluebroth3rok, then then nothing should have been written to the player.
12:31:32pixelmathen it's no Rhapsody as far as I know. The USB settings are called something like "Plays for sure" or somesuch
12:32:00andromedeaI figured
12:32:41andromedeav 1.02.24
12:32:44bluebroth3rhave you already tried resetting the player? I.e. hold Power until it turns off, then power it up again?
12:33:07pixelmabluebroth3r: the error "bootloader mismatch" sounds a bit like there already is a Rockbox bootloader on it, or not?
12:34:13andromedeagoogle turns up depressingly little
12:34:17bluebroth3rno, that is an internal check of the downloaded file
12:34:52bluebroth3rafter the bootloader has been downloaded it gets its version information checked.
12:35:02bluebroth3randromedea: have you saved the system trace of Rockbox Utility?
12:35:51JdGordondo we currently allow %xd() and %vd() to reference a not yet loaded image/viewport?
12:36:04JdGordoni.e before the coresponding %xl/%vl ?
12:36:19S_a_i_n_tJdGordon: Yes
12:36:28andromedeaI'm not following
12:36:32 Join DataGhost [0] (
12:36:32 Quit DataGhost (Changing host)
12:36:32 Join DataGhost [0] (~dataghost@unaffiliated/dataghost)
12:36:40JdGordonS_a_i_n_t: ok thanks
12:36:42JdGordonbugger :p
12:36:57S_a_i_n_tsorry, I just answered the question :P
12:37:02bluebroth3randromedea: after that error appears there should be a "save log" button in the log windows. Did you use that?
12:37:30pixelmaJdGordon: %Vd before the %Vl was said to be a must with the first implementation
12:37:48JdGordonok, yeah, just making sure the new parser owuldnt break that
12:37:55JdGordoneasy enough to work around it
12:38:18S_a_i_n_tI find it good practice to load before the call, but it isn;t necessary.
12:38:38bluebroth3rif not, have you closed and opened Rockbox Utility in the meantime?
12:39:14andromedeafail.log is empty
12:39:23andromedeaand yes, i have
12:39:31bluebroth3rempty? That's strange.
12:39:36bluebroth3ris Rockbox Utility still running?
12:39:48 Quit cjcopi (Read error: Operation timed out)
12:40:05bluebroth3rok, can you go to the Help / Troubleshooting / System Trace menu? It should show the most recent trace
12:42:55bluebroth3ryou have selected the c200 as player but it's in fact an e200
12:43:21bluebroth3rinstalling the wrong bootloader would cause serious problems, so bootloader installation verifies this and refused to install the wrong bootloader.
12:43:34 Quit JdGordon (Quit: Leaving.)
12:43:35andromedeawhat an idiot
12:43:36 Quit JdGordon| (Remote host closed the connection)
12:43:40bluebroth3r(this at least works for e200 vs c200, it's a bit different for other players)
12:43:51bluebroth3rplease try again with the correct player setting :)
12:45:25andromedeai just want to point out that "c" and "e" bear a strong resemblance at 3am
12:46:09 Join halmi [0] (
12:46:30*S_a_i_n_t hands andromedea a screen magnifier ;)
12:47:08*pixelma blames Sandisk ;)
12:47:13bluebroth3rdon't worry, such mistakes happen. Though autodetection should have worked done the correct thing for you ;-)
12:50:06andromedeaoh good, iwas hoping this would fix the jumpy wheel thing, but it didn't
12:50:43 Join efyx [0] (
12:52:41andromedeathanks for the elp guys :)
12:58:56 Join pamaury [0] (
12:59:04 Quit pamaury (Changing host)
12:59:04 Join pamaury [0] (~quassel@rockbox/developer/pamaury)
13:03:43andromedeahoo boy, now my computer doesn't recognize the sansa
13:05:05pixelmawhat operating system does your computer run?
13:05:16amiconnpixelma: Building manuals automatically might be desirable, but there is at least one fundamental problem
13:05:44andromedeawindows 7
13:06:14amiconnBuilding them only after commits to manual/ isn't enough, because afaiu the manuals also use the config/ stuff to figure out features
13:06:23amiconnThey also use tools/configure.
13:06:50amiconnBut building manuals after every commit is overkill, imo
13:07:32pixelmaandromedea: weird... your Sansa and its microSD slot should get a drive letter in explorer, I'm not aware of USB problems with Windows7
13:08:07*S_a_i_n_t still needs to turn off HID on the Nano1g sometimes on Win7
13:09:15pixelmaamiconn: I thought of that too (the features parsing), but it will still be much better than now and if a commit to features.txt will break a manual then it needs a commit to manual/ to fix it anyway (or to features.txt again)
13:11:32amiconnThe features file is generated as part of the build (btw, that's the reason why a manual build as well as a voice build needs the target crosscompiler)
13:11:38 Join JdGordon [0] (~jonno@rockbox/developer/JdGordon)
13:12:18amiconnSo if a code change that changes features breaks a/the manual(s), you won't notice if it doesn't trigger a manual build
13:13:04pixelmabut now you sometimes even won't know if a manual commit breaks a manual
13:13:40pixelmayou have to wait for the next day and check (and not forget in the meantime)
13:14:18amiconnSince we don't have automated manual builds now, I'd expect the committer to test more thoroughly before commit
13:14:47pixelmaI test a lot but still overlooked one manual last time
13:15:19 Join Buschel [0] (
13:15:44amiconnI'm not saying automated builds are bad - but defining the conditions which should trigger a build seems to be complicated
13:16:57amiconnBtw, the automated binary builds have the same problem, albeit to a lesser degree
13:17:53pixelmaand as you said yourself, manuals can break even when committing something to the rest of the code which happened last time with some hotkey enabling. Then almost no-one notices
13:18:33 Join t0rc [0] (~t0rc@unaffiliated/t0rc/x-5233201)
13:19:00amiconnAnother problem is scanning the build output for errors, given the ton of messages a manual build produces
13:19:38pixelmaIMO doing commits on changes to manual/ would already be a great help - or another idea discussed sometimes: a frontend that lets committers trigger a new daily manual in case it was broken
13:19:47amiconnThis output contains numerous 'warning' and 'error' messages which are meaningless - but some will be real
13:20:17amiconnThis is different from the main builds
13:21:12 Nick Horschti is now known as Horscht (~Horscht2@xbmc/user/horscht)
13:21:17pixelmain my experience it'll do with checking if it compiles at all and not using non-stop mode. On an error that really breaks a manual, the build process stops (and asks you what it should do)
13:21:42amiconnYes, and that's even worse
13:22:01pixelmaeh, why?
13:22:07amiconnYou cannot automate that - if you have a manual that doesn't build, the process won't end
13:22:31amiconnNobody is there who could answer that request. You could kill the process
13:22:56amiconnBut you won't see what messages it generated. Logging all of them is not an option I'm afraid
13:23:16amiconnSo all you get is a simple yes/no answer - did it generate an output file or not?
13:24:12bluebroth3rwith non-stop mode building should break on errors, not ask for user input
13:24:37bluebroth3rthe problem is when you try to manually interrupt building because LaTeX will then go into interactive mode
13:24:44amiconnYeah okay - that still leaves the problem that you don't get to see the messages
13:24:46pixelmabluebroth3r: yes, but how do you find the "offending" error then?
13:25:10bluebroth3rpixelma: seems like those guys managed to get an output parser done:
13:25:34bluebroth3rI was thinking in looking into it in the hope to generate an output filter script
13:25:51bluebroth3rbtw, the output itself is also written into the log file in the build folder.
13:26:47bluebroth3rone option to address the dependency issue would be to automatically generate dependencies for the manual, and trigger a rebuild if any of those files changes. This will of course fail if the dependencies change, but IMO this would be a minor issue
13:26:52amiconnDid you compare the log size of a manual build and a main build?
13:27:34bluebroth3rno, why?
13:28:11bluebroth3rI know LaTeX prints heaps more of (partly unimportant) stuff
13:28:21pixelmafor upload purposes I guess
13:28:39pixelmaand for people to look through
13:29:29bluebroth3rthat's the reason why I wanted to try getting a parser for the log files done
13:31:41bluebroth3r66kB for this somewhat outdated manual build log file
13:31:41 Part andromedea
13:31:47pixelmaand if Zagor or someone else won't fix whatever is broken in the voice and manual daily build script then we won't have manuals from tomorrow on or so I guess :\
13:32:01bluebroth3rwouldn't consider that too bad. It's noticably worse than the build logs of course
13:32:50bluebroth3rwe won't have manuals anymore? Aren't the old ones kept in case updating fails?
13:33:03pixelmaif there is an error in it besides the warnings in non-stop mode how will you find it
13:33:51pixelmabluebroth3r: not that I've seen the last few times manuals were broken long enough (longer than 3-4 days recently)
13:36:09bluebroth3rpixelma: ouch, that's bad. If things fail I can build a round of current manuals and upload them to my webspace though.
13:37:34bluebroth3rpixelma: grep the log file for lines starting with !
13:38:39bluebroth3rhmm, might even be sufficient to filter the LaTeX output
13:39:15bluebroth3rfor normal filtering that is
13:39:17pixelmamaybe that could be done on the build client side before uploading the log?
13:40:52bluebroth3rI'm thinking about filtering the stdout messages for that. Would make the output much more useful.
13:41:12bluebroth3rwill hide Overful / Underful messages though, but I don't think anyone cares about them anyway.
13:43:50 Join webguest59 [0] (
13:44:34pixelmathe only thing I "use" them for is to see if compiling still goes on ;)
13:44:35 Quit webguest59 (Client Quit)
13:46:30 Join TheSeven [0] (~TheSeven@rockbox/developer/TheSeven)
13:50:10AlexPI'd at the least like a table that shows whether each manual built during the last round
13:50:39AlexPamiconn: It just isn't feasible to always catch all errors
13:51:20AlexPSpeaking for myself I always build lots of different ones, but they can slip through and at the moment there is no easy way to see if they are broken or not
13:51:49*bluebroth3r should dig out his old buildall script for the manuals
13:52:27bluebroth3rpixelma, AlexP: you're aware that the WPS tags in the manual haven't been updated to the new syntax yet?
13:52:40AlexPbluebroth3r: I'm about to commit that
13:52:57bluebroth3rAlexP: the tags? Nice.
13:53:01AlexPCurrently test building
13:53:46AlexPIs there any way in the tagmap tables to break the first column but still remain in a cell? Some of the tags are so long that they leave about 4 letters for the explanation
13:54:13bluebroth3rAlexP: IIRC \newline did that. Not completely sure though
13:54:22AlexPthanks, I'll try
13:54:40pixelma\newline works great in button tables for that
13:54:52pixelmaCustomWPS hasn't been updated either and I thought JdGordon would at least do that
13:55:10AlexPI've also added multifont
13:55:51TheSevenpamaury: I had some thoughts about the drive cache tonight
13:56:15TheSevenhow should we deal with write accesses?
13:56:41TheSevenshould they lock out other accesses to that sector until the sector gets committed? should they read the old state? should they read the new state?
13:56:52TheSevendo we need an "exclusive read" mode?
13:57:29AlexPhmmm, doesn't seem to work
13:58:19TheSevenwhat happens with pre-existing read handles to a sector if it gets opened for writing? allocate a new buffer for that?
13:58:25TheSevenor block until the read handles are closed?
13:58:57bluebroth3rwtf? Why does the grep filtering don't work on the first run in a clean folder?
13:59:51TheSevenor even panic in that case?
14:00:18TheSeven(can it ever happen that a sector is being written while it's opened for some file/dir?
14:02:43 Join playing-the-fool [0] (
14:06:55***Saving seen data "./dancer.seen"
14:06:56S_a_i_n_tI was wondering what happens with the manual and the wiki regarding wps syntax.
14:07:10AlexPNo \newline doesn't work, it just eats the space
14:07:11S_a_i_n_twhat happens if someone has the release build?
14:07:21S_a_i_n_t(and wants to check wps syntax)
14:07:22AlexPThey use the release manual
14:07:29pixelmaAlexP: what?
14:07:42 Quit playing-the-fool (Quit: Ухожу я от вас)
14:07:43AlexPpixelma: r.e. \newline?
14:07:48S_a_i_n_taha...right, that's a bit of a facepalm on my part ;)
14:07:55*S_a_i_n_t creeps back into his corner
14:08:18AlexPpixelma: I have this: \config{\%V(x,y,[width],[height],[font])}\newline \config{\%Vf([fgcolour]) \%Vb([bgcolour])} & See section \ref{ref:Viewports}\\}
14:08:54*bluebroth3r figured it
14:09:02AlexPpixelma: I want a line break where \newline is, but all that happens is that I get "font])%Vf with no space
14:09:17pixelmawhat happens with just the space?
14:09:24AlexPThe space is there
14:09:40AlexPi.e. ...font]) %Vf...
14:10:40pixelmaI mean: doesn't LaTeX break the line on a simple space anyway if necessary (without the \newline)
14:11:13pixelmamaybe there is a difference between the table environment for button tables and the one you are dealing with currently
14:11:37AlexPno, it doesn't break
14:12:11AlexPmust be, I'll have a look what
14:12:54 Quit stoffel (Ping timeout: 265 seconds)
14:13:48TheSevenis there any documentation on the source tree?
14:14:15TheSevenwhat does go to common, what to drivers, and what's directly in firmware?
14:14:30TheSevenare there any guidelines for this?
14:14:38 Join cjcopi [0] (
14:15:25 Part Jerom
14:15:43bluebroth3rpixelma: ah, found something: the LaTeX run can stop in interactive mode when doing the first run in an empty folder. That's because a different rule triggers first and is lacking $(LATEXOPTS) :(
14:16:20JdGordonpixelma: that would be fair if I had touched it at all...
14:17:03JdGordonTheSeven: what are you asking about in particular?
14:17:16TheSevenI'm about to add a disk cache layer
14:17:26pixelmaJdGordon: you did not change the WPS syntax - or what are you referring to?
14:17:33TheSevenor rather sector buffer pool
14:17:34JdGordonAFAIK there isnt any actual guidelines... whereever it makes sense
14:17:46TheSevenand what *does* make sense?
14:17:46pixelmabluebroth3r: sorry, I can't follow
14:17:55JdGordonpixelma: I meant because I hadnt fixed the doc at all that wasnnt fair
14:18:14JdGordonit is a wiki after all...
14:18:23TheSevenstorage.c is in firmware, fat.c is in drivers, file.c, dir_uncached.c and dircache.c are in common
14:18:29bluebroth3rpixelma: the Makefile for the manual has a bug. It requries running LaTeX several times (which is needed because of LaTeX way of breaking text).
14:18:48TheSeveni don't really see any scheme behind this
14:18:57JdGordonthere probably isnt any
14:19:07bluebroth3rhowever, the first run of LaTeX triggers a different Makefile rule. These also calls pdflatex, but fails to pass the $(LATEXOPTS) value thus not enabling nonstopmode.
14:19:13TheSevenso that's just a "put it wherever you like"?
14:19:24TheSevenwhat are those folders for then?
14:19:37bluebroth3rso for this first run (which only gets triggered if the .aux file doesn't exist) building *can* end up in interactive mode
14:19:41TheSevenwhy don't we just throw everything into a single one?
14:20:05*bluebroth3r wonders if there's a way to make LaTeX not break its output at 80 characters
14:20:07JdGordonit probably made sense 5 years ago
14:20:22pixelmaJdGordon: I didn't follow everything that had changed (which thing is which now) and since you invented it you have the best knowledge. Since you are not touching the manual at all and kept CustomWPS up-to-date the last times you changed things...
14:21:42pamauryTheSeven: sorry, I just saw you talked to me :)
14:22:23pamauryI would say that there are two different ways to do it:
14:22:48pamaury1) Sector cache: a layer between storage api and actual storage api, you need to solve all these issues
14:23:27pamaury2) Sector pool: just a specialized kind of malloc, we just need to use it instead of allocating on stack, no change in the storage layer
14:23:59CIA-8New commit by theseven (r26982): Add a STORAGE_ALIGN_ATTR macro, similar to CACHEALIGN_ATTR, for consistency
14:25:06TheSevenpamaury: I'd like to do something in between
14:25:41CIA-8r26982 build result: All green
14:25:54pamauryTheSeven: how is this possible ? :)
14:27:42TheSevenkeeping it as simple as possible while at least re-using buffers for multiple read accesses to the same sector
14:28:10TheSeveni'm wondering if it should be merged with the fat cache or not
14:28:27pamauryI would personally go for the first one because I'm not sure a sector "cache" is useful. We don't do much disk accesses so it's will basically be a cache miss on every read/write. The FAT is the only exception imo
14:28:56pamauryI might be proved wrong of course.
14:29:22CIA-8New commit by theseven (r26983): Use STORAGE_ALIGN_ATTR for instead of hardwired alignment in the Nano2G FTL and NAND driver
14:30:16TheSevensearching directories might have a few hits
14:30:52pamauryhm, that's right, but I *think* the fat_dir does caching by itself for directory scanning, let me check
14:31:01CIA-8r26983 build result: 88 errors, 0 warnings (theseven committed)
14:31:06pamaury(anf there is dircache of course)
14:31:09TheSevenit only has a single buffer for that
14:31:43pamauryyes, one buffer per dir, that's sufficient, it's sequential accesses
14:31:46TheSeveni just think it can't hurt to keep recently used data around until to be able to re-use it until we need the space
14:32:40*S_a_i_n_t often wondered about the source subdirs, for someone like myself its a real PITA to find anything.
14:33:00pamauryNo, it's just that a cache is more complicated than a simple pool
14:33:01S_a_i_n_tthere doesn't seem to be any rules defining where things should be placed.
14:33:19S_a_i_n_tand, IMO, there should be.
14:33:33pamauryS_a_i_n_t: basically, on the least common ancestor of uses :)
14:33:40pamaurywith exceptions :)
14:34:10CIA-8New commit by theseven (r26984): Also define STORAGE_ALIGN_ATTR if STORAGE_WANTS_ALIGN isn't defined
14:34:15S_a_i_n_tits "the exceptions" that confuse me :P
14:34:25pamauryTheSeven: so you would go for a api that allows you to read a sector and give you a handle to a buffer, that's right ?
14:34:54S_a_i_n_treorganising the code so everything was in its propper subdir would be a real PITA, but worthwhile.
14:35:03S_a_i_n_tthen creating rules, and sticking to them.
14:35:19pamauryS_a_i_n_t: what are the exception ? I can see the drivers/ one which has strange subdirs in it but that's all
14:35:31CIA-8New commit by theseven (r26985): Fix misspelled STORAGE_WANTS_ALIGN in s5l8700/system-target.h. Also add PROC_NEEDS_CACHEALIGN.
14:35:44CIA-8r26984 build result: All green
14:36:01pamauryOk, I need to leave for a few minutes, continue to give your thoughts on the sector cache please
14:36:05TheSevenwhat's the exact meaning of the "common" dir?
14:36:10S_a_i_n_tthings like rbutil for instance (a lesser example)..why is this split out? why not have it in tools?
14:36:11TheSeven"common" between what exactly?
14:36:14S_a_i_n_tis it not a tool?
14:36:39TheSevenS_a_i_n_t: tools are build scripts and similar stuff
14:36:49TheSevenif anything, it should be in util/
14:37:08CIA-8r26985 build result: All green
14:37:12 Quit Rob2223 (Quit: Rob2223)
14:37:31TheSevenpamaury: yes, that's basically what I'm planning to do
14:37:37JdGordonit's all just one big bloody mess!
14:38:24 Join stoffel [0] (
14:38:26S_a_i_n_tJdGordon: This is true.
14:38:28TheSeventhe question is just if the fat cache should be merged with that, so that it can use the currently-free sector buffers as additional cache. this means that we have to handle writes somehow of course, but that could be done in a similar way as the fat cache is doing it currently
14:38:35S_a_i_n_tit may as well be one directory :/
14:39:03S_a_i_n_tAt least then I'd know exactly where to look for things :P
14:39:25CIA-8New commit by bluebrother (r26986): Apply LATEXOPTS to all LaTeX runs. ...
14:40:56CIA-8r26986 build result: All green
14:41:06pamauryS_a_i_n_t: the common/ dir is a bit strange, I admit that. It's more a user interface to the firmware but yeah...
14:41:26pamauryTheSeven: if you do it this way, get rid of the fat cache, that's basically the same behaviour no ?
14:41:55TheSevenyes, with some small additions
14:42:40TheSevenyou need the ability to lock buffers etc.
14:43:26pamauryand the ability to read without using a cache buffer also, when you already have a buffer
14:43:43pamauryand you know there is little point in caching
14:45:16 Join Rob2222 [0] (
14:45:21TheSevenIIUC there will always be only a single user for everything that may be written (either the FAT, a dir or a file)
14:45:24bluebroth3rpixelma: can you give that change a try on cygwin?
14:45:47pamauryS_a_i_n_t: anyway, the subdirs are mainly here to avoid having a big chunk of .c files, it's suboptimal but better than a plain dir imo
14:46:06 Join fdinel [0] (
14:46:24bertrikTheSeven, can we put an assertion on that assumption (e.g. panic if it's not the case)?
14:46:45TheSevenbertrik: that would make sense
14:46:49pamauryTheSeven: that needs to be checked, but except if you open the same file twice of writing (which is undefined), I think yes
14:47:06TheSevenwe could also try to just sequentialize those accesses through locking
14:47:18pixelmabluebroth3r: just test if a manual still builds?
14:47:47bluebroth3rpixelma: yes. Should give you much less output on stdout (the log file won't be affected in case you need more information)
14:48:06bluebroth3rit does some truncation (or to be more precise, it doesn't do reflowing) but that shouldn't hurt.
14:48:24TheSevenpamaury: how are files read/written?
14:48:57TheSevenis that always done through that file's sector buffer?
14:48:58pamauryI think file.c has a buffer per file
14:49:08pamauryAnd that's all
14:49:15pamauryfor reading at least
14:49:30TheSevenwill that be bypassed for big accesses that need to read a whole sector?
14:50:09TheSevenhm, even if it is, the current code will already have to deal with the r/w conflict anyway
14:50:58TheSevenso it looks like we basically don't need to care about write accesses, they'll just have to be done in a safe manner, which should already be the case
14:51:33pixelmabluebroth3r: patch fails (on an updated SVN r26986)
14:51:50pamauryLooking at the code, it seens read>one sector trash the cache and just read
14:52:20TheSevenare file writes cached?
14:52:34TheSeven(if you only write a few bytes at a time until the sector is full)
14:52:37AlexPbluebroth3r, pixelma: Any objection to me changing the tagmap environment to "\tabularx{\textwidth}{>{\raggedright\arraybackslash}p{.42\textwidth}X}\toprule\rowcolor{tblhdrbgcolor}"
14:52:48TheSevenor is the whole sector rewritten for each write?
14:52:56AlexPIt means that the second column doesn't end up with space for only a couple of characters
14:53:08pamauryI'm not sure but I think it is cached until trashed, so it can write a whole sector at once
14:54:00pixelmaAlexP: can't think of any ;)
14:54:00bluebroth3rpixelma: urgh? Have you used -p1?
14:54:16*JdGordon is stumped
14:54:33*bluebroth3r is puzzled
14:54:51bluebroth3rI've made that patch from an r26986 checkout :o
14:55:01JdGordonhow the heck do I incoportate the new skin parser (which dumps a completly different data structure) without having to recode everything at once?
14:55:18JdGordonthis patch is turning mammoth and I havnt actually got anything workling yet
14:56:39pamauryTheSeven: if I read the code correctly, if the write is < one sector, it's just a memcpy if it fits into the sector cache, otherwise, it's write through for all bytes except the tail which is cached
14:57:01TheSevenok, so that's a clever single-sector cache
14:57:17pamaurythat would be stupid otherwise
14:57:50pixelmabluebroth3r: .rej looks like I could apply the two failed hunkjs (#2 and #3) by hand
14:58:01pixelmahunks too
14:58:11CIA-8New commit by alex (r26987): Add multifont to the manual. Closes FS #11294.
14:58:26CIA-8New commit by alex (r26988): Adjust the tagmap table environment so that the second column doesn't end up tiny. It can be adjusted further if necessary.
14:58:43CIA-8New commit by alex (r26989): Convert the manual to the new theme syntax. As a result, adds the viewport colour tags which closes FS #11372. Correct multifont tag from %F to %Fl
14:58:48bluebroth3rpixelma: just tried it on mac, patching fails here too. I really don't understand why :(
14:58:49pamauryTheSeven: it would be a good thing to have an api before starting coding
14:59:40pixelmabluebroth3r: my guess is because patch is stupid about $ or * or a sequence with these
14:59:47CIA-8r26987 build result: All green
15:00:49pixelmaan uneducated guess though
15:01:25CIA-8r26989 build result: All green
15:01:29bluebroth3rprobably. But if that's the case wouldn't others have hit that issue before and (hopefully) it gotten fixed years ago?
15:01:30 Quit antil33t (Read error: Connection reset by peer)
15:01:36 Join antil33t [0] (
15:01:44bluebroth3rfortunately it's not a big deal for this patch.
15:01:57AlexPWe appear to be building on all manual changes already
15:02:09AlexPGiven those last three commits of mine were all manual folder only
15:02:17bluebroth3rAlexP: afaik the build system doesn't filter on directories at all
15:02:33CIA-8New commit by Buschel (r26990): Fix FS #11416. Resuming mpc was broken since r26032 due to false assumption of amount of buffered data.
15:02:34AlexPIt used to I think
15:02:36pixelmaAlexP: but we don't build manuals
15:02:54AlexPBut it might not do since the change
15:03:02AlexPpixelma: Which makes it a bit pointless :)
15:03:05bluebroth3ryes, the old one did filter.
15:03:46pixelmait used to but that's already a while ago, IIRC it changed with the new build system which was invented during DevCon 2009 IIRC
15:03:52pixelmaso about a year ago
15:03:53TheSevenpamaury: do you know by chance how many open dirs/files are supported and what the fat cache size is?
15:04:10CIA-8r26990 build result: All green
15:04:17pamauryIt's written in the headers, it depends on mem size
15:04:52bertrikWhat happens if you call udelay(0) ?
15:05:28pamaury11 open files (file.h)
15:05:55pamaury8/12 open dirs (dir_uncached.c)
15:06:44pamauryfat cache: 0x20 sectors (fat.c)
15:07:23pixelmabluebroth3r: it's a bit weird to see nothing. I can't tell if your change works or not :\
15:07:25TheSevenok, so the sector cache should probably be something like 64 sectors in size
15:07:39TheSeventhat's 32/128K, sounds reasonable
15:07:54TheSevensearching 64 entries when looking for a sector also shouldn't hurt too much
15:08:57pixelmabluebroth3r: I tried a "make clean && make manual" in a build directory that worked before and it was not cleared yet
15:08:59bluebroth3rpixelma: you should see something. Most output for me is about overfull and underfull boxes
15:09:17pixelmaso I guess it doesn't work
15:09:52pixelmabut I didn't get an error message, haven't tried with your changes in SVN right before though
15:10:25bluebroth3rstrange. I'd expect it to behave identical than on my linux box
15:11:30pixelmahmm... pressing Ctrl+C to abort the process makes it go on with cleaning and building
15:11:38pixelmasomething has hung before?
15:11:45pamauryTheSeven:, does that sum up the wanted operation for the cache ?
15:12:28bluebroth3rno idea.
15:12:39 Join grawity [0] (
15:12:43*grawity looks around.
15:12:51grawityWhere do I post feature requests? Flyspray?
15:13:51*TheSeven just writing the very same on pastie :-P
15:13:51pixelmagrawity: no, you can only discuss feature ideas in the forums
15:13:55bluebroth3rpixelma: output looks like this for me (only parts of it):
15:14:10*grawity tries to find an apropriate forum for requesting a graphical volume bar.
15:14:21grawityI'd have a really nice WPS with that.
15:14:31pixelmayou can already have it
15:14:35 Quit Buschel (Ping timeout: 276 seconds)
15:14:51TheSevenpamaury: i would split the writing thing into "flag a buffer as dirty" and "clean a buffer"
15:15:07pixelmagrawity: AFAIK, haven't tried myself but S_a_i_n_t has I believe ;)
15:15:30grawitypixelma: I need something that works in the same way as %pb progress bar - is it already available?
15:15:36TheSeventhe fat code will probably want to release buffer handles without cleaning them
15:15:50pamaurywhat is cleaning for you ? write back ?
15:15:50pixelmagrawity: yes, I think so
15:15:55TheSevenpamaury: yes
15:16:16pamauryYeah, that make sense, leave the write back handling to the cache
15:16:38bluebroth3rpixelma: I'd really like to commit that change. But if it's not working on cygwin that's a blocker :(
15:16:39TheSevenyes, while keeping the ability to forcible write something
15:16:47pixelmabluebroth3r: yes, works now but the hang is really weird
15:16:53pamauryOk, agree.
15:17:02bluebroth3rpixelma: can you reproduce that hang?
15:17:04grawitypixelma: What tag is it, then? (I hope it's not %xl/%xd, that would need lotsa RAM.)
15:17:26CIA-8New commit by jdgordon (r26991): fix ffwd/rewind on touchscreen targets using regions
15:17:47pamauryTheSeven: can you edit the pastebin ? I need to leave now, but I'd appreciate if you could write the prototypes in it, so I can begin working or other people can give an opinion
15:18:09JdGordongrawity: there are two ways you can do a graphcal volume display
15:18:25pixelmaI *believe* you have to assign the %pv the parameters you would assign to %pb but as I said I haven't used it myself yet
15:18:31JdGordoneither like a progressbar %pv(... same tags as %pb ... )
15:18:45grawityThanks, I'll try
15:18:48JdGordonor using the %?pv<> conditional which is complicated
15:18:54CIA-8r26991 build result: All green
15:19:03 Quit petur (Quit: *plop*)
15:19:17TheSevenS_a_i_n_t seems to have used "%pv(0,0,122,11,Volumebar.bmp)"
15:19:28JdGordonthats the simple way
15:19:39pixelmabluebroth3r: I'll try, also without the patch, just the things you already checked in
15:21:09bluebroth3rpixelma: ok, great
15:21:24*bluebroth3r goes building a complete manual round
15:21:42CIA-8New commit by jdgordon (r26992): woops, this chaneg too
15:21:48S_a_i_n_tJdGordon: The "complicated" way crashes and burns :P
15:21:54 Quit bmbl (Quit: Bye!)
15:22:18S_a_i_n_t(if you try to use more than one bar)
15:22:26grawityBtw, when is %Vb actually used? I did not notice it ever having any effect...
15:22:56JdGordonit will only work if you dont have a background image
15:23:22CIA-8r26992 build result: All green
15:24:03S_a_i_n_t%?pv<Mute|%pv(0,0,-,-,Volumebar.bmp)|Line Level|%pv(0,0,-,-,Volumebar.bmp)> failed drastically last time I tried it.
15:24:10S_a_i_n_thence the "simple" way.
15:24:41grawityI thought the simple way was just a %pv() without being in a conditional.
15:24:48 Quit t0rc (Remote host closed the connection)
15:24:54S_a_i_n_tyes, it is.
15:24:59S_a_i_n_t(read above)
15:27:17pixelmaS_a_i_n_t: what happens without the bar for +0dB?
15:27:54S_a_i_n_tpixelma: It never draws, and its presence messes up drawing the -0dB bar :/
15:28:09 Join thefirstM [0] (
15:28:22AlexPgrawity: This is partly why we don't have feature requests - 1) It is pointless, devs work on what they want and 2) People request things that already exist and are in the manual
15:28:28pixelmakeeping the conditional "branch"?
15:28:38grawityAlexP: Sorry, didn't notice it in the manual :\
15:28:45grawity(And I've read it a few times.)
15:28:46S_a_i_n_tpixelma: ...?
15:29:06JdGordonrasher: ping?
15:29:24S_a_i_n_tthe line as I posted earlier, which I would expect to work, fails in the mannor I described.
15:30:15 Quit Bagder (Ping timeout: 260 seconds)
15:30:32pixelmaS_a_i_n_t: %?pv<Mute|%pv(0,0,-,-,Volumebar.bmp)|Line Level|%pv dB> for example, just replacing the last "bar" definition, not removing the last part completely
15:30:58S_a_i_n_tpixelma: Yes, that works as expected.
15:30:59pixelmaI just read it "crashes and burns", so I'm not sure how it fails
15:31:35S_a_i_n_t"It never draws, and its presence messes up drawing the -0dB bar :/"
15:31:36pixelmanow I'm confused but will leave it this way
15:32:38S_a_i_n_tit just won't draw the +0db Bar, and if you try to, it draws the -0dB bar incorrectly.
15:33:44grawityAww :( So I can't have a different background behind the volume bar...
15:33:58S_a_i_n_tof course you can
15:34:08S_a_i_n_twe're talking about something entirely different
15:34:55grawityI don't know what you're talking. I just tried %xd(V)%pv(0,0,162,16,vb.bmp)
15:35:02S_a_i_n_tgrawity: The easiest way to do that is to include the backgroung for the volume bar in the backdrop.
15:35:13S_a_i_n_tthen you don;t waste any buffer on a seperate image
15:35:33grawityWhat I was trying to do is an ipod-like volume bar
15:35:38S_a_i_n_t%xd(V)%pv(0,0,162,16,vb.bmp) <−−will not work at all :/
15:35:46grawity(Appears when changing volume, disappears in a second)
15:35:46pixelmabluebroth3r: seems I can't reproduce anymore, weird
15:36:06S_a_i_n_tgrawity: Try looking at the tags first.
15:36:16S_a_i_n_tyou want "%mv" for that.
15:36:19grawityI do have that.
15:36:29bluebroth3rpixelma: does the SA9200 manual build?
15:36:40grawityBut when I change the volume... the %xd(V) background blinks for a second, then volume bar appears and the image is gone.
15:37:02S_a_i_n_tgrawity: pastebin the code.
15:37:03bluebroth3rthe platform file uses a file name differing to the target name. I guess that will break
15:37:10S_a_i_n_tI suspect you're doing it wrongly.
15:37:15pixelmabluebroth3r: don't think so and I believe I asked if someone knows if it ever did during DevCon
15:37:24S_a_i_n_tI have an Apple_OF theme working very well.
15:37:38grawityI'll upload the entire thing in a moment
15:37:40S_a_i_n_tTheSeven will testify to that ;)
15:37:40pixelmabluebroth3r: no-one answered... ;)
15:37:50TheSevenS_a_i_n_t: no, the bars don't come flying in from the side :-P
15:37:58S_a_i_n_tgrawity: No, please done. Jst pastebin the WPS
15:38:00bluebroth3rok, the sa9200 also misses the keymap file.
15:38:10bluebroth3rso it's definitely broken.
15:39:50pixelmabluebroth3r: how does a broken manual look with this output, or do I have to try myself?
15:39:59S_a_i_n_tViewport (p) is just: "%al%pc%ac%pp of %pe%ar%pr"
15:40:10grawityS_a_i_n_t: There are two viewports with the same identifier.
15:40:25S_a_i_n_tdo they overlap?
15:40:44grawityS_a_i_n_t: It's allowed by the parser. But it's not the problem. It occurs even with the second p and v viewports removed.
15:41:00S_a_i_n_tyes, I they overlap?
15:41:13S_a_i_n_tif so, you should just call them in the same viewport
15:41:16bluebroth3rpixelma: sa9200 manual after renaming the platform file with my changes:
15:41:22grawityHmm. Maybe I could
15:41:31S_a_i_n_tyes, you can.
15:41:39grawityBut anyway, the problem is only with the first one −− below "Conditional Progress/Volume bar".
15:41:41S_a_i_n_tI suspect it is trying to draw both at once.
15:42:27grawityS_a_i_n_t: Fine:
15:42:37grawityThat's what I had at the moment of complaining :/
15:42:44bluebroth3rpixelma: for a manual that usually builds you get something like this (excerpt):
15:43:05S_a_i_n_tPlease don't get pissy with those trying to help:/
15:43:09 Join Bagder [0] (~daniel@rockbox/developer/bagder)
15:43:11S_a_i_n_tThey give up quickly.
15:43:23bluebroth3rso it's much easier to spot the error as a lot of uninteresting stuff gets filtered away.
15:43:48grawityS_a_i_n_t: I'm not... I just think you misunderstood me when I said that the "playlist position" viewports are not the problem I have.
15:43:52bluebroth3rthe solution is not the nicest but it's a quick one :)
15:43:54grawityAnyway, sorry.
15:47:26S_a_i_n_tthe solution is to either draw two viewports in the same place, and display them conditionally (hard-ish), or just encorporate the background for the volume bar in the backdrop.
15:47:52AlexPbut doesn't work for having an appearing bar
15:48:00AlexPwell, not nicely
15:48:08S_a_i_n_tI have a working solution, but mine is different as it shows the progressbar, and the volume only if being changed.
15:48:09grawityWell... the backdrop already has a background at that location.
15:48:21S_a_i_n_tthen, do the first.
15:48:25AlexPS_a_i_n_t: Yes, that'd do it
15:48:40grawityIsn't the first what I'm doing? :(
15:48:43S_a_i_n_ttwo identical viewports, one with the volume-bar, the other with the backdrop.
15:48:57S_a_i_n_tshow one when %mv is true, the other when it is false
15:49:26pixelmabluebroth3r: does a usual "make manual" not stop on errors now?
15:49:34pixelmawith the patch
15:49:44pixelmaor even without
15:50:24grawityS_a_i_n_t: The problem is, it is exactly what I am doing :(
15:50:29S_a_i_n_tgrawity: you're working one the Nano right?
15:50:36S_a_i_n_tdo you want me to save you some time?
15:50:37*grawity uploaded the thing to grawity/files/"> if you have rockboxsim/ipodnano2g
15:51:15*grawity even made sure it isn't running out of memory or something.
15:51:20 Quit flydutch (Quit: /* empty */)
15:51:31pixelmabluebroth3r: I even got the pdf now but the place where I placed an error is blank
15:52:05bluebroth3ryou don't get a "! <something>" for that line?
15:52:06pixelma(used a \ButtonSelect in an H10 manual which has no \ButtonSelect)
15:52:35pixelmaI prefer if it stops when compiling for myself, saves me from digging through the log :\
15:53:39pixelmaI can't find it
15:53:54bluebroth3rtoo bad.
15:54:34bluebroth3rI guess I have to give that issue a more detailed look then :'-(
15:56:23pixelmaI don't want non-stop mode for myself... and there is no ! in the complete log (except in the grep rules at the top), pasted in a text editor and let it search
15:57:06pixelmawell, you escape ^Unde
15:57:42pixelmawhich will suppress the "Undefined control sequence" too which probably occurs in my case, IIUC
15:57:47bluebroth3rafaik every error report starts with a ! as the first character.
15:58:14bluebroth3rso that should definitely show up.
15:58:43pixelmahaven't seen error reports in non-stop mode
15:58:45 Quit hebz0rl (Ping timeout: 265 seconds)
15:59:00pixelmaI can paste you the log if you want
15:59:15bluebroth3rwould be interesting. I'll try here as well.
16:01:20 Join bmbl [0] (~Miranda@unaffiliated/bmbl)
16:01:26pixelmain any case, I don't want non-stop mode
16:02:18bluebroth3ryou can always empty LATEXOPTS again. We have nonstopmode since quite a while now.
16:02:37bluebroth3rthough it might be a good idea to use some V=1 method to override that
16:03:05pixelmaI never used it and was never forced to
16:03:26bluebroth3rit's been in the Makefile :)
16:04:03bluebroth3rthe only problem is (was) that due to some bug nonstopmode wasn't activated for the first LaTeX run if the aux file was missing
16:05:38thefirstMHow do you build GCC 4.5 for the arm-eabi toolchain? There were some conversation about that here yesterday. I tried, but I always get the error "checking for library containing strerror... configure: error: Link tests are not allowed after GCC_NO_EXECUTABLES."
16:06:59***Saving seen data "./dancer.seen"
16:07:46bluebroth3rusing interactive mode in conjunction with filtering would be bad. And IMO building shouldn't fall into any interactive mode.
16:08:18pixelmabluebroth3r: I really prefer a stop when working for myself... seems that I also understand now why I even got a manual or at least a running build and why I started always cleaning the build dir before because it confused me
16:09:44 Quit rasher (Read error: Connection reset by peer)
16:10:09pixelma"it" - that the build ran and I got a pdf (only in the "manual" subdir, but there were case where the last copy step failed for me before which is why I started to not care about that)
16:11:01bluebroth3rpixelma: how about something like then? Just set STOP=1 when running make and you're back to the old behaviour
16:11:10pixelmabluebroth3r: (log)
16:11:46 Join hebz0rl [0] (
16:12:49pixelmabluebroth3r: guess I have to revert the first patch?
16:15:58bluebroth3rto get back the old (broken) behavious: yes.
16:16:33bluebroth3roh, giess I misunderstood you :o
16:16:34pixelmaI mean before applying the patch you just posted
16:16:49bluebroth3ryes, it includes the older patch.
16:17:06*JdGordon needs a bit of maths help...
16:17:07bluebroth3r(sorry, didn't pay attention to that when creating the diff)
16:17:13bluebroth3rJdGordon: 1 + 1 = 2 :)
16:17:32pixelmabtw. does your already checked in change mean we'll always have a manual even if they are "broken" (and except the uploading or daily scripts are broken)?
16:17:46JdGordonI need to setup a system for the touchscreen buttons where a "position" inside the region is returned (for progressbar/volume bar)... whats the best way to do this so it makes sense?
16:17:57JdGordonusing the width/height of the region
16:18:02 Quit bieber (Ping timeout: 252 seconds)
16:18:11JdGordondo I just calculate the percentage position of the touch?
16:18:20 Join bieber [0] (
16:18:26JdGordonthen the user converts that to a usable number
16:18:34bluebroth3rpixelma: no, if building fails no output file will get generated. However, building won't hang (which I guess could be a problem with nightly builds)
16:19:52bluebroth3rJdGordon: depends on what exactly you want to tell. Maybe how exact the user hit the middle of the area? Then make center = 100%, and the edges 0%
16:20:15bertrikAnyone with a c200v2, e200v2, fuze willing to test a patch (updated i2c driver for radio) ?
16:20:37bluebroth3ror return an x and y value calculated from one of the edges.
16:20:46JdGordonbluebroth3r: right now it is hard coded for use with the progressbar and volume bar... it doesnt need to be precise
16:20:54pixelmabluebroth3r: but I got an output file now (with the first of the other patches though), not only in the manual subdir you can see that the copying didn't fail in the log
16:21:16JdGordonan egde offset would certainly be simpler :)
16:21:19JdGordonI'll do that
16:22:17bluebroth3rpixelma: hmm. Right, that would then be an incomplete file (i.e. only the manual part up to the error)
16:22:30*bluebroth3r should get more coffee
16:23:03pixelmajust the cell in the button table with the wrong button was empty
16:23:45bluebroth3rok, that surprises me. Wouldn't be too bad of a behaviour though, at least for nightly builds
16:24:31pixelmanot sure, then you'll probably get a bunch of errors before someone notices :Y
16:24:37bluebroth3rbtw, in case the manuals disappear I've uploaded a set of current ones to
16:24:39 Join rasher [0] (~rasher@rockbox/developer/rasher)
16:24:43pixelmaor :|
16:25:34pixelmaand less incentive to get the manuals into the build system ;)
16:26:30JdGordonbluebroth3r: bah, no a single offset isnt enough, I need the full width/height value also somehow
16:26:45bluebroth3rwell, catching the errors should work though. We then only need some kind of build table
16:27:11grawityOne more question. What does the "0%|" thing mean in this line: %X(backdrop.bmp)0%|0%|
16:27:22grawityI find it in some WPSes...
16:27:54pixelmalooks like a wrong conversion because someone used it wrong in the old syntax
16:28:24 Join halmi_ [0] (
16:29:13pixelmathe 0%|0%| shouldn't be there
16:30:05*grawity removes
16:30:29 Quit halmi (Ping timeout: 240 seconds)
16:30:35pixelmaa backdrop doesn't need setting the coordinates, some people used %X|backdrop.bmp|0|0| before for some reason. The automatic conversion "thought" | was meant to be displayed "literally" there which needs escaping with %
16:36:12 Join funman [0] (~fun@rockbox/developer/funman)
16:37:13S_a_i_n_tpixelma: Funny thing is, it passes checkwps like that.
16:37:24S_a_i_n_tits the ElectricBarsOfColour thems.
16:37:38grawity(And Cabbie v3.)
16:38:17pixelmait's not wrong I guess, not sure you could find a 0|0| on screen somewhere
16:38:31S_a_i_n_tpixelma: it is wrong.
16:38:36S_a_i_n_tit fails on device now.
16:38:40JdGordonit wont be displayed because the current(old) parser eats the whole line
16:38:48S_a_i_n_tit didn't *used* to be wrong though.
16:39:10S_a_i_n_teverything past the | after the bitmap image gets ignored.
16:39:13S_a_i_n_tor used to.
16:39:55funmanTheSeven: isn't aligning stack buffers simpler?
16:40:11S_a_i_n_tBut now it manages to pass checkwps, yet fail on this works I have no idea.
16:40:22grawityS_a_i_n_t: It works on the device too.
16:40:34TheSevenfunman: this means we have to reserve like 4K for sector buffers on several threads' stacks
16:40:54S_a_i_n_tgrawity: Perhaps a recent thing.
16:41:11S_a_i_n_tIt failed on device as little as a week ago (I tested it on two targets)
16:41:36S_a_i_n_ttopik brought it to my attention.
16:41:49funmanTheSeven: ah this explains the stkov on nano2g
16:41:59grawityS_a_i_n_t: Huh. I used Cabbiev3 for several months, with that line.
16:42:19funmanbertrik: udelay(0) should sleep for ~17 seconds
16:42:28S_a_i_n_tgrawity: with the *old* syntax?
16:42:37bertrikah :)
16:42:38S_a_i_n_tmust syntax isn't that old.
16:42:45grawityS_a_i_n_t: Oh, now that you mentioned it... *facepalm*
16:43:13TheSevenfunman: yes, I want to finally get rid of the root cause for them
16:43:15S_a_i_n_tIt *should* fail the new syntax.
16:43:35S_a_i_n_tyet it passes checkwps, not a big deal, but misleading.
16:44:22pixelmabluebroth3r: I'm confused now, I thought I applied your newer patch but the output looks like the "old" one
16:44:39grawityMeh, I give up. Whatever I try to put behind a progress or volume bar doesn't get displayed, even though half of that bar is 'empty'. ...I guess I'll make it something like original appleware screen.
16:44:57funmaniiuc fat/sd accesses are already protected by a mutex
16:45:19funmani'm not sure if you can access 2 drives at the same time
16:48:10 Join panni_ [0] (
16:49:47pixelmabluebroth3r: ah, the "STOP=1" will give me the "old" output? Thought it would also cut the warnings but just not go on when an error occurs
16:53:46 Quit Highlander (Quit: Quitte)
16:55:08 Join Jerom [0] (~heidi@
16:56:57bluebroth3rpixelma: yes, that was the idea. Maybe naming it INTERACTIVE would be more clear. Changing the filtering could be done as well.
16:57:07 Quit thefirstM (Remote host closed the connection)
16:57:41bluebroth3rjust move the LATEXFILTER in LaTeX.Rules
17:03:39JdGordonanyone want to test fms touchscreen skin support patch?
17:08:39 Quit stoffel (Ping timeout: 252 seconds)
17:08:44pixelmabluebroth3r: did you see and following by the way?
17:10:41pixelmaand... applying the filter options to the STOP=1 case is weird: it stops but you don't see why
17:11:03 Join greg987 [0] (
17:11:31 Quit greg987 (Client Quit)
17:11:32pixelmafor some reason it still supresses errors in my case
17:16:34funmanbertrik: you have an idea on the fuzev2 problems with µSD ?
17:16:50funmanit took me several boots to see it happen on my fuzev2
17:18:33pixelmabluebroth3r: error messages do not start with a ! here. In my example running non-stop mode without the filter gives me "./plugins/pegbox.tex:102: Undefined control sequence. \n <argument> \ButtonSelect" 3 times in a row
17:18:40bertrikno, but didn't you mention something was a bit out of spec in the sd driver? maybe increasing the Fcpu pushed the timing over the margin
17:19:30funmanbertrik: i'm not sure if it's out of spec but it definitely works like it is, the same thing is used on AMSv1
17:19:42bertrikOR maybe the special case for the sdram / PCLK for the fuzev2 has something to do with it (although this was tested)
17:19:53funmanbit6 of CGU_PERI ?
17:20:20funmanperhaps adding some logf in fat code would help
17:20:58 Join greg987 [0] (
17:21:16JdGordonHow do I do the (c) notice for a new file which I'm copuing code out of a old one for?
17:21:50funmanbertrik: also I want to make sure the recent changes to SD are 100% working
17:22:43funmanalright, there is something fishy in text_viewer preferences
17:22:49JdGordonI think it is mostly origionally my code, but I dunno...
17:23:05bertrikfunman, OK. I have no problem with you reverting the 248 MHz change, by the way :)
17:23:07funmanmaking tv_set_preferences() return immediately stop crashes (i can't see anything on screen however)
17:23:14funmanok nice
17:23:34funmanhopefully we will get an eureka soon on this :)
17:26:10 Join GeekShadow [0] (~Antoine@reactos/tester/GeekShadow)
17:27:45bertrikfunman, I have the strong suspicion that we can't really make 240 MHz out of 384 MHz with the 5/8 pre-divider. I think the pre-divider just skips cycles, so the cycles that it doesn't skip will be at 384 MHz (i.e. too fast)
17:28:04funmaniirc it worked on AMSv1
17:28:54funmanwouldn't this be a problem if we set pre-divider to 5/8 and postdiv to 1/2?
17:29:15funmanthe cycles would be below 240MHz
17:30:56greg987Hey, I'm running my ipod without battery (since it didn't hold any charge and also seemed to be bloated somehow). Now I encouter continuous resets/crashes at pretty precisely the same time after boot-up (about 30s into the song). I wouldn't like to risk buying a new replacement battery that then might just bloat up again (I have it plugged in all the time, anyways). Anybody some ideas how to fix these crashes?
17:31:22bertrikyes that would work, but wouldn't Fcpu be rather slow then (120 MHz)?
17:31:41JdGordondo we have a wiki page for how the touchscreen is supported?
17:31:50pamauryTheSeven: what is the consistent flag ?
17:31:56funmanbertrik: yes it would be slow, i just meant that scheme as experiment, not something to be committed
17:32:30TheSevenpamaury: that indicates that the state of the sector buffer is currently consistent enough to be written back if needed
17:32:52TheSeventhis is only needed for that clean_all thing which we might as well just drop
17:33:33pamauryand sectorcache_wait_block ?
17:33:59TheSeventhat will block the current thread until some buffer was released or flagged consistent
17:34:16TheSeven(should actually be static)
17:34:28 Quit grawity (Quit: Leaving.)
17:34:54 Quit DerPapst1 (Read error: Connection timed out)
17:35:58TheSeveni'm still not sure how to handle hotplug though
17:36:02 Join DerPapst [0] (
17:36:44TheSevencan we assume that all locks to that drive's sector buffers will have been removed before the remounting happens?
17:36:51pamauryThat's the problem with caching, if the user remove the card, we are dead :)
17:37:27TheSeveni don't care about the user deliberately unplugging the card
17:37:37TheSevenbut we need to invalidate everything when he plugs a new card
17:37:43pamauryI would say no, you can't assume that
17:37:55pamauryDo you plan to use a timer to write back after x seconds ?
17:38:11TheSeveni'm not sure about that yet
17:38:18TheSevenhow does the fat cache currently handle this?
17:39:59pamauryit does nothing about it I think
17:41:13TheSevenso if the fat gets written to, that won't be written to disk until it runs out of cache space or the player is shut down?
17:41:21TheSeven(how does it actually intercept the shutdown?)
17:41:58TheSevenor is the fat cache write-thru?
17:42:56pamauryNo, the fat is written back when a file a created/closed(and modified)/removed/renamed
17:43:04pamaurySo it's reasonably safe
17:43:39TheSevenok, so cleaning is triggered actively
17:43:55TheSevenwe could do the same for the sector cache
17:44:21bluebroth3rpixelma: gnah, then this filtering obviously won't work :( What LaTeX distribution are you using?
17:44:41bluebroth3ras for the ligature issue: it doesn't appear anymore in recent manuals because of the switch to T1 fontenc.
17:45:25pixelmaprobably an old one... how to find out
17:45:44pixelmaaha, about the other answer
17:47:57pixelmabluebroth3r: pdfeTeX 3.141592-1.21a-2.2 (Web2C 7.5.4) <- do you mean that one?
17:48:50bluebroth3rwell, I was rather referring if you are using MikTeX, TexLive or something else. Cygwin packages I guess?
17:49:19pixelmathe latter, tetex and tetex-devel I believe it was
17:49:33bluebroth3rbut my system reports pdfTeX using libpoppler 3.141592-1.40.3-2.2 (Web2C 7.5.6) so it's definitely newer
17:50:01bluebroth3rtetex is outdated and no longer maintained. AFAIK all linux distributions now use TeXlive
17:50:12pixelmaI know
17:50:25bluebroth3ranyway, this leaves us with the problem that we can't do this filtering that way. Too bad :(
17:52:00bluebroth3rmight work by adding "|.+\:[0-9]+:" before the closing bracket for LATEXFILTER
17:52:12*pixelma looks around for someone running the newer cygwin 1.7
17:56:27 Quit greg987 ()
17:56:49 Join Antibuddha [0] (
17:56:56Antibuddhahow do i play videos with rockbox on ipod 5g?
17:57:08Antibuddhai need mpegplayer?
17:58:08funmanAntibuddha: check the manual, mpegplayer is explained
17:58:23Antibuddhaim setting up an ipod for my mom right now
17:58:33Antibuddhawould it be better for video quality to dualboot?
17:59:10funmansee for yourself
18:00:02TheSevenpamaury: if we run out of sector buffers, should we 1) panic, 2) return an error or 3) block until one gets freed?
18:00:42Antibuddhai dont have time funman
18:00:46pixelmaAntibuddha: on the Ipod Video it's still better to use the original firmware for videos I believe. Apple can use a special chip for it which Rockbox can't
18:00:55Antibuddhahave to leave home to take mom to airport in 30 min
18:01:00 Quit bucko (Ping timeout: 258 seconds)
18:01:01pamauryI would say block, or even better, block with an optional timeout after which is fails
18:01:49pamauryTheSeven: what is sectorcache_replace_idx ?
18:02:03TheSevena rotating index where to start to look for free buffers
18:02:33pamaurywhat is the difference between valid and free ?
18:03:28TheSeveninvalid = garbage, free = contains valid and up-to-date data for that sector, but isn't expected to be used again and should thus be re-used preferably before killing other buffers
18:04:31funmani found text_viewer problem: char[XXX] = (struct font *)
18:04:40pamauryYou should comment each flag to make it clear what do they them
18:04:43funmandunno why gcc didn't warn
18:05:35funmanargh no it's correct >< i can't read
18:07:00***Saving seen data "./dancer.seen"
18:11:39 Join t0rc [0] (~t0rc@unaffiliated/t0rc/x-5233201)
18:14:55 Join fml [0] (
18:16:50fmlAlexP: hello. I see some typos in the recent manual commits, but don't have svn access to fix them. One is in advanced_topics/main.tex ("declerations"). The other is in advanced_topics/viewports/colour-vp-syntax.tex: "and 2-9 are additional are skin loaded..."
18:17:13AlexPThanks, I'll fix now
18:17:37amiconnHmm, static variables in a function for no special reason?
18:17:51*amiconn is trying to solve the weird lines problem in PF
18:18:23fmlAlexP: other variants of the VP syntax have the same typos.
18:19:47AlexPfml: Incidentally, I copy/pasted these from the wiki :)
18:20:55fmlAlexP: also, in appendix/wps_tags.tex, the text \config{\%Vi(`label',\dots)} should not use fancy apostrophes since it's code.
18:21:13fmlAlexP: yes, copy-paste, I guessed it :-)
18:21:18AlexPfml: Either way is wrong
18:21:27AlexP'' doesn't just do ' either
18:22:25fmlAlexP: I don't understand. Before "label" is a fancy apostr.
18:23:35 Quit fml (Quit: CGI:IRC)
18:23:44pamauryTheSeven: shouldn't an entry with 0 locks be free ?
18:23:49CIA-8New commit by alex (r26993): Fix typos/c&p errors.
18:24:11TheSeven0 locks, free = false => can be reclaimed, but is likely to be re-used
18:24:26TheSeven0 locks, free = true => should be reclaimed first, not likely to be re-used
18:24:40pamauryand how do you switch between these two states ?
18:24:59TheSevensee the keepincache argument to sectorcache_unlock
18:25:16TheSevenmaybe also in writethrough/readthrough
18:25:33CIA-8r26993 build result: All green
18:27:43CIA-8New commit by alex (r26994): Use normal apostrophes for code. Thanks to fml for pointing out (also for previoys commit).
18:29:04CIA-8r26994 build result: All green
18:29:17pixelmabluebroth3r: the "firmware" in the mentioned headline still has the "fi" ligature
18:29:38pamauryTheSeven: what did you change ?
18:29:44TheSevenadded some comments
18:29:49pixelmabluebroth3r: in a daily pdf (June 17th) off
18:29:51pamauryjust saw
18:31:42 Quit t0rc (Remote host closed the connection)
18:32:27 Quit hebz0rl (Quit: Ex-Chat)
18:32:47funmanyou want to remove ligatures?
18:33:02CIA-8New commit by amiconn (r26995): Make unnecessarily static local variables non-static. It's more efficient, and it also fixes the weird lines with EABI on ARM7TDMI, although it ...
18:33:40bluebroth3rpixelma: really? Even if you copy the text with Adobe Reader? I can't reproduce this with the svn manuals I did earlier today.
18:34:17 Quit BHSPitMini (Read error: Connection reset by peer)
18:34:30CIA-8r26995 build result: All green
18:34:32*bluebroth3r wonders what TeX distribution the build server uses
18:35:07pixelmayes, well with the pdf plugin in firefox (which I think is acrobat reader, as I haven't installed any other PDF readers)
18:35:11 Join hebz0rl [0] (
18:35:47*bluebroth3r is puzzled (again!)
18:36:06bertrikamiconn, weird ...
18:36:14 Join BHSPitMini [0] (
18:36:16AlexPfunman: For things like "fi" in the word "firmware", yes - they should of course be separate letters
18:36:20pixelmabluebroth3r: same with my own PDFs
18:36:36amiconnThe lines themselves are most likely a gcc bug, which is just avoided now
18:36:38bluebroth3rhmm. Maybe caused by an older LaTeX installation?
18:36:50pixelmawouldn't surprise me
18:36:52funmani thought ligatures were cool
18:37:18amiconnHaving simple local vars static is weird though, except for special reasons (which don't apply here)
18:37:30 Quit Antibuddha (Quit: ChatZilla 0.9.86 [Firefox 3.6.3/20100401080539])
18:38:03bluebroth3rBagder: can you figure which LaTeX version the build server uses?
18:38:19AlexPfunman: Not when you try to search for words :)
18:39:07bluebroth3rLaTeX is supposed to use Ligatures. However, it should embed the text without ligatures for searching / copying.
18:39:40 Join pamaury_ [0] (
18:40:23AlexPI haven't tried this personally incidently
18:43:31funmantext_viewer data aborts are linked to footer/header
18:43:46 Quit pamaury (Ping timeout: 265 seconds)
18:44:07 Nick pamaury_ is now known as pamaury (
18:44:21 Quit pamaury (Changing host)
18:44:22 Join pamaury [0] (~quassel@rockbox/developer/pamaury)
18:44:40TheSevenpamaury: what are you doing right now? who of us should continue fleshing this out?
18:47:07pamauryright now, whatching a film :) If you are boring, put in in a pastebin and I'll continue it (well I'll try). We don't have a deadline anyway, so it's not for tonight :)
18:47:36 Join flydutch [0] (
18:47:39TheSevenok, then i'll continue to work on this
18:50:41S_a_i_n_t"If you are boring" :P
18:51:40pamauryS_a_i_n_t: that's why I I rectified ;)
18:52:06amiconnSomehow current rockbox is very unstable
18:52:29S_a_i_n_tIn what way?
18:53:02S_a_i_n_tI haven't noticed anything untoward with 26990...yet
18:53:07amiconnData aborts, bootloader unable to find rockbox.ipod, ...
18:53:16amiconnThat's on ipod color
18:54:07S_a_i_n_tbugger, that was actually a typo "r26980"
18:54:18S_a_i_n_tI'll build latest and see what I find.
18:56:09amiconnDatabase commit not working 'live' even though dircache is enabled, scan running again and again...
18:57:46pamaurydatabase always had problem no ?
18:57:51pamauryThat's not new
18:57:53S_a_i_n_tParhaps what TheSeven and pamaury have been working on has had an unexpected effect?
18:58:03amiconnpamaury: It used to work a while ago
18:58:14TheSevenS_a_i_n_t: that's only being planned, nothing committed yet
18:58:52amiconn(a while == a few months; I don't use the database most of the time so I won't notice bugs quickly)
18:58:58 Join stoffel [0] (
18:59:00S_a_i_n_tAh, right. And your earlier commits were nano2g specific...yes?
19:00:00pamauryI didn't notice any new database bug recently
19:00:28pamauryAnd I use it (only for basic things though)
19:02:53 Join MethoS- [0] (~clemens@
19:05:35pamauryamiconn: if you look at the log for tagcache.c, there has been no real change for more than 9 months, only cosmetics and external fixes
19:05:44TheSevenS_a_i_n_t: those commits just can't have broken something... only preprocessor stuff and higher alignment values for some arrays, can't hurt
19:06:12 Join anewuser [0] (anewuser@unaffiliated/anewuser)
19:06:35amiconnpamaury: Since it's not *only* the database, I wouldn't even expect the problem there. It must be something more fundamental
19:09:04 Quit guymann (Quit: brb)
19:09:06bertrikamiconn, eabi changes maybe? Also gevaerts made some changes to prevent contention between threads wanting disk access, maybe this had some influence?
19:19:14funmanTheSeven: re r26985, I think PROC_NEEDS_CACHEALIGN is not correct
19:19:28TheSevenwhat exactly does this do?
19:19:41funmanit's only needed for dual core where cache alignement is needed, not just desired for performance
19:20:06TheSevenhm, there are CACHEALIGN_ATTRs on some things that need to be aligned on nano2g, mostly DMA stuff
19:21:48funmansee r25339 log
19:22:28TheSeven this isn't only storage stuff, also PCM and the like
19:22:49funmanyeah i know but i had talked about thsi with amiconn
19:22:56pixelmabluebroth3r: adding the things you mentioned to the filter rule lets the error appear but in non-stop mode it's still not easy to find or needs some getting used to (the only thing that does not start with Underfull or Overfull)
19:23:41pixelmaand I get some font warnings
19:24:13funmanTheSeven: hm no it was only about the STORAGE macros
19:24:58funmanTheSeven: DMA alignement isn't related to cache alignement, is it ?
19:26:28funmanwhich things use CACHEALIGN_* on nano2g? grep target/arm shows nothing
19:28:04 Join CaptainKwel [0] (
19:30:45TheSevenfunman: sorry, someone called me on the phone
19:30:57TheSevenaha, so you were the one who broke storage alignment on nano2g by misspelling that define? :D
19:31:06funmanprobably O:-)
19:31:15TheSeveni already wondered how this could have slipped in and though I did it myself :P
19:32:27 Quit stoffel (Ping timeout: 264 seconds)
19:35:05CIA-8New commit by theseven (r26996): Remove PROC_NEEDS_CACHEALIGN for Nano2G again, it's only needed for PP
19:36:02funmanthe D2 has 2 different cores, it would be nice if we can use the 2nd one
19:36:28CIA-8r26996 build result: All green
19:36:45 Quit Bagder (Quit: connection reset by beer)
19:40:40 Quit funman (Quit: free(random());)
19:40:47 Join Bagder [0] (~daniel@rockbox/developer/bagder)
19:41:14 Quit DerPapst (Read error: Connection reset by peer)
19:45:50 Join DerPapst [0] (
19:46:17 Join jannef [0] (
19:47:14jannefhi everybody, are there any news about the new v2 fuzes which refuse to installed rockbox-modified firmware files?
19:49:14 Join ubuntu-nathan [0] (ubuntu-nat@
19:49:24ubuntu-nathanHi all!
19:49:29*ubuntu-nathan is back
19:49:57ubuntu-nathanThere is any RockBox developer online?
19:50:57TheSevenprobably ~100 of them.
19:52:25ubuntu-nathanI want to know where I can get the sources of a very simple(extremelly) simple bootloader for iPod Video
19:52:33ubuntu-nathanI don't want the RockBox one
19:52:57TheSeventhere might also be ipod bootloaders from ipodlinux.
19:53:04ubuntu-nathanbecause that is so much integrated with RB, then I want something that is only the boot loader
19:53:32ubuntu-nathanTheSeven: Have a link to their repository?
19:53:47TheSevenwhat exactly are you planning to do?
19:56:01 Join guymann [0] (~charlie@
19:58:50ubuntu-nathanI've already developed a OS for x86 and for Integrator-CP(ARM)
19:59:01ubuntu-nathanbut now I want to dive into the iPod world
20:01:05 Quit hebz0rl (Ping timeout: 245 seconds)
20:01:39 Join saratoga [0] (463f90ed@gateway/web/freenode/ip.
20:01:47*TheSeven would suggest starting with code *loaded* by the bootloader then
20:05:28 Join stoffel [0] (
20:07:01***Saving seen data "./dancer.seen"
20:07:21ubuntu-nathanwhat you're saying is that is better to forget the boot loader for a minute
20:07:22 Quit halmi_ (Quit: halmi_)
20:07:30ubuntu-nathancreate a Hello, World OS
20:07:36TheSevenbasically yes
20:07:45ubuntu-nathanand use some boot loader to load it
20:07:59 Quit stoffel (Remote host closed the connection)
20:08:04ubuntu-nathanbut, did you have any suggestion of docs for me?
20:08:08ubuntu-nathanto start
20:08:25TheSeventhat way you don't need to care about apple's firmware partition and can just put your code as a regular file on the HDD
20:08:49TheSeventhe best docs on that hardware are probably the rockbox and ipodlinux code
20:08:52ubuntu-nathandid you have any docs for me, or a starting point?
20:09:09ubuntu-nathanI have RB sources here
20:09:27ubuntu-nathanwhere is the "main loaded file" source?
20:09:37saratogaubuntu-nathan: it sounds like what you actually want to do is work on a rockbox plugin not a bootloader?
20:09:54saratogasince a bootloader is mostly device drivers and not an application
20:10:01ubuntu-nathanI want to build an iPod OS
20:10:09ubuntu-nathanbecause I have a good base on this
20:10:14saratogathats probably unrealistic
20:10:17CIA-8New commit by mc2739 (r26997): Revert r26350 - fixes FS #11421
20:10:36ubuntu-nathanI've already done an OS for x86(my first one), and one for Integrator-CP(ARM)
20:10:42ubuntu-nathanNow I want to go mobile ;)
20:11:36TheSevenI'd suggest looking at the rockbox bootloader source, if you want to dig into the rockbox driver architecture
20:11:41saratogawell all the drivers are in the bootloader, knock yourself out
20:11:42CIA-8r26997 build result: All green
20:11:50TheSevenit shares the drivers with the main os
20:12:02ubuntu-nathanI'm on it now
20:12:27TheSevenpamaury: what should we do if somebody tries to WriteThrough a sector that has an exclusive cache handle?
20:12:46Jeromkill him ?
20:13:41 Quit Barahir_ (Read error: Connection reset by peer)
20:13:43pamauryTheSeven: panic
20:13:59 Quit leavittx (Read error: Connection reset by peer)
20:14:01TheSevenwe could also just return an error code
20:14:15pamauryThat will hide the problem no ?
20:14:23 Join petur [0] (~petur@rockbox/developer/petur)
20:14:37ubuntu-nathanI got a better idea!
20:15:07pamauryTheSeven: when such a situation could arise ?
20:15:22TheSevenit shouldn't, afaics
20:16:14pamaurySuch a situation in unsafe. It means someone has a dirty buffer for a sector and someone is trying to write it. this is a major programming bug imo
20:16:29TheSevennot neccessarily dirty
20:17:17pamauryYes but it will at some point, otherwise, exclusive access is useless
20:17:44TheSevenexclusive access might also be used just to lock out writes
20:17:59TheSeven(to make sure the sector buffer stays consistent across yields)
20:18:43TheSevenor should we say in general that write accesses *must* use exclusive accesses, so that they'll fail if someone else has a handle on it?
20:19:11ubuntu-nathanI'm going to do a launcher!
20:19:18ubuntu-nathanlike those ones for Palm OS
20:19:28ubuntu-nathanbut now I will do one for RockBox
20:19:33pamauryTo me, a write should 1) either be done though writethrough 2) either be done with exclusive access, so yes
20:20:21TheSeventhat implies that we need to panic if writethrough hits open handles, even if they're non-exclusive read-only
20:20:41ubuntu-nathanok, I'm out
20:21:00ubuntu-nathanI'm going to watch the Brazils game(since I'm brazilian)
20:21:03pamauryYes, if someone has a handle that has not been release yet, that's a same programming error no ?
20:21:04ubuntu-nathanbye guys
20:36:18 Join M3DLG [0] (
20:40:10 Join will__ [0] (~chatzilla@
20:41:06will__How much work would it take to make the text viewer display rotated 90 degrees?
20:41:37TheSevenwill__: that depends on several things
20:41:51TheSevenif your lcd is square and the driver supports rotating, it should be rather easy
20:42:09will__sansa e200v1
20:42:11TheSevenbut i doubt that you'd want to do that if your display would be square
20:42:24will__indeed :)
20:51:33 Quit will__ (Quit: ChatZilla 0.9.86 [Firefox 3.6.3/20100401080539])
20:56:38 Quit flydutch (Quit: /* empty */)
21:00:47 Join katg [0] (
21:04:37 Join stripwax [0] (
21:12:07 Quit bieber (Ping timeout: 276 seconds)
21:12:22 Join bieber [0] (
21:17:22 Quit Xerion (Ping timeout: 240 seconds)
21:20:15 Quit Viaken (Ping timeout: 245 seconds)
21:20:31 Quit saratoga (Quit: Page closed)
21:20:50 Join Viaken [0] (
21:21:35 Quit sharp (Read error: Operation timed out)
21:21:52 Join sharp [0] (
21:22:02 Join domonoky [0] (~Domonoky@rockbox/developer/domonoky)
21:22:20 Nick sharp is now known as Guest53364 (
21:25:20 Quit Topy44 (Ping timeout: 260 seconds)
21:32:47TheSevendo we accept ignoring storage function return codes?
21:33:11 Join Topy44 [0] (
21:34:37pamauryIt depends on how many lines you want to rewrite :)
21:37:28peturwow... there's a pretty nasty bug in recording: auto-split on 2GB no longer works O_o
21:39:51mitkpamaury: tested clip+ for battery draining after shutdown from: rockbox, OF, charging in OF then shutdown. I haven't noticed battery draining. To time I assume we had only impression about that or it changes between versions. Tested with r26959, bootloader r26730. Now testing shutdown from rockbox after charging in rockbox. Is it possible it depends on which version of bootloader we are using? Which one you used?
21:40:36TheSevencan I generally panic if storage_write_sectors returns a nonzero RC?
21:42:10bluebroth3rpixelma: you could also remove the underfull / overfull warnings from the output. I'm not sure if it's better to filter them out or not
21:42:18pamauryIt might be because of a panic. I basically get a panic every minute when working on usb so perhaps there was a panic but the screen stayed black so I thought it was off but it wasn't
21:42:43gevaertsTheSeven: that seems harsh. It can easily do that on hotswap systems if you unplug at the right moment I think
21:43:09pamauryTheSeven: most of the code should check for the return values, doing a panic on that is a bit violent
21:43:26TheSevenI've seen quite some code that doesn't
21:43:29AlexPbluebroth3r: Do they ever mean anything useful?
21:43:57TheSevenand what should I do if i need to write back a dirty sector because I need buffer space, and get an I/O error?
21:45:46pamauryhum, no easy answer...
21:46:11TheSevenshould I make the cache allocation fail? that will make an arbitrary read fail...
21:46:42TheSevenpanicing seems to be the only option
21:46:50pamauryAnyway, at this point it's too late, that would fail an unrelated operation
21:47:06pamauryOn the other end, there is this hotswap problem...
21:47:40mitkpamaury: It's possible so I'm leaving this. It was only impression :)
21:48:35TheSevenanyway, if such a thing happens, something was corrupted for sure
21:49:05pamauryTheSeven: anyway, most of code will do explicit cleaning/flushing, so the error will be catchable
21:51:11 Quit katg (Remote host closed the connection)
21:51:20 Quit mitk (Quit: Leaving)
21:51:27pamauryAnd in this precise case, well, you can just panic for now
21:54:25bluebroth3rAlexP: of course they do!
21:54:49bluebroth3rthe thing is that if you want to make proper typesetting you're definitely interested in the overfull / underfull warnings.
21:55:03AlexPMmmm, but for us?
21:55:05bluebroth3rhowever, for something like our manual they could get ignored in almost all cases.
21:55:11AlexPaha :)
21:55:43bluebroth3rit might get interesting if the values get too large, especially for overfull boxes (as this means it prints to the margings, and even outside the margins if necessary)
21:57:00pixelmano-one notices if there are just too many and 99% of it is ok
21:57:15AlexPI never pay any attention to them at all
21:57:20AlexPThere are just too many
22:07:03***Saving seen data "./dancer.seen"
22:11:22TheSevenhm, why does passing a null pointer to storage_read_sectors silently succeed?
22:14:07bluebroth3rwell, given the fact that even after removing all other output there is still a load of warnings I wouldn't mind removing them too.
22:16:36 Join mitk [0] (
22:19:43 Quit Jerom (Read error: Connection reset by peer)
22:22:54 Quit CaptainKwel (Quit: Ex-Chat)
22:24:17 Quit ubuntu-nathan ()
22:24:29 Quit stripwax (Quit:
22:49:04 Join stripwax [0] (
22:50:36TheSevenpamaury: do we want to keep the fat cache (but probably smaller), or get rid of it?
22:50:58TheSeveni just spotted an advantage of it: it only needs to mirror a single fat
22:51:04 Join kramer3d_ [0] (~kramer@unaffiliated/kramer3d)
22:51:43TheSevenreducing the fat cache size to like 8 sectors and stacking it on top of the sector cache seems to be the best solution
22:52:18peturhmm I tested recording auto-split with the latest build and found it not splitting. To try stuff I did svn up, changed the max filesize to 2MB and that works fine. Wtf?
22:52:46pamauryTheSeven: I don't see why this an advantage ?
22:52:59TheSevenit only needs half the memory if there are 2 fats :-)
22:54:02pamauryi don't get it
22:54:09pamauryAh yes, ok
22:54:14pamauryI'm a bit slow :)
22:56:43pamauryI think we should get rid of it
22:57:08 Join Barahir [0] (
22:58:44TheSevenwoah, it even seems to work
22:58:55TheSevenbut something is slow as hell
23:04:25bertrikmc2739, thanks for testing!
23:06:41TheSevenpamaury: oh, that one in write_long_name is really nasty
23:06:54 Quit mitk (Quit: Leaving)
23:09:42 Quit M3DLG (Ping timeout: 260 seconds)
23:09:50pamaurywhy ?
23:10:31TheSevenbecause that buffer is used with fat_readwrite
23:12:30TheSeveni'll need to invent something that gets me a cache handle to a sector of a file
23:15:49pamauryOr you need to be able to use a sector buffer just as a sector buffer, that's is as storage but not necessarily at a precise sector
23:16:35 Quit jannef (Remote host closed the connection)
23:18:03TheSevenbut that means losing the cache
23:18:47TheSevenactually i only need to make that fat_readwrite thing return me a sector number :-)
23:21:30pamauryNo, just be able to use one entry of the cache as a scratch buffer, something unrelated to a precise sector of the disk. It does not prevent caching because caching will already be done indirectly by fat_readwrite
23:22:01 Quit BHSPitMini (Ping timeout: 264 seconds)
23:39:18 Quit petur (Quit: Zzzz)
23:41:04 Join BHSPitMini [0] (
23:41:54 Quit bmbl (Quit: Bye!)
23:48:45 Quit fdinel (Quit: Miranda IM! Smaller, Faster, Easier.
23:49:10 Quit TheSeven (Ping timeout: 272 seconds)
23:49:34stripwaxjust added a bunch more music to my ipod, booted into rockbox, and decided to watch the Debug -> Database screen. 117% complete? -494731232/-444391366 B used ?
23:52:14 Join TheSeven [0] (~TheSeven@rockbox/developer/TheSeven)
23:52:32 Quit BHSPitMini (Ping timeout: 240 seconds)
23:52:54stripwaxhrm , and a massive amount of duplicate entries in the db
23:53:49pamauryThat's probably one of the know-bug-that-no-database-dev-managed-to-get
23:53:50CIA-8New commit by funman (r26998): text_viewer: cleanup & bugfix ...
23:54:46stripwaxit seems to have added each *track* as a separate entry in the *albums* tag in the database, somehow. e.g. an album with 5 tracks, has the album name listed 5 times in db.
23:54:54stripwaxmaybe that's an unrelated bug?
23:55:02 Join funman [0] (~fun@rockbox/developer/funman)
23:55:12stripwax[selecting each such entry shows just one track]
23:55:32CIA-8r26998 build result: All green
23:58:43stripwaxi'll just try (re)initializing the db. looks like the update freaked out.

Previous day | Next day