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

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

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

Notice: Only Gecko based browsers prior to FF4 support the multipart/mixed "server push" method used by this log reader to auto-update. Since you do not appear to use such a browser, this page will simply show the current log, and not automatically update.

#rockbox log for 2009-08-30

00:00:06AlexPIt looks to me like there is a gap between the colour of the last row and the line in the new version that wasn't there before (clearly only visible if the last row is blue)
00:00:45fmlAlexP: this was also the case with the original booktabs. I really like it.
00:00:56 Quit stoffel (Remote closed the connection)
00:00:57AlexPI much prefer it without the space :)
00:01:41AlexPBut I will bow to the majority view: pixelma, bluebrother?
00:02:32AlexPfml: Is the line between the header and the table easily movable?
00:02:48AlexPbetween the header and the body of the table that is
00:03:06fmlAlexP: that would be hard to implement (IMO) since at the point of the macro def you don't know the color of the row. Its number is available through some macro but it's too much magic for me. I'm not a texnitian! :-)
00:03:32fmlAlexP: that will be another patch :-)
00:03:39AlexPfml: OK, cool :)
00:06:58fmlAlexP: actually, here's the patch that does both (first and the last rows): Please test. But don't tell me it doesn't work!!!
00:07:09AlexPhehe, I'll try :)
00:08:05fmlAlexP: all is much simpler than I thought!
00:10:02fmlAlexP: sorry, you don't need that line with tabrowstrut. It's a rest from an earlier try.
00:10:48AlexPwhich line?
00:10:55AlexPIs this latest from clean svn?
00:11:22fmlAlexP: line #15 in the patch. Yes, it's relative to the last svn
00:13:42AlexPlooks good!
00:14:09fmlWhat about cygwin (pixelma)?
00:15:48fmlI'll leave soon. SHould I commit and hope that cygwin won't complain?
00:16:19***Saving seen data "./dancer.seen"
00:16:21AlexPNot sure :)
00:16:42pixelmaI'm testing
00:17:01fmlOK. I'm going in busy waiting mode
00:18:07CIA-43New commit by 03kugel (r22551): Samsung YH: Many keymap fixes for keyboard screen, wps, set time/date and quickscreen. Little kludge for quickscreen: Entering the quickscreen ...
00:18:51kugelyes, I do like the quickscreen
00:19:43pixelmafml: I wonder a bit why I get cr/lf with your patches although I select "download raw" from pastebin
00:19:57pixelmanothing major, just out of interest
00:20:34fml pixelma: because pastebin first lets me enter the patch in a text field in the browser (I think)
00:20:44 Join Llorean [0] (n=DarkkOne@rockbox/user/Llorean)
00:20:49kugelthat's because pastebin adds them. Just about any paste service does that for some reason
00:21:15fmlpixelma: sill compiling? Not complaining? ;-)
00:21:25funmanyou should paste them base64 encoded :)
00:21:42pixelmano, but I wanted to have a look at the result
00:23:46fml pixelma: now? Or tomorrow?
00:23:56CIA-43New commit by 03kugel (r22552): Samsung YH: Make going back to the main menu by keeping cancel press also work. For some reason, this needed on Samsung YH, but not on an e200 (it ...
00:25:54pixelmaI just looked at the time I had to answer your questions first... I think the spacing at the table end looks a bit weird but all that has the small advantage that the black horizontal lines are more likely to show up - in some cases they don't and I couldn't discover a pattern why and which tables are affected
00:26:29pixelmaas you can understand from this, building works fine still
00:27:03pixelmafml ^
00:27:14fmlWhat places look weird? Can you tell the page no?
00:28:24fmlI'll commit then.
00:29:25CIA-43New commit by 03alle (r22553): Manual: Fix the vertical spacing in the first and the last rows in tables
00:30:42fmlWhat??? It didn't build after the commit for me!
00:31:15fmlAh, now it does. Weird.
00:31:34n1spixelma: thanks for testing
00:31:53fmlpixelma: same from me!
00:32:27pixelmae.g. *some* plugin button tables (doesn't have to be all tables on a page), e.g. there are 2 on page 65 of an OndioFM manual (both brickmania ones), but e.g. in the split editor chapter one is affected, the other is not
00:33:39fmlpixelma: I don't build the manual for those players. Are the tables made using the environment (btnmap)?
00:33:47pixelmaAlexP: did you take the jackpot screenshot of an Ondio for all Archoses?
00:34:06*fml Leaves. Bye!
00:34:09 Quit fml ("CGI:IRC 0.5.9 (2006/06/06)")
00:34:21AlexPpixelma: I can't remember if it was an Ondio, but it was the same one (recorder I think)
00:35:34AlexPOr rather, I just took one screenshot for the screen size
00:36:26pixelmafml (for the logs): yes, in the split editor case - both are btnmap, one shows up completely correct the other doesn't
00:38:02pixelmaAlexP: you must have used an Ondio one sim then (or the screenshot function is broken) - as this screenshot has the Ondio "backlight" background colour. The screenshot has more target specific colours for a while now but this hasn't been adapted for the manuals yet
00:38:27pixelmabut this screenshot stands out now because it's blueish not greenish
00:38:40AlexPah right, I just picked a random target of that screen size
00:39:04AlexPare all the others done with a different target?
00:39:54pixelmaso either we would always take Recorder ones for Archos manuals or start updating screenshots per target. Well, before the Ondio screenshots showed the same background colour
00:40:31AlexPI don't envy the person that redoes every single screen shot for a specific target
00:40:37pixelmasome had to be made with/for the Ondio, e.g. all that don't show a clock...
00:41:21pixelmaer.... I mean - ones that show a clock on RTC targets
00:41:55 Quit Llorean ("Leaving.")
00:44:52pixelmaAlexP: same applies to H100/M5/Ipod Greyscale screenshots (though the differences are more subtle)
00:46:37 Quit stripwax ("")
00:46:52 Join CaptainKwel [0] (
00:47:06 Quit domonoky1 (Read error: 104 (Connection reset by peer))
00:48:03pixelmawould be nice if screenshots for the manual could somehow be automated. Doing those screenshots for same sized displays but different targets (with different colours) would also mean a lot of almost duplication :\ but would be a bit nicer for the readers of the manual
00:50:01n1spixelma: i guess it could be done by executing a set of actions but it would be hard for plugins
00:51:30 Quit funman ("free(random());")
00:55:35 Join GeekShadow [0] (n=Antoine@reactos/tester/GeekShadow)
01:01:11 Quit GeekShado_ (Read error: 60 (Operation timed out))
01:03:41rasherI bet you could do something with xte
01:04:31 Quit n1s ("Lmnar")
01:05:55 Join rotzbouw [0] (
01:06:34rotzbouwso how do I do this to make my ipod be identified as a usb mass storage device rather than the ipod upon connecting/mounting it?
01:08:14rotzbouwit's worked before but today I had to reset it with itunes and now it always gets identified as apple ipod rather than 'rockbox audio player' in dmesg (on debian)
01:10:50kugelgrml, the lcd init on the yh925 isn't correct I think
01:11:03kugelit seems we actually flip the display in rockbox
01:11:29kugelomitting 1 lcd command in the init makes the OF work correct, but that obviously breaks rockbox
01:14:45 Quit robin0800 (Remote closed the connection)
01:15:01rotzbouwah, i think i got it.. it was disabled in 3.3? current build's looking better
01:22:33 Join FOAD_ [0] (
01:27:39 Quit rotzbouw (Remote closed the connection)
01:31:17 Quit FOAD (Read error: 145 (Connection timed out))
01:31:17 Nick FOAD_ is now known as FOAD (
01:31:56 Quit Klowner (Remote closed the connection)
01:31:59 Join Klowner [0] (
01:50:45 Join stripwax [0] (
01:52:39 Join Llorean [0] (n=DarkkOne@rockbox/user/Llorean)
01:54:15LloreanDid the release schedule change at some point?
01:57:32 Quit stripwax ("")
02:10:47 Part Buschel
02:10:51rasherDon't think so. Are we due?
02:11:09saratogaanyone want to test the latest ATRAC3 patch on CF?
02:11:21LloreanBut it was posted in the forums by Saratoga that it's every 4 months, so I wanted to know if that was a typo or if we changed it.
02:11:21saratogawell preferably an iaudio i guess so the IRAM changes apply
02:12:07saratogathat should have been 4 times a year not 4 months
02:12:20rasherPretty sure it's still 3 months. How long was the freeze/branch periods? Last one was June 16th
02:12:23rashereh, 19th
02:12:31saratogaok fixed the post
02:12:35LloreanSo, fairly soon.
02:13:06LloreanAlright. I was just curious that I might've missed some schedule changing.
02:13:39pixelmasaratoga: if you tell me where to get an atrac file from and have a few minutes time I volunteer testing with my M5
02:16:21***Saving seen data "./dancer.seen"
02:18:30saratogawith and without v7 of this patch:
02:18:33saratogaif possible
02:18:49saratogasomeone on the forums asked about tagging NSF files
02:18:55saratogado we even support that
02:19:34LloreanSince a single NSF file has multiple tracks in it, which may be composed by different people and have unique names (sometimes even just being sound effects) there's no suitable way to database them
02:19:56saratogalooking at the code i don't even see a metadata parser, but he says theres tags when he plays it
02:21:00LloreanI thought there were.
02:21:15LloreanI haven't played an NSF in ages though, and don't have the player they're on with me.
02:21:37saratogathe codec itself seems to know about the file tags, is it possible for it to update teh WPS during playback (well after buffering)
02:23:07 Quit bertrik (Read error: 113 (No route to host))
02:25:31 Join jboy_ [0] (
02:27:52pixelmasaratoga: I used the "Goodbye Caroline" track and without the patch playback stalls every second, buffering takes longer than usual, put on a patched build now
02:30:46saratogapixelma: could you try test_codec?
02:30:52saratogathats mostly what I'm curious about
02:33:11tmztis that a standalone program?
02:33:15pixelmaI'll try but I hope it won't take too long
02:34:18pixelmais that FUN_RM a real track I could use for testing purpose?
02:36:46 Quit GeekShadow (Read error: 110 (Connection timed out))
02:37:31saratogapixelma: I don't see why not
02:38:20saratogatmzt: its a rockbox plugin, but there is a patch that allows it to compile outside of rockbox, although the process is not trivial
02:42:12pixelmais it not possible to interrupt test_codec?
02:42:39 Quit langzeitstudent_ (Read error: 110 (Connection timed out))
02:42:51*pixelma curses not having chosen a shorter track
02:43:13pixelmaas I wanted to
02:44:13saratogaforce a reboot
02:48:40pixelmaspeed test result of the fun_rm track with path: 19.31% realtime, 643.11 MHz needed
02:48:49pixelmapatch too
02:49:11 Quit AndyI (Read error: 104 (Connection reset by peer))
02:53:12pixelmawithout patch: 18.24% realtime, 680.84 MHz
02:55:07 Join AndyI [0] (i=AndyI@
03:00:13 Quit z35 (Read error: 110 (Connection timed out))
03:02:00 Join z35 [0] (
03:05:02*pixelma reads something about realtime on PP in the patch description...
03:07:02 Quit Thundercloud (Remote closed the connection)
03:07:43pixelmaI'm off for sleep now though, bye
03:08:17saratogapixelma: its due to lack of ASM for fixed point functions on CF
03:08:25saratogawell in large part
03:13:21 Quit Llorean ("Leaving.")
03:31:42 Quit AndyI (Read error: 104 (Connection reset by peer))
03:34:30 Quit chandoo ("Leaving")
03:34:33 Join AndyI [0] (i=AndyI@
03:38:37 Join toffe82 [0] (n=chatzill@
03:57:41 Join xavieran [0] (
04:01:33 Join xavieran_ [0] (
04:01:49CIA-43New commit by 03kugel (r22554): Samsung YH925: Implement lcd flipping. Although it's a questionable feature, it should enable us to fix the problem that the OF's display is flipped ...
04:02:05kugelso, finished 4h of trial and error :p
04:04:46 Quit xavieran (Read error: 60 (Operation timed out))
04:04:50 Nick TheSeven is now known as UselessTron (
04:04:58 Nick UselessTron is now known as TheSeven (
04:06:00 Quit TheSeven (Nick collision from services.)
04:06:18 Join The_Seven [0] (n=theseven@
04:06:24 Nick The_Seven is now known as TheSeven (n=theseven@
04:07:38 Join SimSimma [0] (
04:08:58saratogakugel: celebrate by posting yh bootloaders
04:13:19kugelwill do, later :)
04:13:35saratogawow someone posted an NSF metadata parser
04:14:59CaptainKwelsomeone needs a life.
04:16:23***Saving seen data "./dancer.seen"
04:16:43kugellimiter preamp isn't an exactly cheap feature :/
04:18:59saratogaCaptainKwel: reviewed your patch
04:22:01kugelsaratoga: I'm not sure why we're flipped compared to the OF, the comments in the file suggest it's based on disassembly
04:22:26kugelbut there's a big copy&paste business for that controller going, the comments are probably horrible inaccurate
04:22:48saratogakugel: maybe whenever the code was reverse engineered it happened to be flipped by accident and no one ever fixed it
04:22:53CaptainKwelsaratoga: thanks for the catch, copied and pasted without thinking.
04:23:12saratogaclean it up and i see no reason not to commit it
04:29:37CIA-43New commit by 03kugel (r22555): Dreaded last minute changes at 4am :( Fix red.
04:38:50 Join dys` [0] (n=andreas@
04:42:50CIA-43New commit by 03kugel (r22556): Fix x_offset for YUV blitting.
04:43:22 Quit Rondom (Nick collision from services.)
04:43:32 Join Rondom [0] (
04:43:46kugelsaratoga: I think I'll just change our behavior instead of calling lcd_set_flip() in the bootloader
04:47:16saratogaCaptainKwel: looks good but I wonder where the 120 seconds thing comes from for length?
04:48:18SimSimmaAnybody know the progress of rockbox for the sansa fuze?
04:48:35saratogacheck the status link on the front page
04:48:47SimSimmaokey dokey
04:48:58saratogai guess the 120 is just a dummy variable taken from mod since we don't know the NSF length?
04:50:05CaptainKwelJust thought it would be better than leaving it at 0. I think SPC tracks loop? Not sure if there's a way to calculate length without actually playing through the file.
04:51:13saratogaok it seems reasonable enough
04:51:38saratogais this your first patch accepted?
04:52:17saratogaok i'll add you to credits too
04:52:40CaptainKwelCool. Thanks.
04:53:12 Quit dys (Connection timed out)
04:53:18CIA-43New commit by 03saratoga (r22557): Accept FS #10570 by Jason Yu. Adds metadata parsing for NSF files.
04:53:27 Quit sinthetek (Read error: 54 (Connection reset by peer))
04:54:16 Quit SimSimma ("Leaving")
04:55:40 Quit panni_ ("( :: NoNameScript 3.81 :: )")
05:14:38CaptainKwelsaratoga: I think you left out metadata/nsf.c
05:15:45CIA-43New commit by 03saratoga (r22558): Patch doesn't SVN add . . .
05:26:39 Join ademille [0] (
05:37:06kugelsaratoga: gaah, the OF is also flipped horizontally
05:39:01kugelI don't really bother to fix that also :/
05:39:28kugelwe have 1, maybe 2 or 3 yh925 users. I doubt anyone of them wants to go into the of
05:43:30 Quit gevaerts (Nick collision from services.)
05:43:32saratogaCaptainKwel: still awake?
05:43:42 Join gevaerts [0] (n=fg@rockbox/developer/gevaerts)
05:44:26saratogaelse if (memcmp(buf, "NESM",4)) should be else if (!memcmp(buf, "NESM",4)) right?
05:45:06 Quit kadoban (Remote closed the connection)
05:45:27 Join kadoban [0] (n=mud@
05:46:25CaptainKwelwe want to continue if it is NESM
05:47:13CaptainKwelYeah, I think I screwed up my logic.
05:47:35*kugel likes being explicit for that reasons for functions returning 0 on success for that reason
05:48:01saratogayes the return values here are confusing
05:48:10kugele.g. with if (blah != 0) or something
05:49:48saratogalook good?
05:52:10 Quit kugel (Remote closed the connection)
05:52:15CIA-43New commit by 03saratoga (r22559): Logic was backwards on the check for NSF file format. Fix that and save one memcmp.
05:59:41 Join BHSPitLappy [0] (n=BHSPitLa@unaffiliated/bhspitmonkey)
06:00:48 Join Ronoh55 [0] (n=41bda16d@gateway/web/cgi-irc/
06:01:25 Join Llorean [0] (n=DarkkOne@rockbox/user/Llorean)
06:01:29saratogaCaptainKwel: looking at this closer, does the strlen logic work?
06:02:01saratogap += strlen(p)+1; is only going to increment the length by whatever the length of the first string was, not the length of the one you just added
06:03:14saratogaoh nevermind
06:03:55saratogaits too late for me to be thinking about this
06:04:12saratogathough i do wonder if its a good idea to assume the input is a null terminated string
06:04:57 Quit saratoga ("Page closed")
06:06:14 Quit Ronoh55 ("CGI:IRC (Ping timeout)")
06:07:26 Quit Llorean ("Leaving.")
06:16:27***Saving seen data "./dancer.seen"
06:28:22 Part toffe82
06:44:00 Quit fdinel ("Miranda IM! Smaller, Faster, Easier.")
06:50:58 Part ademille
07:06:09 Join Horschti [0] (n=Horscht2@xbmc/user/horscht)
07:24:48 Quit Horscht (Read error: 110 (Connection timed out))
08:06:28 Join saratoga [0] (i=463f90ed@gateway/web/freenode/x-uitgberxsnkzgvlj)
08:07:09saratogatheres apparently a new Sansa Clip revision floating around with a microsd slot
08:10:21 Join NooB23 [0] (
08:11:02 Quit saratoga (Client Quit)
08:16:29***Saving seen data "./dancer.seen"
08:34:45NooB23how can i write chars like ',,,...' on the display with 'rb->lcd_puts(0, 0, XXX);'?
08:38:30 Quit alexbobp (Read error: 110 (Connection timed out))
08:39:13 Join Rob2223 [0] (
08:50:04 Join stoffel [0] (
08:51:39 Join n1s [0] (n=n1s@rockbox/developer/n1s)
08:56:40 Quit Rob2222 (Read error: 110 (Connection timed out))
09:14:40 Join stoffel_ [0] (
09:20:05 Join advcomp2019_ [0] (n=advcomp2@unaffiliated/advcomp2019)
09:20:36 Quit timc (Read error: 110 (Connection timed out))
09:22:43 Join timc [0] (n=aoeu@
09:28:45 Quit stoffel (Read error: 110 (Connection timed out))
09:32:57 Quit advcomp2019 (Read error: 113 (No route to host))
09:47:41 Quit NooB23 (Excess Flood)
09:47:44 Join flydutch [0] (
09:47:57 Join NooB23 [0] (
09:57:22 Join mt [0] (n=MTee@rockbox/developer/mt)
09:57:22 Quit MG_Man ("...GOT AWAY SAFELY!")
10:03:07 Quit flydutch ("/* empty */")
10:07:47n1si wonder if the masking should be done everywhere we convert from bcd, it only seems to be necessary for som rtc's
10:13:29 Quit timc (Read error: 110 (Connection timed out))
10:16:32***Saving seen data "./dancer.seen"
10:25:28 Quit crashd_ ("leaving")
10:25:37 Join crashd [0] (
10:37:34n1sbluebrother: if you're awake, could you test the new version of fs#10569 on that rtc modded h100?
10:40:13NooB23is it possible to print ',,,...' on the lcd?
10:44:36n1srockbox supports unicode
10:47:07NooB23on my simulator i have the ISO-8859-1 active and this side ( say that 0xE4 is the code for ''... so i write in my string \xe4 but it does not work
10:48:18amiconnRockbox always uses utf-8 internally
10:49:10 Join decayedcell [0] (
10:49:42amiconnThe codepage setting applies to tags only, as rockbox cannot figure out the encoding itself for some tagging formats
10:50:59NooB23ahhh ok, so i have to print the utf8-code of the char i want
10:53:52NooB23is this command the right for me 'utf8decode(ASCII,OFFSET)' (dont know if i really understand it)
11:01:12amiconnJust use an utf-8 string. Decoding a constant at runtime would be a waste
11:05:42 Join bertrik [0] (
11:09:02 Join flydutch [0] (
11:11:51NooB23utf8encode(0xe115, buf); thisone?
11:13:39amiconnNo. just rb->lcd_puts(l, c, "ÄÖÜäöü");
11:14:02amiconnYour editor needs to be set to utf-8 when writing this
11:22:05 Nick Horschti is now known as Horscht (n=Horscht2@xbmc/user/horscht)
11:25:29 Join domonoky [0] (n=Domonoky@rockbox/developer/domonoky)
11:31:54 Quit NooB23 ("Nettalk6 -")
11:40:56pixelmarasher: about your automatic screenshot patch - would it be possible to make the script hook up a bit "deeper/earlier", i.e. at the moment you would need probably one script per target, if it could send the ACTION_SOMETHING or BUTTON_SOMETHING in the plugins you could use one script for almost all targets, no?
11:42:56 Quit BHSPitLappy (Remote closed the connection)
11:43:07rasherPretty sure it'd require an entirely different approach
11:43:45rasherThat hooked into Rockbox itself. Which I don't really have any idea how would work
11:47:38 Join Lynx_ [0] (
11:51:04 Join bmbl [0] (n=Miranda@unaffiliated/bmbl)
12:00:27 Join petur [50] (n=petur@rockbox/developer/petur)
12:16:34***Saving seen data "./dancer.seen"
12:19:28 Join GeekShadow [0] (n=Antoine@reactos/tester/GeekShadow)
12:38:42 Join __lifeless [0] (n=lifeless@
12:46:16 Join ender` [0] (
12:51:00 Nick fxb__ is now known as fxb (
12:52:56 Quit _lifeless (Read error: 110 (Connection timed out))
12:53:22 Nick fxb is now known as fxb__ (
12:53:30 Nick fxb__ is now known as fxb (
13:01:10 Join Bagder [241] (n=daniel@rockbox/developer/bagder)
13:04:05 Join Buschel [0] (
13:13:56 Join Lear [0] (i=chatzill@rockbox/developer/lear)
13:28:29 Join GeekShado_ [0] (
13:45:24 Quit petur ("lunch!")
13:45:25 Join Thundercloud [0] (
13:47:08 Quit GeekShadow (Read error: 110 (Connection timed out))
14:03:24 Join pamaury [0] (
14:04:43Buschelanyone with an M5/X5 willing to support me with testing a patch?
14:07:24pixelmaif you mean the one in the tracker, I tested this night when saratoga asked - should be in the logs
14:08:11Buschelpixelma: yes, I already saw it. but now I have some more coldfire stuff to test
14:08:57pixelmaok, where can I find the new things?
14:10:33pixelmashall I apply this instead of the old or additionally
14:10:42Buschelinstead of the old one
14:10:53 Join _lifeless [0] (n=lifeless@
14:12:19pixelmaok, building
14:15:08pixelmaI get build errors (and warnings) though
14:15:20Buschelhmm, can you pastebin?
14:16:38***Saving seen data "./dancer.seen"
14:19:17Unhelpfulthe sims just use SDL for keyboard events, right? would it be hard to make them accept commands from the console or via a named pipe that they're passed as an argument?
14:19:55Unhelpfulthat would get rid of the focus issues with the screenshot script
14:22:55pixelmaforgot the last line - which was just that make fails
14:23:40Buschelpixelma: i'll need some time now... thanks so far. you stay online for a while?
14:24:13 Quit pamaury ("Parti")
14:24:45pixelmaonline - always (except in case of technical failure), at keyboard - most probably
14:25:52rasherUnhelpful: sounds like a useful solution. The focus thing isn't a great problem - just don't touch the computer while it runs :-)
14:26:55 Quit __lifeless (Read error: 113 (No route to host))
14:27:42Unhelpfula pipe could also accept commands that map directly to actions, instead of to simulated buttons that map to actions, so that you wouldn't have to worry about per-target keymaps at all.
14:28:57Unhelpfulie, "cancel\n" on the pipe could send ACTION_STD_CANCEL directly :)
14:32:07 Join BdN3504 [0] (n=55b235c8@gateway/web/cgi-irc/
14:33:11 Quit flydutch ("/* empty */")
14:34:07BdN3504is it possible that i build the manual without having a vm installed on a winxp machine?
14:34:56UnhelpfulBdN3504: if all of the needed tools are available for cygwin, i don't see why not.
14:34:58BdN3504i have miktex, texniccenter and tortoisesvn installed and already checked out the latest revision into a folder with tortoisesvn
14:35:37pixelmathe PDF manual can be build with cygwin
14:36:09BdN3504ok, i'll make myself familiar with cygwin then. thanks
14:36:42pixelmathe HTML one probably *could* if someone figures out (and tells me) how to manually install the tex4ht package - or it becomes available for automatic installation
14:37:45 Nick YpsyZNC is now known as Ypsy (
14:38:27BdN3504so when i have installed cygwin right, i will be able to run ../tools/configure from within the windows commandline? or does it work differently?
14:39:10pixelmafrom within the cygwin shell
14:39:32Unhelpfulisn't cygwin *slower* than the emulators?
14:40:41pixelmaVM is probably quicker but it's ok for building the manuals
14:45:26 Join pamaury [0] (n=pamaury@
14:46:46BdN3504pixelma: which option will i have to take to install unicode suppport on cygwin?
14:49:37pixelmathe first (manual), I put the files in the place where LaTex complained it couldn't find them after a first make manual (although I think it doesn't matter much) - and I had to run mktexlsr
14:50:57rasherUnhelpful: sounds good. I look forward to the patch!
15:16:06Buschelpixelma: can you try this one? ->
15:18:31CIA-43New commit by 03mcuelenaere (r22560): Fix Onda VX777 boot extension
15:21:50 Quit TheSeven (Excess Flood)
15:21:56 Join lyngaas [0] (
15:22:28 Join TheSeven [0] (
15:24:49bertrikfunman, I looked a bit closer at the c200v2 button readout, it gets 9 bits in total (bits 2..6 and bits 12..15) from the DBOP IN, the power button is read separately
15:31:38 Join flydutch [0] (
15:32:51 Part decayedcell
15:34:13 Quit parafin (Read error: 60 (Operation timed out))
15:36:32 Join parafin [0] (
15:38:32 Quit pamaury (Read error: 60 (Operation timed out))
15:41:41 Quit BdN3504 ("CGI:IRC")
15:42:13 Quit bluebrother (Nick collision from services.)
15:42:18 Join bluebrother [0] (n=dom@rockbox/developer/bluebrother)
15:43:50 Join fdinel [0] (
15:47:28 Join funman [0] (n=fun@rockbox/developer/funman)
15:48:57 Join pamaury [0] (n=pamaury@
15:49:39pamauryhello, I have some questions. The first one is about tagcache: I noticed that there is a tagcache_retrieve method to retrieve a string tag of any entry indexed by it "idx_id". But why there is no equivalent method for numeric tags ? Then, what is the size of the audio buffer or the minimum size I can assume ?
15:49:47pixelmaBuschel: will do
15:51:08 Join timc [0] (n=aoeu@
15:51:40 Join fyrestorm [0] (
15:52:52Buschelpixelma: perfect :) at least it should compile now
15:55:04 Join cfp [0] (
15:55:27pixelmaindeed it does. I assume you want a speed test again?
15:56:00Buschelpixelma: yesm and of course a quick listening test whether it still decodes fine
15:57:09funmanpamaury: i don't know for tagcache but the audio buffer can range from 30kB to 30MB
15:57:47pixelmawell, currently
16:00:17funmanah right, basically rockbox runs on targets with 2, 8, 16, 32, 64 MB of SDRAM. Some only have 1MB but I believe they are not supported and porting on them is now inactive
16:03:15pixelmaBuschel: wow, speed test (with the same track as earlier) - 134.76% realtime, 92.15 MHz CPU needed. Listening test - yes, I can actually listen to the song now and recognise it! Everything in the WPS looks alright and seeking works too. :D
16:03:35Buschelstrrrike! :)
16:04:16*Buschel dances the realtime dance
16:04:29funmanpamaury: depending on how much buffer you need to allocate, you can make your code conditional on memory size
16:06:08pamauryThis is MTP related. The problem is that tagcache seems really sloooooooooooow. So perhaps some caching would be useful
16:06:24 Join panni_ [0] (
16:06:55pamauryBut caching the whole database requires toomuch some memory I think
16:08:04*mt hands Buschel some nice stuff for his optimizations :)
16:08:31 Quit bzed (Read error: 60 (Operation timed out))
16:11:03 Quit stoffel_ (Remote closed the connection)
16:11:16bertrikfunman, DBOP bits 12..15 are not mapped to a GPIO pin so that probably explains why we could not find them yet
16:11:42bertrik(DBOP bits 12..15 are connected to some buttons)
16:12:16 Join bzed [0] (n=bzed@
16:13:51funmani remember having checked all the dbop bits, currently we read bits 2 to 6
16:14:04funman(assuming you speak about the c200v2)
16:14:26CIA-43New commit by 03Buschel (r22561): Further performance optimization of the atrac3 decoder. Rework the internal sample representation and usage of dsp routines. For now a quick and dirty ...
16:14:51bertrikthe c200v2 OF also does something button-related with DBOP bits bits 12-15
16:15:44moos+639%, wow !!
16:16:40***Saving seen data "./dancer.seen"
16:17:00 Quit n17ikh (Connection timed out)
16:19:04Buscheland there is still room for more ;o)
16:19:18funmani'll look again in a few moments, my c200v2 was completely discharged
16:19:26funmanbertrik: where do you see this code ?
16:19:35moosBuschel: impressive, indeed! :)
16:19:57funmani'm using c200v2 OF v3.02.05
16:19:59 Join saratoga [0] (i=463f90ed@gateway/web/freenode/x-sbjjjhtbrdvoxrxt)
16:20:49rasherUnhelpful: such a command interface could even contain commands such as starting a specific .rock
16:22:01bertrikthe button read routine is at 0x3e6c, it reads the DBOP and re-arranges some bits of the value read, then adds the power button bit
16:24:48funmanat 0x3E80 it only keeps bits 6:2
16:25:39rasherDoes the sdl sim simulate button-reading delays somehow? When sending automated keypress/release events, I have to sleep between each, or it gets eaten rather than added to the button queue
16:25:51Buschelmt/saratoga: I forgot to mention that I have compared rockbox atrac3-decoding output to ffmpeg's (as a reference). The channels were definately switched in svn before r22561. Also: the current rockbox output is inverted (caused my change of the imdtc-window). But the worst is: rockbox's output differs a lot from ffmpeg's (even before my optimizations). The noise is ~20dB below the signal.
16:26:54 Join _zic [0] (
16:28:15 Quit Buschel ()
16:28:58bertrikfunman, at 0x3e86 it shifts the dbop_in bits right by 7 bits and at 0x3e90 it ANDs it with 0x1E0, thereby keeping bits 12..15
16:31:00funmanah right, and then it's OR'd with the bits 6:2
16:32:46 Quit mt (Read error: 113 (No route to host))
16:39:07bertrikfunman, do you know how the buttons are actually connected to the DBOP lines in the ams sansas?
16:39:38bertrikI mean, do the button switches enable some kind of weak pull-down resistor on the DBOP lines?
16:39:54funmanno idea i'm very stupid wrt electrics
16:41:25 Join Buschel [0] (
16:43:36funmanbits 15:8 are always 0 whenever i press buttons or not
16:43:50bertrikit seems a bit silly that we send some LCD command to write outside of the visible area in order to pre-charge the DIN lines
16:44:30bertrikI think it's possible to manipulate the DBOP data lines directly without having to send an actual LCD command, simply don't toggle any of the command lines
16:45:14funmanin fact the OF writes some bits after each read
16:49:33bertrikweird that you don't see anything on DBOP bits 12..15, I'm convinced something must be connected there
16:50:01bertrikalso, don't you see anything on bit 2?
16:51:20funmanbit 2 is left button
16:52:26bertrikah ok, the SansaAMSHardwareMappings wiki is not up-to-date
16:53:06funmanwell perhaps it was not possible to read the left button on the GPIO
16:54:47bertrikI think I should add bit 2 to the table, even though it's not readable through GPIO, it's at least somehow connected to that pin
16:56:33bertrikI'll check if we're using 16-bit mode when reading DBOP, that could explain our current inability to read bits 12..15
16:57:35bertrikI would like to borrow a c200v2 for cleaning up the c200v2 lcd and button reading
16:57:58 Join n17ikh [0] (
16:58:04funmanwell i could send you the one ej0rge sent me
16:59:03funmanindeed we are using 8 bits mode
16:59:08bertrikI would appreciate that
16:59:56 Quit pamaury ("exit(*(int *)0 / 0);")
17:04:16funmanbertrik: bit 12 of dbop_ctrl is for *output* data width
17:13:05funmandata input is always 16 bits according to the register description
17:13:29funmanalthough the summary of 7.3.12 mentions 8 and 16 bits
17:16:07bertrikthe code starting at 0x4450 does seem to explicitly enable 16-bit mode before reading DBOP_DIN
17:16:44 Quit _lifeless (Remote closed the connection)
17:17:02funmani haven't detailed the exact bit sets in my disassembly
17:17:18funmani have set bit 12 in button_read_dbop though and bits 15:8 still are 0
17:18:13 Quit cfp ("Quitte")
17:18:27 Quit Lynx_ (" HydraIRC -> <- It'll be on slashdot one day...")
17:19:42 Join merbanan [0] (
17:22:12 Part Ypsy ("WeeChat")
17:23:08 Join kugel [0] (n=kugel@rockbox/developer/kugel)
17:23:24kugelfunman: the c200v2 uses 8bit mode? Why that?
17:23:55funmankugel: you can't switch between 8 and 16 bits for dbop _input_
17:24:15soapsaratoga, I really really appreciate the way you leave a note at the end of the threads you send to the trash. Thanks.
17:25:31kugelfunman: I mean for output
17:25:52funmanwell i don't know
17:26:25funman DBOP_DOUT = *data << 8 | *data >> 8;
17:26:28kugelit's a 16bit display, I'd expect at least 16bit written at a time. the fuze/e200v1 also use 16bit mode
17:26:35funmanthat means it's at least using 32 bits
17:28:06kugelit just the bytes in data, doesn't it?
17:28:20kugelfb_data is 16bit
17:28:51bertrikoh so it's doing a byte swap
17:29:20funmanhm right
17:29:26funmansince DBOP_DOUT is a short*
17:29:40funmando we have macros for fast byte swapping?
17:30:34funmangcc adds 2 useless instructions :/
17:30:51funmanlsl 16, lsr 16 to clear the 16 msb, and then it uses strh
17:32:18bertrikI've often wondered about memory-mapped IO pointers being smaller than 32 bit
17:32:41funmanthe pointers, or the memory pointed to ?
17:32:44kugelDBOP can be 32bit too, but we don't use it so far
17:32:52kugelDBOP out, at least
17:33:16funmankugel: ah the FIFO can be 32 bits, but not the real output
17:33:24funmanperhaps we can use 32 bits only with DMA
17:34:36kugelfunman: the table under table 71 shows 2 32bit modes
17:35:10kugelthe FIFO is always 32bit, so depending on DBOP_DOUT 24, 16 or 0 bits are wasted per transfer
17:35:15bertrikfunman, one write cycle takes 16 PCLK cycles, so it's quite slow. Also we're keeping the DBOP FIFO topped up, so the byte swap probably isn't slowing things down, I expect most of the time is "wasted" in the while-FIFO-full loop
17:36:15bertrikfunman, I mean the IO memory locations
17:38:51bertrikI can imagine that you get useful compiler warnings if you try to access a peripheral register with more bits than actually used by the peripheral, but I expect that most of the time this just results in useless truncations/casts
17:39:15saratogahas anyone heard from FlynDice recently about the SD clocks on AMS?
17:39:30saratogait seemed like he had good results
17:40:06Unhelpfulrasher: that might be hard to do in the sim...
17:40:09kugelbtw, what are you thinking about boosting during SD transfers in favor of the CPU voltage scaling?
17:40:58kugelmost transfers are boosted anyway, such as during buffering or copy/move/delete actions
17:41:06bertrikif there is a narrow bus between the processor and the peripheral, it probably helps to also keep the peripheral accesses as narrow as possible
17:41:13funmansaratoga: according to the opened flyspray patch, he had made a mistake in the last patch
17:42:02funmanbertrik: compiler will cast according to the pointer type, but here it's doing useless truncation because strh only read/writes the 16 lsb
17:42:52 Nick advcomp2019_ is now known as advcomp2019 (n=advcomp2@unaffiliated/advcomp2019)
17:44:31 Quit Lear ("ChatZilla 0.9.85 [Firefox 3.5.2/20090729225027]")
17:44:50rasherUnhelpful: Well, I'm looking into the first part at least
17:46:35kugelno opinions??
17:47:02 Join mikkas [0] (n=cbdbe3a8@gateway/web/cgi-irc/
17:47:10mikkasI love you guys.
17:51:24*kugel slaps funman, saratoga and bertrik
17:52:48saratogakugel: without proper SD clocking i don't think we can really think about that?
17:53:54kugelit enables cpu voltage scaling
17:56:39bertrikI think voltage scaling isn't that important yet
17:56:41saratogai thought there was an issue with voltage scaling and the sd card?
17:59:56 Quit mikkas ("CGI:IRC (EOF)")
18:05:30kugelsaratoga: yea, boosting gets around that
18:05:47kugelI don't think it's bad as virtually any transfer is boosted anyway
18:06:00 Join gregzx [0] (
18:06:27 Join AndyIL [0] (i=AndyI@
18:06:39saratogai would prefer to work out the clocking issues and see if boosting is still needed
18:06:50saratogathe whole issue may just be that we're overclocking the sd cards
18:07:10 Quit AndyI (Read error: 60 (Operation timed out))
18:08:18bertriksaratoga, you measured it, right? so we should know now if we're overclocking or not
18:08:25funmanwe do
18:08:31 Join stoffel [0] (
18:08:51funmanbasically the card is clocked with pclk, and attempts to lower the clock make transfers fail
18:09:39saratogawonder if its AS3531
18:10:32saratogathe whole voltage thing may just be that when run out of spec the cards require higher voltage (or the sd interface requires higher voltage)
18:15:07kugelthat's what I'm thinking, in that case boosting would almost proper
18:15:25funmansaratoga: we could guess by having a firmware file
18:16:28saratogafunman: so far i haven't found one
18:16:43***Saving seen data "./dancer.seen"
18:17:03funmanthere's no point for them to put a firmware file available for download if they don't have released a newer version of what the clip+ ships with
18:17:34funmansince the clip+ features folder browsing, we might see clipv1/v2 updates in a short future, not sure about a clip+ update
18:17:58 Join alexbobp [0] (
18:18:20saratogathe firmware sounds very similar feature-wise to the fuze, so its probably a 3531
18:19:24funmanwhich new features did they add except µSD/slotradio ?
18:19:44funmanclipv2 has no new features wrt the clipv1, but I suppose AMS just stopped producing as3525
18:20:50saratogaits got the folder browser
18:21:03saratogai don't think the clip had that, though maybe they thought it wasn't useful enough without SD
18:21:22kugelbotj have that
18:21:34kugelI think. the fuzev1 definitely got it with an update
18:21:44funmanfolder view was added in a firmware update of the fuze
18:22:16funmanclipv1/v2 still don't feature it, but i expect it soon since it has been requested a lot of times (on sandisk forums)
18:29:32FlynDicesaratoga: I find myself "chasing windmills" through the ams sd code and have just stopped screaming like a senile fool every time I think I've found something..... I think thats an improvement? Quite frustrating. Still trying though.
18:31:14kugelmaybe the clip doesn't have enough RAM (for sandisks implementation of folder view)
18:31:37saratogaFlynDice: what do you think the main problem is? I would have tought slightly changing the clock speed would have been straightforward
18:31:48 Join BHSPitLappy [0] (n=BHSPitLa@unaffiliated/bhspitmonkey)
18:31:54kugelhrm, the mr500 is on the current build page. does that mean it's supported?
18:32:06 Quit Buschel (Read error: 54 (Connection reset by peer))
18:32:17 Join Buschel [0] (
18:32:19funmanthe question was asked some days ago but i don't remember the outcome
18:32:28kugelif yes then it should be added to the front page; if no then I'd like to have e200v2/fuze there too
18:32:43funmansaratoga: changing it is straightforward, figuring why we can't use the card after doing that is not :/
18:40:19 Join domonoky1 [0] (
18:40:30 Nick fxb is now known as fxb__ (
18:40:45 Quit saratoga ("Page closed")
18:51:39kugelJdGordon: have you thought of moving the wps backdrop into the skin buffer? currently it's somewhere else which doesn't really allow for fms backdrops for example
18:51:51JdGordoni have
18:52:05*pixelma wonders since when exactly returning from the WPS to the browser when stopping playback (end of playlist, manual stop) takes aaagges on the Ondio :\\
18:52:36kugelbisect it :)
18:52:38JdGordonthe problem with that is the size needed for the skin buffer....
18:53:23kugelwell, moving the wps backdrop alone is +-0, adding others will be a huge ram usage hit, on color at least :)
18:53:32JdGordonmaybe the better way to do it is store the filename for the backdrop in the gui_wps struct and reload it when it needs to be
18:54:02kugeldisk spinnup when going from wps to fms sounds sucky
18:54:05 Join PaulJam [0] (
18:54:54 Quit _zic (Read error: 60 (Operation timed out))
18:56:08kugelgevaerts: I think I can manage adding multiple album art sizes, I've had a look
18:56:49pixelmatesting builds on the Ondio is not much fun because disk access is slow so that copying different builds takes a while, here at least the firmware file is enough :/
18:57:38kugeldo you have fade on stop/pause enabled?
18:58:02 Quit domonoky (Read error: 110 (Connection timed out))
18:58:03kugelturning it off shows how long it really takes
18:59:53pixelmait's one thing I always turn off if for some reason I got the default settings again - and playback stops quite quickly, just drawing the browser doesn't. If I go to the browser or menu while music is still playing it's not slow either
19:00:40kugelyea, showing the browser is a bit slower than showing the main menu, I noticed that also
19:01:27gevaertskugel: nice. I can't wait :)
19:01:35kugelstopping playback writes resume info onto disk (IIRC) which delays it even more
19:01:54pixelmabut it's slower than it was
19:03:45pixelmameh, and then there was this problem with testing old builds and having newer flashed... it doesn't work (I get crashes) :(
19:04:03 Join polobricolo [0] (
19:05:50JdGordonmultiple AA sizes (and images) is technically very doable from the skin P.O.V... the trickyness is the loading side of it
19:09:00 Join jfc [0] (
19:09:06 Join pamaury [0] (
19:11:50kugelwhich is what I looked at
19:12:08JdGordonthe loading side?
19:12:10 Quit pamaury (Client Quit)
19:15:36kugelloading is not hard actually, the trick is to get the correct handle by aa size
19:19:55JdGordonI'd also like it to look for back.jpg (or similar)
19:27:15pixelmaah, part of the problem seems to be that I had still turned "cuesheet support" on from my tests after the rework last month - can't tell how slow it was with it turned on before...
19:27:34pixelmaat least turning it off again seems to have sped things up a bit again
19:40:07 Part glass-eye
19:47:19bluebrothern1s: new RTC patch works on h100. It gives a warning though, I've added it to FS again.
19:52:23 Join Ubuntuxer [0] (
19:53:10 Part Ubuntuxer
19:55:03bertrikfunman, can you try setting bit 12 in the DBOP_CTRL register in button_read_dbop?
19:55:10bertrikI can provide a patch if you want
19:56:06bertrik(or anyone else with a c200v2 who is ready to test something)
19:56:29funmani did already
19:56:36bertrikand it didn't help?
19:56:39*kugel too slow
20:14:59n1sbluebrother: great, thanks for testing :)
20:16:11bluebrothern1s: you're welcome :)
20:16:13kugelamiconn: you said the lcd controller of h10/etc is documented, where can I find this documents?
20:16:31*n1s wants a way to make builds stop on warnings, they are way too easy to miss :/
20:16:45***Saving seen data "./dancer.seen"
20:17:35bertrikIIRC, gcc has a flag that treats warnings as errors
20:20:59kugeleasy to miss? on an eeepc? You should've plenty time to look at every single line :)
20:23:17JdGordonpixelma: you sure its the cuesheet setting making it slow? that sounds rather odd....
20:23:18n1sbertrik: but that means manually editing each makefile :(
20:23:55bertrikI think you can do that once in tools/configure and then do make reconf for each target
20:25:52 Quit BHSPitLappy (Remote closed the connection)
20:28:22n1shmm, maybe
20:28:58kugelCFLAGS=$CFLAGS" -Werror" make ?
20:29:09kugelmaybe with a +, don't know
20:29:13 Quit merbanan (Remote closed the connection)
20:31:15 Quit moos ("Allez l' O.M !!!")
20:31:32 Join rds [0] (n=rogelio@
20:33:50n1skugel: tried, it does nothing
20:33:50 Join banan_ [0] (
20:34:27n1seh noting else than a regular make, warnings just scroll by and the buid continues
20:35:10rdsHi. would it be possible to route sound through a rockbox device to the PC? I mean, connect the rockbox device (ipod in my case) by USB to my PC, play some mp3 _in_ the rockbox device and listen the sound in my PC speakers?
20:35:43 Quit banan_ (Client Quit)
20:35:46 Join merbanan [0] (
20:37:28 Join robin0800 [0] (
20:38:32 Quit PaulJam (Read error: 113 (No route to host))
20:39:50 Quit jchillerup (Remote closed the connection)
20:40:51 Quit Thundercloud (Remote closed the connection)
20:41:37domonoky1rds: maybe somtime in the future, when we have usb-audio. But at moment its not possible.
20:42:53 Quit amiconn (Nick collision from services.)
20:42:56 Join pixelma_ [0] (i=quassel@rockbox/staff/pixelma)
20:42:56 Quit pixelma (Nick collision from services.)
20:42:57 Join amiconn_ [0] (i=quassel@rockbox/developer/amiconn)
20:43:04 Nick amiconn_ is now known as amiconn (i=quassel@rockbox/developer/amiconn)
20:43:13 Nick pixelma_ is now known as pixelma (i=quassel@rockbox/staff/pixelma)
20:44:00rdsdomonoky1: It would be great. Is somebody currently working on it?
20:44:48domonoky1rds: i am not sure. feel free to help out :-)
20:45:38robin0800FS #8663 patch is this still applicable? as so many changes have now been made now to SD handling and I personally have never needed this patch
20:47:10 Part bertrik ("Ik ben weg")
20:47:38rdsdomonoky1: I will give it a look, thanks for replying
20:54:02n1s bluebrother the warning was caused by a sourious ; on line 129, which should mean that read_datetime should give a wday that was 1 too high...
20:54:40 Quit kugel (Read error: 110 (Connection timed out))
20:54:52 Join kugel [0] (n=kugel@rockbox/developer/kugel)
20:57:34 Join BdN3504 [0] (n=5ce2251e@gateway/web/cgi-irc/
20:58:42BdN3504this is slightly off topic but: i want to write a bash script that'll cope with the manual building process. i am using the case variable my question concerns case.
20:59:00BdN3504i have options called e200) giabeat) etc.
20:59:42BdN3504how do i put it into case that two strings can be matched? is [e200 gigabeat]) correct?
21:00:13amiconnkugel: If you're talking about the big h10 - it uses the same controller as the H300, one version of the ipod color etc: Renesas HD66789R
21:01:09amiconnThat datasheet has been in our wiki for ages
21:02:09 Join getchar [0] (n=qwaszx@
21:02:20 Quit rds (Client Quit)
21:02:29 Part getchar
21:09:01rasherBdN3504: e200|gigabeatf)
21:09:52 Quit kugel (Nick collision from services.)
21:10:00 Join kugel [0] (
21:10:33 Quit Topy (Read error: 104 (Connection reset by peer))
21:10:53 Join Topy [0] (
21:12:23 Join bertrik [0] (
21:13:42rasherIf anyone who knows how buttons are handled could take a look at my attempts at simulator "remote control" using a named pipe, that'd be grand: FS #10575
21:19:32JdGordonrasher: can you pass the button from the pipe directly into the sims regular button handler?
21:19:47 Quit stoffel (Remote closed the connection)
21:20:17rasherJdGordon: No, because the sim's regular button handler expects SDL keys
21:20:20rashernot target keys
21:20:29JdGordonotherwise, yes you need to handle button_repeat and button_rel events manually
21:20:44bertrikdo we know for the ams sansas (c200v2 for example) how the the DBOP control pins are connected?
21:20:48rasherJdGordon: I don't need to *handle* them, I just need to fake them :)
21:20:50rasher(for now, anyway)
21:21:33rasherThat is, when I receive "BUTTON_RIGHT" from the pipe, I want it to look to Rockbox like right was "clicked" (press+release)
21:21:34JdGordonits certainly an interesting idea :p
21:21:51rasherI could hook it up to my LIRC remote with this!
21:22:13rasherOr other interesting things
21:22:15JdGordonwell.. the simplest solution is to post the release event staright after the press
21:22:41rasherWhat does a release event look like?
21:22:49 Join GeekShadow [0] (n=Antoine@reactos/tester/GeekShadow)
21:23:16rasherAh, so simple
21:23:23JdGordonthat may not work well with scroll buttons though
21:23:41bertrikas far as I can see for the c200v2, control C0 = always 1, C1 = LCD reset?, C2 = data/instruction select, C3 = write strobe
21:24:03*JdGordon waits for an abomination in the sim with the beasts' LCD and the h300 buttons
21:24:24rasherJdGordon: It does in fact work. Thanks
21:24:36BdN3504rasher: thanks!
21:24:39JdGordonBUTTON_REPEAT also needs to be "handled" somehow
21:25:32rasherYeah, the cmdpipe reader should accept more complicated events, but right now this will do
21:25:45JdGordonthere is a unix command to see if a file handle has input to read yeah? you might want to loop forever and easily handle button_rel if there is nothing to be read, and button_REPEAT if its the same button twice
21:26:25JdGordonbutton_defines.h should be auto-gened if you take this further...
21:26:59rasherJdGordon: Why? It's rarely a new button type gets added
21:27:18JdGordonfor completeness's sake.. but yeah maybe not
21:27:26rasherA new target that uses the current button define names wouldn't need any adjustment
21:27:39rasherDoesn't seem terribly critical
21:28:30 Quit GeekShado_ (Read error: 60 (Operation timed out))
21:29:49bluebrotherrasher: "other interesting things" as in "write a test framework"? ;-)
21:31:53*JdGordon runs away screeeming
21:32:45*gevaerts thinks that we need people who have experience in testing for that :)
21:33:35bluebrotherexperience with the GUI? I'm only experienced with Qt GUIs.
21:33:56gevaertswith testing
21:33:59bluebrotherthough I was thinking about using QTestLib (and how it could be used) for testing rbutil.
21:34:02*JdGordon goes about writing a unit testing framework for rockbox in c#
21:34:17*bluebrother slaps JdGordon with a #
21:34:56bluebrotheryou're definitely using the wrong language at work. If you'd go writing it in whitespace, ok, ...
21:35:52 Quit jordan` (Read error: 131 (Connection reset by peer))
21:39:08 Join jordan` [0] (i=gromit@
21:40:12 Quit funman (Remote closed the connection)
21:40:24 Join funman [0] (n=fun@rockbox/developer/funman)
21:43:13rasherWhat's the **pd in sim_plugin_load? I can't make sense of what it's doing
21:47:41funmanwhich file?
21:50:03JdGordonsim plugin loading is different to target..
21:50:23funman*pd = NULL; *pd = bla; < looks wrong to me
21:51:05 Join ender [0] (
21:51:17funmanrasher: probably to keep the handle to give to sim_plugin_close
21:51:18bertrikfunman, how well does the c200v2 display work at all?
21:51:32funmanbertrik: fine, but sometimes (often) i have no display at all in the bootloader
21:52:11funmanrasher: s/probably//
21:52:17rasherfunman: ah
21:52:42 Join _lifeless [0] (n=lifeless@
21:52:47JdGordontoo many *'s for me :p
21:53:57 Quit n1s ("Lmnar")
21:53:58 Quit bmbl ("Bye!")
21:55:13Unhelpfulrasher: i think part of the problem is going to be that the simulator, as i understand it, is in part an implementation of the rockbox kernel on top of SDL and system I/O, threads, etc. so we can fake key inputs to it, and maybe directly fake actions if we mess with the code for fetching them a bit, but since the plugins are actually run in-process by the UI thread, i'm not sure how we'd force starting one.
21:56:07rasherUnhelpful: I tried running sim_plugin_load("/blabla", &somewhere); but it didn't seem to load
21:56:13rasherJdGordon: that doens't apply
21:56:58rasherautorock starts *on boot* we want to start plugins on demand (from the pipe)
21:57:24Unhelpfulrasher: it has to run in the UI thread as i understand it. what i *think* might work would be to have code in the main menu, only compiled for sim, that recognizes a special event. when that event is received instead of one of the usual key-triggered events, read a plugin to load from a shared memory buffer.
21:57:57JdGordonyes, but call plugin_load() instead of doing it manually
21:59:12 Quit BdN3504 ("CGI:IRC (EOF)")
21:59:51funmanrasher: why not plugin_load() ?
21:59:51 Quit Sajber^ (
21:59:51 Quit bah_ (
22:00:03NJoinbah_ [0] (
22:00:49NJoinSajber^ [0] (
22:01:55rasherfunman: if I #include plugin.h, everything blows up
22:02:29funmanwhat if you just give plugin_load prototype?
22:03:03rasherfunman: let me try that
22:03:16JdGordonrasher: Unhelpful is right.... you need to load the plugin from the ui thread... so putting handling in the main menu (root_menu.c) is the way to go
22:05:55CIA-43New commit by 03Domonoky (r22562): rbutil: correct usb-id reading on windows.
22:06:58bluebrotherdomonoky1: what was broken there? Windows does also return a revision number here.
22:07:37domonoky1bluebrother: yes, but somehow it didnt like the Rev of my e200v2, and as it isnt used i removed it.
22:07:56bluebrotherhow does the registry string look on that player?
22:08:48 Quit ender` (Read error: 110 (Connection timed out))
22:09:19bluebrotherif it doesn't like the revision for your player it would be better to figure why it doesn't like that instead of removing it ...
22:09:27domonoky1"USB\Vid_0781&Pid_7423&Rev_4" dont know why this _stscanf wouldnt pick that up, but it doesnt. now it works :-)
22:09:54bluebrotherthat's a bad solution.
22:10:11domonoky1why, if we dont need the revision, why extract it ?
22:10:42 Quit Buschel ()
22:10:44bluebrothertrue, but if it breaks then something is fishy so we should figure why it breaks.
22:10:52bluebrotherI can't see anything problematic with that code.
22:12:26bluebrotherbtw, you didn't add a prefix to the qDebug() call. Somewhat nasty given that now users can access the debug output.
22:13:01rasherWell, I managed to load matrix.rock
22:13:06domonoky1true, will fix the debug output.
22:13:21*bluebrother really hates things getting called "fix" if they don't actually fix an issue but work around one
22:14:17 Quit gregzx ("ChatZilla 0.9.85 [Firefox 3.5.2/20090729225027]")
22:14:30CIA-43New commit by 03Domonoky (r22563): rbutil: correct debug output.
22:14:49rasherBut that seems to have taken over control
22:15:52 Join mt [0] (n=MTee@rockbox/developer/mt)
22:16:39bluebrotherdomonoky1: does this also break? And how?
22:16:48***Saving seen data "./dancer.seen"
22:18:17domonoky1bluebrother: yes it also doesnt find the e200v2
22:19:14bluebrotherwhat revision numbers does it find for other devices? Numbers less than 10?
22:22:04domonoky1it doesnt list a usb-device with revision < 10 but, there is also no such device connect (without the e200v2)
22:22:39bluebrotherwell, do other devices like mice etc have a revision number > 9?
22:24:15domonoky1i have no such device, so i dont know.
22:25:13domonoky1ah, the otherway round: other devices have revision numbers listed, all > 0x100
22:25:47bluebrotherwell, it might be related to that.
22:29:28 Join stoffel [0] (
22:30:36 Join Zagor [242] (
22:33:24bluebrotherinteresting. A small test program not using unicode works fine.
22:34:30 Quit merbanan (Remote closed the connection)
22:36:40bluebrotherbut the string read out from the Registry is correct?
22:41:44domonoky1yes, i already pasted what rbutil gets from the registry. just this _stscanf seems to have problems.
22:42:07bluebrotherthis is really weird. Have you checked the return value from _stscanf?
22:42:24bluebrotherplus the contents of the variables it writes to
22:43:50 Quit timc (Read error: 110 (Connection timed out))
22:47:57domonoky1return value is 2.
22:48:14bluebrotherhmm. Does the rev value have the correct value?
22:49:13bluebrotherinteresting. So it didn't match the last number correctly.
22:50:47 Quit stoffel (Remote closed the connection)
22:53:00*bluebrother tried again with another test program, still works.
22:53:51 Join Llorean [0] (n=DarkkOne@rockbox/user/Llorean)
22:53:57*bluebrother wonders when this Qt build will finally finish
22:53:59 Quit flydutch ("/* empty */")
23:01:20 Quit niekie (Read error: 104 (Connection reset by peer))
23:27:14 Quit _lifeless (Read error: 54 (Connection reset by peer))
23:28:17 Join GeekShado_ [0] (
23:30:32kugelZagor: I noticed the "pessimistic" points system is slower?
23:30:45 Quit GeekShadow (Read error: 60 (Operation timed out))
23:31:03Zagorkugel: I'm testing tweaks on lots of parameters, not just that
23:31:31kugelI looked at the client table, and builds where gevaerts had only ~500 have been generally much slower
23:32:20Zagorwell, the last 188 second build is also using pessimistic speed calculation. the speeds vary a lot depending on lots of factors.
23:32:45kugelthe most recent one?
23:33:17kugelgevaerts has 900 on that one, so I thought it wasn't a pessimistic one
23:34:25Zagorthe pessimitic calculation uses a "33% median" speed value instead of normal average
23:34:47Zagorso it naturally will change too. especially for gevaerts since his clients run so many builds each round.
23:35:32funmangevearts has a lot of clients ?
23:35:43Zagori meant "client"
23:36:01kugelfunman: it's gevaerts!
23:36:06*gevaerts is a successful entrepreneur!
23:36:15 Quit funman ("free(random());")
23:38:18 Join moos [0] (i=mostafa@rockbox/staff/moos)
23:41:49 Join safetydan [0] (n=deverton@rockbox/developer/safetydan)
23:42:54Zagoryay, 179 seconds. that's just 15 seconds more than the projected time.
23:46:35Zagorhmm, I need to redo the build allocation. slow clients cannot be allowed to run a single build the whole time. they always get overrun.
23:47:09Zagor(since fast clients finish faster than estimated, and start the slow clients' builds speculatively)
23:50:41Unhelpfulrasher: where exactly *is* the main event loop for the UI thread? wherever it's *normally* calling get_action and such is where we probably want to hook it.
23:52:44kugelUnhelpful: root_menu()
23:56:41 Join low_light [0] (i=ad58bb86@gateway/web/freenode/x-xyezrwdddvzcatmp)
23:58:58low_lightkugel: The yh925 OF handles the screen rotation in software. It needs the of reset back to how the bootloader does it. This is the first OF that I've seen not re-init the lcd.

Previous day | Next day