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 2009-08-16

00:00:29kugelwhy that?
00:01:57 Join T44 [0] (
00:02:34bluebrotherwhy not? It sometimes "interleaves" with log messages from the build client, and the details about the running upload aren't that interesting anyway.
00:03:31kugelImo they're interesting
00:06:15 Quit bluebrother ("enough headache for today")
00:06:45CIA-61New commit by mt (r22337): Move bitstream.[ch] to codecs/lib/ffmpeg_bitstream.[ch] to avoid duplicate copies across codecs.
00:06:49 Nick geo|kvirc is now known as geo|afk (
00:15:13kugelseems the build system is struggling a bit without gevaerts server :)
00:17:45Zagorwow, that was a crappy round
00:18:36 Join kirilla [0] (
00:20:05 Quit Topy44 (Read error: 110 (Connection timed out))
00:20:28CIA-61New commit by mt (r22338): Fix red.
00:20:37kugelqnap-jdgordon doing massive parallel building agian. His client seems particularly hit by that
00:22:03*kugel wonders if ccache can handle the constantly changing build directory
00:22:55Zagorthat's a valid question. and we learned yesterday that we can't trust the ccache manpage, so a test would be good
00:23:20kugelthe cache hit / cache miss ratio is horrible, 15/66
00:23:58kugelI'm getting *very few* hits duing this round, but I don't know if I get the same builds
00:25:48kugelZagor: can we have a −−ccache parameter for that's passed to configure? I don't feel like messing with it locally :(
00:26:24Zagorwhat would it do?
00:27:04kugelyou can deactivate the use of ccache with configure IIRC
00:27:50Zagoryou are right, ccache only caches compiles in the same path
00:28:34kugelwow, qnap-jdgordon is seriously broken this round
00:28:50kugelah no, it did fine actually
00:29:12kugelZagor: so ccache can't help us actually?
00:29:26Zagorit seems like the bug is accelerating
00:29:36Zagorkugel: it seems not
00:29:45 Part kirilla
00:30:24 Quit AsaelReiter ("CGI:IRC")
00:30:46kugelah, not "fine", but it did do his duty (queued) build at least before starting the massive parallel building this time at least
00:31:24kugelZagor: is the client still asking for builds after one is done? maybe that's not working correctly anymore (i.e. asking for a new one even if it didn't complete yet)
00:32:36kugelrasher's work is really helpful
00:32:50 Join HBK [0] (
00:35:21ZagorI think it is the generated .h files (autoconf.h and lang.h for example) that break ccache.
00:36:08Zagorthough I can't think of a way to get around that
00:37:20 Quit jernejovc ("I divided by zero. Oh shi--")
00:37:37kugelwe could deactive ccache for all clients to see if it makes a difference
00:37:37Zagorkugel: yes they do. but I doubt that is what is broken. the server is much more likely.
00:38:13Zagorkugel: that's a very crude way of testing. client speeds vary anyway.
00:38:46kugelyea true :(
00:39:05kugelbut pts/s don't change, does it?
00:39:20Zagorsimply create two different build dirs and run "time make" in both. you'll see they take nearly the same amount of time. then run "make clean; time make" in one, and see how fast it goes...
00:39:29kugelactually, I had 220 pts/s last week, and since about amiconn's fix for ccache I only have 180
00:39:36 Quit efyx ("Quitte")
00:40:15Zagorwe discovered yesterday that different compilers run at different speed on different machines. so the average speed is actually not terribly correct.
00:41:17kugelmeh, if I disable ccache in configure locally it will produce M builds :(
00:41:22Zagoron the reference machine they all run att 100 pts/sec, but on one of my machines the arm gcc is nearly twice as fast as the m68k one.
00:41:37Zagorkugel: you don't have to disable ccache to test this
00:42:05kugelbut I'd be interested to see if I'd reach the 220 again after deactivating
00:42:19Zagoraren't you reading what I'm writing?
00:42:25kugeland 180 after reactiving, and so on
00:42:38kugelI am
00:42:50Zagorccache is not a factor in this
00:42:58Zagorsince it's disabled anyway
00:43:28Zagoryour average speed sunk because you were given other builds
00:43:39kugelit's disabled?
00:43:54kugelit's clearly running here, I watched ccache -s during the round
00:44:01 Quit GeekShadow ("The cake is a lie !")
00:44:19Zagorkugel: do the test
00:44:26kugelI will
00:48:03 Quit mt (Read error: 113 (No route to host))
00:50:35kugelit's obvious that ccache doesn't handle differing build dirs
00:51:53 Quit dfkt ("-= SysReset 2.53=- Ph'nglui mglw'nafh Cthulhu R'lyeh wgah'nagl fhtagn.")
00:51:58Zagorit's quite embarassing that we haven't noticed earlier...
00:52:15ZagorI did think it was strange that amiconn
00:52:38kugellet's see how fast it goes w/o ccache
00:52:40Zagor's fix didn't seem to improve the times any significantly. but I was preoccupied with other issues.
00:53:54kugeljust a tad bit though
00:54:02Zagoryes, ccache is rather efficient
00:55:13kugelthe build round was 1:48, the second 1:38, the last (no ccache) 1:31
00:57:26kugelmy machine is comparably fast, with big ram, I think the overhead isn't as high as on slower clients
00:57:50Zagorwell, the fix isn't to remove ccache. the fix is to remove the different build dirs.
00:58:23Zagorall we have to do is run "make zip" before starting the next build
00:58:27kugelI guess that'd do the trick better indeed
00:58:53Zagor(and, obviously, moving away the zip so it can be uploaded while the new build is running)
01:00:15Zagorthere's still the parallell builds server bug, too
01:00:22kugelbut ccache still only helps if getting the same builds (which is quite likely though)
01:02:31Zagoronly if you have a very small cache. the default size is 1GB which is room for quite a lot of builds
01:02:48kugel2GB should be plenty :)
01:02:57BdN3504bluebrother: what's your opinion on this? also to the others, what do you think if we split up the viewers section of the manual or include such a table?
01:02:57 Quit roolku ()
01:08:49kugelZagor: I'm now wondering how proper ccache use would complicate best-fit
01:11:11Zagorit will make speeds vary more, which will of course make prediction more difficult.
01:12:06Zagorbut over time (as clients populate their ccache) the average speed will still be a speed indicator
01:12:31Zagorit just won't match each build during the round as well
01:13:12Zagorit will certainly be interesting! :-)
01:15:19***Saving seen data "./dancer.seen"
01:15:24 Quit Zagor ("bed time")
01:19:10 Quit robin0800 (Read error: 54 (Connection reset by peer))
01:23:49 Join mylogic [0] (
01:25:58T44i am trying to manually apply the (outdated) histogram patch to the current svn
01:30:10T44but i am having trouble to see where one certain recording.c reject is sopposed to go...
01:35:07 Join efyx [0] (
01:35:44 Quit w1ll14m (Read error: 60 (Operation timed out))
01:36:07 Join w1ll14m [0] (
01:45:48 Quit gevaerts (Nick collision from services.)
01:46:00 Join gevaerts [0] (n=fg@rockbox/developer/gevaerts)
01:54:04 Join bubsy [0] (i=Bubsy@
02:01:18 Quit Rondom (Nick collision from services.)
02:01:29 Join Rondom [0] (
02:07:32 Quit z35 (
02:38:11 Part DarkDefender ("Leaving")
02:41:13 Join saratoga [0] (i=9803c6dd@gateway/web/freenode/x-dpnyzspyadfxjiog)
02:53:49 Join Voluntaryist [0] (
02:54:16Voluntaryistif i install RB on a fuze there is 100% surety that DRM will not work after?
02:55:07kugelI guess yes. I haven't seen it myself, but that's because I have 0 DRM'd files
02:55:28Voluntaryisti have a e200 w/ rb on it, but i just bought a fuze to listen to audible
02:55:36Voluntaryistso i guess i wont do that :-/
02:58:00 Quit martian67 (Remote closed the connection)
03:00:06 Quit domonoky1 (Read error: 104 (Connection reset by peer))
03:07:46 Join martian67 [0] (n=martian6@about/linux/regular/martian67)
03:10:30 Quit martian67 (SendQ exceeded)
03:12:27 Join martian67 [0] (n=martian6@about/linux/regular/martian67)
03:15:22***Saving seen data "./dancer.seen"
03:19:54 Quit SUSaiyan (Read error: 104 (Connection reset by peer))
03:20:16 Join SUSaiyan [0] (
03:25:21Unhelpfulis there a reason it wouldn't still work in the OF? some kind of firmware checksumming or such?
03:26:07saratogaUnhelpful: its been speculated that its due to changes to the RTC, which is used for determining if DRM keys are still valid
03:26:27saratogaprobably fixable if anyone was really interested in it
03:29:54Unhelpfulsaratoga: oh! yes, i believe you're not actually allowed to *change* the RTC with the current gen of DRM. you're supposed to handle "setting" the clock by modifying a display offset for the real RTC time, or something like that, to prevent the user adjusting the time to make time-validated DRM work after it expires.
03:34:08kugelUnhelpful: I don't know if we still do, but I think we changed the rtc in the early days (re-used v1 code before we noticed the different offset)
03:36:14saratogaatrac seems to like to copy things between buffers that could probably be done in place
03:38:27 Quit BdN3504 ("CGI:IRC (EOF)")
03:39:17saratogaugh and why are "channel_unit" structs refered to by the variable "pUnits"
03:51:12 Quit ehntoo (Read error: 110 (Connection timed out))
03:51:15Unhelpfulkugel: but the offset is probably variable, too, since as i understand it it's the only thing you're allowed to vary. the basic idea is that your subcription content is validated for the current RTC date plus one month, or whatever. :/
03:54:06kugelIIUC we changed (or still change) the rtc date (at least when setting the time)
03:54:16kugelno idea really
04:02:37 Nick daurn is now known as daurnimator (i=daurn@freenode/staff/daurnimator)
04:15:05 Quit chandoo ("Leaving")
04:15:32saratogamt: (for the logs) do you have a real media test file that uses joint stereo? none of the ffmpeg ones do and I'd like to test some changes with JS files before commiting
04:21:20 Join SUSaiyan` [0] (
04:21:31 Quit SUSaiyan (Read error: 104 (Connection reset by peer))
04:22:38kugelUnhelpful: I'm looking into anti-aliased font right now and struggle again :(
04:24:03kugelstupid mpegplayer :(
04:24:13CIA-61New commit by saratoga (r22339): Make the IMDCT temporary working buffer a local variable instead of keeping one for each channel. Saves 4KB of RAM.
04:29:29saratogahmm I bet we could compute ATRAC3 spectra on a per QMF band basis rather then doing them all upfront and having to store the output
04:30:56 Join Hillshum [0] (
04:35:21 Nick intrados_ is now known as intrados (
04:35:36 Join SUSaiyan [0] (
04:40:51 Quit kugel (Remote closed the connection)
04:41:19 Quit efyx (Remote closed the connection)
04:46:55 Quit SUSaiyan` (Read error: 104 (Connection reset by peer))
04:53:39 Quit bah_ (Read error: 104 (Connection reset by peer))
04:53:51saratogaheh if the ATRAC3 decoder doesn't have a subband coded, it zeros the buffer and then adds it anyway, instead of just not adding . . .
04:59:33 Join bah [0] (
05:06:37 Quit J-23 (Read error: 110 (Connection timed out))
05:09:52 Quit bah (Nick collision from services.)
05:10:07 Join bah_ [0] (
05:15:23***Saving seen data "./dancer.seen"
05:31:06 Join Mig [0] (n=chatzill@
05:31:07 Quit SUSaiyan (Read error: 104 (Connection reset by peer))
05:31:11 Join SUSaiyan` [0] (
05:31:29saratogado plugins need to handle USB inserts specifically? i notice test_codec tends to freeze on insert
05:31:33 Nick Mig is now known as Guest44401 (n=chatzill@
05:34:53 Part Guest44401 ("I'm not here right now.")
05:35:22 Quit parafin (Nick collision from services.)
05:36:45 Join parafin|away [0] (
05:54:58 Quit Hillshum ("ChatZilla 0.9.85 [Firefox 3.5.2/20090729225027]")
06:02:44 Join webguest76 [0] (n=4a681d95@gateway/web/cgi-irc/
06:05:07 Quit webguest76 (Client Quit)
06:05:42CIA-61New commit by saratoga (r22340): Put QMF coefficients into IRAM. Update test program to still compile with IBSS_ATTR defined.
06:07:27 Quit mylogic (Client Quit)
06:11:15 Quit geo|afk ("good night everyone!")
06:13:50saratogathe atrac IQMF is going to be impossible to make fast as its currently written
06:20:14saratogahmm where is buschel when i need him
06:22:34 Quit saratoga ("Page closed")
06:41:16 Join inigomontoya [0] (i=621cc255@gateway/web/freenode/x-sokuwxrrovaqxsfw)
06:42:29inigomontoyaHello, I am having a problem with new builds, I can't turn off crossfade, something i'm doing wrong? should i file a bug report?
06:42:52marazstop listening to mixtapes
06:53:33 Quit inigomontoya ("Page closed")
06:56:19 Join mylogic [0] (
06:56:52 Join ehntoo [0] (
06:57:44 Quit mylogic (Client Quit)
06:58:00 Join mylogic [0] (
07:15:24***Saving seen data "./dancer.seen"
07:19:07 Quit JdGordon ("Leaving.")
07:36:13 Quit CaptainKwel (Remote closed the connection)
07:55:36 Join robin0800 [0] (
07:57:45 Join Horscht86 [0] (
08:00:36 Join n1s [0] (n=n1s@rockbox/developer/n1s)
08:05:00 Join Anu [0] (n=chatzill@
08:08:12Anuhello everyone
08:08:52AnuI failed to remove password using ATA PWD for dell optiplex 330 so what should i do ?
08:09:29 Join Grahack [0] (n=chri@
08:12:41 Join Rob2222 [0] (
08:15:51 Join jayren [0] (
08:16:03 Join Marvin_ [0] (n=5abfca64@gateway/web/cgi-irc/
08:16:11 Quit Horschti (Read error: 110 (Connection timed out))
08:16:33Marvin_remember the nano 1g i told about ... that froze and was "dead" well it started to charge after a week of being dead
08:16:38Marvin_and this night this happened
08:16:51Marvin_i am lucky that i woke up, saw the table literlly burning
08:17:00Marvin_with open fire
08:17:37Marvin_seems that it did not know when to stop charging
08:23:13jayreni am assuming the english.voice file is combined speech fragments for the various voice prompts and such, is there some way to uncombine it? the link to the tour of the rockbox for the blind gave me an invalid server and i've been wanting to hear the voice of the thing since I read about it about a year ago.
08:24:16*jayren is back.
08:25:30Marvin_the room still smells, god knows how much cancerogen it released to air
08:30:30 Quit Rob2223 (Read error: 110 (Connection timed out))
08:32:14 Quit jayren ("Time wasted on IRC: 18 minutes 49 seconds")
08:32:38 Quit Anu (Read error: 60 (Operation timed out))
08:33:12 Quit daurnimator (Read error: 110 (Connection timed out))
08:33:14 Join jayren [0] (
08:34:04 Part jayren
08:42:25 Join daurnimator [0] (i=daurn@freenode/staff/daurnimator)
08:46:25markunMarvin_: wow
08:48:09Marvin_i'll take it to apple shop on monday for laughs before breaking it apart to see the damage inside
08:48:34Marvin_being 2 years old it has no warranty damn
08:52:00T44i think as it WENT UP IN FIRE i believe you should AT LEAST get a new device out of it :)
08:52:13T44if not compensation or whatever
08:53:58Marvin_yeah, but we'll see
08:55:31Marvin_i'm lucky that i was not drinking, god knows that it might have ignited
08:56:50 Quit n1s (Read error: 110 (Connection timed out))
09:14:31 Quit fyrestorm ("lamers envy me like they envy bill g -- main boot xp, just the way it should be!")
09:15:26***Saving seen data "./dancer.seen"
09:33:14 Join mt [0] (n=MTee@rockbox/developer/mt)
09:49:01 Join TheCoolGman [0] (n=446e5e7d@gateway/web/cgi-irc/
09:49:10 Join flydutch [0] (
09:50:54TheCoolGmanHi I was wondering the possiblites of MPEG Player being able to read the size of the video and then if it is bigger than 320x240 it would play it at a temp resized version
09:52:42 Quit Tristan (Nick collision from services.)
09:53:23 Join Tristan [0] (
09:53:42 Quit Tristan (Nick collision from services.)
09:54:29 Join Trista901 [0] (
10:14:30 Quit TheCoolGman ("CGI:IRC (EOF)")
10:16:50 Nick Horscht86 is now known as Horscht (
10:31:51 Join n1s [0] (n=n1s@rockbox/developer/n1s)
10:32:41 Quit Grahack (Read error: 110 (Connection timed out))
10:44:41 Join daurn [0] (i=daurn@freenode/staff/daurnimator)
10:49:35 Join domonoky [0] (n=Domonoky@rockbox/developer/domonoky)
10:50:32mtAny coldfire owners around and willing to test a patch ?
11:02:11 Quit daurnimator (Read error: 110 (Connection timed out))
11:07:13 Quit n17ikh|Server ("moving day")
11:09:21 Quit n17ikh ()
11:09:59 Join JdGordon [0] (n=jonno@rockbox/developer/JdGordon)
11:12:15 Join GeekShadow [0] (n=Antoine@reactos/tester/GeekShadow)
11:15:28***Saving seen data "./dancer.seen"
11:18:14 Join Moomoc [0] (
11:19:32T44mt: h340 is coldfire, right? yeah, sure
11:20:04T44just tell me what you need
11:20:18mtT44: Yeah, Thanks. here's the patch :
11:20:44T44firing up vm...
11:20:48mtyou know how to do speed tests for codecs ?
11:21:20T44no, are there instructions in the wiki?
11:21:28T44wait, i think i saw it yesterday
11:22:32mtJust add "test_codec.c" to apps/plugins/SOURCES
11:23:41mtit will compile the test_codec plugin with your build, then you will open the target file with test_codec.
11:24:11T44sec, have to update first
11:29:58T44so... applying the patch
11:30:36amiconnmt: I may try some things later
11:31:20mtamiconn: Great, thanks.
11:32:00 Join bmbl [0] (n=Miranda@unaffiliated/bmbl)
11:35:26T44hm... for some reason its not accepting the patch, **** Only garbage was found in the patch input.
11:35:56mtweird .. one minute.
11:36:18T44sec, meaby the copypaste added cr/lf...
11:36:33T44nope, thats not it
11:36:58pixelmadid you use "download", that'll give you the raw text
11:37:28T44nope, copypasting into the vm seemed easier, let me try
11:38:33T44ah, did the trick
11:39:23pixelmafor c+p, be sure to copy from the lower window (the above has line numbers added) and cr/lf can also be a problem on Windows
11:39:52mtOK good. Did you add test_codec.c to apps/plugins/SOURCES ?
11:39:55T44yeah, probably thats the problem, even though i told my editor to use unix style newlines
11:39:59T44yeah, compiling
11:41:25mtHere are some samples of atrac3_rm files in case you don't have any : - Just one file should be enough.
11:43:09T44transferring to device
11:43:41 Join Grahack [0] (
11:45:49pixelmaT44: you don't have to let us know every detail, especially on seperate lines... remember that the channel is logged. Thanks
11:46:08 Join bluebrother [0] (n=dom@rockbox/developer/bluebrother)
11:47:16 Join fyrestorm [0] (
11:47:38T44hm, just realized make produced an error... sec
11:49:50T44i am getting a region PLUGIN_IRAM is full error, when compiling atrac3_rm.codec i believe
11:50:15T44can you compile it? (device 11, iriver h3x0) might be on my side
11:51:05*pixelma will try for M5 (same CPU)
11:52:53mtT44: No the problem is with my patch, putting more buffers in iram than there's actually available, as it seems. Thanks for testing. :)
11:53:18T44tell me when it compiles, i'll test it again
11:56:26 Quit bubsy ("I'll be back somewhere in time...")
12:03:34 Quit evilnick_home (Read error: 113 (No route to host))
12:10:24 Join dfkt [0] (i=dfkt@unaffiliated/dfkt)
12:23:26mtamiconn: Why would this be wrong : ? ( This is supposed to do 32bit x 32bit -> return result >> 16 bit )
12:44:14 Join bubsy [0] (i=Bubsy@
12:46:01 Quit lyngaas (Read error: 110 (Connection timed out))
12:48:51mtpixelma, T44: (Whoever is free) another patch that needs coldfire testing - :)
12:50:20n1sis it important for us to make uses of gcc-specific features conditional on __GNUC__ etc
12:53:14mtIt's not needed imho.
12:58:28 Quit mt (Read error: 104 (Connection reset by peer))
12:59:06 Join mt [0] (n=MTee@rockbox/developer/mt)
13:02:49 Quit Moomoc ("leaving")
13:04:07amiconnmt: I already told you what's wrong with that ~14 hours ago
13:05:00mtamiconn: Yes, I don't see why I should do a 15-bit left shift rather than a 16-bit one ?
13:05:24amiconnBecause that's how emac works
13:05:58amiconnFor this to work, emac needs to be in fractional mode. If you use it for iteger, accN contains bit 62..31 of the result, not bit 63..32
13:06:40*amiconn hands mt a copy of CFPRM.pdf
13:06:47mtOh, OK. Thanks. :)
13:07:09amiconnAnd you need to make sure (in a suitable place) that the emac actually is in that mode
13:07:58mtCould that be in the init_codec function for example?
13:15:30***Saving seen data "./dancer.seen"
13:18:53amiconnHmm, saratoga isn't here :\
13:19:40amiconnPutting the imdct buffer on the stack was a bad idea
13:19:56amiconnIt causes an immediate Stkov codec
13:20:18 Join DarkDefender [0] (
13:26:33 Join bertrik [0] (
13:27:40 Join BdN3504 [0] (n=5ce53b7b@gateway/web/cgi-irc/
13:30:40BdN3504Who wants to comment/commit this manual patch?
13:32:36 Quit BdN3504 (Client Quit)
13:32:45CIA-61New commit by amiconn (r22341): Revert r22339 as it causes immediate stack overflow on target, at least on coldfire.
13:35:48amiconnmt: Ahem, fixp_math.c has only inline functions. You know they will never be used inline this way?
13:36:37amiconnThey should go into fixp_math.h and made 'static inline'. fixp_math.c can be deleted
13:37:45mtNo, didn't know that.
13:38:10amiconnFunctions can obviously not be inlined if they reside in another .c file
13:38:36*amiconn starts playing around with fixp_math
13:38:50amiconnUnfortunately I didn't take an arm target with me
13:39:42mtI could test on my e200
13:48:01 Quit bertrik (Read error: 110 (Connection timed out))
13:48:31 Nick parafin|away is now known as parafin (
13:48:53 Quit T44 (Read error: 104 (Connection reset by peer))
13:48:55 Quit parafin ("So long and thanks for all the fish")
13:48:58 Join z35 [0] (
13:48:58 Join parafin [0] (
13:49:10 Join AndyI [0] (i=AndyI@
13:51:24*amiconn needs to wait for test_codec to finish
13:51:27amiconnI need base data
13:52:42 Join Lear [0] (i=chatzill@rockbox/developer/lear)
13:53:24 Join kugel [0] (n=kugel@rockbox/developer/kugel)
13:54:21 Join efyx [0] (
13:56:09 Join robin0800_ [0] (
14:01:15 Quit AndyIL (Read error: 110 (Connection timed out))
14:02:57 Quit robin0800 (Read error: 60 (Operation timed out))
14:07:50 Join robin0800__ [0] (
14:07:52 Quit pandrew (Remote closed the connection)
14:09:27 Quit robin0800_ (Read error: 60 (Operation timed out))
14:10:55 Join teru [0] (
14:11:00 Quit robin0800__ (Remote closed the connection)
14:11:26 Join robin0800 [0] (
14:13:27amiconnmt: Could you test whether this works on arm?
14:13:41amiconnI get roughly 20% speedup on coldfire
14:14:38kugeljust due to inlining?
14:14:40kugelnot bad
14:15:29amiconnWill be even better with cpu specific optimisations
14:15:46 Join bertrik [0] (
14:17:59 Quit pixelma (".")
14:18:46 Join billenium [0] (
14:18:51billeniumDoes rockbox play ogg files?
14:19:19 Quit z35 (
14:19:46 Join pixelma [0] (i=quassel@rockbox/staff/pixelma)
14:19:56billeniumHm, they seem not to play for me... maybe i have to update
14:20:19bluebrotherwhich version are you running? Ogg is supported since ages
14:20:30mtamiconn: Building now - Could you try this for the CF : (it just puts the lookup tables into iram).
14:20:37bluebrotheryou should make sure the files don't contain ID3 tags.
14:21:49billeniumVersion r18669-080929
14:21:58billeniumHow would i check?
14:22:55bluebrotherdepends on the system you're using, obviously. On linux try id3v2
14:24:52CIA-61New commit by bluebrother (r22342): Remove svn:executable on source file and set keywords.
14:25:59kugelbillenium: that's really old, you should update nevertheless
14:26:31amiconnmt: The patch doesn't apply. "patch: **** Only garbage was found in the patch input."
14:26:49amiconnThe leading @@ of all chunks seem to be missing
14:27:07bluebrotherit should still play ogg files. Ogg not playing is not related to a version several months old
14:27:24CIA-61New commit by teru (r22343): fix crossfade setting broken by r22192.
14:27:46kugelbluebrother: yes, hence I added "nevertheless"
14:28:55*amiconn fixed it manually
14:30:42amiconnmt: Your patch gives about 2% speedup on cf
14:31:15n1smt: it might be a good idea to look at using more compiler optimizations for the codecs you've added as iiuc they currently use the same optimization level as the core rockbox except for librm that uses O3 (which sometimes is slower than O2 btw)
14:36:50Learamiconn: Replacing fixmul31 with an assembler macro looks useful (and easy, since there's MULT31 in Tremor)...
14:37:24amiconnLear: I will play with cf macros soon. Just want to know whether the inline stuff is committable and then commit first
14:37:54*amiconn is playing with fat time+date in the meantime
14:38:50mtamiconn: about 10% speedup on arm.
14:39:24n1samiconn: am i correct in thinking that both libcook ad libatrac are compiled with the same optimizations as core?
14:41:07CIA-61New commit by amiconn (r22344): Make those functions actually inline. Around 20% speedup on coldfire, 10% speedup on arm.
14:42:27amiconnmt: Will you commit the iram change for the tables?
14:50:02kugelbluebrother: yes, hence I added "nevertheless"
14:50:14kugelbluebrother: oops sorry
14:51:07 Quit cg_ (Remote closed the connection)
14:51:20CIA-61New commit by mt (r22345): Put the lookup tables into iram ; 2% speedup on ColdFire 1% on ARM
14:52:50 Quit billenium (Read error: 110 (Connection timed out))
15:00:54 Join cg_ [0] (
15:03:20mtusing the am multiply functions on arm gives a 32% speedup !
15:15:27 Join qurvel [0] (
15:15:34***Saving seen data "./dancer.seen"
15:18:03 Join petur [0] (
15:18:23mtand a 2.5% speedup by putting ATRAC3Context in iram - This should be more for CF I think, amiconn ?
15:20:01 Part qurvel
15:21:16mtHow much IRAM is available on the CF targets ?
15:26:35amiconnFor codecs you can use 48KB IRAM on MCF5249 (iriver H1x0, H300, iAudio M3), and 64KB IRAM on MCF5250 (iAudio M5, X5)
15:30:33 Join BdN3504 [0] (n=5ce53b7b@gateway/web/cgi-irc/
15:31:02*n1s wonders why wma data aborts on the beast with all the samples he's tried
15:32:11BdN3504can you please have a look at ? i value your opinion.
15:34:50bluebrotherBdN3504: why are you adding hard newlines in the LaTeX code? One should not do that unless absolutely necessary.
15:34:52mtn1s: The beast is an arm core right ?
15:35:25amiconnYes, ARM1136EJ-S (ARMv6)
15:35:25BdN3504because the notes automatically stick to the text if you don't
15:35:27bluebrotheralso, I find the description for the png viewer confusing. It doesn't open files "in the file browser"
15:36:07BdN3504bluebrother: it doesn't? i didn't know that. sorry.
15:36:14bluebrotheryou still shouldn't to this. If you need a new paragraph put a new paragraph.
15:36:36n1si'll try to work out when it broke later gtg
15:36:49bluebrotherwell, you can't open anything *in* the file browser. You open it *from* the file browser ;-)
15:37:37mtn1s: I guess I know why it broke.
15:37:44bluebrotherbtw, \note{} not including a \par is intended. I can't see any reason why a note should be a new paragraph in every case.
15:39:09 Join Thundercloud [0] (
15:41:00BdN3504bluebrother: i looked at the other parts of the manual and they looked different, than what i made. in section 1.3 for example is a note and that is clearly on a new paragraph, so i went that same route.
15:41:29bluebrotherhow is the new paragraph done there? You put in a newline, which is *not* a new paragraph.
15:42:16bluebrother\\ forces a newline. If you want to force a new paragraph use \par. Besides, starting a new paragraph is usually done by two consecutive newlines (i.e. leaving one line empty).
15:42:47BdN3504so using \par would be ok?
15:43:42bluebrotherwell, it would be at least better. You really should not mess with LaTeX' layout as that is done by the document class
15:47:09 Join IuDeX [0] (n=52a0f8f7@gateway/web/cgi-irc/
15:47:15 Quit IuDeX (Client Quit)
15:47:32 Join geo|ppc [0] (
15:47:39bluebrotherBdN3504: and an ellipsis is \dots, not ... (it really looks different)
15:50:01bluebrotherhmm, something is strange here. \note *does* add a \par unless in inner mode.
15:50:09bluebrotherwhich is not the case here.
15:50:28BdN3504but actually that's not an ellipsis, the setting in the context menu read open with...
15:50:48bluebrotherit does, but in typography it's an ellipsis.
15:51:04BdN3504aha ok
15:51:13bluebrotheran ellipsis is three dots, slightly far spaced than three consecutive dots.
15:51:44bluebrotheras I said, it looks different (but you might not notice it if you're not aware of it :)
15:52:00n1smt: good, will you fix or should i try to gigure it out?
15:52:30mtn1s: I'm on it now .. could you just sen me links to the samples you're using ?
15:53:26bluebrotherBdN3504: Better explanation that I would be able to give ;)
15:54:10n1smt: i took some random samples from mplayerhq and from supposedly fixed taks in our tracker
15:54:13 Join qurvel [0] (
15:54:31 Part qurvel
15:57:18BdN3504all but vbrfix look nice with /par
16:00:14BdN3504two newlines don't work, /par doesn't work, only // makes the note in vbrfix look consistent to the other notes in that section.
16:00:45bluebrotherif you put in slashes instead of backslashes it won't work at all ...
16:01:41 Join Lss [0] (
16:02:14bluebrotherfor the jpegviewer, the note does not make sense with the note directly following it.
16:02:26BdN3504yea, you know what i mean stuüid typo in irc
16:03:17bluebrotherwell, I've seen people writing links using backslashes and wondering why the server has problems, so I don't know what you mean. At least not if you're aware of the difference.
16:04:33bluebrotherok, seems I know now :)
16:04:35BdN3504i'm using a tex editor and it marks \ blue.
16:06:36BdN3504ok, /par doesn't create a new line on any instance. that's not how i wanted it to be, but if you say the guidelines are not to use \\, so be it.
16:06:44BdN3504i mean \par
16:07:13 Quit Hadaka (
16:07:13 Quit AB3JU (
16:07:28*bluebrother fails to recognize the colors in BdN3504's editor
16:07:59NJoinHadaka [0] (
16:07:59NJoinAB3JU [0] (
16:08:24bluebrothersomething seems to be broken in the plugins sections. Someone needs to figure what ...
16:08:27bluebrotheranyway, bbl.
16:09:07 Quit kachna (Read error: 113 (No route to host))
16:09:11 Quit geo|ppc ("Pocket IRC 1.2.5 ( UNREGISTERED")
16:09:42 Quit mt (Remote closed the connection)
16:10:29BdN3504i don't understand this. now all the other notes don't look right, only the one on vbrfix... i thought latex is not like word, where so much inexpliccable stuff happens :(
16:11:08 Join mt [0] (n=MTee@rockbox/developer/mt)
16:12:43 Join shaggy-h [0] (
16:13:28 Join rkjnsn [0] (n=4c69cb73@gateway/web/cgi-irc/
16:13:56BdN3504maybe it has something to do with my compiler.
16:17:35rkjnsnDo you guys have time for a quick user question?
16:18:03pixelmatry it ;)
16:18:42n1sBdN3504: are you not using the regular manual build system?
16:19:38rkjnsnOkay. I am trying to get album art to work using the "everything in 'albumart' under '.rockbox'" approach, but I am having trouble with artists which non-ascii characters.
16:21:33pixelmanon-ascii but valid in FAT, like cyrillic or ä or so?
16:22:43BdN3504i am using miktex to build the manual
16:22:50rkjnsnHmm... not sure on that. I think my OS transferred it as UTF-8 encoded (which may be the problem).
16:23:03pixelmarkjnsn: or do you mean things like questionmarks?
16:23:24rkjnsnNo, I did all the conversion to _ for those.
16:24:38BdN3504nls: the vmware image debian distro gives me tons of errors although i have utf 8 support installed and builds a manual which does not have working links. miktex' output is fine though
16:24:46n1sBdN3504: if that gives different results it sounds like a bad solution, any reason you don't do it the regular way?
16:24:50pixelmarkjnsn: hmm... sounds like a codepage problem then. How were your files tagged and does the filename show correctly in the browser (and the tag in the WPS or database?
16:25:10rkjnsnI mean things like "Claude-Michel Schönberg" and "植松伸夫"
16:25:35 Join evilnick_home [0] (
16:25:38rkjnsnThe files are tagged in UTF-8, which shows up correctly in the WPS/Database
16:25:56markunrkjnsn: sounds like a problem with your mount options
16:26:05markunVFAT is unicode
16:27:38BdN3504i updated the patch
16:27:52rkjnsnIt's the automount thingy with Gnome/HAL
16:30:05Tornemarkun: he said they were tagged in utf-8
16:30:14Tornemarkun: that's fine (and preferable)
16:30:22markunTorne: but the tags are not the problem, right?
16:30:28pixelmahow do the filename look in Rockbox's filebrowser (you might need to set "show files" to "all" to look inside the .rockbox folder?
16:30:30Torneand vfat is *always* unicode
16:30:31markunthe filenames
16:30:49Torneat least in theory
16:31:17*pixelma needs some more )
16:31:35rkjnsnpixelma: The filenames are incorrect. I looks like what happens when you view UTF-8 as Latin-1
16:32:01markunrkjnsn: then it sounds like the default codepage for mounting is latin-1
16:32:27rkjnsnmarkun: In Rockbox or on my PC?
16:33:26 Quit efyx (Remote closed the connection)
16:33:32Torneyou need 'utf8' in the mount options, slightly counterintuitively
16:33:46Tornewell, maybe.
16:33:55Tornevfat is always unicode on disk
16:34:12Tornebut the fs needs to know what to translate that to for the purposes of the unix API
16:34:12markunrkjnsn: PC
16:34:26Torneby default it may try latin1 or similar.
16:34:31Tornelinux generally wants utf8
16:34:35Torne(these days)
16:35:43Torne(on disk it's utf16 which is not suitable for POSIX APIs)
16:37:08markunbut I don't know how to change these settings for HAL
16:37:20Torneindeed :)
16:37:25TorneWorks For Me, also
16:37:36 Quit BdN3504 ("CGI:IRC")
16:37:43Tornemy player gets automounted by hal and utf8 translation is already on
16:38:12Tornelinux works in utf8, vfat driver writes to disk as utf16, all is happy.
16:39:35rkjnsnOkay. I guess I'll have to look into getting HAL to use the utf8 option.
16:39:36rasherrkjnsn: Open gconf-editor and go to system > storage > default_options > vfat
16:41:12rkjnsnrasher: Thank you, that did the trick.
16:41:34rkjnsnThank you all. Everything seems to be working now.
16:47:01 Join mt_ [0] (n=MTee@
16:47:13 Quit robin0800 (Remote closed the connection)
16:48:49mt_n1s: The problem is with my commit which created one set of bitstream.[ch] files for libwma/cook/atrac in codeclib. Reverting this change for libwma fixes the data aborts, do you think I should just revert it for now for libwma ?
16:54:22 Quit mt (Read error: 113 (No route to host))
16:54:37 Join robin0800 [0] (
16:55:54kkurbjunHow do we go about adding a new theme section for a new player?
16:57:18CIA-61New commit by teru (r22346): pluginlib_bmp: fix bug that bitmap isn't saved correctly if width is not a multiple of 4. optimize a bit.
16:58:57n1smt_: i don't know, maybe a revert is best unless it gets fixed pretty quickly
17:09:37 Join CaptainKwel [0] (n=jason@
17:15:35***Saving seen data "./dancer.seen"
17:16:08 Join Omlet [0] (
17:16:17 Quit Thundercloud (Remote closed the connection)
17:20:59 Quit Omlet (Client Quit)
17:29:10 Quit teru ("Quit")
17:30:32 Quit faemir ("Leaving")
17:30:54 Join keef [0] (n=5b547ee0@gateway/web/cgi-irc/
17:33:09keefHello. I have installed Rockbox (release 3.3) on my Iriver H10 6GB MTP, together with the H10 MTP Bootloader (from The installation itself seems to be successful - on startup Rockbox loads and reaches the main menu. But then no buttons or keys have any effect, including the on-off button - so I have to remove the battery to shut
17:35:29 Join raphi [0] (
17:36:35CIA-61New commit by mt (r22347): revert r22337 for libwma for now since it broke wma playback on ARM due to alignment issues.
17:45:13 Join faemir [0] (n=faemir@
17:46:32 Join GeekShado_ [0] (n=Antoine@
17:51:48 Quit raphi ("leaving...")
17:57:36 Join FOAD_ [0] (
17:58:56 Quit FOAD (Read error: 60 (Operation timed out))
17:58:57 Nick FOAD_ is now known as FOAD (
18:02:43 Quit mt_ (Read error: 110 (Connection timed out))
18:03:50 Quit GeekShadow (Read error: 110 (Connection timed out))
18:08:44 Join efyx [0] (
18:19:23 Quit jfc (Read error: 104 (Connection reset by peer))
18:19:47 Join jfc [0] (
18:20:10 Quit jfc (Read error: 104 (Connection reset by peer))
18:20:31 Join jfc [0] (
18:20:55 Quit jfc (Read error: 104 (Connection reset by peer))
18:21:17 Join jfc [0] (
18:21:41 Quit jfc (Read error: 104 (Connection reset by peer))
18:22:02 Join jfc [0] (
18:22:28 Quit jfc (Read error: 104 (Connection reset by peer))
18:22:49 Join jfc [0] (
18:27:41 Join domonoky1 [0] (
18:27:41 Quit domonoky (Read error: 104 (Connection reset by peer))
18:29:43 Quit efyx (Remote closed the connection)
18:42:50 Quit Lear ("ChatZilla 0.9.85 [Firefox 3.5.2/20090729225027]")
18:48:27 Join mcuelenaere [0] (
18:49:21mcuelenaerekkurbjun: (if you have an account)
18:50:25mcuelenaereonly thing is, themes having the same resolution will show up for different players (which isn't good for touchscreen- vs non-touchscreen-compatible themes)
18:50:58kkurbjunmcuelenaere: thanks for the link - I don't think I have an account - who would I talk to to get one?
18:51:16mcuelenaereanyone with SSH access to that server I guess (which I have)
18:51:19kkurbjunthat won't be a problem for the M:Robe right now since it is the only player that runs at 640x480
18:51:40mcuelenaerehmm and it isn't a problem for the Onda either (400x240), but that doesn't fix the problem :)
18:51:46mcuelenaereeh 240x400
18:51:48kkurbjun:), true
18:52:19kkurbjunactually, one other thing - I modified the onda and D2 touchscreen drivers to hopefully act like the m:robe
18:52:38kkurbjunI think reversi and calculator should be working on those targets now
18:52:54mcuelenaereI saw the commit, but hadn't look into what it actually did :)
18:53:19kkurbjunon calculator I just enabled the work that you had done - minesweeper did not seem to work for me
18:54:02 Quit Tuplanolla (Read error: 110 (Connection timed out))
18:54:27kkurbjunhow much work is it to add a target for If it is just a few settings that need to be enabled it might be easier to do that than for me to get an account
18:54:57mcuelenaereit seems you need a picture of the player
18:55:10mcuelenaerebut I can't seem to find my logs, better to ask rasher for an account
18:55:24kkurbjunok, I'll do that
19:02:52 Quit AB3JU (
19:02:52 Quit Hadaka (
19:02:55mcuelenaerekkurbjun: regarding that commit, I'm wondering whether it shouldn't be done in firmware/drivers/touchscreen.c instead
19:03:45NJoinHadaka [0] (
19:03:45NJoinAB3JU [0] (
19:05:37 Join mt [0] (n=MTee@rockbox/developer/mt)
19:09:57 Join ender` [0] (
19:11:38 Quit robin0800 (Remote closed the connection)
19:15:39***Saving seen data "./dancer.seen"
19:17:56 Quit cg_ (Read error: 110 (Connection timed out))
19:18:49CIA-61New commit by amiconn (r22348): FAT timestamp handling improvements for non-RTC targets: ...
19:19:46 Quit AB3JU (
19:19:46 Quit Hadaka (
19:20:48NJoinHadaka [0] (
19:20:48NJoinAB3JU [0] (
19:21:11 Quit AB3JU (
19:21:11 Quit Hadaka (
19:21:14NJoinHadaka [0] (
19:21:15NJoinAB3JU [0] (
19:21:15 Quit keef ("CGI:IRC (EOF)")
19:21:38JdGordonanyone got a ipod video handy? does the icatcher theme work correctly in svn?
19:26:21 Join cg_ [0] (
19:27:29JdGordondamn... it does
19:30:49JdGordonah FUCK! it works in the sim fine with my patch, but not on target
19:31:40JdGordoncan I get some testers for 8996 please? I really want to get it in today
19:35:19JdGordonsweet :) using a vairable before initialising it causes wierd shit to happen on target :)
19:35:33 Join stripwax [0] (
19:35:38CIA-61New commit by amiconn (r22349): Save a bit of binsize by using a more suitable datatype and making the array static.
19:36:47 Join mitk [0] (
19:37:59 Quit kugel (Remote closed the connection)
19:38:06 Join kugel [0] (n=kugel@rockbox/developer/kugel)
19:39:51stripwaxmt - is fixmul16 in atrac really the same as MULT32 in codeclib? MULT31 and MULT32 differ by one bit shift but fixmul16 and fixmul31 differ by 15bit shift, so not sure I understand how they can be used as drop-in replacements
19:41:34mtNo it isn't this was wrong, but the asm stuff were reverted some revisions ago ?
19:42:44stripwaxoh, so it was - sorry, I'd missed the revert.
19:43:21mtnp - could you perhaps, ahem, profile atrac ? :)
19:43:36kugelah, which reminds me
19:44:04stripwaxmt - not easily (not for a couple weeks I'm afraid)
19:44:10kugelI tried to play a bit with profiling recently, but it didn't work at all
19:44:20kugelplugins just freezed
19:44:32mtstripwax: Sure, whenever you feel like it.
19:44:34stripwaxkugel - which plugins? which functions did you want to instrument?
19:45:04stripwaxkugel - did you at least get a profile.out when you were done?
19:46:17kugelno, I started a plugin and rockbox freezed
19:46:49kugelI used rb->profile_thread() and rb->profstop(), and added the cflag to plugins.make
19:47:15stripwaxkugel - which plugin
19:47:18mtstripwax: here are the correct arm multiplies
19:47:21stripwaxoh, all plugins?
19:47:33kugelall, although I just inserted the stuff into bubbles
19:47:45amiconnmt: ~20% speedup on coldfire due to ATRAC3Context q IBSS_ATTR;
19:47:52amiconn(in atrac3_rm.c)
19:48:11mtThat's nice ! only 2.5% for arm.
19:48:24amiconnYou mean PP502x
19:48:43amiconnOn PP5002 it will also be quite a bit more (but I can't test that now)
19:49:20stripwaxkugel - you should only put the cflags into the plugin you want to profile, ideally. but presumably the plugin that froze rockbox was the bubbles plugin, is that right? not some unrelated one
19:49:39kugelevery plugin froze
19:49:46kugelincluding bubbles
19:50:00kugelthere's only 1 make file for all plugins
19:50:16kugel(except for the subdir ones I think)
19:50:46mtamiconn: Are you going to commit it, or should I do it ?
19:51:23stripwaxoh, right. and then you had to reboot rockbox, no way to quit or anything? it shouldn't do that [obviously] but I haven't seen it do that before so I don't know why it didn't work for you
19:51:42stripwaxI've only tried profiling codecs in this way, not all plugins
19:51:55stripwaxwhich device?
19:52:27kugelsame happened when profiling the core btw, although I don't know if that was ever possible
19:52:43stripwaxkugel - does profiling codec work for you?
19:53:02kugelI didn't try
19:53:09stripwaxbtw I don't see why codec would work and other plugins not; but I know codec profiling works [for me]
19:53:23kugelwithout modifications?
19:54:42stripwaxyou still need to put $PROFILE_OPTS in cflags and ci->profile_thread in codec_main
19:55:13kugelI mean other modifications, like fixing profiling :)
19:55:35stripwaxI'm pretty sure it works without any other modifications, at least on iriver h120/140 and ipod 5g .... but I can't test right now. I don't think I needed to make any changes to profiling code to fix anything
19:55:50kugelI'm wondering if I did a complete rebuild, enabling profiling breaks the api
19:55:58stripwaxthat's likely
19:56:23kugelcodecs worked though
19:56:26stripwaxI would recommend a complete rebuild - since you need to reconfigure with Profile anyway. I used a separate build dir for that.
19:56:37 Join Creposucre [0] (n=53c3f041@gateway/web/cgi-irc/
19:57:09kugelI mean, I didn't try profiling a codec, but the codec api breaks as well, and since they worked I think I did a complete rebuild
19:57:53CreposucreJdGordon: still need a tester for video?
19:58:49JdGordontester on any target would be good
19:59:12JdGordonI tihnk its all good now though
20:00:07 Part mitk ("Leaving")
20:00:14kugelJdGordon: are you sure you really need testers for 8996?
20:00:26CreposucreJdGordon: which task? FS #8996 is "Usb charge-only mode is not documented"
20:00:28stripwaxkugel - I would still recommend a totally separate build directory if you want to be able to switch between regular and profile builds
20:00:35JdGordon9886 maybe?
20:00:48 Join mitk [0] (
20:00:53JdGordon:p yeah 9886
20:01:34bluebrotherany thoughts on the wording about the icons in the about dialog?
20:03:44JdGordonseems ok?
20:05:41stripwaxmt - on arm, you should be able to combine the orr and the shift into one instruction. fixmul31 would be the same as MULT31. but I thought it doesn't affect the condition code so "cc" should not be in the clobbers list.
20:06:52 Join Tuplanolla [0] (n=jani@unaffiliated/tuplanolla)
20:07:48JdGordondisk access should cause the calling thread to yield yeah?
20:09:07 Join robin0800 [0] (
20:13:36CreposucreJdGordon: patch hunks with latest build
20:14:10JdGordondw, that patch is old.. commiting anyway
20:15:31CreposucreJdGordon: anything that I can do?
20:15:59JdGordonget the new build on your ipod once its built and let me know if anything breaks :)
20:16:42JdGordonis apps/gui/skin_engine/skin_engine.h in svn?
20:17:37Creposucrethe .h is not
20:17:52JdGordonits not?
20:18:56Creposucreskin_buffer.c, skin_buffer.h, skin_engine.c and skin_tokens.h only are in skin_engine
20:19:40*kugel slaps JdGordon again for not doing the renaming in a seperate commit
20:19:59JdGordonI didnt know.... *cries* I didnt know...
20:23:04CIA-61New commit by jdgordon (r22350): more wps->skin engine work.. ...
20:23:55 Part Voluntaryist
20:26:04Creposucrejdgordon: is there something special to do to test the functionalities of this patch?
20:26:15JdGordonbah, forgot to mention that its advisable to load themes from .cfg files and NOT .wps files
20:26:33JdGordonCreposucre: there is no new functionality... you shouldnt notice any badness
20:26:39JdGordonor anything different at all hopefully
20:26:49kugel"Unlimited "string" tokens now (limit was 1024 which was rediculously wasteful) " really sounds funny
20:27:27JdGordonwell.. hmm.. yeah, considering hardly any themes would use more than 20.. the 1000 was silly
20:28:11JdGordondamn fine delta if you ask me :D
20:28:19 Quit stripwax ("")
20:28:34JdGordonand thats not cheating by using buffer_alloc() for the buffer :)
20:28:48kugelthat'll give a decent delta won't it?
20:29:07JdGordonchanging to tha will be a HUGE fake delta
20:29:11kugel1bit remotes with tiny displays basically get the same buffer as a big color screen, right?
20:30:48kugelheh, people with remotes suffer a bit :)
20:30:52*amiconn slaps JdGordon around with 82 points
20:31:03JdGordonyeah yeah, I'm fixing it
20:31:13JdGordonbut come one.. the delta more than makes up for that :)
20:32:13kugelJdGordon: I think it would've been wiser to not just multply with NB_SCREENS, but rather something like "+((LCD_REMOTE_HEIGHT*LCD_REMOTE_WIDTH*LCD_REMOTE_DEPTH/8) + (2*LCD_REMOTE_HEIGHT*LCD_REMOTE_WIDTH/8))
20:32:27CIA-61New commit by jdgordon (r22351): fix player
20:32:51JdGordonare those defined to 0 for non remote targets?
20:32:52 Join lyngaas [0] (
20:33:03CreposucreJdGordon: looks correct
20:34:27UnhelpfulJdGordon: not really fake when you consider how little of the static buffer size most real themes need. removing artificial limits *and* saving memory for the common case is a win all around. :P
20:34:34kugelJdGordon: no, but you can easiy do that as "#define REMOTE_BUFFER [...]" within #ifdef HAVE_REMOTE_LCD, then buffer_size = IMG_BUFSIZE + REMOTE_BUFFER
20:34:59 Quit rkjnsn ("CGI:IRC (EOF)")
20:35:24JdGordonUnhelpful: yeah :)
20:36:32CreposucreJdGordon: i don't see anything different after test, i guess it's a sucess :)
20:36:41 Join DarkFlash [0] (
20:37:26*kugel still thinks the buffer on targets with remote is unreasonably higher than on those without remote
20:38:08CreposucreJdGordon: by the way, if you want to have some rest, you can test 10494, i added a function for your dock
20:38:18Creposucrehopefully it'll work
20:38:35pixelmaJdGordon: unfortunately the delta is quite some plus on Archos
20:38:53pixelma(about the "make up for it" bit
20:39:02 Join stripwax [0] (
20:39:18JdGordonblame its remote :)
20:39:50JdGordonthe talk above about the buffer size... maybe kugel wants to fix it and take credit :)
20:40:21pixelmaOndios and Player (?) don't have a remote and the Recorder one is screenless
20:41:32DarkFlashHi everybody. I want tu buy an mp3 player where a can install rockbox. I want an player with about 40 GB of memory... Any advice?
20:41:33JdGordonoh right
20:43:04JdGordonCreposucre: ok, I'll try to test it today
20:43:12bertrikI think skin_buffer.c should also #include skin_buffer.h
20:43:59JdGordonwhy? it doesnt actually need to
20:44:47 Join efyx [0] (
20:44:51bertrikit's ensures that the implementation is consistent with the declaration (gcc warns if it is not)
20:45:49rasherJdGordon: urgh
20:46:34rasherYou can no longer pick a wps independently of the theme?
20:47:13JdGordonno, you can
20:47:36JdGordonjust if you keep doing it, it will stop working untill you load a .cfg
20:49:36 Quit Grahack ("Leaving.")
20:51:49rasherThat sounds pretty terrible, to be honest
20:52:28JdGordonumm why? how often would you manuall load a .wps? the theme site even forces the correct .cfg structrue
20:52:49linuxstbJdGordon: Why is loading a .wps a problem now?
20:53:00JdGordonits not a problem...
20:53:09linuxstbYou just said it was...
20:53:11rasherIt just might randomly stop working
20:53:18rasherI'd say that qualifies as a problem
20:53:22JdGordonyou will just run out of room after loading image heavy .wps files mulitple times
20:54:07rasherSo basically if you only switch wps, but not the entire theme, you're likely screwed
20:54:19JdGordon3 or 4 times depending on the theme
20:54:21kugelis that only true if you don't reboot inbetween?
20:54:44JdGordonyou can just load a .cfg (ANY .cfg.. it can even be blank) and it will work again
20:54:48*linuxstb thinks that's clearly a problem, but has a flight to catch...
20:55:08rasherJdGordon: That sounds spectacularly braindead
20:55:17kugelit sounds weird indeed
20:55:29pixelmaJdGordon: did you perhaps higher the WPS buffer on Archos for some reason?
20:55:32JdGordonthen you make it work correctly
20:55:41rasherJdGordon: I wasn't the one who broke it
20:55:43JdGordonpixelma: not on purpose
20:55:49JdGordonITS NOT BROKEN
20:56:05rasher"if you keep doing it, it will stop working"
20:56:06pixelmait sounds like it is though
20:56:07rasherthat's broken.
20:56:14JdGordoncome up with a way to safely reload every .wps if just one changes (im tihnking remote targets here) and by all means it wil be done
20:56:51 Quit linuxstb (Read error: 60 (Operation timed out))
20:57:01 Quit moos ("Rockbox rules the DAP world")
20:57:03rasherJdGordon: Again, I wasn't the one who broke it.
20:57:11 Quit domonoky1 (Read error: 54 (Connection reset by peer))
20:57:14JdGordonits not broken
20:57:19rasher"if you keep doing it, it will stop working"
20:57:37rasherYou should know better than to commit half-finished work.
20:58:02pixelmaJdGordon: will you answer the reports of "my WPS suddenly stopped working, what's wrong"?
20:58:16JdGordonthats not going to happen
20:58:20 Quit stripwax (Read error: 110 (Connection timed out))
20:58:35JdGordonrasher: I posted on the ml.,. if you dont care enough to comment then thats your bad luck
20:58:37pixelmasure "the one I loaded yesterday doesn't work when I load it now"
20:59:12JdGordonthat will only ever happen if you load .wps files from the file browser MUTLIPLE TIMES and they are image heavy
20:59:23JdGordondoing a .cfg load avoids that entirely
20:59:25rasherJdGordon: That's no excuse for committing broken code. Apparently you knew of this brokenness, yet just dumped it in anyway
20:59:29pixelmaumm... some people do
21:00:21pixelmaand those who change their WPSs multiple times usually have image heavy ones too
21:00:31rasherI'm pretty sure I wouldn't be using the theme feature on my H120 if I still had it
21:00:41rasherJust loading wpses and fonts as I saw fit
21:01:13 Join domonoky [0] (n=Domonoky@rockbox/developer/domonoky)
21:02:11kugel-84k sounds a bit too much for just making remote buffer smaller
21:04:36 Quit flydutch ("/* empty */")
21:05:23 Join J-23_ [0] (
21:05:36kugelJdGordon: what's the problem with just reloading every wps? why's loading an empty cfg magically fixing things?
21:05:37JdGordonpixelma: if yo change your wps multiple times, chances are you are changing the whole damn theme in one hit
21:05:55JdGordonkugel: resetting the buffer on the 2nd one blows away the first one
21:06:10pixelma*I* usually don't
21:06:13JdGordonright now this only affects remote targets, but think about radio skin or rec skin
21:06:15 Nick J-23_ is now known as J-23 (
21:06:47 Quit CaptainKwel (Remote closed the connection)
21:06:51pixelmabut I don't change my WPS that often
21:07:22pixelmait still sounds like broken by design and I'm awaiting the first bug report
21:07:30 Quit J-23 (Remote closed the connection)
21:07:46rasherJdGordon: *you* might change the whole theme at once, I guarantee not everyone does.
21:07:50 Join J-23 [0] (
21:08:13 Quit mt (Read error: 101 (Network is unreachable))
21:08:21JdGordonare you suggesting that people are so stupid they cant work the rediculously simple work around?
21:09:25kugelI still don't understand this
21:09:37pixelmaJdGordon: you can't be serious
21:09:46kugelwhy can't you just use the code that works for loading an empty .cfg for loading wpses to?
21:10:08JdGordonkugel: loading an empty .cfg will call settings_apply() which reloads everything
21:10:45JdGordonthere is a single large buffer for all .wps files... there is no safe way without really big complication to reset that buffer and load skins again
21:11:28rasherJdGordon: But you figured you'd just go ahead and commit broken code?
21:12:02JdGordonfuck... its not broken
21:12:21rasherThen why are there suddenly things we can't do anyomre?
21:12:45rasherYou latest mail doesn't really make things any better.
21:15:20CIA-61New commit by kugel (r22352): Recalibrate the skin buffer size calculation so that targets with remote get a reasonable extra buffer (depending on the remote) instead of just ...
21:15:43***Saving seen data "./dancer.seen"
21:16:19kugelmy rockbox-info.txt numbers were weird
21:16:59JdGordonkugel: yeah thats the mess I wanted to avoid...
21:19:02kugelI can't spot mess
21:19:45 Join keef [0] (n=5b547ee0@gateway/web/cgi-irc/
21:20:06JdGordondid you fuck something up? those deltas look way too big
21:20:30JdGordonactually what was the old buffer size?
21:21:03kugelno idea
21:21:30JdGordonit must have been a worse waste on remote targets than I thoguht
21:22:06kugelit's definitely not wrong that the mr500 needs additional 700k just for its 64x16x1 (???) remote
21:22:13kugelnot right*
21:23:01 Join BdN3504 [0] (n=4e3435ff@gateway/web/cgi-irc/
21:23:20pixelmaand what's with the difference between e.g. the M5 and the X5 - it's the same remote
21:23:23JdGordonyeah, but the delta from my commit shuld have been different
21:24:04kugelpixelma: it was twice the main screen before
21:24:06pixelmaah, colour main screen or greyscale mainscreen
21:24:42pixelmathat's why it was so big on the M:Robe500 before - huge main screen
21:25:23kugelJdGordon: well, right now I can't figure out how the remote was taken into account before your commit
21:25:37 Quit petur (Remote closed the connection)
21:25:43JdGordonpretty sure its 2*IMG_BUF
21:25:53BdN3504i'm with you jd. i already made a couple of WPSs that i had to recode or change because of the buffer size restricitions. and as we have a strict themes upload policy requiring intact folder structures and cfg files
21:25:55JdGordonbucause its an array of wps_data which is statically sized
21:27:25BdN3504and i dont think it's likely anybody would change the wps file 4 times in a row. in the end themes are called themes, because the creators thought about which bg will accompany the wps best
21:27:33kugelah right, wps_data[NB_SCREENS]
21:27:48JdGordonBdN3504: thank you.. exactly
21:28:43kugelBdN3504: that doesn't mean the code is intact. it's based of rather subjective assumptions
21:30:07BdN3504is this only happening with the wps file? what about large font files? say i load a unifont ten times in a row, will that cause something weird?
21:30:20JdGordonno, *only* .wps files
21:30:35JdGordonwell.. and rwps
21:30:37kugelwhy don't you just call settings_apply() when loading a wps?
21:30:37n1sif we don't want people to load wsp's directly we should drop that functionality IMO
21:30:54 Join funman [0] (n=fun@rockbox/developer/funman)
21:30:56 Join petur [0] (
21:31:06JdGordonn1s: yeah, I dont think we need to go that far
21:31:13funmanhi, i wanted to let you know i had no news from this AMS guy today ...
21:31:40kugeln1s: I think that feature is questionable anyway, wps are designed around a certain font size for example, mostly loading another wps doesn't work out well for that reason
21:31:49BdN3504do they work on sundays? :)
21:32:13kugelbut I wouldn't remove it :>
21:32:17kugelfunman: no call?
21:32:22amiconnkugel: Not all wps are designed for fixed font sizes. Also, it's a handy feature if you're testing your wps
21:32:38*amiconn s standard wps doesn't care about font size
21:33:01 Quit Creposucre ("CGI:IRC (Ping timeout)")
21:33:24kugelJdGordon: is this overflow at least detectable?
21:33:45JdGordonits not an overflow :) its protected..
21:33:48kugelas in, throwing a "please reboot to enable" in case it happens
21:33:56JdGordonbut yes it can be checked for
21:34:08JdGordonI plan on adding a "skin ram usage" line to the rockbox info screen
21:34:11kugelI'd rather see it fixed, though
21:34:19JdGordonits not a reboot to fix issue
21:34:49 Join Creposucre [0] (n=53c3f041@gateway/web/cgi-irc/
21:34:51 Join toffe82 [0] (
21:35:06kugela reboot wouldn't get around it?
21:35:21JdGordonit would,m but all that is needed is to run an empty .cfg file
21:35:40kugelI wonder why doing settings_apply() on wps loading doesn't work
21:35:59JdGordonyou want to reapply every setting to reload one .wps file?!
21:36:07JdGordonslooow and wasteful and sucky
21:36:29 Quit Creposucre (Client Quit)
21:37:33 Quit Lss (Read error: 104 (Connection reset by peer))
21:37:37 Join Riku [0] (
21:37:41kugelthen hack a small version together
21:38:17 Quit amiconn (Nick collision from services.)
21:38:19 Join amiconn_ [0] (i=quassel@rockbox/developer/amiconn)
21:38:39 Nick amiconn_ is now known as amiconn (i=quassel@rockbox/developer/amiconn)
21:39:57 Quit pixelma (Nick collision from services.)
21:39:57 Join pixelma_ [0] (i=quassel@rockbox/staff/pixelma)
21:40:15 Nick pixelma_ is now known as pixelma (i=quassel@rockbox/staff/pixelma)
21:40:15 Quit BdN3504 ("CGI:IRC (EOF)")
21:41:56kugelJdGordon: like this
21:42:42JdGordonlunch bbs
21:43:48funmankugel: no call / no SMS
21:44:53 Quit petur (Remote closed the connection)
21:44:54 Quit keef ("CGI:IRC (EOF)")
21:45:36 Join petur [0] (
21:46:16kugelfunman: :'(
21:46:44CIA-61New commit by bluebrother (r22353): Add Save icon and add a note to the about dialog about the origin of the icons. Should have been there much earlier ...
21:48:54 Quit funman ("free(random());")
21:50:45mcuelenaereWhy is there an Archos AV300 target in config.h when the config-av300.h file doesn't exist?
21:51:10 Join Grahack [0] (
21:51:26kugelmcuelenaere: historical reasons? :)
21:51:36mcuelenaereprobably :) so it can be removed?
21:52:19pixelmaask linuxstb and scorche
21:53:07mcuelenaerelinuxstb, scorche: ?
21:56:23kugelJdGordon: that code actually works
21:57:05kugelmcuelenaere: I think we never remove ports, and I really hate this situation in case of clearly dead ones
21:57:31kugelespecially since re-adding is a matter of calling svn diff
21:57:39mcuelenaerekugel: it's not that I want to remove the port, but it just doesn't have a config-av300.h file and that breaks my script :)
21:57:52mcuelenaere(although I could work around it, but I would prefer not to)
21:58:53kugelI'd prefer to remove a bunch of others too in that run :>
22:02:56 Join aaron424 [0] (
22:04:32rasherThe gminis were removed
22:10:08 Quit DarkFlash (Read error: 113 (No route to host))
22:10:52 Quit petur (Remote closed the connection)
22:12:20CIA-61New commit by kugel (r22354): Fix loading of multiple .wps/.rwps files in a row, by reloading all wpses upon loading one so that the skin buffer can be reset properly.
22:12:45JdGordona bit of pateince would kill you?
22:12:48JdGordonTHATS A BAD IDEA
22:13:04kugellol, I just looked before hitting the button if you answered
22:13:17 Join petur [0] (n=peter@rockbox/developer/petur)
22:13:17kugelit's better than what we have now
22:14:08JdGordon1) its not.. 2) its a bad idea
22:14:28JdGordonits a very bad band aid for a issue which will never show up
22:14:42rasherJdGordon: Just because you keep saying it will never show up doesn't make it true.
22:15:16JdGordonof course, the bad side effects of that change wont show up for most poeple for a while
22:15:17rasherWe don't leave bugs in just because someone thinks it'll never happen. And we certainly don't knowingly commit them. Oh wait...
22:15:36kugelJdGordon: lol, I just reproduced the problem very quickly
22:15:51JdGordonin regular use?
22:16:01kugelwhat's bad about the fix?
22:17:14JdGordonits not really bad right now... at most there are 2 skins being loaded... think about what happens when fm gets skins... and on remote targets... thats 4 loads every time
22:17:20JdGordonthen add other screens
22:17:22JdGordonits a bad idea
22:17:48kugelit isn't
22:17:53 Quit Riku (Nick collision from services.)
22:18:11bluebrotherOMG can't you guys get your changes *finished* before committing?
22:18:13 Join Lss__ [0] (
22:18:21kkurbjunJdGordon: I load wps files direct...
22:18:25kugelthe problem you created will get even worse if you add more skins, the the user just needs to load a wps, a rwps and a wts and BANG it's broken
22:18:33kugelthat doesn't even require reloading another skin
22:19:07kugelthe disk is already spinning if you load one skin, it doesn't even create much delay
22:19:18*JdGordon gives up
22:19:49*bluebrother shakes head
22:20:14 Part Grahack
22:21:31kugelkkurbjun: happy about the 700k ram saving?
22:22:01kkurbjunrasher - on themes, I was wondering if it would be possible to add a section for the m:robe 500? mcuelenaere pointed me to the admin panel, but I don't have an account - is that something you could help me out with?
22:22:14 Join moonscapex [0] (n=4b9fb4d0@gateway/web/cgi-irc/
22:22:15kkurbjunkugel: :), yes that is nice
22:22:18rasherkkurbjun: certainly, hang on
22:22:44 Quit moonscapex (Client Quit)
22:23:50kkurbjunkugel: oh, I didn't see the impact that changing the remote allocation made, that's REALLY nice :-D
22:25:25kugelkkurbjun: you basically had enough space to run a fully featured 640x480 wps on the remote too :p
22:25:48kkurbjunyeah, and for such a small screen that made no sense
22:25:57JdGordondont give him all the credit.... it was my apparently devasting change which made that saving possible!
22:26:22 Join kadoban [0] (
22:26:22kkurbjunrasher, how do I add an image for the m:robe
22:26:41rasherkkurbjun: get it into - that's where they come from
22:26:55kugelJdGordon: "JdGordonthe talk above about the buffer size... maybe kugel wants to fix it and take credit :)" :D
22:27:17rasherkkurbjun: is where they are expected to be
22:27:18kugelyes I take it :p
22:28:49kkurbjunrasher, ok, so I just need to check out the WWW section in svn and add the player image?
22:29:20JdGordonkugel: credit yes.. not *all* the credit :D
22:29:21pixelmakkurbjun: do you have an svg?
22:29:26rasherkkurbjun: and then poke the swedes, since doesn't run directly out of svn
22:29:33kkurbjunpixelma: I do
22:29:41JdGordoncan some ams targets be added also?
22:29:41CIA-61New commit by learman (r22355): Fix realloc bug in TLSF.
22:29:47JdGordonthey are no less suported than the mr500
22:31:16kugelI think so yes
22:31:32kkurbjunpixelma: is there a good place to store svgs publically?
22:31:41 Join GeekShadow [0] (n=Antoine@reactos/tester/GeekShadow)
22:31:51kugelalthough e200v2 doesn't need an extra image
22:32:00JdGordonthere hasnt been any desicisons to not put working-in-progress targets on theme site has there?
22:32:01 Join pamaury [0] (
22:32:09rasherkkurbjun: the manual holds the other svgs
22:32:12bluebrotherkkurbjun: put it with the proper naming for use in the manual svn :)
22:33:10 Quit aaron424 (Remote closed the connection)
22:33:10 Quit suom1 (Remote closed the connection)
22:33:54JdGordonkugel: back to that commit... part of the motivation for the whole work was to make some screens easier to draw.. the time/date screen is a good example which wastes lots of code drawing the time when the skin engine could do it free... reloading the skins like that is a waste when the time screen doesnt always need to be loaded
22:34:02JdGordonwhat I was going to do was add an event which is triggered when the buffer is reset so screens can decide to reload or not
22:34:43kugelbut since you didn't add a possiblity to reload skins seperately, it's the only way to fix *right now*
22:35:50 Quit thegeek_ (Read error: 104 (Connection reset by peer))
22:36:21kugelthere's nothing holding you back from doing a better fix
22:37:58kugelI also think the buffer could managed better. for example a pointer for each skin pointing to the buf_start for that skin in the buffer, and the size. Then the skins could be memmoved when another skin needs reloading to make it fit
22:38:24kugelI think Unhelpful's buflib also wouldn't do such a bad job here
22:38:28JdGordonthat doesnt guarentee it will always git
22:39:03CIA-61New commit by mcuelenaere (r22356): Rockbox Utility: add preliminary support for installing the bootloader (+ dual boot) on ChinaChip targets
22:39:04 Quit GeekShado_ (Read error: 145 (Connection timed out))
22:39:08kugelwell, it does, unless all of your skins exceed the skin buffer
22:39:16mcuelenaerecrap, forgot to mention FS#..
22:39:40CIA-61New commit by kkurbjun (r22357): M:Robe 500 add SVG for player
22:39:53JdGordonand no... the memroy cant move, unless lots of work is done to use offsets instead of pointers
22:40:03kugelyea, why not?
22:40:09JdGordonbut sure, if you want to do the work, be my guest
22:40:15 Join Thundercloud [0] (
22:40:21kugelnah, I don't feel like it, that's your job
22:40:22JdGordonwhy not hwat?
22:40:28kugelstoring offsets
22:40:53JdGordonoffsets are slower and more likeloy to be done wrongly
22:41:15kugelthen I prefer being slow *once* on loading, instead of being slow always at drawing
22:41:16mcuelenaerehmm I should probably also add the vx777 and vx747+
22:41:42JdGordonoy gevalt
22:41:47*JdGordon is off again
22:42:02kugeloy gevalt?
22:42:40JdGordonyiddish :)
22:42:44JdGordonthats how annoyed I am :p
22:43:41rasherkkurbjun: That's certainly a very basic svg!
22:43:56kkurbjunrasher, I can't say that I'm an artist :-D
22:44:15kkurbjunIs there more that should be in it?
22:44:42rasherThe manual expects button labels for any physical buttons. I assume there is at least one?
22:45:00kkurbjunyeah, there is a power button on the top
22:45:21kkurbjunany preference on the font used?
22:45:26JdGordonhmm... is any validation done on themes when they are uploaded?
22:45:31rasherkkurbjun: Take a look at the other svgs.
22:45:37kugelI hope that's getting scaled down a bit for the theme site :)
22:45:45rasherJdGordon: A good deal.
22:45:46JdGordonis checkwps or whatever its called run?
22:46:01rasherIf it builds these days..
22:46:28JdGordonI'll try getting that to output the ram usage stats for the skins, would be useful
22:46:43kkurbjunrasher - I was, but it doesn't look like the font type is consistent
22:46:58rasherJdGordon: Looks like you'll need to fix checkwps building first..
22:47:27rasherIt's been broken for months
22:48:16domonokymaybe we should try to make automated build of checkwps so it doesnt get broken ?
22:48:39pixelmakkurbjun: I also drew side buttons on the sides at least a bit so that you could see where they are even if you couldn't see them in reality from the top view. The font introduced in the beginning and which I used (so *most* svgs have it from the Palatino family
22:48:47bluebrotherhaving a checkwps target in the build system would be nice. Plus a bootloader and simulator target :)
22:49:19mcuelenaerehmm is there a difference between TARGET_ID and MODEL_NUMBER?
22:50:36mcuelenaeredomonoky: that was going to be added in the build server rework, but because checkwps was so badly broken; it didn't got added
22:50:50 Join mcscruff [0] (
22:52:09 Join Omlet [0] (
22:53:35 Join matsl [0] (
22:55:35kkurbjunpixelma: thanks I'll add the button on the top along with the text
22:56:49JdGordonrasher: also, are remotes going to be added to the theme site?
22:56:57CIA-61New commit by kkurbjun (r22358): M:Robe 500: Add power button to SVG.
22:57:18*JdGordon doesnt even know how to build checkwps
22:57:56rasherJdGordon: Ideally they would be :|
22:58:35rasherJdGordon: cd tools/checkwps and run
22:58:35domonokyJdGordon: if wasnt broken, it would just be a tools\checkwps\
22:58:48rasherdomonoky: \!?
22:58:58JdGordonok, i'm looking at fixing it now
22:58:59domonokywindows :-)
22:59:07pixelmaM3 (Iaudio remote) could also be added - at least as a main screen
22:59:25mcscruffhi all, is there any development going on for Sony Walkmans
22:59:56JdGordonyikes, seems very very broken
23:01:44 Quit mitk ("Leaving")
23:02:44bluebrothermcscruff: no
23:03:08*bluebrother wonders how domonoky plans to invoke a sh script using windows path delimiters
23:03:25CIA-61New commit by mcuelenaere (r22359): * Onda VX747/VX767: fix target_id & MODEL_NUMBER ...
23:04:32JdGordoncheckwps is very very fubar
23:11:16CIA-61New commit by kkurbjun (r22360): M:Robe 500: Add small image.
23:14:12 Join suom1 [0] (
23:15:20kkurbjunok, I got the theme section added and I added the picture - who should I follow up with to get the www updated so that the theme site shows the M:robe 500 small image?
23:15:46***Saving seen data "./dancer.seen"
23:16:12 Join moos [0] (i=mustapha@rockbox/staff/moos)
23:17:11 Quit n1s ("Lmnar")
23:18:07JdGordonkkurbjun: hhe the mr500 sim is broken if you show the backdrop :)
23:18:30JdGordonand, can you test a quick patch? my mr500 is out of action...
23:18:40JdGordonfiddle with the region array in the wps
23:20:37*JdGordon wonders if there is a nice way to deal with all the different struct skin_token_list * members in the wps_Data struct
23:21:38 Quit mcscruff ("Leaving")
23:23:00CIA-61New commit by jdgordon (r22361): remove the limit of touch regions in touchscreen skins
23:26:39 Quit HBK ()
23:26:46kkurbjunJdGordon: region array?
23:27:24JdGordonUnhelpful: would it be theoretically doable to have more than one AA image loaded? (either the same image at different res, or different files
23:27:26JdGordonkkurbjun: ?
23:27:34kkurbjunwhat is the region array?
23:27:39CIA-61New commit by moos (r22362): Use a more appropriated word than the string commited in r21223.
23:28:23 Quit ehntoo (Read error: 110 (Connection timed out))
23:28:28 Quit Omlet ()
23:28:57kkurbjunor I guess more importantly how would you like it tested? Just to see if the user defined buttons work?
23:29:29JdGordonit shouldnt have broken anything, just changed how they are stored
23:30:15CIA-61New commit by jdgordon (r22363): remove an unused #define
23:30:15pixelmaI don't believe my buildclient did 2 bootloaders in 1 second each
23:30:45pixelmait can't be true
23:30:50JdGordonccache... its entirely cached
23:30:58JdGordonthere was no change at all for them
23:31:14bluebrotherand does it work if the output folder changes on each run?
23:31:19pixelmabut that's a VM on a not too fast laptop
23:31:45JdGordonbluebrother: ccache? no idea.. and apparently its man page cant be trusted
23:32:03bluebrotherJdGordon: yes, that's the reason I'm asking.
23:32:19bluebrothermaybe it would make more sense to name the output folder build-<targetname>?
23:32:29Bagderthe location of the output or source surely can't matter
23:32:39JdGordonyeah, it sounds really wierd if it did
23:32:59bluebrotherthat way ccache could also catch differently preprocessed sources.
23:33:19moosoops, /me wanted to refer to r21364 in his commit message :(
23:35:48JdGordonlets say Unhelpful says yes to above... how would people feel about allowing that?
23:36:19JdGordonit would be pretty impressive to show front+back covers.. no idea how it would work in the .wps though.. but would be cool
23:36:48mcuelenaerewhen one taps on the image, it could be switched? (touchscreen only)
23:37:26JdGordonboth on the screen at the same time maybe also
23:38:05pixelmarotate the cover in 3D?
23:38:16JdGordonrasher: whats the correct way to deprecate a string?
23:38:35JdGordonI'm gong to remove the AA size line in the rockbox info because its unseless now with resizing aa
23:38:55bluebrotherhmm. If ccache doesn't bother if the output directory changed, why does my ccache show that much cache misses?
23:39:05kkurbjunJdGordon: the buttons still work
23:39:13JdGordon*phew* :)
23:40:04rasherJdGordon: tools/ −−deprecate −−id LANG_ALBUMART −−inplace apps/lang/*.lang
23:40:10bluebrotheruseless? The AA size line still shows the size of the AA the theme is resizing too. The use of that line is quite limited, admittedly.
23:40:27JdGordonso dont remove it?
23:40:57JdGordonit had its use, now its just a waste imo
23:41:08bluebrotherwell, I don't object to removing. I just don't think it's completely useless.
23:41:08JdGordonbut I'm happy to leave it if others want it
23:41:28bluebrotherit's just calling it useless :)
23:41:31pixelmaIMO it could be removed
23:41:45JdGordonI cant spell unnesacery, so i try to avoid using the word :)
23:41:57bluebrotherI found that line quite confusing in the past :D
23:41:59rasherI agree, it doesn't really add much
23:42:39 Quit bertrik (Read error: 110 (Connection timed out))
23:43:06bluebrotherJdGordon: remove it! (just to make that clear :)
23:43:12 Join bertrik [0] (
23:43:15JdGordondone :)
23:44:30moosyou want to remove it for replace it with a skin buffer line? ;)
23:46:39JdGordonyes, the usage line is useful
23:46:52JdGordonor at least, mildly interesting :p
23:47:07mooshehe :)
23:47:37JdGordonand wrong :)
23:47:43JdGordon"skin usage: 18.0MB" !
23:47:53pixelmais there a way to check if it's really the WPS/skin buffer which is responsible for the bin size growth on Archos?
23:48:03pixelmaand maybe why?
23:48:26pixelmaall others saw a decrease
23:48:37JdGordonwell.. a fair bit of code was added... and because its small screen and generally small arrays anyway, not enough of the new code was offset
23:48:44JdGordonI wouldnt worry about 500b
23:49:00pixelmait could break my voice file(s)
23:49:47JdGordonyou can shrink the buffer if you really want
23:49:54JdGordonassuming you dont use the full buffer for your wps
23:50:05pixelmamaybe I could but others not
23:50:45amiconnIt's almost exclusively bss increase. Binsize itself is +36 bytes (recorder)
23:51:15amiconnHint: hover over the delta to see the details
23:53:10pixelmaso it's not the added code but somehow more space is reserved for the WPS/skin buffer?
23:54:10 Quit matsl ("Leaving")
23:54:29JdGordonthe buffer size didnt change so that shouldnt be possible...
23:55:04 Quit ej0rge (Remote closed the connection)
23:55:25JdGordon"buffer_pos-buffer_start" is the correct count of usage isnt it? (_pos is incremeneted with usage, _start is constant)
23:55:31kugelpixelma: bload-o-meter or objdiff may help (in utils/analysis), just throw the rockbox.elf before and after the commit at one of them
23:58:59pixelmahave to pass

Previous day | Next day