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 2010-06-23

00:00:55amiconnThat would be quite some added diskspace requirement for every user
00:01:06funmanranma: true there is some common code already but it's mostly helpers
00:02:35 Quit domonoky (Read error: Connection reset by peer)
00:03:19CIA-6New commit by kugel (r27067): Convert plasma to use pluginlib actions.
00:03:25 Quit blaaa (Quit: CGI:IRC (EOF))
00:04:45CIA-6r27067 build result: All green
00:05:49 Quit hebz0rl (Ping timeout: 240 seconds)
00:07:20simonrvnhm, might be a pebkac in my case
00:07:30funmanranma: this is on my TODO list actually :) do you want to work on it?
00:07:53bertrikThe database is acting up on my clipv1 on the 27066, first finding about 10 songs/s, then 500 song/s (I have about 200 on the player). Also I couldn't enter the Files menu anymore.
00:08:03***Saving seen data "./dancer.seen"
00:08:19ranmaI'm kind of interested, yes. May have to wait until next month though.
00:08:47CIA-6New commit by kugel (r27068): Use correct opt in plasma.tex.
00:10:02kugelbertrik: it doesn't print the songs/s but the files/s (or maybe (files+dirs)/s even)
00:10:02 Quit Luca_S (Quit: CGI:IRC (EOF))
00:10:26CIA-6r27068 build result: All green
00:12:43amiconnbertrik: Just checked an arbitrary target (ipod color). It has 146 .map files. Worst case fat32 cluster size is 32KB (disk >32GB use this size). That means ~4.5MB wasted just for .map files on such a disk
00:12:52 Join fml [0] (
00:13:14*bluebrother guesses that he was meant :)
00:13:19funmanwhat about taring them?
00:13:58bluebrotherdo we need all map files at all? Shouldn't be a good start for most cases?
00:14:07fmlbluebrother: hello. Do you use vim for development? What plugins do you use?
00:14:13bluebrotherfml: yes.
00:14:38fmlbluebrother: and the second question?
00:15:14bluebrothertaglist is the most important one :) Sometimes NERDtree, and Vim Outliner
00:15:34 Part AzureWurk
00:18:17fmlbluebrother: ok, thnaks!
00:19:57simonrvnyup, pebkac. please ignore. thank you.
00:20:09ranmaI see the STORAGE_WANTS_ALIGN problem even with -O and without -mthumb
00:20:33 Quit M3DLG (Ping timeout: 240 seconds)
00:20:48bertrikI just found a lot of corrupted files on my clip v1, and after correcting those I got problems with the database again, I suspect either 27065 or 27066
00:20:59 Quit fml (Quit: ChatZilla 0.9.86 [Firefox 3.6.3/20100401080539])
00:21:07 Quit bertrik (Quit: De groeten)
00:21:08ranmafunman: Maybe you can try to reproduce it? Even with mp3 it starts blipping and skipping within a minute or so on my player
00:21:21funmannot now sorry i'm on the sloooow laptop
00:22:15 Join M3DLG [0] (
00:24:35 Quit CGL (Read error: Connection reset by peer)
00:24:37 Quit T44 (Quit: Leaving)
00:27:31bluebrotherTheSeven: AFAICS r27027 causes theme changes to crash
00:27:55bluebrotherat least reverting that change fixed it to me. The crash before occured in font_unload()
00:35:23 Quit halmi_ (Read error: Connection reset by peer)
00:41:29 Quit funman (Quit: leaving)
00:44:19 Join BHSPitMonkey [0] (~stephen@unaffiliated/bhspitmonkey)
00:49:24 Join marines [0] (
00:49:32marineshi guys
01:03:00 Quit giovanni (Quit: Sto andando via)
01:05:28 Quit ender` (Quit: The word 'politics' is derived from the word 'poly', meaning 'many', and the word 'ticks', meaning 'blood sucking parasites'. -- Larry Hardiman)
01:15:03mc2739saratoga: e200v1 volume changes across the entire volume range, with FS #11367, there was no change for the v1 and the v2 line out volume changes when it goes below -40dB. With the OF (both v1 and v2), the volume is fixed.
01:15:18 Quit bluebrother (Disconnected by services)
01:15:22 Join bluebroth3r [0] (~dom@rockbox/developer/bluebrother)
01:21:05 Quit jfc (Ping timeout: 264 seconds)
01:22:32CIA-6New commit by kugel (r27069): sim_icons.c is actually only for charcell display simulation.
01:24:11 Join jfc [0] (
01:24:16CIA-6r27069 build result: All green
01:26:33 Quit DerPapst (Quit: Leaving.)
01:31:14kugelhm, down to 3 files, that are needed from uisimulator/* for RaaA
01:31:56kugelsim_tasks.c, stubs.c and io.c
01:32:12kugelerr, no, 4, powermgmt-sim.c also
01:33:33 Join t0rc [0] (~t0rc@unaffiliated/t0rc/x-5233201)
01:34:00 Quit GeekShadow (Quit: The cake is a lie !)
01:43:52 Quit DataGhost (Ping timeout: 260 seconds)
02:01:55 Join fdinel [0] (
02:02:35saratogabertrik: IMO we shouldn't commit the "high quality" mode settings unless someone can show a situation where they make a measurable difference
02:02:56saratogathey may only improve output under certain other settings or supply voltages
02:03:20saratogaI'd like to see some evidence that apply to the sansas before we commit
02:03:45 Quit marines (Ping timeout: 240 seconds)
02:04:54saratogamc2739: looking at audiohw_set_master_vol, it looks like the driver changes the line out voltage
02:05:16saratogawait no thats line in
02:08:05***Saving seen data "./dancer.seen"
02:12:33CIA-6New commit by kugel (r27070): Silence the line during compilation a bit (to "GEN version.h").
02:13:58CIA-6r27070 build result: All green
02:16:04kugelsomething is wrong with feature.txt, the build output mentions PP feature.txt twice
02:17:28saratogaanyone have a fuzev2 handy?
02:18:18 Join anewuser [0] (anewuser@unaffiliated/anewuser)
02:20:23 Join [Saint] [0] (S_a_i_n_t@
02:20:33 Quit S_a_i_n_t (Ping timeout: 265 seconds)
02:21:41saratogaI'm curious what GPIOA says on the debug menu when its not plugged into anything
02:22:58 Quit kugel (Remote host closed the connection)
02:36:29mc2739saratoga: On dock connect, e200v1 OF volume level is fixed (both headphone and line out, maybe they are connected internally) - e200v2 OF disables headphone output and line out volume is fixed
02:40:44Battousaisaratoga: GPIO A = 25 when unplugged
02:50:15 Join Buschel [0] (
02:53:57saratogaBattousai: thanks
02:54:19saratogamc2739: on ABI someone posted the GPIO outputs while docked for a Fuzev2, but the results are inconclusive
02:54:44saratogaGPIO A4 is a possibility, but the wiki says thats already "FM i2c SDA variant ?"
02:55:14saratogaunless the Fuzev2 also has DBOP input pins?
02:56:22 Nick fxb is now known as fxb__ (
02:57:55mc2739if it does, we are not using them in the button driver
02:58:12saratogawe use it for display though
02:58:20saratogaat least i see references to it in the target tree
02:58:33saratogabut its disabled in the debug menu?
02:59:45saratogaassuming the DBOP in pins are still there, its possible the dock detect is on one of those
03:00:42 Quit Buschel (Ping timeout: 245 seconds)
03:11:43 Join freddyb [0] (
03:16:39 Quit ssorgatem (Remote host closed the connection)
03:16:41 Quit freddyb (Quit: CGI:IRC (Ping timeout))
03:17:01 Join freddyb [0] (
03:40:15 Quit mt (Ping timeout: 258 seconds)
03:48:22 Quit steve|m (Ping timeout: 276 seconds)
03:49:34 Join S_a_i_n_t [0] (S_a_i_n_t@
03:49:50 Quit [Saint] (Ping timeout: 258 seconds)
03:53:35 Quit storm` (Ping timeout: 240 seconds)
03:54:37 Join storm` [0] (
03:56:00 Quit t0rc (Quit: Leaving)
04:03:22 Join steve|m [0] (
04:03:54CIA-6New commit by jdgordon (r27071): redo r27027 to hopefully fix data aborts when loading a theme
04:05:51CIA-6r27071 build result: All green
04:08:06***Saving seen data "./dancer.seen"
04:08:31*JdGordon isnt sure if that actually fixes the problem or not....
04:08:41JdGordonbluebroth3r: can you confirm?
04:11:01 Join Barahir_ [0] (
04:14:11 Quit Barahir (Ping timeout: 248 seconds)
04:15:18mc2739JdGordon: will that mix multi font also?
04:16:52mc2739r27070 - I have a large font as font #2 in fms and it displays as font #1
04:18:18 Quit S_a_i_n_t (Ping timeout: 252 seconds)
04:19:56 Join S_a_i_n_t [0] (S_a_i_n_t@
04:20:53JdGordonhas that been reported as a bug?
04:21:18JdGordonit might fix that problem, but as that is the first im hearing of it, probably not
04:21:31mc2739JdGordon: no, just now noticed it
04:21:58mc2739and it looks like it is just on one device, so maybe just a problem on my end
04:22:20JdGordonit doesnt look like it owuld fix hat
04:24:22mc2739I'm getting the data abort on theme change - let me load the latest and I'll see if it stiil happens
04:28:44mc2739r27071 - still gets data abort on theme/skin change
04:30:49mc2739and multi font works ok after fresh creating a fresh .rockbox directory
04:31:55 Quit amiconn (Disconnected by services)
04:31:57 Join amiconn_ [0] (quassel@rockbox/developer/amiconn)
04:32:00JdGordoncheck the map file.. where is the abort happening?
04:32:15 Nick amiconn_ is now known as amiconn (quassel@rockbox/developer/amiconn)
04:32:38JdGordonand, can you revert 27071 nd 27027 and see if it still aborts?
04:32:42 Quit pixelma (Disconnected by services)
04:32:43 Join pixelma_ [0] (quassel@rockbox/staff/pixelma)
04:32:51mc2739don't have a map file - downloaded from build page
04:32:59 Nick pixelma_ is now known as pixelma (quassel@rockbox/staff/pixelma)
04:33:10mc2739but I'll build it locally and retest
04:37:38*JdGordon whacks TheSeven with a 2x4
04:41:38 Quit Rob2223 (Quit: Rob2223)
04:41:54JdGordonbassically that usb change to unload the fonts wasnt very good.. it unloads all the skin fonts but doesnt reload them after exit, you need to call settings_apply_skins() as well as settings_apply() to have them reloaded
04:42:05JdGordonbut doing that to just load the fonts is le suck!
04:42:27 Join mt [0] (~mtee@rockbox/developer/mt)
04:44:14mc2739JdGordon: Data abort at 0005AE90 (0)
04:44:48JdGordonI think a more correct way to do it is font_reset(NULL) and unload the remote ui font on enter, and manually close all the skin font fd's (but NOT reset the font structs), and then on usb exit we load the ui fonts manually and manually reopen the skin font fd's and seek to the old position? (I'm not sure if the seek is needed)
04:44:57JdGordonmc2739: which is where? can you check the map?
04:45:13mc27390x0005ae74 font_unload
04:45:33mc2739thats with 27071
04:46:04 Join Rob2222 [0] (
04:48:39mc2739JdGordon: no data abort with 27071 and 27027 reverted
04:51:33JdGordonthat is rather odd...
04:51:47JdGordonI have no idea how font_unload() is even being called
04:51:57mc2739JdGordon: interestingly, my e200v2 is not getting the data abort, just the e200v1
04:52:16 Quit freddyb (Remote host closed the connection)
04:52:37JdGordonI wonder if this is unaligned access..?
04:53:02 Join binaryhermit_ [0] (
04:53:37 Quit binaryhermit (Disconnected by services)
04:53:43 Nick binaryhermit_ is now known as binaryhermit (
04:54:57 Quit TheSeven (Ping timeout: 260 seconds)
04:56:41 Join binaryhermit_ [0] (
04:56:41 Quit binaryhermit (Disconnected by services)
04:56:45 Nick binaryhermit_ is now known as binaryhermit (
04:57:05CIA-6New commit by jdgordon (r27072): revert r27027,27028, 27071 - need to find a correct way to close font fd's. 27027 caused data aborts, 27028 caused multifont issues after returning ...
04:57:13JdGordonyay palindrome :)
04:58:28JdGordonTheSeven: I'll try have a fiddle today if I get time to get a better solution
04:58:31CIA-6r27072 build result: All green
04:59:02 Join kramer3d [0] (~kramer@unaffiliated/kramer3d)
04:59:48 Join TheSeven [0] (~TheSeven@rockbox/developer/TheSeven)
05:01:38 Join funman [0] (~fun@rockbox/developer/funman)
05:11:19JdGordonmc2739: just a snity check, does svn still crash?
05:12:28mc2739no, r27072 is not crashing
05:14:32 Join CGL [0] (~CGL@
05:15:24JdGordonit looks like the font fd just needs to be open, it is always seek()-ed before reading so this should be doable
05:17:01JdGordonhmm... umm
05:24:16 Join llua [0] (
05:25:22 Quit kramer3d (Quit: Leaving)
05:27:54lluausing rockbox on a sansa e260v2, when i change songs or resume from pause there is always a delay before it happens. i cant seem to find a way to stop that
05:28:16funmanfading on pause
05:31:24funmanipod4g has a problem when built with eabi or has it just not been tested yet?
05:31:42lluafunman, ty
05:32:58lluais there a way to disable the pause between track changes?
05:33:00 Join wad11656 [0] (
05:33:19 Quit fdinel (Quit: Miranda IM! Smaller, Faster, Easier.
05:34:13wad11656Hello! Is someone available to help me get album artwork properly on my iPod from my Mac OS X? I can use a virtual Windows XP + iTunes if necessary. My iPod music directory seems to be very different than what the guide on this site claims it should be. Anyone?
05:34:45 Quit S_a_i_n_t (Ping timeout: 240 seconds)
05:35:40mc2739wad11656: have you seen
05:35:48 Join S_a_i_n_t [0] (S_a_i_n_t@
05:35:56wad11656i.e. it's iPod_Control>Music> then a bunch of folders titled "F00", "F01", & "F02", each with 5-7 songs in them, but each one is named like CBAE.mp3 and such
05:35:59wad11656I've been on that page
05:36:11wad11656I'm looking at it now...I guess I'll see if I can find anything else...
05:36:27funmanwad11656: that's how itunes copy the music files; if you don't like that just copy them through a file browser
05:36:52mc2739and don't use itunes :)
05:37:05wad11656Am I still able to put album artwork for each song for them when I DO add songs that way?
05:37:28wad11656(compatible with both rockbox and apple firmware)
05:37:49wad11656or should I just restart and drag & drop through file browser?
05:38:47 Quit Horscht (Quit: Verlassend)
05:39:49 Quit binaryhermit (Ping timeout: 240 seconds)
05:40:31funmanjhMikeS: why set CACHEALIGN_SIZE to sizeof(int) by default in r26011 ?
05:42:12 Join binaryhermit [0] (~binaryher@
05:42:30wad11656meeeeehhhh Well thanks peeps
05:42:52wad11656I guess I'll restart my Music library and just add the music through the file browser/Finder
05:42:55lluafunman, crossfade doesnt affect the pause between track changes. but it did fix the pause when resuming from paused
05:47:07 Quit wad11656 (Quit: ChatZilla 0.9.86 [Firefox 3.6.3/20100401064631])
05:47:27 Join wad11656 [0] (
05:49:06 Quit panni_ (Quit: ( :: NoNameScript 3.81 :: ))
05:50:19 Quit YPSY (Ping timeout: 260 seconds)
05:51:48 Join Ypsy [0] (
05:52:08 Nick Ypsy is now known as YPSY (
05:53:29 Quit bieber (Ping timeout: 240 seconds)
05:53:45 Join bieber [0] (
05:54:49 Quit wad11656 (Quit: ChatZilla 0.9.86 [Firefox 3.6.3/20100401064631])
05:59:55funmanranma: i think i've found the problem with STORAGE_WANTS_ALIGN
06:08:09***Saving seen data "./dancer.seen"
06:10:47 Quit ThomasAH (Ping timeout: 260 seconds)
06:11:18 Join ThomasAH [0] (
06:13:05funmanit comes from playback.c which assumes cacheline size = 16 bytes, we didn't see it on ipods because here it's the case
06:21:47JdGordonTheSeven: I dont know what to do about skin fonts. I think reloading all settings and skins after usb is a waste (although presumably the disk is spinning anyway so maybe it isnt a big deal?) the skin fonts need to be unloaded properly and reloaded so they get the same font_id (which is technically possible but hard to guarentee it wont break in the future)
06:22:07funmanJdGordon: it's necessary if the files have been modified
06:22:41JdGordonnot really... it loads the whole skin to ram so it can change without causing any problems
06:23:24JdGordonthe issue being worked around is the font fd's not being closed, not that themes dont get pdated after usb exits
06:24:35funmanif the font file changes the fd isn't valid anymore?
06:25:39JdGordonI was thinking about manually closing and then reopening the .fnt instead of reloading the whole theme
06:26:44JdGordonbut I'm thinking that might not be so easy after all :/
06:29:45bieberJdGordon: Did you get rid of Xd in the new syntax?
06:34:21CIA-6New commit by funman (r27073): playback.c: don't assume cacheline size is 16 bytes ...
06:34:25CIA-6New commit by funman (r27074): revert r27044 (which was a revert of r26953 & r26954) ...
06:36:01CIA-6r27073 build result: All green
06:37:00funmanhope i don't have to make a revert of a revert of a revert ^^
06:37:34CIA-6r27074 build result: All green
06:38:03funmanhm i do see FS corruption too
06:39:06funmanwith the fonts :/
06:45:22funmanabout USB in sansa AMS bootloader, since it's intended as a recovery method, should we activate it only if a button is pressed? (SELECT for example)
06:45:42funmanso if it's powered by USB it would load rockbox.sansa if possible
06:45:42saratogamakes sense
06:45:53saratogamaybe also if the rockbox.sansa file is missing
06:50:59ranmafunman: The playback.c fix seems to fix the glitches for me
06:51:51ranmaI don't quite get why though, IMHO even without the fix the worst that should happen is an unaligned read or write and thats handled correctly in sd-as3525.c
06:52:27funmani don't know either and i don't want to look in these files
06:52:57saratogaplayback.c makes all sorts of assumptions about alignment
06:53:34funmanso now we should beat OF wrt USB speed
06:54:06 Join wad11656 [0] (
06:54:58wad11656Hey. Can someone answer a question I have? Is there a way for music to show up both in the rockbox firmware and in the original firmware while still supporting album artwork while in either firmware? Thanks!
06:55:38saratogawhat player
06:55:49*ranma idly wonders if the OF tries to switch cards to HS mode too, AFAIK HS is supposed to have different timings (which is why you have to 'enable' it first, so we might be out of spec there
06:56:05wad11656The music in iPod_Control that I added through iTunes on to my iPod shows up in the Database on Rockbox, but I don't believe there's a way to get cover art for those songs. I have music in a "Music" folder as well which is able to have artwork with cover.jpg's
06:56:07wad11656ipod nano 2g
06:58:23wad11656Can I have one directory of music that both the original firmware AND rockbox recognizes?
06:58:28 Join Zigtown [0] (
07:00:55funmanif you use the database rockbox will see all your files
07:01:46wad11656but the original Apple firmware on the iPod seems to not recognize files in the only searches in the iPod_Control folder...
07:02:01wad11656I want to be able to play music from either firmware
07:02:18wad11656without having duplicates of songs. Is this possible?
07:04:33JdGordonbieber: you mean %X(d) right?
07:04:45bieberOh, is that how we're doing it now?
07:04:53 Quit leavittx (Ping timeout: 258 seconds)
07:05:14JdGordonthat is what the hardcoded wps is using
07:05:24JdGordonI tihnk that might be the only theme using it :)
07:05:49 Join AzureSky [0] (
07:06:25JdGordonbieber: did you see my brain dump last night?
07:06:35bieberI'm not sure, lemme check
07:06:36JdGordonyour parser really doesnt work well with the old displayer :)
07:06:47JdGordonnot very important
07:07:11bieberOh, lol
07:07:12JdGordonim pretty much rebuilding the skin code from your tree to pass to the old parse functions for the tokens :p
07:07:36 Part AzureSky
07:07:40JdGordonI've given up rebuilding everyhing in one hit, way to huge
07:08:39CIA-6New commit by funman (r27075): Sansa AMS bootloader: enter USB mode only when needed ...
07:09:52bieberI can imagine
07:10:16CIA-6r27075 build result: 12 errors, 0 warnings (funman committed)
07:10:18 Join leavittx [0] (~leavittx@
07:13:23 Quit binaryhermit (Ping timeout: 260 seconds)
07:13:29jhMikeSfunman: I think it was so it was at least a word size alignment and defined to something
07:16:25CIA-6New commit by funman (r27076): sansa AMS bootloader: build if USE_ROCKBOX_USB isn't defined
07:16:27jhMikeSfunman: also, (from the logs about the pcm lock). the interrupt fires and just says that it did. otherwise *all* dma callbacks would get shut down since all 32 virtual channels share one interrupt.
07:16:57funmansorry, i don't remember the discussion
07:17:59CIA-6r27076 build result: All green
07:18:59jhMikeSfunman: was waaaay back on June 1 (I was rummaging back)
07:19:13funmanhm ok
07:22:41jhMikeSplayback.c codec buffer alignment was geared toward the need for line reads and writes on coldfire. vorbis performed pretty badly there without it. it really didn't have anything in mind about caches.
07:23:08funman'line' isn't cache line here?
07:23:17 Join Buschel [0] (
07:23:49saratogacoldfire has no data cache
07:24:22saratogaso its probably how the memory controller internally accesses blocks of words in memory
07:24:33jhMikeSfunman: not there, no. you only get burst reads using movem.l if the burst starts on a 16-byte boundary though.
07:24:39funmanalso storage alignement isn't necessarily the same than cache line
07:24:53saratoga(i know very little about CF)
07:25:32jhMikeS*burst access (read or write) really
07:25:33funmanperhaps there should be some default alignement value set to the max of cache line, storage align, and 16 for coldfire
07:26:36jhMikeSif cf had a D-cache cache (iiuc later models do), no doubt cache line fetches and stores would be bursts like that
07:27:37funmanmc2739: before making a new mkamsboot we must release bootloaders with USB enabled, let's wait a bit to see if USB is really stable
07:27:52*jhMikeS hasn't looked at storage align and what it's intended to do
07:28:18funmanjhMikeS: align buffers for storage_read/write_sectors() (for DMA)
07:28:37saratogait took us something like a year to get PP USB completely stable, I recommend waiting a while to be sure with AMS USB
07:28:45funmanTheSeven is working on something more complete which ensures alignement
07:28:46jhMikeSfunman: and that's not directly related to cache line size (or rather, the same)?
07:29:29jhMikeSI guess not, DMA might have restrictions too
07:29:36funmanon as3525 DMA alignement depends on the DMA settings. right now it happens to be the same alignement than cache line but that's coincidence
07:30:02funmanranma: perhaps incerasing DMA_S8 to DMA_S32 (burst size) in sd-as3525.c could improve performance? IIRC test_disk showed not much difference
07:37:07 Quit bug2000 (Ping timeout: 260 seconds)
07:42:11 Quit shai (Quit: Leaving)
07:43:41*JdGordon thinks he is being blocked by a pictureflow bug :(
07:44:23 Join AzureSky [0] (
07:47:12 Join shai [0] (
07:47:24JdGordonyay! my patch doesnt cause the bug
07:48:07S_a_i_n_tgevaerts: Anyone else: Scanning tha database without at least one "database acceptable" file will cause it to either a: Scan indefinitely, or b: Crash and Burn.
07:48:28*S_a_i_n_t thinks this is known, but isn't sure now...
07:48:37CIA-6New commit by jdgordon (r27077): r27028 again.. properly close font fd's and reload the skins on usb exit (disk should be spinning so hopefully not disastrous!)
07:48:54JdGordonS_a_i_n_t: thats one for pamuary probably :)
07:49:00JdGordonand yeah, annoying and known bug
07:49:37pixelmaS_a_i_n_t: yes it is, with bug report and not fixed yet
07:49:41jhMikeSdatabase hangs up if it runs out disk space too
07:49:54JdGordonapparently going into usb from inside pictureflow will panic when usb is detached
07:50:08CIA-6r27077 build result: All green
07:51:22amiconnsaratoga: High quality mode seems to sound a bit better. It may hard to measure, since low power mode already produces a flat frequency response according to dfkt
07:51:38pixelmaJdGordon: speaking of pictureflow... why did you commit this "integration" for all targets even though it was not possible to start music from inside pictureflow on hwcodec and the patch didn't touch that at all? ;)
07:51:56amiconnI guess it's intermodulation or sth similar that makes low power mode sound worse. Do we have a way to measure distortions?
07:52:06saratogaamiconn: the tests showed a mostly negligible improvement in distortion
07:52:56amiconnMeasured with real ear-/headphones as the load?
07:54:02S_a_i_n_tpixelma: JdGordon: I only mention it through reading the backlog...was more for gevaerts, he seemed unsure if it was known or not.
07:55:11 Quit wad11656 (Ping timeout: 240 seconds)
07:55:26pixelmaI also read that but just read on, the guy he helped with confirmed that putting one audio file helped so I guess gevaerts understood that it hasn't been fixed yet ;)
07:56:19S_a_i_n_tYeah, just wanted to clarify it wasn't some form of magic ;)
07:56:31S_a_i_n_twhen I first hit that bug it plagued my for days :P
07:56:36saratogathe 16ohm tests are the only ones with vaguely meaningful distortion, and the settings don't really seem to help much
07:57:02saratogae.g. 0.124 % IMD to0.096 %
07:57:09saratogastill worse then the OF as well
07:58:07S_a_i_n_tIs there updated skin syntax on the wiki yet?
07:58:13*S_a_i_n_t is search failing...
07:58:51bieberSkinBreakingChange has some info
07:59:29*S_a_i_n_t nods ( I forget that one..thanks)
07:59:39JdGordonS_a_i_n_t: if you're bored feel like updating CustomWPS? :D
08:00:08S_a_i_n_tI don't think I'm that bored :P
08:00:12 Quit llua (Remote host closed the connection)
08:00:25bieberJdGordon: Just run your skin updater on it :P
08:00:38JdGordonI dont tinhk that would work :)
08:00:41S_a_i_n_twell, I'm actually a little busier than I'd like, but I can certainly look into it later tonight.
08:01:02S_a_i_n_tIt's been on my "todo if no one else gets around to it soon" list.
08:01:10S_a_i_n_t+for some time now.
08:01:13saratogafunman: do the DBOP input pins still exist on AMSv2?
08:02:09funmani guess yes
08:02:25saratogathe debug screen doesn't print them so i wasn't sure
08:03:33S_a_i_n_tI also hope there's a fix for the font loading/unloading thing soon...the USB screen looks "yuck"
08:03:34 Join Bowbles [0] (
08:03:35 Quit BHSPitMonkey (Read error: Connection reset by peer)
08:04:11S_a_i_n_tthere's only 4 options, perhaps to hell with translation and use bitmaps :P
08:04:45S_a_i_n_tthen we get themes USB screens as a result :P
08:05:15JdGordonI've fixed the font unloading thproblem
08:05:25JdGordonusb screen will never allow custom fonts though
08:05:27JdGordonall bmp
08:05:35S_a_i_n_taha, awesome.
08:05:45bieberJdGordon: Is there any kind of transparency supported in skin images?
08:06:11JdGordonpixels coloured 255,0,255 are transparent
08:06:14S_a_i_n_t255,0,255 is transparency in bitmaps
08:06:25JdGordontransparent in that they show the background image, not what is really behind it iirc
08:06:53bieberOh, okay
08:06:56S_a_i_n_tif there's no backdrop, you just get hideous pink also.
08:07:20bieberI just saw some obvious masking in a bitmap, but I didn't see anywhere to specify the mask color, so I was all "wtf"
08:07:33bieberWait, masking is the wrong word. Chroma keying?
08:07:45S_a_i_n_t"magic" ;)
08:08:07JdGordonthere is another magic colour which uses the forground colour iirc
08:08:10***Saving seen data "./dancer.seen"
08:08:12JdGordon255,255,0 iirc?
08:08:25S_a_i_n_t...pass, that's new on me.
08:10:05BowblesIf I use iTunes to sync my music to my rockboxed ipod, will rockbox understand the id3 info in the songs? or will I need to manualy update the rockbox Database every single time I update songs from iTunes?
08:10:06*S_a_i_n_t finds "Viewports cannot be layered transparently over one another" in the manual...not true if one viewport is an image, and the other text.
08:10:10S_a_i_n_tI do this all the time :/
08:11:01bieberFor tags that return percent values, do people actually have to include 100 enumerations to cover every possible percentage value?
08:11:25S_a_i_n_tthere's some form of magic there.
08:11:58JdGordonnot quite true
08:12:09S_a_i_n_tif its %Xx<|||> it will be split into 25% increments IIUC
08:12:12bieberHow's that work? Does it just autoscale to the number of enumerations provided or something?
08:12:39JdGordonyeah, you need to start working with skin_tokens.c :)
08:13:33bieberI reckon so
08:13:36 Quit anewuser (Quit: Sixth edition of 25 samples + 48 hours + detroit party (july 31) (maybe streamed online!))
08:13:58bieberThat or just play around with the skin engine and see how things work. That'd probably be faster than trying to discern everything from the rendering code ;)
08:14:56S_a_i_n_t...maybe ;) between the manual and reading code I've still had some quality "wtf" moments with UI stuff.
08:14:57JdGordonna, skin_tokens.c is pretty simple
08:16:13JdGordonbieber: for conditionals to work the token has to return a number, if the number is greater then the amount of options the last one is used (careful here because %?aa<true> needs to act the same as %?aa<false|true> (or maybe true|false)
08:16:13*S_a_i_n_t just noticed there's only one mention of transparency in the manual, and its a slightly wrong sentence re: overlapping viewports.
08:16:27JdGordonso, if any scaling is done it is done in the token handler
08:16:32JdGordonnot in the renderer
08:16:55bieberI see
08:17:15JdGordon%?xx<true|false> says CustomWPS
08:17:43 Quit leavittx (Ping timeout: 258 seconds)
08:17:45JdGordonthe token value option thing is 1-indexed, not 0-indexed
08:17:54 Join kish [0] (~o2@unaffiliated/spice)
08:18:13JdGordonalthough if you arnt using skin_tokens.c that doesnt really matter for you
08:18:39 Join stoffel [0] (
08:18:49*S_a_i_n_t adds "explain transparency" to his CustomWPS todo list. (Or someone elses)
08:20:38bieberYeah, I just have to make sure that however I implement it matches RB's rendering
08:22:05 Quit Buschel ()
08:24:49 Join leavittx [0] (~leavittx@
08:25:20CIA-6New commit by Buschel (r27078): Clean up comments.
08:26:03ranmafunman: I doubt that'll make a difference. I'm already seeing 10MB/s from internal flash on reads, which is the same I get from the uSD from the fast external card reader
08:26:29ranmaFor uSD it would be nice to lower PCLK to 50MHz, since it then could use 25MHz instead of 15MHz
08:26:52CIA-6r27078 build result: All green
08:30:00funmandynamic pclk would require reclock all peripherals no?
08:30:14funmanand slower memory
08:31:26saratogai thought SD was driven off the ide clock?
08:33:18 Join Luca_S [0] (
08:33:36saratogaclock-target.h says otherwise
08:34:03funmanperhaps flyndice knows the details
08:35:25saratogait looks like its pclk on v1 and plla on v2
08:35:56funmanide clock is needed but i'm not sure why
08:37:28saratogaso its probably driving something and we just dno't know what it is
08:37:41bieberJdGordon: How do you display a specific image from a bitmap strip in the new format?
08:37:52bieberSkinBreakingChange says %Xd only takes one argument
08:38:32S_a_i_n_t%Xd(z), %Xd(b) etc.
08:39:09S_a_i_n_tand "%xd" rather
08:39:29S_a_i_n_tthe "X" is backdrop no?
08:40:02bieberRight, I meant %xd
08:40:26bieberBut that just displays a named image, the old format also allowed you to include an index into a bitmap strip
08:40:53S_a_i_n_tyeah, then "%xd(inentifier of the bitmapstrip)"
08:41:00funmanis it me or default rockbox voice sucks for spelling?
08:41:13simonrvnnot just you
08:41:40S_a_i_n_toh right, I get you..."%xd(Aa), %xd(Ab) etc.
08:42:01S_a_i_n_tA being the bitmapstrip, a/b being the subimage
08:43:25funmanranma: can you make mkamsboot check left button on c200v2 ?
08:46:22 Join ender` [0] (
08:47:18ranmaUsing DBOP, yes
08:50:09 Quit TheSeven (Ping timeout: 240 seconds)
08:54:22pixelmabieber: the old format had e.g. %xdAe which would display the 5th part of the "A" image, doesn't it translate to %xd(Ae) now?
08:54:39bieberYep, I just didn't realize that was how it was done
08:56:57ranmaNice, rockbox.sansa loads noticably faster with 4bit mode
08:57:07funmanbtw we can't use MMU to detect writes to read-only data because we're running in supervisor, and read-only attributes only applies to user mode
08:57:55CIA-6New commit by ranma (r27079): Use DBOP to check for left button on C200v2 like we are supposed to instead of right button
08:58:11S_a_i_n_tJdGordon: Is plain "%x" (load and display an image) new? o_0
08:58:23S_a_i_n_tIf not, wow...I've been doing things the hard way.
08:59:11 Join flydutch [0] (
08:59:24funmanranma: thanks that was fast! can you update README and dualboot.c/h too?
08:59:31CIA-6r27079 build result: All green
09:00:12ranmafunman: Well, I still had that one lying around, just needed a cut&paste, recompile and test :)
09:00:26funmaneheh :)
09:01:40CIA-6New commit by ranma (r27080): Use DBOP to check for left button on C200v2 like we are supposed to instead of right button
09:01:51 Join Zagor [0] (bjst@rockbox/developer/Zagor)
09:01:56pixelmaS_a_i_n_t: %x is ooold
09:02:24amiconnbluebroth3r: I just checked the ipod color .map files. ~3.5MB uncompressed, 423KB zipped (max compression)
09:02:32 Join Szpila [0] (~sszpila@
09:02:33S_a_i_n_twow...I've been %xl/%xd 'ing everything...even if it isn't conditional.
09:03:06S_a_i_n_t(at least me code looks more consistent :P)
09:03:11CIA-6r27080 build result: All green
09:03:32 Join petur [0] (d408b802@rockbox/developer/petur)
09:03:35 Join GeekShadow [0] (~Antoine@reactos/tester/GeekShadow)
09:04:22 Quit JdGordon| (Quit: leaving)
09:04:44 Join JdGordon| [0] (~jonno@rockbox/developer/JdGordon)
09:05:07CIA-6New commit by ranma (r27081): Remove the unused ENRD0_USB_STATUS define, I'll leave the other button defines in, maybe someone wants his bootloader to use one of those instead ;)
09:06:06 Join JdGord [0] (~js@
09:06:08CIA-6New commit by funman (r27082): as3525*: initialize cpu_frequency correctly
09:06:37CIA-6r27081 build result: All green
09:06:51JdGordI would like to change %xd to use a number instead of letters for subimages
09:07:28JdGordEspecially if the new parser will le us do something like %xd(a,%pv)
09:08:07CIA-6r27082 build result: All green
09:08:45S_a_i_n_tthe idea seems yuck in my head, but ok. I can't imagine how "%xd(a,%pv)" would be useful though...
09:09:38JdGordReplace pb with something with only a few values... playback mode maybe
09:10:18 Quit funman (Quit: free(random());)
09:10:28S_a_i_n_tI still don't get it. Adding things like that to an image display line seems messy.
09:12:01JdGord%xd(shuffle,%ps) is much cleaner than a conditional and two CD tags
09:12:08pixelmaJdGord: that would mean quite a few less IDs possible with just one character
09:12:27S_a_i_n_tI don't see it as being cleaner either.
09:12:45JdGordUsing one letter Id's suck
09:12:58pixelmawhy that?
09:13:15 Part AzureSky
09:14:00*S_a_i_n_t would rather see %xd(a) , %xd(b) than %xd(a,b,c,d)
09:14:05JdGordThere is no technical reason for the one char ID's
09:14:18S_a_i_n_tI think its just more consistent, and easier to debug
09:15:20JdGordNo. I'm saying the first parameters is the image I'd. The second is another tag which returns a number. That number IA used to choose the subimage
09:15:46 Join mitk [0] (~mitk@
09:16:17S_a_i_n_toh, in that case I really don't like it. sorry. I working example might sway my opinion.
09:16:40amiconnbluebroth3r: Using 7z it's down to 195KB (max compression)
09:17:08 Quit Bowbles (Ping timeout: 258 seconds)
09:17:20amiconn7za a -mx=9 -r -i\!*.map maps.7z in the build dir
09:17:47pixelmaJdGord: one character is more readable IMO than two digits or so
09:18:24CIA-6New commit by bieber (r27083): Theme Editor: Working on image rendering
09:18:26*S_a_i_n_t agrees
09:18:46JdGordNot in a huge skin where you have 15 images loaded... which is which?
09:18:59S_a_i_n_tI also see that as making it difficult to discren how many subimages are needed.
09:19:04JdGordAnd of course one letter would be legal also
09:19:12S_a_i_n_tperhaps I'm getting it wrong though
09:20:00CIA-6r27083 build result: All green
09:20:21S_a_i_n_tPersonally, I comment my themes to the enth degree...I know what's what in my themes.
09:21:09S_a_i_n_tI'd like to see more comments in themes, but there's no way to enforce this.
09:21:59 Join einhirn [0] (
09:22:10 Quit einhirn (Read error: Connection reset by peer)
09:22:40 Join bor_ka [0] (
09:23:49CIA-6New commit by bieber (r27084): Theme Editor: Fixed image display relative to viewport
09:24:48JdGordWell subimage count is set in the load line anyway. This would let you collapse massive conditiwls which only display an image down to on tag
09:24:57pixelmaJdGord: how would a subimage be addressed in your number proposal? I can't imagine it better readable
09:25:34CIA-6r27084 build result: All green
09:25:46JdGordXd (label, 1)
09:26:08pixelmawhat is label?
09:26:47S_a_i_n_twith a comma? wouldn't it still be %xd(lable1) ?
09:27:02JdGordThe image label
09:30:08 Join leavittx_ [0] (~leavittx@
09:36:01 Join einhirn [0] (
09:37:49CIA-6New commit by bieber (r27085): Theme Editor: Implemented %xd tag with subimages
09:37:54 Quit einhirn (Client Quit)
09:38:25 Join einhirn [0] (
09:39:22CIA-6r27085 build result: All green
09:39:51 Quit einhirn (Client Quit)
09:40:13 Join einhirn [0] (
09:40:58 Quit leavittx_ (Remote host closed the connection)
09:42:08 Join Kitr88 [0] (
09:42:23 Quit JdGord (Ping timeout: 264 seconds)
09:42:36 Join JdGord [0] (~js@
09:44:09 Join leavittx_ [0] (~leavittx@
09:45:35 Quit Kitar|st (Ping timeout: 248 seconds)
09:45:35 Join B4gder [0] (~daniel@rockbox/developer/bagder)
09:46:36 Quit Kitr88 (Ping timeout: 265 seconds)
09:47:21 Quit bor_ka (Quit: CGI:IRC (Ping timeout))
09:52:02CIA-6New commit by bieber (r27086): Theme Editor: Removed some lingering debug code
09:52:08 Join marines [0] (
09:52:16 Join Kitar|st [0] (Kitar_st@
09:53:38CIA-6r27086 build result: All green
09:54:17 Join Rob2223 [0] (
09:57:49 Quit Rob2222 (Ping timeout: 265 seconds)
09:57:57 Join hebz0rl [0] (
10:01:02 Join Bowbles [0] (
10:07:16 Quit r0b- (Read error: Connection reset by peer)
10:08:12***Saving seen data "./dancer.seen"
10:22:21 Join bor_ka [0] (
10:26:04 Join n1s [0] (~n1s@rockbox/developer/n1s)
10:33:52 Join DerPapst [0] (
10:41:19 Quit scorche (Disconnected by services)
10:41:30 Join scorche` [0] (~scorche@rockbox/administrator/scorche)
10:44:32JdGordonS_a_i_n_t: pixelma: this is a random line from cabbie... %?bp<%?bc<%xd(Ba)|%xd(Bb)>|%?bl<|%xd(Bc)|%xd(Bd)|%xd(Be)|%xd(Bf)|%xd(Bg)|%xd(Bh)|%xd(Bi)|%xd(Bj)>> that could be reduced to %?bp<%xd(B,%bp)|%xd(B,%bl)>
10:45:28JdGordonwell, almost... It needs either 2 strips to do it, or an offset value as another param
10:45:34JdGordoneither way, much simpler code
10:45:58*JdGordon personally thinks that using letters as the subimage id was a rather wonky idea to begin with
10:45:59S_a_i_n_tsimpler "looking", I think the former simpler in practice though.
10:46:05S_a_i_n_tperhaps I'm oldschool.
10:46:22JdGordonsimpler looking, and easier to debug
10:46:54S_a_i_n_treally? I don;t like it because I can't look at it and say "X has Y subimages"
10:47:21JdGordonthe %Xl line wouldnt change
10:48:41S_a_i_n_tyeah...I'm not a huge fan of the idea though. But, as you said, if the "old" (original way) is still valid, I needn't care.
10:50:36S_a_i_n_tnumbers for subimages I'm not a huge fan of either, if that is allowed, why not make numbers valid identifiers?
10:50:55S_a_i_n_tthen get rid of the (artificial?) 52 identifier limiy?
11:13:02JdGordonthat limit will be gone soon, yes
11:13:24*JdGordon is having a bit pf trouble getting the new parser into the core
11:18:45 Quit Zigtown (Ping timeout: 276 seconds)
11:26:59 Join hamish_ [0] (~hamish@
11:27:21 Quit hamish_ (Client Quit)
11:40:52JdGordonbieber: I thought you were going to put each tag into a seperate object?
11:41:12JdGordonhaving a massive switch seems like a very messy idea (that is how it used to be in the core btw)
11:43:53 Quit bor_ka (Quit: CGI:IRC 0.5.9 (2006/06/06))
11:44:17 Quit leavittx_ (Read error: Connection reset by peer)
11:46:55 Quit leavittx (Read error: Connection reset by peer)
11:48:16 Quit JdGord (Quit: Bye)
11:49:28 Quit mt (Ping timeout: 265 seconds)
11:58:04 Join bor_ka [0] (
11:59:20 Join pamaury [0] (~quassel@rockbox/developer/pamaury)
12:01:33 Join leavittx_ [0] (
12:02:27 Join robin0800 [0] (
12:02:42 Join leavittx [0] (
12:03:51 Join mt [0] (~mtee@rockbox/developer/mt)
12:08:16***Saving seen data "./dancer.seen"
12:10:00 Join efyx [0] (
12:15:48 Quit antil33t (Read error: Connection reset by peer)
12:20:30 Quit GeekShadow (Ping timeout: 240 seconds)
12:23:04 Join antil33t [0] (
12:26:23 Part Szpila
12:38:54 Join giovanni [0] (~giovanni@
12:43:59giovanni@pixelma I'm trying to edit the ondavx777.tex and keymap-ondavx777.tex for the manual, am I doing it the right way?
12:55:24 Join dfkt [0] (dfkt@unaffiliated/dfkt)
13:02:40 Join funman [0] (~fun@rockbox/developer/funman)
13:05:44 Quit Barahir_ (Ping timeout: 260 seconds)
13:06:09bor_kafunman, hello!
13:07:10bor_kaI still get playback glitches on r27082
13:07:20 Join Barahir [0] (
13:07:30bor_kaand massive data corruption as well
13:07:35bor_kafuze v1
13:08:00bor_kanow testing 27073without thumb
13:08:26funmanperhaps the playback glitch comes with the data corruption? try to run fsck/chkdsk
13:08:36funmanfuze v1 is built without thumb normally
13:09:01funmanthough thumb should work too
13:09:20bor_kaI mean that if I get a playback glitch, then the internal drive is corrupted
13:09:28ranmabor_ka: Since when? Can you try if reverting 27065 or 27066 fixes it?
13:09:33funmannot necessarily but the reverse may be true
13:09:40bor_kamay be
13:09:52bor_karanma, will try in a minute or two
13:10:05ranmaSo far 4bit mode works fine for me, but I mostly just listen to mp3s, not much write activity.
13:10:35bor_kathe glitch is strange, I get a piece 1-2 sec from another song, and the time progress skips back and forth
13:10:47ranmaIf it's 27065, maybe switching back to 1bit on write would work
13:11:07bor_kaIf i rewind to the start of track, the glitch is in the same place
13:11:17bor_kabut if i reboot, it is gone...
13:11:47bor_kaI'd say that the internal fat buffer is corrupted
13:12:01bor_kaif it exists - I don't know internals
13:12:24bor_karanma, glitches are on the 320 mp3's
13:12:49*bor_ka starts building 27065
13:13:12bor_kabtw, I occasionally get data aborts
13:13:32bor_kadata abort at 30800AD8 (domain 0, fault 8) address 0xBEB1C520
13:13:45S_a_i_n_tplease don;t press enter after every few words ;)
13:14:30funmanbor_ka: data abort when doing what? do you verify the filesystem after each crash?
13:14:45 Nick fxb__ is now known as fxb (
13:15:18bor_kathis data abort I posted was when I just listened a song. Yeas, I did verify the FS after data aborts - it is clean
13:15:52bor_kaIt is corrupted only after the playback glitch.
13:15:55funmanwhich codec are you listening to ?
13:16:09bor_kaPlain mp3, 320 kbps, lame-made :)
13:16:14funman30800AD8 is in lcd_bitmap_part()
13:16:56bor_kaalso at at 3003D350 and 30800B68
13:17:19funmanwell bisect until you find a working revision
13:17:36funmanbor_ka: do you have dsp effects enabled?
13:17:41bor_kaok, like with the database freeze :)
13:17:49bor_kano, no eq or dsp
13:18:11funmancan you check in the .map file which function is at 3003D350 ? (it's tdspeed_setup() here but it could be different in your build)
13:20:57pixelmagiovanni: you probably don't need that much in the keymap-onda file, Everything that is standard for all touchscreen devices can be reused from the keymap-touchscreen.tex which you already link to. Latex will just collect everything that it finds in the linked files. You would only need actions where the power button is used
13:21:42giovanniok, I've just modified the keymap of the D2
13:22:09giovanniand what about the ondavx777.tex? there's something I've not understand yet..
13:22:39funmanI do see a lot of "FAT: Invalid FSINFO signature: 0x41615252, 0x00000000 (sector = 1)" in dmesg
13:23:44bor_kafunman, unfortunately I don't remember the build # for the 3003D350 :[
13:24:02funmanwell if you can't reproduce it then it never existed ;)
13:25:24 Join ucchan [0] (
13:27:56ucchanfunman: patch thanks. I am confiming it now. then I comment later.
13:28:34funmanucchan: thanks
13:34:12bor_kabtw, 27055 with the manually enabled USB for fuze worked yesterday Ok for me for some 4 hrs, so I think 27064 will be Ok too. Testing it now - 4min playback, no glitches
13:34:35giovannipixelma: this is part of the output of make manual
13:34:55 Quit funman (Quit: free(random());)
13:38:55ranmaucchan: PGPを使いますか?
13:38:55 Part Zagor
13:41:38 Join Zagor [0] (bjst@rockbox/developer/Zagor)
13:44:26bor_karanma: r27064, 3 songg, no glitches. So I beleieve all is Ok, since usually I got a glitch on the second song, after 5-8 min of playback.
13:44:32ucchanranma: I don't use PGP. but if it is necessary, I will install. why need PGP ?
13:45:46bor_kaGoing to try 27086 with the 27065 commit #if 0' ed
13:46:00ranmaucchan: I was going to ask you if you are interested in keysigning.
13:47:44ucchanranma: OK. please give time to me until installing it.
13:49:06*gevaerts thinks that ranma and ucchan might move this to private messages
13:49:09ranmaNo need to install it :)
13:51:11ucchanranma: Please give mail to me if you need tome private comments (e-mail:
13:51:59ucchanranma: of cource, It is OK in Japanese.
13:53:52ucchanfunman: I think tv.diff is good because text viewer returns to the original conditions.
13:58:50 Join GeekShadow [0] (~Antoine@reactos/tester/GeekShadow)
14:03:37ucchanfunman: In FS #11399, As for the cause of this problem, is the alignment of tv_preference not correct ? i.e., enum foo is not int but char.
14:03:41 Quit einhirn (Read error: Connection reset by peer)
14:04:05 Join einhirn [0] (
14:05:26 Join teru [0] (
14:08:19***Saving seen data "./dancer.seen"
14:09:22 Part Zagor
14:11:44 Quit bor_ka (Quit: CGI:IRC (Ping timeout))
14:14:00 Join Zagor_ [0] (
14:14:07 Quit Zagor_ (Changing host)
14:14:07 Join Zagor_ [0] (bjst@rockbox/developer/Zagor)
14:14:44ucchanfunman: your patch (tv.diff) occurs some warnings. It is necessary to correct a little.
14:16:03 Nick Zagor_ is now known as Zagor (bjst@rockbox/developer/Zagor)
14:17:02JdGordonucchan: Hi, is there any reason the new text viewer isnt using the multi screen api at all?
14:21:06ucchanJdGordon: It is because I did not know the existence of multiscreen api.
14:21:21 Quit einhirn (Quit: Miranda IM! Smaller, Faster, Easier.
14:21:37JdGordonit is used so you can use both screens easily
14:22:43 Join einhirn [0] (
14:22:44ucchanJdGordon: Ok I correct text viewer using multi screen api.
14:34:21 Quit robin0800 (Remote host closed the connection)
14:41:59 Join funman [0] (~fun@rockbox/developer/funman)
14:43:11 Join binaryhermit [0] (
14:43:47 Join robin0800 [0] (
14:44:07 Quit robin0800 (Remote host closed the connection)
14:48:49pixelmagiovanni: is this with both of your new files in place? They look correctly at first glance but it seems Latex couldn't find the info from the keymap files, I don't have time to look at it more closely now though
14:48:58 Join robin0800 [0] (
14:50:16giovannipixelma: now I'm handling with the various main.tex to add special feature for the onda's, it will be hard to implement also the new touch screen mode...
14:50:40giovanniand my english just isn't good enough...
14:51:03pixelmabluebroth3r: does the newer tex installs that show errors after ! not show the file (path) where the error occured)
14:51:27funmanucchan: alignement with int is ok because of rb->set_int / rb->set_option use int
14:52:07funmanucchan: the cause of the problem was not text_viewer; it was in rockbox core
14:52:29funmanI suppose we didn't see the problem you found with enums because we didn't try to change preferences
14:53:35pixelmagiovanni: what special features are there on the Ondas?
14:53:52pixelmaand which new touch screen mode?
14:53:55giovanniI mean the way to boot the official firmware
14:54:44giovanniand I'm also taking screenshots with the sim
14:55:53funmanucchan: i have uploaded new patch which fix warnings
14:55:53 Quit robin0800 (Remote host closed the connection)
14:56:22funmanranma: my fuzev1 panic'd while running battery bench (screen was off), and i reproduced it on clipv1
14:57:24funmansd_select_bank(-1) panics with MCI_TX_UNDERRUN after 10 loops, the -1 means transfers failed at some point so the card is reinited
14:58:18Luca_Suhm my fuzev2 just froze while tuning FM Radio - the radio keeps playing but everything else is locked (backlight on, buttonlight on, no response to buttons) - it's the second time it does this
14:58:39funmanLuca_S: revision?
14:58:41Luca_Si was moving forward with station in Scan mode
15:00:37Luca_Sheh... custom one based on r27078 + cpu frequency switching + audio on pllb + volume increase
15:00:41Luca_Si'll try with plain svn
15:03:02ucchanfunman: thanks new patch. I check later.
15:03:20Luca_Sfunny thing I was looking for some background noise since i have an headache and my coworkers are just too chatty :D there o
15:03:35Luca_Sa white noise generator plugin could turn useful
15:04:02funmanranma: i'm checking with r27066 reverted on fuzev1 (playing mp3 - running battery_bench to trigger some writes)
15:04:24funmanLuca_S: there is a forum thread about it i think
15:05:16funmanthe filesystems on my clipv1/fuzev1 are severely corrupted: i have to format them again
15:05:58ucchanfunman: I think that struct tv_preferences { ... } __attribute((aligned(4));. Is not the problem solved ?
15:06:59 Quit efyx (Remote host closed the connection)
15:07:00funmanucchan: no because the alignement problem is inside the struct, if you use enum they will use the smallest storage type possible (it changed with EABI)
15:07:10 Join efyx [0] (
15:07:26funmanif enum has only 2 values it will be char, if it has between 1<<8 and 1<<16 values it will be short
15:08:59Luca_Suhm, cannot reproduce with svn for now, but I'm noticing something strange with the keymap
15:09:31funmanfm keymap changed
15:09:45Luca_Sif I long press the "bottom button (menu symbol)" it both advances FM station and goes back to the menu
15:09:59Luca_SI don't mind keymap variations, but this does two actions o_O
15:10:51funmanhm indeed
15:12:41 Join komputes [0] (~komputes@ubuntu/member/komputes)
15:12:44ucchanfunman: but error is "data abort". then does not "data abort" occur when prefs->foo = bar is executed? (excepts calls set_option()).
15:13:18funmanucchan: data abort is fixed by r27045 in apps/gui/statusbar.c
15:13:52funmanLuca_S: shouldn't stations advance with left/right ?
15:14:43ucchanfunman: Oh! I don't confirm it. then Does close FS #11399 ?
15:15:00funmanranma: reverting r27066 alone doesn't fix it. now trying with r27065 and r27066 reverted
15:15:09funmanucchan: yes
15:15:41Luca_Sfunman: left / right buttons = FM+/- && left / right scroll = Vol+/-
15:15:57funmanso what is up ? pause?
15:16:19Luca_Sshort press = pause, long press = toggle scan/preset mode
15:16:22funmanhm ok
15:16:46 Join bor_ka [0] (
15:16:53funmanwhat does give ?
15:18:33ucchanfunman: thanks. and the problem of set_option() is corrcted by my patch (in FS #11399). then text viewer's abend bug is clear!
15:19:18funmanucchan: i'm not sure your patch to change set_option is ok
15:19:20Luca_Scompiling (cygwin in vm.. gonna take a while)
15:19:35funmanucchan: sizeof(char) is always 1 but sizeof(short) / sizeof(int) could change
15:20:08bor_kafunman, ranma, I've tested 27064 - no glitches or data corruption. Also I have #if 0'ed the 4bit part commited in 27065, in the current HEAD - all is Ok too
15:21:13funmanbor_ka: thanks, i'm testing svn with 4 bit #if 0 too
15:21:17ucchanfunman: my patch is corrected this problem. i.e., my patch is optiontype add CHAR and SHORT.
15:22:30funmanucchan: enum optiontype { should be BOOL = sizeof(bool); CHAR = sizeof(char); SHORT = sizeof(short); INT = sizeof(int); }
15:22:49funmanbut if 2 types have the same size, the enum will have 2 times the same value, i think it's not possible
15:24:20funmanucchan: other option is to use enum optiontype { BOOL, CHAR, SHORT, INT }; and not use sizeof() in argument, but the enum value directly
15:24:28 Quit Bowbles (Ping timeout: 276 seconds)
15:25:39 Quit yosafbridge (Quit: Coyote finally caught me)
15:28:15ucchanfunman: yes. my patch (see rb_text_viewer_abend.patch in FS #11399) is written by {BOOL=0, CHAR=1, SHORT=2, INT=4}. it is same about funman's idea,
15:29:14funmanucchan: using 0, 1, 2, 4 and then using sizeof() is not good
15:29:57 Quit GeekShadow (Ping timeout: 248 seconds)
15:31:31ucchanI agree should use sizeof.
15:31:49 Join GeekShadow [0] (~Antoine@reactos/tester/GeekShadow)
15:31:50funmannothing guarantess that sizeof(short) != sizeof(int)
15:32:02 Join yosafbridge [0] (
15:33:09CIA-6New commit by funman (r27087): sd-as3525.c: disable wide bus, it corrupts writes
15:34:50CIA-6r27087 build result: All green
15:35:12ucchanIt is correct for a lot of processors. but the standard of C, maybe sizeof(short) == sizeof(int) is admitted.
15:35:43funmanso i think it's better to avoid it and use set_option for variable with known storage size
15:39:50Luca_Suhm funman now long press down means FM+, is it what you intended?
15:40:13funmannot sure
15:40:17Luca_Swhile short press goes to the menu continuing playing
15:40:26Luca_Sto the *main menu
15:40:27CIA-6New commit by funman (r27088): sd-as3525.c: get rid of MCI_ARG ...
15:40:43funmanideally long down should do nothing but it's part of another keymap
15:40:46ucchanAnd #if sizeof(short) != sizeof(int) ... #endif is wrong usage. it is difficult.
15:41:07funmanucchan: i think using int for all options is simpler, no ?
15:41:48 Quit bieber (Ping timeout: 264 seconds)
15:42:03 Join bieber [0] (
15:42:07CIA-6r27088 build result: All green
15:42:09funmanLuca_S: ?
15:42:24Luca_Spersonally i'd expect: short home = main menu while playing - long home = main menu stopping playback, short down = preset list, long down = nothing
15:43:07funmani prefer down for just leaving the menu
15:43:18funmanHOME is quit for just everything
15:44:00ucchanyes I think only tv_preferences are written char or int without using enum.
15:44:07 Quit mc2739 (Ping timeout: 240 seconds)
15:44:20funmanwe could use char because the settings file format use bytes
15:44:22Luca_S(AFAIK, long home is quit in the plugins - with my proposal this would be respected - but I don't really care; now compiling your latest diff)
15:44:39funmanucchan: you want to commit the last patch on FS #11399 or should I do it myself?
15:45:00funmanLuca_S: some plugins use short HOME (although the plugins could be change to detect BUTTON_REPEAT)
15:45:53ucchanOK enum => char. When a better method will be found in the future, we'll correct it.
15:46:21 Join mc2739 [0] (~mc2739@rockbox/developer/mc2739)
15:46:37 Part Zagor
15:48:25ucchanyou commit please.
15:50:15 Join t0rc [0] (~t0rc@unaffiliated/t0rc/x-5233201)
15:52:25Luca_Sok funman, your last diff is ok, long down == short down == main menu while continuing fm playback
15:53:17 Join kachna [0] (
15:53:26ucchanfunman: When the problem is found after you commit, I correct the part and commit.
15:53:28funmani'll check what the manual says and commit it
15:54:24 Quit mitk (Quit: Leaving)
15:56:10CIA-6New commit by funman (r27089): FS #11399 by me: fix r26998 for text_viewer ...
15:56:25ucchanfunman thanks.
15:57:42CIA-6r27089 build result: All green
15:58:27ranmafunman: Maybe this will work:
15:58:38ucchanfunman: check the text viewer tomorrow. I will go to bed. bye.
15:58:46funmanucchan: good night!
15:58:53ucchangood night.
15:58:57 Quit ucchan (Quit: Leaving...)
15:59:05CIA-6New commit by funman (r27090): fuze: make sure down exits the FM screen, and nothing else
15:59:10ranmafunman: So it's reproducible with battery_bench? I'll try that.
15:59:34funmanranma: only because it triggers repeated writes i think
16:00:01funmanplayback stops quite fast because filesystem isn't accessible (if you have short songs of course)
16:00:12funmans/songs/audio buffer/ so it won't be a problem on c200v2 ;)
16:00:14ranmaUSB writes work fine though, I've updated rockbox from over rb usb a few times with 4bit enabled :)
16:00:33funmanranma: perhaps it's worth checking if this affects only internal or µSD too
16:00:47CIA-6r27090 build result: All green
16:00:47funmanbtw i don't understand what the ACMD42 (CLR_DETECT) does exactly
16:01:26ranmafunman: Theres a pullup on DATA3, that can be used to detect cards I think
16:02:10funmanis it linked to wide bus?
16:02:19ranmafunman: Define 'quite fast' :)
16:02:28ranmaWell, DATA3 is only used in wide bus mode
16:02:39funmanah right
16:03:31ranmaFrom the Simplified Physical Layer Spec: "SET_CLR_CARD_
16:03:41 Quit B4gder (Quit: It is time to say moo)
16:03:49ranma"Connect[1]/Disconnect[0] the 50 KOhm pull-up resistor on CD/DAT3 (pin 1) of the card."
16:03:58funmanyep i didn't think a pull-up would prevent using the line
16:04:37bor_kawell, 27086 without the 4bit bus, got "prefetch_abort at 736E7546 FSR 0xFF (domain 15, fault 15)" when updating the database, no playback or button pressing...
16:04:38ranmaI doesn't prevent it, but it might skew timings
16:04:45funmani'm trying wide bus enabled only for µSD now
16:04:57funmanplaying from µSD + battery_bench logging on µSD
16:05:45funmanbor_ka: sounds bad, i didn't try database update recently
16:06:34bor_kaI have a map for this build, btw
16:06:56funman736E7546 is not in a valid region so it wont' help
16:07:01bor_kalet's see if it is reproducible
16:07:12bor_kaoh, stray pointer.. ?
16:08:22***Saving seen data "./dancer.seen"
16:10:17funmanperhaps we could save the lr from supervisor mode and print it
16:10:25funmanif you can reproduce it i'll give you a patch
16:11:25funmanwide bus seems to be stable on µSD
16:12:04bor_kagot it, in ~ 1min after the start of database update
16:12:17bor_kathe same address
16:15:42funmanoops sorry it's buggy
16:15:57 Join kugel [0] (~kugel@rockbox/developer/kugel)
16:17:27funman (edited)
16:17:32 Quit teru (Quit: Quit)
16:23:18funmanranma: i don't see the OF using acmd42 on clipv1
16:25:13 Join _spachy [0] (~spacaru@
16:27:03funmanranma: your patch gives START_BIT_ERROR
16:27:19 Part _spachy
16:28:05bor_kafunman, now it is data abort at 3010366C FSR 0x8 (domain 0, fault 8) address 0xDEADBEEF LR: 0x00049757
16:28:58 Join Bowbles [0] (
16:29:28funmanwhat's 3010366C in your .map ?
16:29:55bor_ka .bss 0x300ffe70 0x4b88 /root/rock/obuild-27086-thumb-no4/apps/tagcache.o .bss 0x301049f8 0x0 /root/rock/obuild-27086-thumb-no4/apps/keymaps/keymap-fuze.o
16:30:01 Join Szpila [0] (
16:31:06funmantagcache.o bss, could be a problematic file ?
16:32:17bor_katheoretically - yes, since there were data corruptions, some music files could be garbled
16:34:02*funman summons FlynDice
16:34:35bor_kaI have run chkdsk, so there is no FS corruption
16:34:56bor_kaWill try to find offending file...
16:35:47funmanif you have the files somewhere else run md5sum to see if the content have changed (the filesystem could be valid but the file corrupted)
16:36:02funmanand keep well the faulty file (if there is one)
16:36:59bor_kayes, I understand the file corruption - but most files are converted from flacs and deleted right after copying to fuze
16:37:34bor_kaI mean that resulting mp3s are moved to the fuze and deleted from the PC
16:38:06funmankugel: ping
16:38:23kugelfunman: pong
16:38:56bor_kaI have started database update without the mSD card - all seems to go well. Strange, since there were no FS corruption on the mSD during the 4bit bus troubles
16:39:37bor_kawow, player had locked, just like in the old pre-3.6 bug
16:39:44funmanwhen running the sim in valgrind the console is flooded with ALSA underruns, any idea on how to avoid those ? (i didnt look at the code but i suppose you know it by heart now :P
16:40:28kugelfunman: no, not from SDL
16:40:33 Part Szpila
16:40:58funmando we use SDL audio output or directly alsa ?
16:41:29 Join bmbl [0] (~Miranda@unaffiliated/bmbl)
16:41:42kugelSDL uses an alsa backend (libasound2), but you cannot do much about it. in alsa directly I think you can configure threshold and different streaming mechanisms
16:41:51kugelfunman: the sim uses SDL
16:42:14funmanrunning vlc with SDL audio output gives teh same thing :/
16:43:16 Join Szpila [0] (
16:45:03funmanranma: 40 mins playback+battery_bench on µSD seems to work fine
16:46:28bor_kafunman, trying db update without the mSD card: once player locked, once gave "Undefined instruction at 00C1A444". At the same file count 910. The same subdirectory
16:46:51funmangood thing if you can pinpoint it to a single file
16:51:26 Join anewuser [0] (anewuser@unaffiliated/anewuser)
16:51:37bor_kahappy me, that USB is already working in the RB ;)
16:54:32funmanranma: iirc i tried test_disk with wide bus and it was fine : perhaps a problem when switching back and forth reading and writing ?
17:10:43 Join kaylinsigswort [0] (
17:10:45 Quit antil33t (Read error: Connection reset by peer)
17:10:51 Join antil33t [0] (
17:15:38bor_kafunman, I have deleted the database and copied it from the backup, all is Ok now. I think that it was not a song, that was corrupt, but the database itself
17:16:05 Quit antil33t ()
17:16:16funmanhm ok
17:16:28funmanlet's pretend this bug never existed, ok for you ? O:-)
17:16:48bor_kaof course Ok, I'm in software business too :-D
17:18:22 Quit bor_ka (Quit: CGI:IRC (EOF))
17:23:46 Quit GeekShadow (Quit: The cake is a lie !)
17:24:07 Quit funman (Quit: free(random());)
17:25:10 Join AzureWurk [0] (
17:27:55giovannipixelma: I've got taken a lot of screenshots (240x400x16) for the onda vx777 manual I'm working on, I need some help to how to upload to SVN
17:28:06 Join Jaykay [0] (
17:29:03 Join MethoS- [0] (~clemens@
17:40:58 Quit M3DLG (Ping timeout: 276 seconds)
17:47:19 Join r0b- [0] (
17:48:35 Part Szpila
17:52:08 Join Jerom [0] (~jerome@
17:53:02 Join Szpila [0] (
17:55:33 Quit pjm0616 (Ping timeout: 240 seconds)
17:55:48 Join pjm0616 [0] (~user@
17:59:12 Quit einhirn (Quit: Miranda IM! Smaller, Faster, Easier.
17:59:18 Join GeekShadow [0] (~Antoine@reactos/tester/GeekShadow)
18:00:47 Part Szpila
18:01:06 Join Szpila [0] (
18:01:23 Quit petur (Quit: *plop*)
18:06:17 Join bertrik [0] (~bertrik@rockbox/developer/bertrik)
18:08:23***Saving seen data "./dancer.seen"
18:12:16 Join Xerion [0] (
18:14:02S_a_i_n_tis there a note on the wiki for cywin saying that there's no need to pis around with the PATH anymore?
18:14:18*S_a_i_n_t can't seem to reach it in his browser(s) at the moment.
18:15:17S_a_i_n_t "/usr/local/bin/" should be in the PATH by default, editing will probably only cause confusion.
18:16:29 Quit Kitar|st (Ping timeout: 260 seconds)
18:16:59krazykitdoesn't place the compilers into their own subdirectories?
18:18:31 Join Stummi_ [0] (
18:19:24 Quit Barahir (Ping timeout: 260 seconds)
18:19:55 Join Barahir [0] (
18:21:04S_a_i_n_tbut, default PATH for CygWin includes "/usr/local/bin:/usr/bin:/bin:" so, its all covered.
18:21:31Tornekrazykit: no, it installs to /usr/local now
18:21:46Tornethe arch subdirectories are part of the gcc install, they don't need to be in the path
18:21:59Torne(not everything in the toolchain is prefixed with the arch on a per-file basis)
18:22:14 Join Kitar|st [0] (
18:22:19 Quit Stummi (Ping timeout: 260 seconds)
18:23:49S_a_i_n_tTorne: Its "usr/local/bin" no?
18:24:01TorneThe binaries, yes
18:24:30Tornea toolchain includes a lot more stuff than just binaries, the prefix for the install is /usr/local, whereas it used to be /wherever/arm-elf or similar
18:25:10S_a_i_n_taha...well, you've just confirmed my theory about there being no need to edit the path anymore.
18:25:27Torne(which meant you would get a /wherever/arm-elf/arm-elf directory - you now get a /usr/local/arm-elf directory, but that's not where your path points)
18:25:52S_a_i_n_tOn a hunch I deleted my custom PATH line, and echo'd $PATH, and there was no change.
18:26:06Torneindeed, you don't need to edit the path, but the compilers are not installed into tir own directories; that's why you don't need to edit it :)
18:27:37S_a_i_n_tIIRC it ( still pops up a message saying "Add X to $PATH" which I thought might be confusing.
18:27:57Tornewell, not all unixes have /usr/local/bin in the path by default
18:28:10Torneand you can still install it somewhere else if you choose to, it's just the default install location that changed
18:28:23Torneand it doesn't *matter* if you redundantly add a directory to the path that's already there..
18:28:34S_a_i_n_tno, this is true.
18:28:50 Quit kish (Remote host closed the connection)
18:28:55S_a_i_n_tyou'd just see it in $PATH twice if you ever chose to check it.
18:29:12 Join kish [0] (~o2@unaffiliated/spice)
18:29:19 Quit pamaury (Remote host closed the connection)
18:30:05 Join evilnick_B [0] (0c140464@rockbox/staff/evilnick)
18:30:06bluebroth3rgiovanni: you can't "upload" stuff to svn. Write access for svn is invite only so you have to go through the usual tracker first: create a patch for your changes (see, put all new images in an archive, preferrably keeping the folder structure), then submit it as patch to Flyspray.
18:30:56giovannibluebroth3r: ok, thanks. I will do this maybe tomorrow :D
18:33:42bluebroth3rpixelma: no, it doesn't show the filename in the same line. However, the processed files are grouped like (./some/file.tex <various multiline output>)
18:34:02bluebroth3rso if you get an error you need to scroll back to find the filename.
18:36:19 Join M3DLG [0] (
18:47:36 Quit hebz0rl (Ping timeout: 260 seconds)
18:48:12 Nick fxb is now known as fxb__ (
18:54:23 Join domonoky [0] (~Domonoky@rockbox/developer/domonoky)
18:56:16 Join pamaury [0] (~quassel@rockbox/developer/pamaury)
18:57:47 Quit Luca_S (Quit: CGI:IRC)
19:04:31bertrikHas anyone seen Boris Gjenero (dreamlayers) lately?
19:04:34 Join u42p [0] (
19:04:47u42pi love you guys. can' wait to put rockbox on my sansa clip+
19:06:11 Quit M3DLG (Read error: Connection reset by peer)
19:06:25 Join M3DLG [0] (
19:06:59u42pcurrently it is still pretty dangerous to put it on there, correct?
19:07:00 Quit kaylinsigswort (Read error: Connection reset by peer)
19:07:09u42pi dont want a 3 day wait for the battery to die ;)
19:07:57bertriku42p, works fine for me here, I haven't run into the 3-day wait problem yet
19:08:50JeromI did ^^
19:12:22JeromBut still it doesn't seems to be frequent
19:14:49 Quit M3DLG (Ping timeout: 264 seconds)
19:20:30 Quit DerPapst (Quit: Leaving.)
19:25:43 Join funman [0] (~fun@rockbox/developer/funman)
19:27:18 Join Horscht [0] (~Horscht2@xbmc/user/horscht)
19:28:13amiconnbertrik: After using the hi-q patch for quite a while today my impression is that it's definitely an improvement
19:28:39amiconnThze question remains which of the two changes helps most. My IEMs are 16 ohms so it could be both
19:29:55 Quit kugel (Remote host closed the connection)
19:30:07bertrikamiconn, nice to hear
19:32:14 Join M3DLG [0] (~M3DLG@
19:37:59S_a_i_n_tu42p: Which device are you wanting to install on?
19:38:06u42psansa clip+
19:38:16S_a_i_n_tIf you mentioned that already, I must have missed it somewhere
19:38:37u42pi did :)
19:38:50S_a_i_n_tAh, that is *fairly* has certainly come a long way in the recent past.
19:39:09bertrikamiconn, maybe dfkt can do another test with 16 ohm earphones.
19:39:39 Quit stoffel (Remote host closed the connection)
19:44:00funmani tried a series of 10 plugging/unplugging charger on clip+: charger presence and charging status is always correct
19:49:54 Quit flydutch (Quit: /* empty */)
19:55:26 Quit M3DLG (Quit: RAGE QUIT)
19:57:30dfktbertrik, amiconn - repeating the rmaa test with the two separate alterations, with 16 ohm phones?
19:57:59bertrikdfkt, yes, please I you will
19:58:45bertrikI found the hq comparisons a bit weird, because it basically showed no difference at all, while your earlier measurements did seem to show an improvement with the hi-q patch.
19:59:21dfktsure, i'll do it in an hour or two (my pasta puttanesca needs tending ;)
20:00:22CIA-6New commit by funman (r27091): Clip+ manual: use clip screenshots
20:02:36CIA-6r27091 build result: All green
20:07:11bertrikI guess we'll also need a volunteer to do a battery benchmark with these hi-q patches
20:07:40 Join markkoski [0] (~chatzilla@
20:08:26***Saving seen data "./dancer.seen"
20:10:07markkoskiI have a sansa c200V2, which is unstable. I was wondering if the first gens are able to run mpeg and rockboy?
20:12:59S_a_i_n_tmarkkoski: Check the pluginindex on the wiki
20:13:04S_a_i_n_tthat should tell you.
20:16:34saratogathey'll probably run both but hte screen is so tiny its probably not useful
20:16:57S_a_i_n_twiki says it runs on both
20:17:38markkoskiso aside from adjusting the plugin buffer is there anything else that would be stopping either plugin from running on a V2?
20:18:34gevaertsthe c200v2 has a lot less RAM than the v1
20:18:35bertrikthe c200 (at least mine) has a very slow, rather bad quality LCD, so mpeg is probably not going to look good
20:18:56S_a_i_n_tROM size can make a difference for rockboy IIUC, depending on available RAM of the DAP
20:19:48S_a_i_n_tah, and gevaerts just mentioned the RAM so I'd say that is a factor
20:21:30markkoskiwell, I was just wondering since I have two rockbox binaries at the moment; one with the default settings for music and a second one with a large plugin buffer at this moment exclusively for espeak
20:21:38CIA-6New commit by alle (r27092): Do not use fancy apostrophes in the verbatim code
20:23:17CIA-6r27092 build result: All green
20:27:13 Quit markkoski (Remote host closed the connection)
20:27:32CIA-6New commit by funman (r27093): Build doom on clipv2 and clip+
20:28:27funmanbeat that, Sandisk!
20:28:35 Join petur [0] (~petur@rockbox/developer/petur)
20:29:07CIA-6r27093 build result: All green
20:30:07saratoganoooo now compiling is going to be slow
20:30:55funmanthat's right
20:31:12 Join SamsungTVnoDispl [0] (
20:31:14funmanyou can still run make bin codecs to skip plugins though?
20:31:45S_a_i_n_tIf sandisk was in the market for "beating" anyone or thing, they never would have released the Clip(s)
20:33:41u42phow does doom look like on the clip display?
20:34:32funmanhm that's not possibly true, this is the image before rendering it on screen
20:34:42u42pbut still nice
20:39:29amiconnfunman: This is the image in the internal representation used by the greylib. i.e. already reduced to 129 shades
20:40:31amiconnIn fact for Clip the screendump function reduces it further to 128 shades, in order to fit both differently coloured parts of the screen into a 256 colour palette
20:42:32amiconnRegarding the flicker on the clips, I have an idea that might actually help to reduce it, but isn't easy to implement
20:43:47funmanamiconn: btw have you noticed that there are different shades of blue/yellow? the scrollbar is darker at the intersection with the line selector
20:44:24funmani noticed it while looking at the screen in a dark room, but it's also visible with some ambient light if you know what you're looking for
20:44:33amiconnYou're right... didn't notce that before
20:45:01amiconnI think it's a load effect
20:45:53bertrikI think it basically lights one row at a time
20:46:09amiconnThe selector has many bright pixels, which means a high load for the display's power supply. This reduces voltage, effectively reducing brightness
20:46:32amiconnYes, the display scans from top to bottom
20:46:57amiconnIt's the same with LCDs. In fact I know this effect from the 1st/2nd Gen iPods
20:48:17bieberJdGordon, anyone else who's interested: For handling tags in the renderer, my thought at is that I'll use a set of nested switch statements to identify any tags that do things (loading images or fonts, altering subline times, etc.), and the defaults for those switch statements will just go to a generic function that adds a line of text to the current viewport with the value of the tag, which I'll have a separate handler for
20:48:37*amiconn tweaked the LCD controller's voltage converter setting in order to reduce this row/column crosstalk on 1st/2nd Gen
20:49:08amiconnThat's the reason why we cannot use the OF's contrast settings on those - they would be wrong for the changed voltage converter settings
20:49:17bieberMaking a separate class for each tag would lead to an awful lot of boilerplate in an awful lot of files, and I'd need a pretty hefty switch statement anyways to decide which specific tag object to construct, so I think it makes the most sense just to do the work for the tags while I'm identifying them, instead of in separate classes
20:49:32funmanamiconn: btw you have an ipod4g ?
20:49:46amiconndomonoky has one
20:50:09funmandomonoky: did you try an eabi build on ipod4g yet?
20:50:25amiconnRegarding the flicker: While I often describe the way the greylib works as "bresenham in time", it can also be looked at as a delta-sigma modulator (for each pixel)
20:50:54amiconnMy idea is to try a 2nd order delta-sigma. This may or may not help....
20:51:29domonokyfunman: nope, didnt found time to try that.
20:53:06amiconnfunman: Btw, in the doom screenshot it looks like the bottom pixel rows contain garbage
20:53:15amiconnIs there something wrong with the scaling?
20:54:16 Quit SamsungTVnoDispl ()
20:55:10funmanamiconn: nop that's "text"
20:56:26Jerompamaury: had you any time for the usb driver ?
20:56:38 Join kaylinsigsworth [0] (
20:57:50 Join Luca_S [0] (
20:58:50 Nick kish is now known as kish|gha5-0ger (~o2@unaffiliated/spice)
20:59:37 Quit krazykit (Quit: wark)
21:03:27 Nick kish|gha5-0ger is now known as kish|gha8-0ger (~o2@unaffiliated/spice)
21:04:56 Nick kish|gha8-0ger is now known as kish|gha10-4ger (~o2@unaffiliated/spice)
21:07:04 Nick kish|gha10-4ger is now known as kish (~o2@unaffiliated/spice)
21:09:47 Quit jfc (Read error: Connection reset by peer)
21:10:41 Join jfc [0] (
21:14:18 Quit t0rc (Remote host closed the connection)
21:24:55 Quit Topy44 (Ping timeout: 260 seconds)
21:25:24 Part Szpila
21:26:16 Join Topy44 [0] (
21:26:53 Join Szpila [0] (
21:36:14pamauryJerom: not much, I worked on it a bit yesterday but I have a problem with controllers transfers. It works randomly. I think this is because at some point a setup packet arrive before we received the completion of the previous one and the driver does crap. I feel it is related to the order in which the core ack transfers but amusingly, TheSeven wants this order for his driver to work :)
21:37:35Jerompamaury: TheSeven wrote which driver ?
21:38:10pamaurythe one of the nano2g which has the same controller but at an older revision
21:39:52JeromOk this is the driver I'm copy/pasting most of the code ^^
21:41:24JeromI will see if I can get anything to work
21:41:32pamauryHonestly there are some strange things in his driver. And it doesn't support transfer splitting which limit the size of the maximum transfer size. And it doesn't support iso.
21:41:35CIA-6New commit by bieber (r27094): Theme Editor: Added rendering support for some more tags
21:41:47pamauryI will commit something soon anyway because I have a local modification
21:43:07CIA-6r27094 build result: All green
21:49:40funmanbertrik: any hint on why the wakeup RTC sometimes fail to power off on clip+ ?
21:50:11bertrikI have not looked at how the RTC wakeup thing works
21:50:22bertrikMaybe there's a pending alarm that needs to be cleared first
21:50:42funmanrtc alarm is off if the device was powered off normally
21:50:53funman(according to the datasheet)
21:51:06bertrikThis is in the as3543 datasheet, right?
21:51:14funmanyes, register 0x22 iirc
21:52:02 Quit MethoS- (Remote host closed the connection)
21:52:09 Join Virusless [0] (
21:52:18 Quit Virusless (Client Quit)
21:52:41 Join Virusless [0] (
21:53:52Viruslessi have a question about RockBox on my Ipod Nano 1g (4gb)
21:54:29Viruslessi have looked in "FAQ" and couldnt find the answer to the question.
21:54:41*bluebroth3r goes for the crystal ball to figure the question
21:55:12JeromVirusless: ask away
21:55:34Viruslesson my ipod, i have usb modes on my ipod, to use it as a media keypad and other HID's..but when i plug my ipod in (with them enabled), it reboots.
21:55:38Viruslessany idea why?
21:55:55bluebroth3rlet me guess ... you have Rockbox 3.6 installed
21:56:17Viruslessmy bad, iPod Mini...and yes, RockBox 3.6
21:56:29bluebroth3rRockbox releases reboot into the Apple firmware for USB connection. This is due to missing charging support via USB.
21:57:10bluebroth3rThat could end up with the Ipod draining more than its charging, eventually ending up with a discharged Ipod and USB connection failure because of that (which then could result in filesystem problems)
21:57:47bluebroth3rcurrent builds have charging support, and current builds don't reboot into the Apple firmware for USB connections since long.
21:58:13bluebroth3rso either wait for the next Rockbox release (which is pretty likely to use Rockbox USB mode) or use a current build.
21:58:54Viruslesswould the current builds work on the 1g mini? if so, how would i download them? (sounds like a noobish question, but i have had the utility find the latest software)
21:59:10bluebroth3rRockbox Utility told you how to do that :)
21:59:44bluebroth3rif you use the automatic installation it asks you if you really want to install the latest release. There's this tiny unimportant tab "Installation" besides "Quick Start".
21:59:45bertrikfunman, what is the problem with RTC/poweroff exactly?
22:00:11Viruslessim having a look now.
22:00:19bluebroth3rIf you check that out you can invoke a separate installation window from that tab, and this allows you choosing if you want to use a current build or the latest release.
22:00:26funmansometimes (1 times out of 10?), when powering off after having enabled the alarm, the player never shutdown
22:01:00ViruslessFunMan: I dont use the ipod for an alarm.
22:01:47Viruslessim having the utility install the "current build"
22:02:23Jerompamaury: What does the usb core do after a usb_core_control_request(&ep0_setup_pkt); ? does it send a packet or something ?
22:03:07bertrikfunman, does the problem of not powering off only occur in combination with the wakeup-alarm, or also in other cases? (I guess only in combination with wake-up because I haven't seen it yet)
22:03:38ViruslessThank you for the help, bluebroth3r. :) and thank you to everyone who contributed to the RockBox Project. I will recomend it to everybody I know with a compatible media player.
22:04:39funmanbertrik: alarm only
22:04:40pamauryJerom: usb_core_control_request queues the setup packet the usb event queue. what the usb sees it, it acts depending on the request. Either is ack or transfer some data and then ack. Have a look in usb_core.c
22:04:53funmanbertrik: because when the RTC is enabled, the power off procedure is not the same
22:06:45 Quit Virusless (Quit: CGI:IRC (EOF))
22:08:27***Saving seen data "./dancer.seen"
22:11:02 Quit Zarggg (Ping timeout: 248 seconds)
22:12:50pamauryGrr, this usb analyzer would be perfect if the software that comes with it actually worked properly !
22:13:08 Join Zarggg [0] (~zarggg@2001:0:4137:9e76:0:fbf7:beb1:ba3d)
22:13:49JeromIsn't wireshark good enough ?
22:13:49 Quit u42p (Quit: Leaving)
22:15:10 Join kaylinsigswort [0] (
22:15:18CIA-6New commit by bluebrother (r27095): Detect when building for 64bit Linux and adjust tarball name.
22:15:21 Nick kaylinsigswort is now known as Zigtown (
22:15:50 Quit Zigtown (Remote host closed the connection)
22:15:50 Quit kaylinsigsworth (Ping timeout: 248 seconds)
22:16:58 Join kaylinsigswort [0] (
22:17:06CIA-6r27095 build result: All green
22:17:26 Join notlistening [0] (
22:17:27pamauryOkay, I reliable got to the set address but something doesn't work with it
22:17:57notlisteningdomonoky, hows the printer?
22:18:34CIA-6New commit by bieber (r27096): Theme Editor: Added dummy font class and implemented font load tag
22:18:52bertrikfunman, do we really have to use the heartbeat watchdog to power down, or would using the PWR_HOLD method work too (and have wake up still work)?
22:20:08 Part Szpila
22:20:10CIA-6r27096 build result: All green
22:20:56funmanbertrik: see condition #3 of Table 29. at page 43 of as3543 datasheet
22:21:27funmancondition #1 says it clears wakeup
22:21:45 Quit n1s (Quit: Lmnar)
22:21:56bertrikfunman, the datasheet says to first disable the heartbeat, then enable the heartbeat watchdog, we do it the other way around it seems
22:22:27 Quit CGL (Ping timeout: 272 seconds)
22:23:46funmani changed the order in some commit, i don't think it's important
22:24:23bertrikright, we do need the heartbeat powerdown
22:24:30 Quit Zarggg (Ping timeout: 240 seconds)
22:24:46 Join playfirst [0] (
22:24:50bertrikI'll experiment with it a bit (and try using the correct order)
22:25:02 Join Zarggg [0] (
22:26:46funmanbertrik: i changed it in r26247 (thinking that it solved the problem)
22:26:49bertrikmeh, this wakeup time thing will take a bit of time to test
22:29:22bertrikCould it be that it locked up because of a blocking call in ascodec_write (with interrupts disabled) ?
22:29:23funmanwakeup itself works
22:29:39funmanafaik i2c should work fine with interrupts disabled
22:29:53funmanit just busy loops for the busy bit if disabled
22:30:06bertrikFor example, if some mutex is locked, will it get unlocked when interrupts are disabled?
22:31:03funmanmutex are only used in adc
22:36:04bertrikwow, this ascodec i2c stuff has become quite complicated ...
22:37:35bertrikI'd like to be sure that it really reaches the while (1); when powering down for alarm
22:39:18 Join [Saint] [0] (S_a_i_n_t@
22:41:38 Quit S_a_i_n_t (Ping timeout: 260 seconds)
22:46:44CIA-6New commit by bieber (r27097): Theme Editor: Began working on device status dialog
22:47:34 Quit Jaykay (Ping timeout: 276 seconds)
22:48:25CIA-6r27097 build result: All green
22:50:11 Quit mc2739 (Ping timeout: 260 seconds)
22:51:47 Join mc2739 [0] (~mc2739@rockbox/developer/mc2739)
22:51:54bertrikfunman, it happened again. First time it was OK, let it wake up, now it hangs. It could be that it hangs when the current poweron was done by wake-up.
22:52:58funmanyou can try (set the alarm, power off, power on immediately) cycles to trigger it
22:54:00 Join gogoflashdrive [0] (
22:54:00 Quit playfirst (Quit: CGI:IRC (EOF))
22:54:04 Quit gogoflashdrive (Client Quit)
22:56:22 Join TheSeven [0] (~TheSeven@rockbox/developer/TheSeven)
22:57:34bertrikfunman, either I'm very lucky in triggering it, or I really see a pattern: it has now failed to power down 3 out of 3 times when alarm-powering down just after previous wakeup
22:58:32 Quit ender` (Quit: Replication with a nonconfigured partner is not allowed. -- net helpmsg 4006)
22:59:34*pamaury is stupid to invalidate to cache AFTER reading the value....
23:21:17 Quit mt (Ping timeout: 258 seconds)
23:22:37notlisteningGreat news on the Sansa usb :)
23:27:29 Join mt [0] (~mtee@rockbox/developer/mt)
23:28:21 Quit kaylinsigswort (Quit: Leaving...)
23:28:45 Quit petur (Quit: reboot)
23:29:52 Join antil33t [0] (
23:30:24 Join petur [0] (~petur@rockbox/developer/petur)
23:30:32 Join godzirra [0] (
23:30:35godzirraHeya folks.
23:30:48godzirraOut of curiousity, does rockbox support TV out on the ipod?
23:37:44 Quit petur (Quit: reboot)
23:38:08bertrikfunman, I can reproduce the failure to power off very consistently here: it always happens when powering off with the wake-up alarm set when the current poweron was also done by alarm wake-up
23:39:13 Join petur [0] (~petur@rockbox/developer/petur)
23:43:17 Join JohannesSM64 [0] (
23:43:26 Quit JohannesSM64 (Client Quit)
23:43:28 Quit petur (Client Quit)
23:45:10 Join JohannesSM64 [0] (
23:45:27 Join petur [0] (~petur@rockbox/developer/petur)
23:46:08funmanhm i didn't remember that, i might have changed it in my last commit
23:50:16 Quit Jerom (Quit: Leaving.)
23:50:22 Quit bertrik (Remote host closed the connection)
23:50:58 Join maffe [0] (
23:51:06 Quit maffe (Client Quit)
23:53:32 Join fdinel [0] (
23:54:00 Quit petur (Quit: more tinkering required)
23:55:10 Quit efyx (Remote host closed the connection)
23:56:17 Join bertrik [0] (~bertrik@rockbox/developer/bertrik)
23:56:28dfktbertrik -
23:56:38dfktand no load, for good measure -

Previous day | Next day