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

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

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

#rockbox log for 2010-08-05

00:00:10 Quit jgarvey (Quit: Leaving)
00:01:45n1samiconn: can i find any better docs on the emac than the CFPRM and mcf5249UM? they don't really tell much about the fractional stuff, sadly
00:02:53 Quit ender` (Quit: If at first you don't succeed, skydiving is not for you.)
00:03:51amiconnCFPRM.pdf tells everything you need
00:05:35n1si probably just can't read then
00:05:39amiconnSection 1.4, plus the pseudo code for move.l from acc and movclr.l in chapter 6
00:06:30*kugel has already 13h15m running time on his phone
00:07:11S_a_i_n_tkugel: That's good I guess?
00:07:51kugelit just shows how much I used rockbox on it since I got writing to storage working
00:12:11n1samiconn: thanks
00:13:54 Quit r0b- (Read error: Connection reset by peer)
00:14:17 Join stripwax [0] (
00:15:51n1samiconn: ah, so fractional mode is always treating the operands as s0.31 (or s 0.15)
00:23:46amiconnYep, that's why it's called fractional mode
00:25:04amiconnUsing it the way we do e.g. in fixmul16 you get bits 31..63 in the accumulator
00:25:24amiconnErr, bits 31..62
00:26:00amiconnIf you need true 32x32->64, you can reconstruct bit 63 using the flags, but that costs 2 cycles extra (iirc)
00:26:14n1sah, yes
00:28:50n1sam i correct in thinking that fixmul16 could be done with emac in integer mode and constructing the result from the 16 bit extension word and the acc?
00:29:44 Quit domonoky (Read error: Connection reset by peer)
00:29:52CIA-6New commit by mt (r27701): Use codeclib's mdct in wmapro. Input coeffs to the transform needed to be scaled down first by (log2(frame_size) - 3). Increases decoding speed by ...
00:31:16amiconnn1s: no
00:31:25CIA-6r27701 build result: 36 errors, 0 warnings (mt committed)
00:32:02amiconnThe emac multiplier is 32x32->40. In order to do fixmul16 with it, it would have to be 32x32->48 at least
00:32:17amiconnThe accumulator part being 48 bits wide doesn't help
00:33:19n1sah, the figure had me confused, thanks for explaining :)
00:35:35n1smt: so using the code from fixmu16l in that VECT_MUL_SCALAR_KERNEL macro seems to be the best option for coldfire too
00:36:59mtn1s: alright
00:37:49 Join r0b- [0] (
00:39:20n1sbtw, good job on using the codeclib mdct
00:41:14 Nick Ypsy is now known as YPSY (
00:41:45mtn1s: Thanks ... although I can't spot what exactly made IRAM overflow again. :(
00:44:13n1smt: the sincos_lookup tables in lib/mdct_lookup.c are in iram as well as the code in lib/mdct.c
00:45:18n1sthese things were unused by wmapro before so they were not included by the linker
00:46:30*mt forgot the mdct tables in wmapro weren't in iram too ...
00:49:01 Quit r0b- (Read error: Connection reset by peer)
00:50:02***Saving seen data "./dancer.seen"
00:50:49 Join JdGordon [0] (~jonno@rockbox/developer/JdGordon)
00:53:32stripwaxmt - nice job
00:54:16mtamiconn, n1s : Are you able to do a speed test on CF ?
00:54:24n1smt: sure
00:54:42mtstripwax: Thanks ... buschel did lots of things though :)
00:55:14stripwaxBuschel - nice job too! [for the logs :-) ]
00:55:16mtn1s : Thanks, Could you measure the slow down from removing the 512 sine window table from iram ?
00:55:25n1smt: ok
00:57:27stripwaxmt - wmapro needs a sine window *as well* as the mdt_lookup tables?
00:57:46*stripwax wonders what the '512 sine window' is
01:01:35n1sstripwax: it uses sine for windowing
01:02:25stripwaxn1s - is it "just sine"? if so, it could use the mdct_lookup tables
01:02:52stripwaxwhich are sin,cos pairs for 0<= t <= pi/8
01:03:08n1safaik, yes but i havent' looked at the math
01:03:42 Join DSGuest46963 [0] (
01:04:47stripwaxdo we evaluate av_log2 more than once per frame? e.g. inside loops or anything? if so we should probably store it alongside s->subframe_len (e.g. add an "s->subframe_len_log2" member)
01:04:49mtstripwax: That is the next step actually (dropping the window tables if possible) .. but previously the the sin_cos lookup for the mdct was a scaled one.
01:04:57 Part DSGuest46963
01:05:11mtso I couldn't merge them easily
01:05:25stripwaxmt - ah ok. should be pretty easy now though i think
01:05:36mtshould be yes
01:05:44 Join DSGuest46963 [0] (~nnscript@
01:06:25 Quit DSGuest46963 (Client Quit)
01:07:18 Join r0b- [0] (
01:10:07n1smt: on the 55kbps sample it made a 0.03MHz difference, dunno if it uses that window much though
01:10:20mtslowdown is very tiny for PP5024 (+0.05MHz -40%) on a 320kbps sample
01:11:25mtn1s: In the samples I tested the larger sizes where used more often.
01:12:11n1si'd go ahead and drop that from iram then
01:14:33CIA-6New commit by mt (r27702): Fix red by removing sine_512 from iram (2KB). minor slowdown for PP5024 (~0.05MHz) on a 320kbps sample, and ~0.03MHz slowdown for Coldfire on a 55kbps ...
01:15:00freddybI think FS #11168 is as good as I can get it if someone wants to take a peek. The fonts seem to load pretty quickly.
01:16:16CIA-6r27702 build result: All green
01:17:59 Quit Jerom1 (Read error: Connection reset by peer)
01:21:38CIA-6New commit by mt (r27703): Increase wmapro decoder output sample depth to 24 bits, makes the mean error between our output and the original floating point decoder ~5e-8. ...
01:21:49stripwaxhm, is it possible to have the iram-or-not-iram usage of certain tables in *codeclib* determined by the codec that uses codeclib? e.g. let's say wmapro doesn't use blocksizes > 1024 (or maybe rarely), so doesn't need sincos_lookup1 at all, so that doesn't need to fill up iram.
01:22:10stripwaxwhereas vorbis does regularly, so vorbis gets sincos_lookup1 in iram.
01:22:21*stripwax doesn't really know what's possible with the linker
01:23:08CIA-6r27703 build result: All green
01:23:16n1sstripwax: i *think* it's possible but it's probably easier to just compile the lib separately for each codec then
01:23:32stripwaxn1s - that's a good point
01:23:36n1s(with some differenc configuration of #defines)
01:23:40stripwaxmt - what units are "~5e-8" in?
01:24:02mtdamn .. missed that.
01:24:04stripwaxwon't rockbox rescale the 24-bits into 16-bits on target anyway?
01:24:13 Quit JdGordon (Ping timeout: 240 seconds)
01:24:15mtstripwax: that's the absolute difference
01:24:31 Part toffe82
01:24:44stripwaxmt - but assuming what scale? -1<=x<1 ?
01:25:08*stripwax thinks in bits
01:25:47 Quit dfkt (Quit: -= SysReset 2.53=- Ph'nglui mglw'nafh Cthulhu R'lyeh wgah'nagl fhtagn.)
01:26:04 Join maffe [0] (
01:26:47stripwaxso for a 16-bit quantised output, that translates to a noise difference you absolute will not see since < 1bit?
01:29:26 Quit maffe (Client Quit)
01:30:38stripwaxI guess I thought all codecs output in 32bit, didn't notice before it was configurable. 24-bit codec output (rather than dsp output) makes total sense. can we do 32-bit codec output for wmapro (would it save anything?)
01:32:45 Quit stripwax (Quit:
01:36:05mtstripwax: (logs) This is the only answer I can think of right now - my brain has already beat me to sleep.
01:36:20*mt goes to sleep
01:36:23mtgood night
01:44:06 Quit mt (Ping timeout: 240 seconds)
01:45:22 Quit simonrvn (Disconnected by services)
01:46:05 Join simonrvn [0] (
01:56:23S_a_i_n_tkugel: Would working on a large iconset be worthwhile for your cause?
01:56:36S_a_i_n_tif so, what dimensions?
02:05:38 Quit S_a_i_n_t ()
02:05:56 Quit sasquatch (Ping timeout: 240 seconds)
02:07:02 Join froggyman [0] (~me@unaffiliated/froggyman)
02:09:57 Join DataGhost [0] (~dataghost@unaffiliated/dataghost)
02:10:05 Quit DataGhost (Client Quit)
02:10:25 Quit n1s (Quit: Lmnar)
02:13:48 Join JdGordon| [0] (~jonno@rockbox/developer/JdGordon)
02:13:55 Join Judas_PhD [0] (
02:15:11JdGordon|crap, did I break touch wps?
02:15:17JdGordon|that change worked fine in the sim
02:16:01 Join S_a_i_n_t [0] (S_a_i_n_t@
02:16:08 Quit fdinel (Ping timeout: 265 seconds)
02:21:38 Join sasquatch [0] (
02:24:27JdGordon|arg, anyone else noticed the clip+ static in the audio when it buffers?
02:25:01*S_a_i_n_t has noticed clicks when the buffer is refilled in his Nanos recently.
02:25:53 Part froggyman
02:28:52 Join bunnyboi [0] (
02:29:11kugelS_a_i_n_t: 24x24 would be fine I guess
02:29:21kugelyes, it definitely would
02:29:30kugel(be worthwhile that is)
02:29:54kugelS_a_i_n_t: you can use the sdl application, you can configure it for 320x480 (or any other resolution)
02:30:06kugelI haven't really checked it it works on non-linux though
02:30:16kugelI'm currently working on cabbie 320x480 a bit
02:38:53 Nick bunnyboi is now known as Jennifur (
02:38:56JdGordon|does the clip+ have a quickscreen?
02:41:41JdGordon|kugel: to build rockbox do I just need the android sdk and ndk?
02:43:06*S_a_i_n_t wonders what/where "the SDL application" is.
02:43:09kugelJdGordon|: yes
02:43:24JdGordon|and can you commit the button code before calling it a night? (unless i missed the commit already?)
02:43:26kugelS_a_i_n_t: ../tools/configure tells you
02:43:36CIA-6New commit by bieber (r27704): Theme Editor: Added code generate/undo functionality to SkinViewer
02:45:14CIA-6r27704 build result: All green
02:47:34kugelJdGordon|: you need at least the ndk and sdk, but I recommend eclipse too because figuring out the commands to make an apk are not trivial (especially with the key signing stuff)
02:48:04JdGordon|that isnt done in the build? :(
02:48:23kugelnot yet
02:49:15 Join lestatar [0] (
02:50:01JdGordon|bugger... I couldnt get eclipse to play nicely with the adt plugin last time I tried
02:50:04***Saving seen data "./dancer.seen"
02:50:15 Quit Jennifur (Ping timeout: 240 seconds)
02:56:53lestatarhi folks, noticing odd db count on gigabeatf40 vs actual music files. have already run initialize/update db several times
02:57:21lestatardisparity is quite large: 4400 mp3 files and 6500db this correct?
02:57:35S_a_i_n_tinitialising the DB counts more than just music files IIRC
02:57:48S_a_i_n_tthough, only music files are added.
02:57:49lestatarhey saint :-)
02:58:07lestatari the above examples are not totally off base?
02:58:49S_a_i_n_tNot at all.
02:59:01lestatarok, then i will stop worrying :-)
02:59:12lestatarone other thing, re: albumart please
02:59:32lestatarhow does one handle forbidden characters like /
02:59:33S_a_i_n_tI only have a few hundred music files on my DAPs and the database init always counts into the thousands...I believe it is counting total files.
02:59:52lestatare.g. music tagged as AC/DC as artist
03:00:17S_a_i_n_tPass. I always just use "Folder.jpg"
03:00:28lestatararrgggh, some help you are :-)
03:00:43S_a_i_n_tAs I find per track AA a complete waste...
03:01:20S_a_i_n_tAre naming conventions for AA not covered in the manual?
03:01:45lestatartrue that re: waste of size, but my whole collection has embedded AA
03:02:02S_a_i_n_twhich, is pointless on your DAP ;)
03:02:03lestatarand yeah, manual is quite good, but do not recall how it handles stuff like "/"
03:02:30lestatar^^ again, true.... but i am too lazy to do otherwise :-D
03:02:47lestatarno worries, just wanted to ask...
03:02:55S_a_i_n_tafter I move my music to my DAP I always use mp3Tag to strip embedded AA, its just a waste of space.
03:03:09lestatar99% accurate aart is perfectly acceptable to me
03:03:43lestatar^^ this is interesting....i assume i can use mp3tag while gigabeat is connected and strip the art out? i might be not lazy enough to do that :-D
03:04:13lestatarsaint, you are awesome. thanks much as always :-)
03:04:28S_a_i_n_tI don't get that too often ;)
03:05:11lestatarcheers and stop being modest - all you rb people deserve medals
03:05:16lestatarbe well and thanks again :-)
03:05:24 Quit komputes (Quit: I haven't slept for ten days, because that would be too long.)
03:10:08*S_a_i_n_t wonders if there is extensions missing from
03:10:26S_a_i_n_tAFAIK new formats have been added since it was created/updated.
03:10:52S_a_i_n_t(it's the Filetype Colours thing in the wiki)
03:12:52 Join AzureWurk [0] (
03:12:52JdGordon|wouldn't surprise me
03:13:44S_a_i_n_tDo the "supported filetypes" need to be present in the viewers.config for the .colours file to work with an extension?
03:13:56S_a_i_n_tI always thought you could just feed it any extension.
03:14:55JdGordon|supported filetypes are sepcifed in 2 ways, either a hard coded list in filetypes.c (IIRC) or in the viewers.config file
03:15:13JdGordon|a colour should be settable to any extension in either of those lists
03:15:28JdGordon|oh atually maybe not
03:15:33S_a_i_n_tI have a huge "viewers.config of doom" list with like 500 or so I assume I'm covered for any event.
03:15:54JdGordon|you can use ??? for unknown exntensions...
03:16:39S_a_i_n_tbut, that would be all unknown, anything it doesn't "know" is "supported" obviously...I wonder if adding them to viewers.config is enough.
03:17:12JdGordon|do you have the file browser set to "supported"?
03:17:21JdGordon|oh never mind
03:17:26S_a_i_n_tAlso, why are you not supposed to edit viewers.config?
03:17:37S_a_i_n_tI remember seeing that in the wiki, and always wondered why.
03:17:52JdGordon|because it will be overwritten if you update from the zip
03:17:55JdGordon|otherwise no reason
03:18:11AzureWurkfound a bug with the fuze build, it nolonger shows progress bars for long files such as long audio book files and some themes like noln(no ones laughing now) was working fine last week strangly enough.....
03:18:12S_a_i_n_tAh, right. I thought it would break file handling somehow.
03:18:39 Quit lestatar (Remote host closed the connection)
03:18:52JdGordon|yeah, I have a funny feeling that the filecolours are done based on a magic flag set on known files that only come from filetypes.c. the viewers.config file is ignored for that
03:19:11S_a_i_n_tJdGordon|: Bummer :/
03:19:22S_a_i_n_tOh well, I was just playing anyway.
03:19:50JdGordon|I'd need to check the code to confirm... have a look in apps/gui/bitmap/list.c and see if you can find the colours code
03:20:49JdGordon|no, not that file
03:21:35JdGordon|ah no :) it does actually work on the file exntension
03:22:27S_a_i_n_tso, if its in viewers.config then I can set a colour? I was just looking but you'rer a lot quicker than I.
03:23:56JdGordon|ah hmm...
03:25:32 Join fdinel [0] (
03:25:43JdGordon|the colour must be in the known types list but it is done on the extension so any from viewers sohuld work
03:25:55*JdGordon| has to get back to real work...
03:49:44 Join captainkewll [0] (cf26d77e@gateway/web/freenode/ip.
03:55:54 Join Drise [0] (18d62202@gateway/web/freenode/ip.
03:56:41DriseThi task should be a quick fix.
03:59:19 Part AzureWurk
04:03:24 Quit Drise (Quit: Page closed)
04:06:46 Quit milz (Ping timeout: 276 seconds)
04:13:05 Quit DerPapst (Quit: Leaving.)
04:13:07 Join anewuser [0] (anewuser@unaffiliated/anewuser)
04:13:13 Quit TheSeven (Read error: Connection reset by peer)
04:13:21 Join TheSeven [0] (~TheSeven@rockbox/developer/TheSeven)
04:16:42 Quit Barahir_ (Ping timeout: 240 seconds)
04:17:55 Quit anewuser (Ping timeout: 265 seconds)
04:18:28 Join Barahir [0] (
04:21:44 Join anewuser [0] (anewuser@unaffiliated/anewuser)
04:23:35 Quit Rob2222 (Read error: Connection reset by peer)
04:24:07 Join Rob2222 [0] (
04:25:06 Quit freddyb (Quit: ChatZilla 0.9.86 [Firefox 3.5.9/20100401213457])
04:26:08 Quit amiconn (Disconnected by services)
04:26:11 Join amiconn_ [0] (quassel@rockbox/developer/amiconn)
04:26:30 Nick amiconn_ is now known as amiconn (quassel@rockbox/developer/amiconn)
04:27:09 Quit pixelma (Disconnected by services)
04:27:12 Join pixelma_ [0] (quassel@rockbox/staff/pixelma)
04:27:27 Nick pixelma_ is now known as pixelma (quassel@rockbox/staff/pixelma)
04:27:36 Quit yosafbridge (Quit: Coyote finally caught me)
04:29:22 Join warrow [0] (
04:30:16 Join yosafbridge [0] (
04:32:25 Quit anewuser (Ping timeout: 265 seconds)
04:34:29 Part warrow ("Leaving...")
04:40:03 Join [J] [0] (~juju2143@gateway/shell/
04:40:42 Join kugel_ [0] (
04:41:04 Quit kugel (Disconnected by services)
04:41:09 Nick kugel_ is now known as kugel (
04:41:14 Quit kugel (Changing host)
04:41:14 Join kugel [0] (~kugel@rockbox/developer/kugel)
04:43:37n17ikhMoving along.. I *would* like to be able to build it, though, since there's no autobuild system yet
04:45:53 Quit kugel (Remote host closed the connection)
04:49:28 Join milz [0] (
04:50:06***Saving seen data "./dancer.seen"
04:58:44 Quit captainkewll (Quit: Page closed)
05:01:54 Join AzureSky [0] (
05:05:51 Quit AndyI (Ping timeout: 248 seconds)
05:09:23n17ikhso, uh.. How do I actually build the APK?
05:09:48n17ikhmake zip gives me what looks like it should be in /data/
05:10:45tmzttry just make
05:10:49tmztthat works on other targets
05:11:00tmztthe whole things in svn then?
05:11:11tmztdo you need an sdk to build it?
05:11:18n17ikhYou need the SDK and NDK
05:11:33JdGordon|once oyu figure out the commands let us know :)
05:11:58n17ikhthat is somewhat less than helpful
05:12:10n17ikhso how does kugel do it?
05:12:19JdGordon|probably in eclipse
05:12:31 Join AndyI [0] (~pasha_int@
05:15:42 Nick YPSY is now known as Ypsy (
05:17:07n17ikhI think installing a complete IDE is a little beyond where I wanted to go for this
05:17:27tmztyou shouldn't need an ide
05:17:30tmztjust ant
05:17:32JdGordon|ok good.. so figure out the command line options
05:17:35tmztand the sdk, ndk
05:17:40JdGordon|someone needs to do that eventually anyway :)
05:17:43tmztI don't have it installed
05:17:49tmztI'm clearing up some room
05:17:57tmzteventually I'm going to port X to ndk :)
05:18:16tmztlet me see what you guys have in svn for now
05:18:57n17ikhwell, I downloaded the apk that was posted earlier, but it looks like it's built off an older version
05:19:06 Nick Ypsy is now known as YPSY (
05:19:15n17ikhno trackball support and it's using a tiny, tiny font by default which is impossible to use with the touchscreen :/
05:19:27tmztit's pretty cool that he already has rockbox as a service
05:19:43tmztI've been looking at it as a multimedia framework on embedded for quite a while
05:19:54tmztsmaller than most given where it came from
05:20:01tmztand a good set of codecs
05:20:49n17ikhwait, what
05:20:55n17ikhI'm not sure I understand the touch controls
05:20:57 Quit sevard (Read error: Connection reset by peer)
05:21:12 Quit rvvs89 (Read error: Connection reset by peer)
05:21:14n17ikhshould it be displaying buttons for me?
05:21:34n17ikhseems like just touching anywhere doesn't do anything, but swiping does.. but results are somewhat unexpected
05:21:36JdGordon|n17ikh: where did you get the impression that it was finished or working enough to be usable except by devs?
05:21:37 Join sevard [0] (sev@
05:22:03n17ikh[21:49] <S_a_i_n_t> it's quite usable by my understanding.
05:22:27n17ikhanyway, I wasn't really expecting it to be finished
05:22:44n17ikhbut I was expecting to be able to at least browse around the menus a bit
05:22:55JdGordon|so go read the manual
05:23:15n17ikhthere's a manual?
05:23:15JdGordon|have you used a touchscreen target with rockbox before?
05:23:40n17ikhNo, I have not.
05:24:10 Nick Strife1989 is now known as Strife89 (
05:24:17JdGordon|then quite frankly, rockbox on android is not ready enough for you
05:26:37 Join bunnyboi [0] (
05:28:41tmzttouchscreen is still the regions of the screen thing?
05:28:48tmztor are both modes going to used
05:29:06n17ikhAt least, now that I know what it is, I can use it
05:29:45tmztJdGordon|: the README seems to explain how to build without eclipse
05:31:39 Quit bieber (Ping timeout: 240 seconds)
05:31:55 Join bieber [0] (
05:33:41 Quit Horscht (Quit: Verlassend)
05:44:06JdGordon|tmzt: buidling shouldnt be a problem. making the apk is
05:50:13 Quit PouncePony (Ping timeout: 260 seconds)
05:51:32JdGordon|kugel: signing the apks is another good reason to split the file out of it. :) no need to have the build system sign and make the apk each time
05:52:23tmztyou can sign with jartool
05:52:29tmztor the android equivalent
06:06:47 Join rvvs89 [0] (
06:08:13 Quit rvvs89 (Changing host)
06:08:13 Join rvvs89 [0] (ivo@pdpc/supporter/base/rvvs89)
06:28:28 Part AzureSky
06:28:58 Join AzureSky [0] (
06:31:24 Quit fdinel (Quit: Miranda IM! Smaller, Faster, Easier.
06:50:08***Saving seen data "./dancer.seen"
06:59:58 Join ruckus [0] (
07:02:00ruckushey peoples. like there's an option to prevent rockbox from entering certain files into the database, is there an option to prevent certain files from being scrobbled by the functionality ?
07:08:25 Quit Stummi (Excess Flood)
07:08:34 Join Stummi [0] (
07:12:11 Quit S_a_i_n_t (Ping timeout: 260 seconds)
07:12:45 Join S_a_i_n_t [0] (S_a_i_n_t@
07:28:25 Join mt [0] (
07:28:39 Quit mt (Changing host)
07:28:39 Join mt [0] (www-data@rockbox/developer/mt)
07:30:24 Quit bunnyboi (Quit: Never look down on someone unless you're helping them up.)
07:35:51 Join [Saint] [0] (S_a_i_n_t@
07:36:43 Quit S_a_i_n_t (Ping timeout: 276 seconds)
07:41:26mtamiconn: ping
07:44:21mtI have 2 CF asm versions of vector_fixmul_scalar (in wmapro) - anyone could review/test if they work correctly ?
07:50:50 Join bros [0] (muffinflav@gateway/shell/
07:51:57brosI am trying to build rbutil on netbsd. I have run into two errors. 1. sh: ../../tools/ not found (appears multiple times) 2. gmake[1]: /bin/uic: Command not found
07:53:38JdGordon| <- pacman in the progressbar :D
07:55:26JdGordon|haha "GAME OVER" when the track ends
07:55:35 Quit [Saint] (Ping timeout: 260 seconds)
07:56:21 Join S_a_i_n_t [0] (S_a_i_n_t@
07:58:29 Join Buschel [0] (
07:59:40 Join angelwolf71885 [0] (
08:00:19bieberbluebrother: The installer looks good. When it downloads the font pack, does it set the registry key for its location?
08:01:14JdGordon|hey bieber, i pasted the build error last night, did you happen to see it?
08:01:34bieberNo, are you still getting it?
08:02:12 Quit angelwolf71885 (Client Quit)
08:02:28CIA-6New commit by mt (r27705): Drop the the mdct files from libwmapro since they are no longer needed.
08:04:17bieberJdGordon|: Just found your message
08:04:22CIA-6r27705 build result: 79 errors, 0 warnings (mt committed)
08:05:07bluebrotherbieber: no, the installer doesn't set anything about the fonts, it just downloads and extracts them to the program folder.
08:05:33bieberThat would be my only concern. If it downloads them but they don't show up in the editor, it could be awfully confusing
08:06:31bieberCan you set registry keys from the installer?
08:07:16ruckushey peoples, just in case nobody saw my question - i know there's an option to prevent rockbox from entering certain files into the database, so is there an option to prevent certain files from being scrobbled by the functionality ?
08:08:41bluebrotheryes, I could do that. Given that the registry keys the theme editor uses don't change ;-)
08:09:29S_a_i_n_tbieber" Pingy-pingy-ping-ping?
08:10:11 Join bertrik [0] (~bertrik@rockbox/developer/bertrik)
08:10:16CIA-6New commit by mt (r27706): oops ! shouldn't have removed mdct_tables.
08:10:45bieberJdGordon|: Could you try compiling the latest build? That flag isn't used in that file anymore, but a different one. It _should_ work fine, since it's in the QGraphicsItem namespace and the class that uses it extends QGraphicsItem
08:10:48bieberS_a_i_n_t: Hey
08:11:07S_a_i_n_tbieber: Why does the themeeditor add a newline after a comment?
08:11:24bieberDo you mean when you generate code?
08:11:50S_a_i_n_tOr if you click "update code" it screws your formatting.
08:11:55CIA-6r27706 build result: 240 errors, 0 warnings (mt committed)
08:12:23Buschelmt: your IRAM change did not affect PP5024 (and PP5022, MCF5250, S5L870X). so, the tests regarding performance loss should have been made on other targets
08:12:57Buschelmt: details about which window types are used how often are placed in FS #11498
08:13:07Buschelmt: nice to see the changes are in now :)
08:13:11bieberS_a_i_n_t: It won't ever keep your code perfectly the same after generating code, because the parse tree doesn't keep all of the formatting info
08:13:15bluebrotherhmm, the uninstaller should also remove the Theme Editor settings now I'm thinking about it.
08:13:39bieberThe newlines after comments are due to the way the parser handles comments and the \n characters after them, iirc
08:13:49 Join Zagor [0] (bjst@rockbox/developer/Zagor)
08:13:53bluebrotherI'll change that tonight.
08:13:57JdGordon|bieber: that was from last night, I'll try again in a few hours when I get home
08:14:22bieberCan anyone else confirm compiling the latest Theme Editor revisions?
08:14:27S_a_i_n_tbieber: I also dislike the way it keeps moving my %Vf's and %Vb's out of my %V and &Vl there a way to stop it from doing that?
08:14:59bieberS_a_i_n_t: I could, but it would require some special code just for those cases
08:15:20S_a_i_n_tAha, I guess I just have to not click "Update Code" ;)
08:15:27S_a_i_n_tI was just checking.
08:15:32bieberYeah, think of it like Dreamweaver
08:15:52bieberYou can edit your documents in WYSIWYG mode, but it's probably not going to end up formatted the way you would type it yourself ;)
08:15:56S_a_i_n_tAaaaargh! No thanks!
08:16:00mtBuschel: according to the histogram 512 and 256 are the least used, and since iram overflowed by 1488 bytes, 512 had to be my choice. I hope I could merge the window tables with the mdct lookup tables or the windowing code from wma though.
08:16:05S_a_i_n_tI actually *like* the themeeditor! :P
08:16:13mtBuschel: Have you seen the CF asm ?
08:16:46bieberFor what it's worth, though, Dreamweaver (at least when I used it some odd years ago) does make a pretty decent code editor
08:16:47bluebrotherbieber: builds fine for me (on linux that is)
08:17:01bieberIt's just that if you let it write your code for you, you'll end up getting laughed at
08:17:19CIA-6New commit by mt (r27707): Fix reds.
08:17:20Buschelmt: your decision was right, but I would like to see measures of performance loss on impacted targets. you tested on PP5024 which was not affected by this change at all (the measured diff seems to be noise)
08:18:01S_a_i_n_tbieber: Did you manage to find a "MS-Paintesque" widget?
08:18:06mtBuschel: ah ok .. hope someone could help us with that.
08:18:19bieberS_a_i_n_t: Haven't looked yet, lemme see if I can find one
08:18:27 Quit ruckus (Quit: Leaving.)
08:18:35Buschelmt: I saw that you've posted CF asm, but I do not really understand CF asm...
08:18:40S_a_i_n_tDon't let it stop you doing important stuff though ;)
08:18:47CIA-6r27707 build result: All green
08:18:58S_a_i_n_tI just thought it would make the themeeditor a more "all-in-one" editor.
08:19:02JdGordon|bieber: well it builds fine on my pc at home, but that one never had a problem
08:19:05JdGordon|my laptop has the issue
08:19:24bluebrotherwhich issue?
08:19:40bieberA quick Googling doesn't find anything, but it's kind of a hard thing to look for, since if you search for "Qt Paint Widget," you get a million pages about the paint() functions of various Qt widgets
08:20:21 Join mitk [0] (~mitk@
08:21:04bluebrotherS_a_i_n_t: what do you mean by "MS-Paintesque" widget?
08:21:28S_a_i_n_tbluebrother: A bitmap editor to chuck in the themeeditor
08:21:44*S_a_i_n_t thought this would be handy
08:22:03JdGordon|sounds out of scope to me
08:22:41bluebrotherah, something for bitmap strips and such?
08:22:41S_a_i_n_tIt means one app can then build an entire theme, sounds entirely in scope to me.
08:22:50S_a_i_n_tbluebrother: Yep.
08:23:09S_a_i_n_tbluebrother: Backgrounds, simple editing, etc.
08:23:16Buschelmt: n1s or amiconn should review your CF asm. both know CF very well
08:23:47JdGordon|it should have OS integration to load the .bmp in a proper image editor
08:23:49bluebrotherthe problem I'm seeing is that you'll end up with a inferiour bitmap editor compared to a "real" bitmap editor. Unless someone spends quite a lot of work making it a fully bitmap editor ß−− which is a bit out of the scope of the theme editor IMO.
08:24:12mtBuschel : yeah but they are not here now apparently, so I'll just have to wait
08:25:00Buschelmt: but both read the logs. if you ask for review and the place the link here, you will most likely get a response
08:25:55CIA-6New commit by bieber (r27708): Theme Editor: Removed empty lines after comments in generated code
08:26:09 Quit Buschel (Quit: ChatZilla 0.9.86 [Firefox 3.6.8/20100722155716])
08:26:33bluebrotherQt has an example called "Plug and Paint" which does some bitmap editor thing (though it's main purpose is to be an example for plugins). I don't think there's a ready-made widget for an editor around
08:26:48JdGordon|bieber: I wonder if qt4 on my laptop is fucked.. I'm still getting the grey unmovable window for the timer window on startup
08:27:03bieberIt's sounding very possible
08:27:32CIA-6r27708 build result: All green
08:28:19bluebrotherJdGordon|: have you already tried removing the Theme Editor configuration file?
08:28:22bieberS_a_i_n_t: Getting rid of the newlines after comments was actually a really easy fix, but the %Vf and etc. would be a little trickier
08:28:55JdGordon|bluebrother: in ~/.config/ ? yes
08:29:40bluebrotheryep. Ok, too bad.
08:32:17JdGordon|bieber: a "generic target" configure option would be very handy... RaaA lets you set any lcd dimension and it wont be in the devicedb
08:32:44JdGordon|i.e choose egeneric and let me then specify the width/height
08:33:06S_a_i_n_tbieber: Don't beat yourself up about it ;)
08:33:17S_a_i_n_tThe newlines way the more annoying of the two IMO
08:33:24bieberI think it still supports #screenwidth and #screenheight in the cfg file, I'll just have to add a "Manual settings" option to the new project dialog
08:33:47 Quit mt (Quit: CGI:IRC)
08:34:02JdGordon|right, remembering them doesnt work :)
08:34:09JdGordon|and that also requires you know about them to begin with
08:35:05CIA-6New commit by bieber (r27709): Theme Editor: Fixed ParseTreeModel's paramChanged function to emit the correct dataChanged range
08:35:52 Join bmbl [0] (~Miranda@unaffiliated/bmbl)
08:36:11 Part AzureSky
08:36:36CIA-6r27709 build result: All green
08:37:56CIA-6New commit by mt (r27710): update wmapro's README.
08:38:18 Quit panni_ (Quit: ( :: NoNameScript 3.81 :: ))
08:39:20CIA-6r27710 build result: All green
08:44:39S_a_i_n_tbieber; What do I need to do to get my animations animatin' in the editor? 0_o
08:44:45S_a_i_n_tThat one has been bugging me.
08:45:29JdGordon|speaking of which... using sublines for animations is wrong :D
08:45:57bieberS_a_i_n_t: Open the timer panel, click start
08:46:18S_a_i_n_tJdGordon|: Yeah, whatever :P
08:46:26S_a_i_n_tNice try in the Forums by the way ;)
08:47:30JdGordon|go on.. try to tell me you dont have your animation strips in the correct order
08:48:12S_a_i_n_tI don't have any animations that could use your implementation I don;t believe.
08:48:22*S_a_i_n_t does NOT want to get into this again.
08:50:12***Saving seen data "./dancer.seen"
08:51:22 Quit Judas_PhD (Quit: Aaarrrggghhh!!!)
08:51:39 Join ender` [0] (
08:54:45 Join Judas_PhD [0] (
08:55:45bieberCan I temporarily revert files in SVN, but still save my changes without committing?
08:57:38JdGordon|svn diff > file; svn revert; patch
08:57:43JdGordon|or swicth to git
08:58:57bieberAw crap
08:59:08bieberIt looks like I broke something in a previous revision and didn't notice it
08:59:09 Join petur [0] (~petur@
08:59:09 Quit petur (Changing host)
08:59:10 Join petur [0] (~petur@rockbox/developer/petur)
08:59:28 Join LinusN [0] (linus@rockbox/developer/LinusN)
08:59:44bieberHow do I get a hold of the commit log?
08:59:53JdGordon|svn log
09:12:43 Join einhirn [0] (
09:13:53bieberWait, can an image be displayed in more than one place in a skin?
09:14:44bieberJdGordon|, S_a_i_n_t: ^
09:15:12JdGordon|once per viewport
09:15:13S_a_i_n_tonce its %xl'd you can %xd it as many times as you like
09:15:15bieberOh, okay
09:15:32S_a_i_n_t(per viewport)
09:15:47JdGordon|no, only once per viewport
09:16:09S_a_i_n_tyeah, I corrected myself.
09:16:11bieberI was just looking at my code for %Vd and thought it was utterly bizarre that it was copying the image object on display, but I guess it makes sense now
09:16:17JdGordon|you can use the tag as many times as you like, but the last one which is actually in a "true" branch will be the subimage displaed
09:18:07 Join bimbel [0] (~Miranda@unaffiliated/bmbl)
09:21:56 Quit bmbl (Ping timeout: 258 seconds)
09:24:52 Nick bimbel is now known as bmbl (~Miranda@unaffiliated/bmbl)
09:26:30 Quit Judas_PhD (Quit: Aaarrrggghhh!!!)
09:28:04CIA-6New commit by bieber (r27711): Theme Editor: Fixed image positioning bug, implemented saveGeometry() in RBImage
09:29:33CIA-6r27711 build result: All green
09:30:56 Join Judas_PhD [0] (
09:38:04CIA-6New commit by bieber (r27712): Theme Editor: Stopped bug that caused segfault if SBS element was moved from within a WPS preview
09:39:35CIA-6r27712 build result: All green
09:41:50 Quit Strife89 (Quit: Night. ^^)
09:43:08S_a_i_n_tbieber: The "FM Radio" device configuration TAB needs a "Radio Art Available" option, at present this setting is defined by "Album Art Available" in the "Playlist/Song Info" TAB which isn't really relevant to the .fms at all.
09:43:46bieberHmm, are they represented by the same tag?
09:43:58S_a_i_n_tHow do you mean?
09:44:01S_a_i_n_tOh, yes.
09:44:23 Quit JdGordon| (Quit: leaving)
09:44:24bieberHmm, I'll have to add a way to duplicate them
09:44:38bieberOut of curiosity, where does "FM Art" come from?
09:44:49S_a_i_n_tThe user needs to add it.
09:45:06S_a_i_n_tThe naming convention is the preset name IIRC
09:45:34S_a_i_n_t...possibly the frequency also, JdGordon is the one to ask there though.
09:46:00 Join Rob2223 [0] (
09:46:11S_a_i_n_tIt just makes *NO* sense having the radio art dependent on a setting in playlist/song info
09:46:31S_a_i_n_tIt took me ages to figure out how to turn radio art off lol ;)
09:46:50S_a_i_n_tI thought "surely it won't use the same setting as regular AA"...but, yep ;)
09:49:01 Quit Rob2222 (Ping timeout: 265 seconds)
09:49:34bieberYeah, all those settings are mapped 1:1 onto the tags, at the moment I don't have a way to map more than one option onto the same tag
09:50:30S_a_i_n_tAha, hmmm...its definitely something worth figuring out.
09:51:05S_a_i_n_tAs radio art and album art are two clearly different things, yet they use the same tag.
09:52:48S_a_i_n_tbieber: Perhaps it is easier to convince JdGordon to add another tag for Radio Art
09:53:04bieberNah, it won't be too much trouble
09:53:12S_a_i_n_tIt doesn't seem to make much sence for them to share the same tag when I really think about it.
09:59:31bieberEh, they're basically the same thing
10:00:22 Join Peddy [0] (
10:01:17Peddyin the 'playlist' menu on my Fuze with the svn rockbox, the filename rather than ID3 tag of music is displayed. Is there a way to display id3 tags instead?
10:01:39 Join JdGord [0] (~jdg@
10:03:17S_a_i_n_tPeddy: Which playlist menu?
10:03:34CIA-6New commit by bieber (r27713): Theme Editor: Implemented saveGeometry() in RBAlbumArt
10:04:46PeddyS_a_i_n_t, the 'view current playlist' window
10:05:23CIA-6r27713 build result: All green
10:05:41S_a_i_n_tMine appears to be displaying the id3 tags, but, it's pretty hard to confirm that my files are named sensibly
10:05:48S_a_i_n_tie: The same as the id3 tags
10:06:05S_a_i_n_tSo, you may indeed be correct.
10:06:44PeddyBizarre. Yes, the majority of my songs have filenames similar to their tags, but sometimes the filename has the track number at the start
10:06:53Peddyand so it appears as 1. 1. Trackname'
10:06:57pixelmathe playlist viewer doesn't display ID3 tags
10:06:59Peddyin playlist menu
10:07:03amiconnmt: Version 2 needs the emac in integer mode, and it will only work correctly if you are sure that each product doesn't overflow 40 bits
10:07:31amiconnOh, and it is (a) incorrect and (b) sub-optimal
10:08:10Peddypixelma, well that confirms it. Is this deliberate, or could I code a patch for consideration?
10:08:56amiconnIn integer mode the right shift by one isn't needed - it is only needed if you use the emac to get the top "32" bits of a 32x32->64 bit multiply
10:09:14amiconnAlso, what are you doing with accext01 after reading it? Nothing, it seems...
10:09:44amiconnIt also misses right shifting by 16 somewhere
10:09:45pixelmaPeddy: it's not possible currently and using tags there would mean looking into each file - there are quite a number of threads in the forums about it. Maybe it's ok-ish on targets with dircache and database loadable to RAM or so
10:10:36pixelmaif a patch would be considered depends on the implementation and the speed penalty, especially for those who don't want or need it
10:10:53S_a_i_n_tIf dircache and 'load database to RAM' is on, it should be fairly "cheap"
10:11:02amiconnOh, and it's sub-optimal because you're causing an emac stall. This can be reduced by moving the leal.l for %[src] up, right after the last mac.l
10:11:07S_a_i_n_tIt would still mean opening each file though.
10:11:25pixelmaPeddy: I think you'll even find patch(es) in the tracker already
10:11:44pixelmaS_a_i_n_t: there are targets without the two things (low-mem)
10:11:54PeddyWould it not be easier to link the playlist the tag database?
10:12:01S_a_i_n_tpixelma: Yes, indeed there are.
10:12:11brosWhen I try to run rbutil as root I get "no protocol specified"
10:12:13PeddyI'll have a look, pixelma. Would be perfect...
10:12:19amiconnIn this case it might even be advantageous to switch modes for blocks of 4 (if version 2 isn't sufficient after fixing it)
10:13:58S_a_i_n_tJdGord: pixelma: anyone: Is someone able to give me an example of a conditional UI viewport? I want to have a different UI viewport for the Rec Screen...but, I appear to be botching it.
10:14:13S_a_i_n_t'Tis the last thing left before this theme is finally finished.
10:14:20JdGordI don't think anyone has one
10:14:27pixelmanever tried
10:14:48S_a_i_n_tJdGord: Didn't you make a WPS/SBS when you first figured it out?
10:14:59S_a_i_n_t(was wondering if you posted it anywhere)
10:15:12JdGordPastebin maybe
10:15:15pixelmaJdGord: is it possible (with expected result) to put %Vb/f into a conditional
10:15:17S_a_i_n_tAh :/
10:15:17JdGordSo it's gone
10:15:34JdGordYeah that should work
10:16:10pixelmaok, I'll try it out. It could potentially simplify my theme
10:16:48S_a_i_n_tpixelma: You actually reminded me of that just now, I was thinking about that this morningbut forgot to mention it to you.
10:17:08S_a_i_n_tI remembered you said you had a theme that changed the fg/bg colours based on the time(?)
10:18:35 Quit JdGord (Quit: Bye)
10:19:25 Join n1s [0] (~n1s@rockbox/developer/n1s)
10:19:26pixelmaS_a_i_n_t: not depending on time, but e.g. battery level (in numbers) gets a more "warning" colour if battery is low - or the volume one if you get into the +0dB area
10:19:31 Join jordan` [0] (
10:20:18 Quit jordan` (Remote host closed the connection)
10:20:54S_a_i_n_tpixelma: %?bl<|||||||%Vf(000000)|||> (just an example) should indeed be possible.
10:21:44pixelmabut I thought about it recently and since the clock tags can be used conditionally, having a theme that e.g. changes to a "night mode" based on time should be doable too
10:22:46pixelmaS_a_i_n_t: I need to try and since the theme is already coded and working with different viewports, it needs some more editing...
10:23:27S_a_i_n_tpixelma: Yeah, I have a theme that does a "night mode" effect somewhere.
10:23:51S_a_i_n_tpixelma: Bonus points if you manage to make it conditional to your birthdate :P
10:24:05S_a_i_n_t(display a backdrop with a present on it) ;)
10:25:02pixelmait doesn't use a backdrop and I don't think that you can use backdrops conditionally ;)
10:25:46S_a_i_n_tPretty sure you can use a conditional backdrop...I seem to remember trying this once.
10:26:23S_a_i_n_tI may have just %xl'd a screen-sized image at 0x 0y instead though.
10:28:39S_a_i_n_thehehe....I managed to crash the theme editor :P
10:28:45S_a_i_n_tbieber: ^
10:28:59 Quit bug2000 (Ping timeout: 276 seconds)
10:29:49bieberS_a_i_n_t: What crashed it?
10:30:45bieberLoading a screen-sized image at (0,0)?
10:30:48S_a_i_n_tTrying to set a UI viewport with the .sbs conditional to %cs/rec screen
10:31:11bieberShow me a code snippet I can copy/paste to reproduce it?
10:31:47S_a_i_n_tIt tells me it parses correctly, but when I change the "current screen" value in the "Misc." TAB the editor just vanished off the screen ;)
10:32:44S_a_i_n_tI'll just send the whole theme, easier (in a lazy mood,'s small)
10:33:39S_a_i_n_tbieber: Can you accept a DCC send with your client?
10:33:46bieberI have no idea, give it a try
10:34:04S_a_i_n_tit's awaiting confirmation
10:34:12S_a_i_n_t(from your end)
10:34:18bieberI don't see anything, so I guess that's a no
10:34:29S_a_i_n_tSweet, one sec.
10:36:38 Join DerPapst [0] (
10:40:02bieberHmm, the problem isn't popping out at me, I'll have to take a closer look in the morning
10:40:20S_a_i_n_tbut, is it crashing for you?
10:40:28PeddyEasytag is top-tier.
10:40:43S_a_i_n_tbieber: ^
10:40:56bieberYeah, it's crashing
10:41:06bieberJust not as nicely as I like :P
10:41:09S_a_i_n_tAha, glad I'm not insane then ;)
10:41:25S_a_i_n_tbut, crap that its crashing...I feel guilty, like I broke it :P
10:42:17S_a_i_n_tbieber: Yes, I wouldn't really call it a crash (though technically it is), its more just abruptly vanishing from view ;)
10:43:20bieberEh, it's an assert failure, so technically it's ending itself, but it's a crash :P
10:43:34bieberAnd it turns out it's not the VI thing that's crashing it, it's another line further down
10:44:43S_a_i_n_tbut, it never crashed until I added the additional UI viewport.
10:45:24bieberYou can comment it out and it'll still crash
10:46:14S_a_i_n_tHmmmm, comment out the %?cs line, *and* the %Vi(a) line?
10:46:16bieberIt was line 132, and I got it fixed now :)
10:46:32S_a_i_n_tHmmmm...what was the problem?
10:46:38S_a_i_n_tIt said it parsed correctly.
10:47:08S_a_i_n_tbieber: Can you please pastebin the corrected .sbs for me?
10:47:16S_a_i_n_t(in entirety)
10:47:17 Join jordan` [0] (
10:47:22bieberNo, it was the editor I fixed, not the SBS
10:48:18bieberThe problem was that when I fixed it to allow you to use conditionals with a single branch basically as if statements, I left a hole open that could lead to a segfault
10:48:54bieberAnd %?Rp<%Sx(Recording)> triggered it
10:48:56S_a_i_n_tGlad it wasn't me, that would have been embarrasing.
10:49:36CIA-6New commit by bieber (r27714): Theme Editor: Fixed bug that crashed conditionals nested in sublines, began work on making progress bars implement RBMovable
10:49:43bieberNope, the embarrasment is all mine :)
10:50:12S_a_i_n_tyou can't call it embarrasment when its a WIP...its called "learning" then ;)
10:50:13***Saving seen data "./dancer.seen"
10:51:00 Join JdGordon [0] (~jonno@rockbox/developer/JdGordon)
10:51:10CIA-6r27714 build result: All green
10:51:34bieberThere we go :)
10:51:39bieberOkay, I'm off to sleep
10:51:43S_a_i_n_tbieber: One thing I haven't noticed...does the editor outline the UI viewport?
10:51:54S_a_i_n_tah, ..sorry. Night.
10:52:07bieberIf you check the "Show viewports" option in the config panel, UI viewport gets outlined in blue
10:52:10bieberAll the other ones get outlined in red
10:52:34S_a_i_n_tAh, many setting I haven't discovered yet. I'm still learning them all.
10:52:43S_a_i_n_tThanks bieber, sleep well.
10:59:03JdGordonpixelma: %?bl<|||||||%Vf(000000)|||> (just an example) should indeed be possible. <- surely %?if(%bl, <, 40)<something> :)
10:59:47JdGordonyou cant doconditional backdrop
10:59:54*JdGordon quickly reading the backog
11:00:34S_a_i_n_tYeah, as an afterthought I remembered I did it with %xl and a fullscreen image at 0x,0y
11:03:29pixelmaif you want to nag about your ideas again - I see your suggestions as two different things. One is the comparison logic, the other the "collapsing" of bitmap strips and which I have differing opinions about either. The collapsing idea isn't new btw. and had already been abandoned once
11:04:22JdGordonof course its two seperate ideas :)
11:18:05 Part bros
11:26:08 Quit Judas_PhD (Quit: Aaarrrggghhh!!!)
11:33:10 Join Judas_PhD [0] (
11:44:32 Quit JdGordon (Ping timeout: 265 seconds)
11:55:56 Join guymann_ [0] (
11:56:48 Quit guymann (Ping timeout: 258 seconds)
11:57:15n1shm, my awesome asm for VECT_MUL_SCALAR_KERNEL is quite a bit slower than just using the fixmul16 macro in the c loop :/
12:01:13 Quit Peddy (Quit: Leaving)
12:04:30n1shmm, could be that gcc is stacking the loop counter
12:07:25 Join JdGordon [0] (~jonno@rockbox/developer/JdGordon)
12:12:05n1shmm, nope, it isn't probably cache effects then :/
12:12:36 Quit saratoga (Ping timeout: 252 seconds)
12:14:03 Join JdGordon1 [0] (~jonno@
12:16:23 Quit JdGordon (Ping timeout: 265 seconds)
12:25:20 Quit mc2739 (Ping timeout: 248 seconds)
12:26:24 Join mc2739 [0] (~mc2739@rockbox/developer/mc2739)
12:26:38 Quit einhirn (Ping timeout: 276 seconds)
12:29:49S_a_i_n_tJdGordon1: What's the maximum timeout that could be specified in %t(<foo>) and how is it written, is it seconds only or does it go to min:sec? hh:min:sec even?
12:30:24JdGordon1seconds, or deci-seconds
12:30:42JdGordon10.1 is the smallest timeout
12:30:45S_a_i_n_tsweet, do you know the max timeout?
12:31:36JdGordon1no,( 2^15-1)/10
12:31:52 Quit detaos (Ping timeout: 260 seconds)
12:32:09JdGordon1no,... ( 2^31-1)/10
12:32:51S_a_i_n_t~270 odd.
12:33:07S_a_i_n_tnot *that* big, unless I'm terribly bad at math
12:33:48S_a_i_n_t~3200 ish?
12:35:08CIA-6New commit by nls (r27715): libwmapro: remove coldfire fixmul24 as it is both incorrect and unused, fix fixmul 31 comment.
12:35:35JdGordon16+ years worth of seconds
12:35:54S_a_i_n_tshould be sufficient ;)
12:36:49CIA-6r27715 build result: All green
12:38:10S_a_i_n_tI'm just wanting to print a "No RDS Available" or similar for a short time every time the .fms is entered and not seen again until the next time the .fms is entered instead of timing out at some point, a year would be a ridiculous amount for that even ;)
12:38:23S_a_i_n_tI thought the number would have been a lot smaller than that.
12:40:44JdGordon1%?mh<on|off> right?
12:41:22S_a_i_n_tfor what? that's correct for "hold on/off", yes.
12:42:37JdGordon1ugg... the image strips for hold are upsidedown for me :(
12:44:43*JdGordon1 really needs to change the token values to start at 0 not 1
12:49:04JdGordon1AlexP: pixelma(regardless of what you think of the change)... can I get some help with the manual text?
12:49:54JdGordon1I'm changing %xd to allow 2 option params, the first being another tag, the second being an offset into the bmp strip. I have no idea how to put that into TeX
12:50:17***Saving seen data "./dancer.seen"
12:51:06JdGordon1what is the difference between \\ and \newline?
12:54:34pixelmain a table? - \\ starts a new table row, \newline keeps the linebreak in the cell
12:57:27JdGordon1 is the manual change
12:58:46 Join kugel [0] (~kugel@rockbox/developer/kugel)
12:58:56kugel"JdGordon1 really needs to change the token values to start at 0 not "
12:59:00kugeljust stop being lazy
12:59:23kugeldon't change how it was over *several* years only because you can't remember it
12:59:53JdGordon1do you have any idea wtf you are talking about? or do you just like spouting crap?
13:00:21kugelyou want to swap the order of binary coniditionals
13:00:55JdGordon1you idiot... I want to change the whole thing to be base 0 indexed like C, not 1 indexed like nonesense
13:01:37kugelindex? $?tag<true|false> is indexed?
13:01:41JdGordon1did I say I want to change the order? no
13:01:45JdGordon1just STFU
13:02:05pixelmaheh, both calm down and don't get personal. Please
13:05:08kugelbinary conditionals are not index, it's just string or NULL. I don't know how that could be changed to some 0 based indexing (not without swapping anyway)
13:06:12JdGordon1"jd, I dont understand what you are saying, kindly explain?" instead of "you moron" perhaps?
13:08:58 Quit JdGordon1 (Quit: Leaving.)
13:16:50 Join Topy [0] (
13:16:56 Quit efyx (Quit: Quitte)
13:20:30 Quit T44 (Ping timeout: 245 seconds)
13:22:27 Join Xerion_ [0] (~xerion@
13:23:03 Quit n1s (*.net *.split)
13:23:03 Quit r0b- (*.net *.split)
13:23:04 Quit bluebrother (*.net *.split)
13:23:05 Quit Xerion (*.net *.split)
13:23:05 Nick Xerion_ is now known as Xerion (~xerion@
13:25:58 Join n1s [0] (~n1s@rockbox/developer/n1s)
13:26:12gevaertskugel: are the current apk builds still screen size specific?
13:26:26kugelgevaerts: yes
13:26:41 Join r0b- [0] (
13:26:47gevaertsCould you do a 240x320 build one of these days?
13:27:01*gevaerts didn't see those in the logs yet
13:27:12kugelsure, will do in a few minutes
13:27:25 Join pamaury [0] (
13:27:25 Quit pamaury (Changing host)
13:27:25 Join pamaury [0] (~quassel@rockbox/developer/pamaury)
13:28:01 Join bluebrother [0] (
13:28:01 Quit bluebrother (Changing host)
13:28:01 Join bluebrother [0] (~dom@rockbox/developer/bluebrother)
13:28:37CIA-6New commit by jdgordon (r27716): Add a T type to the tag table which allows parameters to be a single tag
13:28:50CIA-6New commit by jdgordon (r27717): New feature for the %xd() (display a preloaded image) skin tag. ...
13:30:10CIA-6r27716 build result: All green
13:31:35 Join JdGordon [0] (~jonno@rockbox/developer/JdGordon)
13:31:46CIA-6r27717 build result: All green
13:32:41 Join stooo [0] (
13:32:47 Quit stooo (Client Quit)
13:33:19S_a_i_n_tJdGordon: Where does Radio Art go? needs to be the name of the preset right?
13:33:33JdGordonsorry bieber, without being able to build the theme editor I have no way of even tyring to figure out how to implement that change
13:33:48JdGordonumm, in the .rockbox/fmpresets folder iirc
13:33:52JdGordonand yes
13:35:16JdGordonkugel: the android readme says cp to libs/armeabi... where is that folder?
13:35:51kugelah, it's not there right?
13:36:13kugeljust create it. git doesn't add empty folders to the version control :(
13:36:28JdGordonits in the build dir?
13:36:31JdGordonor android dir?
13:37:19kugelin the android dir
13:37:50kugelyou can build in the android dir if you didn't do it
13:38:06kugelI just need to figure out how to put some gitignores for that one
13:38:09JdGordonah ok, I missed the line that says use the android dir as the build dir
13:38:18JdGordonadd a .placeholder file
13:40:15*JdGordon reads step 1 of that linked url and is already lost
13:40:18 Quit nk_ (Ping timeout: 240 seconds)
13:40:21pamauryHello, I've been fighting with the code for a few days to make the simulator simulate the storage layer too (that is, use a file as storage, just like a real device). Any thoughts on this ? Things you would like to have, etc It is not finished but it's still at a point where I can modify things. I'm also opened to suggestion to build disk images properly. For example, how to create and mount properly a disk image under linux if it has several
13:40:21pamaury partitions, and even harder, how to mount it when the partitions are still not created !
13:40:45kugelJdGordon: aapt is in the sdk
13:41:03kugelit generates a java file from the res/ folder
13:41:51kugelI've gone through this steps once (except the keytool part) and they work but they are horribly described. unfortunately I haven't found something better
13:41:55pixelmaJdGordon: I really wonder why you ask :(
13:42:21JdGordonwell if you acknoledged the message I would have waited
13:43:00pixelmaJdGordon: I meant the whole thing
13:43:16JdGordonoh crud, I dont have that, is that downloaded by the eclipse plugin?
13:43:47kugelhave you only unzipped the sdk?
13:44:07JdGordonpixelma: I thought we were passed this? you said you didnt se the point, I gave up tyring to convince you and commited anyway because I felt it was worth it, end of discussion
13:44:10JdGordonkugel: yeah
13:44:37kugelthat's not enough :(
13:45:41JdGordonand I cant install the eclipse plugin, missing dependancies or something
13:45:56kugelJdGordon: you need to download a bit more. run <sdk>/tools/android (or launch eclipse->window->android sdk and avd manager)
13:46:13JdGordonthis is another reason pulling the .so out of the apk would be very handy, no need for everyone to set all this up
13:46:15 Join dfkt [0] (dfkt@unaffiliated/dfkt)
13:46:45 Join PouncePony [0] (
13:47:26JdGordon73MB download :( not happening tonight
13:47:34kugelthe .so file alone is insufficient
13:48:09JdGordonyes, but can it be downloaded by the java at first run?
13:48:26JdGordonand when the user wants to update
13:48:28kugelah you mean that
13:48:34pixelmaJdGordon: then why did you ask? All 3-4 answers you got (even in the forums) were like "not worth the complication", so you were the only one. I thought someone would only ask if he's unsure and then the few answers you got were almost all negative...
13:48:36JdGordonyes :)
13:53:32JdGordonkugel: I was running raaaaaaa all day today and the only issue I came across was if you wanted to exit with the back button it needed to be pressed 5+ times
13:53:48JdGordonare you starting a new activity too often or something?
13:53:58JdGordonother than that there were no real issues
13:54:05kugelah, I have that too sometimes
13:54:24kugelI don't know what it causes
13:54:54kugelin my experience it works once you've used the back button (instead of the home one) to go to the home screen
14:02:43FlynDicekugel: JdGordon: I think its lib/armeabi/ vs libs/armeabi as the readme says, is that correct? still struggling with eclipse here....
14:02:59kugelno, libs
14:03:13kugellib is another folder :)
14:03:43FlynDicewell maybe that's the prolem then... ;)
14:06:58*kugel wonders why r27717 happened
14:07:03kugelhaven't several people objected?
14:07:37 Join nk_ [0] (
14:07:46FlynDicekugel: so goes into libs/armeabi and goes into lib/armeabi/ (lib vs libs)?
14:07:47kugelJdGordon: I thought "doesn't add anything" was a valid reason
14:07:56kugelFlynDice: both into libs
14:08:08kugeloh the README is wrong
14:08:18pixelmakugel: logs, 13:44
14:08:20JdGordonif that was the case then maybe, but it in fact does add lots
14:08:34pixelmayes, binsize
14:08:44FlynDicewhich one is wrong lib or libs...
14:08:47pixelmaand a redundant way to do things
14:08:59JdGordonoh fucking hell... pixelma if you use that you will have a very big net benefit
14:09:25pixelmaI don't see how
14:09:45JdGordonhardly surprising
14:10:02pixelmaskin buffer is fixed and all my themes are way below the limit
14:14:06pixelmaand I'd really like to see the use case where such shorter lines really gain much, even if they are possible. And about half the examples we found in our WPSs you looked at, you said "that's not the use case" and can't be written in this system
14:15:03JdGordonI'm obviously not going to be able to convince you so I'm not going to waste my time
14:15:32CIA-6New commit by kugel (r27718): Update README a bit.
14:16:12 Join list_accel [0] (
14:17:05*kugel only questions the matter it was committed in
14:17:18CIA-6r27718 build result: All green
14:17:22JdGordonsorry, I thought git svn dcommit was acceptable?
14:18:01kugelI thought it was usual to not do rage commits if there's at least a few objections
14:18:02pixelmaJdGordon: well, why did you ask then at all? That's like, giving the choice and then showing the finger
14:18:17 Quit parafin (Ping timeout: 260 seconds)
14:18:22list_accelHi. Could or would you please increase the list accelaeration speed by adding a few more settings please, currently it max's at 2x10s. pretty please
14:19:19S_a_i_n_tlist_accel: It won't happen in the main builds, you're free to get the source, make the changes, and compile a build yourself though.
14:19:34JdGordonpixelma: go back and look at what I asked and what was commited
14:19:58list_accelaha, so there's no pint in making a suggestion that would benefit everyone
14:20:10JdGordonsomeone needs to do the work
14:20:20JdGordonit might be a trivial change, it might not be
14:20:28list_accelone extra spped step...
14:20:51JdGordonand without looking at the code noone knows how easy that is to do
14:20:55list_accel2x10s >> 2x15s >> 2x20s
14:21:17JdGordonyeah, so?
14:21:30 Quit list_accel (Quit: CGI:IRC)
14:21:33gevaertskugel: this phone seems to suffer from the background bug. Apart from that it seems to work fine
14:21:42pixelmaJdGordon: you added a few things but that doesn't change the basic idea
14:21:49kugelgevaerts: what phone? wildfire?
14:21:49 Quit bluebrother (Disconnected by services)
14:21:53 Join bluebroth3r [0] (~dom@rockbox/developer/bluebrother)
14:23:25JdGordonpixelma: I asked about completly changing the xd tag and breaking compatability. I didnt do that. My mind was always made up about this feature. 2 oud naysayers did nothing but make me want to do it more.
14:23:51JdGordonit is a useful feature and big bloody deal if it duplicates something which is already doable, it is simpler
14:23:52JdGordonend of discussion
14:24:02JdGordonif you dont want to use it then suit yourself
14:24:04 Join dgaf [0] (
14:24:14 Quit dgaf (Client Quit)
14:24:22kugelJdGordon: 4
14:24:52kugelI and gevaerts also objected
14:25:08kugeland the guy on the forum too
14:25:12JdGordonyou want a medal?
14:25:40kugelhow many people does it need for you to think about things again?
14:26:50pixelma"it is simpler", to me it's not. And even if I don't use it, the addition affects me
14:27:26pixelmait's shorter to write, that's all
14:31:00kugelJdGordon: have you had a look at the fms update problem?
14:31:28JdGordonno, I've been busy in pointless arguments
14:32:13kugelno, you've been busy implementing redundant features
14:32:42pixelmawell, you could have dropped the patch - no discussion and more time to code something else
14:32:56JdGordonand you think that sort of comment is going to give me incentive to look into that bug?
14:33:19kugelit was in line with yours
14:34:11kugelI guess nothing gives you the incentive, because it's existing since the fms got in
14:34:57pixelmakugel: which one?
14:36:14pixelmaif you mean preset name and number not updating when switching station then it worked not too long ago with the fms system
14:36:21JdGordonkugel: well are you going to do anything about the string numeric sorting bug?
14:37:32 Join parafin [0] (parafin@2001:470:1f0b:81::1)
14:37:45kugelpixelma: when I last tried the preset name/number did update but not the station art
14:38:28S_a_i_n_tkugel: Now the scan/preset mode does not update.
14:38:46kugelah sweet
14:38:48pixelmaI don't use station art yet, so
14:39:10S_a_i_n_tbut, afaik Radio Art works as expected now.
14:39:17S_a_i_n_tI just saw it working fine in the SIM
14:40:00S_a_i_n_tsome update problems were fixed, which somehow broke other things that *did* update.
14:40:05kugelon a related note, JdGordon: I think the album art is redrawn very often again
14:40:13S_a_i_n_tsublines were broken, but work now.
14:40:58JdGordonI'm pretty sure its only once when needed... any particular reason you tihnk that?
14:41:47 Join Kitar|st [0] (
14:41:47 Quit Kitar|st (Client Quit)
14:41:47kugelI tried the 320x240 touchscreen wps, the album art was drawn over the touch regioins immediately
14:42:06kugelthe s/wps/cabbiev2 wps/
14:42:17JdGordonok, ah yes, I tihnk i see the problem
14:42:18*JdGordon checks
14:43:53kugelthat would possibly explain why the station art updates "properly"
14:44:37 Join Kitar|st [0] (
14:45:43 Join domonoky [0] (~Domonoky@rockbox/developer/domonoky)
14:45:48CIA-6New commit by jdgordon (r27719): Dont redraw the AA every single update, only when needed
14:46:15JdGordonhopefully that doesnt have unwanted side effects...
14:47:32CIA-6r27719 build result: All green
14:49:07 Quit parafin (Quit: So long and thanks for all the fish)
14:49:10 Join parafin [0] (
14:50:19***Saving seen data "./dancer.seen"
14:51:34 Quit Kitar|st ()
15:03:51 Quit antil33t (Read error: Connection reset by peer)
15:03:58 Join antil33t [0] (
15:04:16 Quit bieber (Ping timeout: 245 seconds)
15:04:33 Join bieber [0] (
15:04:57FlynDicekugel: Did you have any problems with debug certificates expired errors in eclipse?
15:05:11kugelFlynDice: no
15:06:46JdGordonkugel: 11234 can be closed?
15:17:18 Join Kitar|st [0] (
15:24:54 Part LinusN
15:26:04 Join stooo [0] (
15:27:22 Quit JdGordon (Quit: Leaving.)
15:28:20 Quit n1s (Quit: Lmnar)
15:31:01 Join freddyb [0] (
15:31:25 Quit stooo (Quit: Leaving.)
15:31:48 Join stooo [0] (
15:32:03 Part stooo
15:32:31 Quit parafin (*.net *.split)
15:37:34 Join panni_ [0] (
15:38:12 Join parafin [0] (
15:39:12 Quit parafin (*.net *.split)
15:41:19 Quit Kitar|st (Ping timeout: 265 seconds)
15:47:36 Quit milz (Remote host closed the connection)
15:49:51 Join parafin [0] (
15:50:28 Join milz [0] (
15:56:22 Join Strife89 [0] (~Strife89@
15:56:45 Quit mitk (Quit: Leaving)
15:59:28 Join jgarvey [0] (
16:08:56 Quit milz (Ping timeout: 265 seconds)
16:12:09 Quit Judas_PhD (Quit: Aaarrrggghhh!!!)
16:14:33 Quit panni_ (Read error: Connection reset by peer)
16:15:52 Join Kitar|st [0] (
16:16:56 Join Judas_PhD [0] (
16:18:55 Join milz [0] (
16:20:59 Join fdinel [0] (
16:29:40 Quit Judas_PhD (Quit: Aaarrrggghhh!!!)
16:40:44CIA-6New commit by kugel (r27720): One more get_user_file_path() fix. Don't store the user dir in config.cfg and ...
16:40:44CIA-6New commit by kugel (r27721): Fix leading slash regex in
16:42:03 Join AzureWurk [0] (
16:42:10 Nick YPSY is now known as Ypsy (
16:43:37CIA-6r27720 build result: All green
16:43:53 Join einhirn [0] (
16:46:44CIA-6r27721 build result: All green
16:50:23***Saving seen data "./dancer.seen"
17:00:30freddybTorne: Would you look at FS #11168 and see if that's what you want?
17:00:41Tornefreddyb: at some point, yes
17:00:48Tornebut i don't have a lot of time for rockbox dev stuff at the moment
17:00:58Torneit sounds good but i need to review the implementation and test it out a bit :)
17:02:44freddybOk, it preserves LRU now and it's loads the fonts quickly, I can't think of anything else to improve it
17:04:04 Join domonoky1 [0] (
17:04:56 Quit domonoky (Ping timeout: 265 seconds)
17:05:23 Part Zagor
17:05:44 Part AzureWurk
17:07:02 Join AzureWurk [0] (
17:07:21 Quit AzureWurk (Read error: Connection reset by peer)
17:07:33 Join AzureWurk [0] (
17:17:37 Quit einhirn (Read error: Connection reset by peer)
17:17:42 Join webguest36 [0] (
17:18:41 Quit webguest36 (Client Quit)
17:24:16 Quit bieber (Ping timeout: 265 seconds)
17:24:29 Join bieber [0] (
17:29:38 Nick guymann_ is now known as guymann (
17:49:47 Quit parafin (Ping timeout: 246 seconds)
17:54:11 Quit bieber (Ping timeout: 240 seconds)
17:54:25 Join bieber [0] (
17:57:45 Join captainkewlll [0] (2669ecc2@gateway/web/freenode/ip.
17:59:37 Join komputes [0] (~komputes@ubuntu/member/komputes)
18:00:32 Quit petur (Quit: *plop*)
18:01:18 Quit Strife89 (Quit: Going home.)
18:01:46 Quit Xerion (Ping timeout: 245 seconds)
18:02:36 Join Xerion [0] (~xerion@
18:06:20 Quit Ramsey[LC] (Quit: (Killed (NickServ (GHOST command used by dissident))))
18:06:24 Quit captainkewlll (Quit: Page closed)
18:06:42 Join Ramsey[LC] [0] (~RamseyLC]@
18:20:44 Join panni_ [0] (
18:22:37 Quit pixelma (Disconnected by services)
18:22:40 Join pixelma_ [0] (quassel@rockbox/staff/pixelma)
18:22:51 Quit amiconn (Disconnected by services)
18:22:53 Join amiconn_ [0] (quassel@rockbox/developer/amiconn)
18:22:55 Nick pixelma_ is now known as pixelma (quassel@rockbox/staff/pixelma)
18:23:13 Nick amiconn_ is now known as amiconn (quassel@rockbox/developer/amiconn)
18:24:52 Quit pixelma (Disconnected by services)
18:24:53 Quit amiconn (Disconnected by services)
18:24:54 Join pixelma_ [0] (quassel@rockbox/staff/pixelma)
18:24:55 Join amiconn_ [0] (quassel@rockbox/developer/amiconn)
18:25:11 Nick pixelma_ is now known as pixelma (quassel@rockbox/staff/pixelma)
18:25:15 Nick amiconn_ is now known as amiconn (quassel@rockbox/developer/amiconn)
18:31:30 Quit Ramsey[LC] (Quit: (Killed (NickServ (GHOST command used by dissident))))
18:31:49 Join Ramsey[LC] [0] (~RamseyLC]@
18:36:14 Join BRi7X [0] (
18:37:41BRi7Xhowdy folks
18:50:24***Saving seen data "./dancer.seen"
18:51:06 Quit solrize (Ping timeout: 264 seconds)
19:01:59 Join stoffel [0] (
19:03:37 Join solrize [0] (
19:08:21 Quit Ramsey[LC] (Remote host closed the connection)
19:11:09 Quit wombat23 (Ping timeout: 252 seconds)
19:12:01 Join bunnyboi [0] (
19:14:40 Quit stoffel (Remote host closed the connection)
19:16:53 Join stoffel [0] (
19:23:17 Join wombat23 [0] (
19:23:44 Join webguest54 [0] (
19:24:14 Quit webguest54 (Client Quit)
19:28:30 Quit DerPapst (Quit: Leaving.)
19:28:41 Join Temik [0] (~temik@
19:28:58 Join Horscht [0] (~Horscht@xbmc/user/horscht)
19:33:52kugeljust noticed a strange skin issue
19:34:04 Nick bunnyboi is now known as Jennifur (
19:34:08kugelI have a AA cond. viewport and a no-AA one
19:34:37 Join Jerom [0] (~heidi@
19:34:38kugelin the no-AA one the area where the AA from the AA-vp would have been was cleared and showed no text
19:35:45kugelthe song had no AA obviously
19:36:08kugelseems to be fixed by moving the AA vp definition above the definition of the non-AA vp
19:36:36kugelhaha, no, that just breaks the AA-vp (shows no AA)
19:44:24kugelanother problem: scrolling lines of disabled cond. vps are not removed (but stay scrolling=
19:46:00 Join parafin [0] (
19:48:28 Join stripwax [0] (
19:50:40 Part AzureWurk
19:54:40CIA-6New commit by kugel (r27722): Cabbiev2 for 320x480. ...
19:55:52 Join DerPapst [0] (
19:56:32CIA-6r27722 build result: All green
19:56:47 Quit pamaury (Remote host closed the connection)
19:57:37CIA-6New commit by kugel (r27723): Add default font for LCD_HEIGHT <= 480
19:59:16CIA-6r27723 build result: All green
20:00:00 Join Strife89 [0] (
20:04:14Temikhello On my X5 with latest build rockbox FM screen white and above scratch
20:07:35TemikWPS working only cabbiev2. Other with blue screen and above scratches or default wps.
20:08:57 Join mt [0] (~chatzilla@
20:11:00bluebroth3rplease define "scratch". That sentence doesn't make sense to me otherwise.
20:12:26TemikMulti-colored points
20:13:09TemikHeight of a strip of 20 points
20:14:37TemikSimilar on incorrect addressing of bitmap
20:15:24pixelmais it really a problem of the build - did you try older ones now? And where did you get the build from?
20:17:05mtamiconn: ping
20:17:13TemikBefore yesterday's updating used build from 16 june. Problems was not
20:18:04TemikAlways building from svn
20:19:23 Join Dreamxtreme [0] (~Dreamxtre@
20:19:40pixelmaah, so you are building yourself. Could you try a fill rebuild - "make clean" and then "make"?
20:20:30pixelmaI do not have an X5, so need to rely on your reports
20:21:27Temikok I will be glad to help
20:22:53 Quit stoffel (Remote host closed the connection)
20:23:28 Join watto [0] (~watto@
20:27:01 Join toffe82 [0] (~chatzilla@
20:27:49 Join saratoga [0] (9803c6dd@gateway/web/freenode/ip.
20:28:11saratogamt, buschel: i repeated my 40MHz CPU clock tests on the clipv2
20:28:28saratoga16.37MHz for WMA, 22.13MHz for WMA Pro
20:29:09saratogaAAC-He is now "only" 70MHz, a full 20% faster
20:29:17saratogaverses last month
20:29:36amiconnmt: pong. Seen the log?
20:30:14 Quit Jennifur (Ping timeout: 260 seconds)
20:30:47amiconnsaratoga: I don't really trust ams speed tests. Something is really fishy on ams (maybe v2 only)
20:31:03saratogaamiconn: too good to be true?
20:31:12mtamiconn: yeah just checked.
20:31:54amiconnThe greylib isr is reported to have a negative cpu consumption by test_fps.rock, i.e. updates are faster with the isr running than without
20:32:22saratogaFWIW I think the results are reasonable if you run the IRAM and core clock at the same speed, that way essentially all memory access is free and you get to use the fast 9E multiplier
20:32:39amiconnThis is outright impossible. It almost looks like the core is changing clock on demand somehow, even though we don't do our cpu frequency scaling on amsv2
20:32:41saratogaits of course much slower per clock at 250MHz since then you actually have memory latency :)
20:34:15 Join stoffel [0] (
20:36:12 Quit Dreamxtreme (Read error: Connection reset by peer)
20:37:43saratogawe need to compare PP with the clock locked at 30MHz, i bet its much faster too
20:38:08amiconnI don't think it will be
20:38:29saratogaDRAM latency will be much less (measured in clocks)
20:38:48 Quit stoffel (Remote host closed the connection)
20:38:49amiconnYes, but proper iram usage and the cache will cover that pretty well
20:39:10saratogaPP has lousy cache, and while IRAM covers everything in MP3, it doesn't for the MDCT codecs
20:39:40amiconniram and cache are both single cycle (always, i.e. they're running with core clock)
20:39:42bertrikbattery life is already pretty good on the clip+ even when running continuously at 240 MHz
20:39:55mtamiconn: the "accext01" line was a mistake that I overlooked, shouldn't be there. But yeah now I see why it's wrong.
20:40:05amiconnWell, PP5002 cache is two-cycle, as are some blocks of PP5020 (not 5022+) iram
20:40:17saratogayes but the cache is small and shared between data and instruction
20:40:22amiconnStill, that's independent of actual core clock
20:40:41 Join Dreamxtreme [0] (~Dreamxtre@
20:40:51saratogaMDCT codecs typically step through >8KB of data per loop (since they use 2048 sample blocks), so they'll flush all ops from cache as well
20:41:04 Join stoffel [0] (
20:41:13amiconnThose buffers should really be in iram
20:42:11amiconnI've observed caching behaviour on pp in libdemac
20:42:58 Quit stoffel (Remote host closed the connection)
20:44:43amiconnThe sample buffers aren't a problem, since they're usually written once, or a accessed a few times only. What really kills performance in libdemac is if the filter buffers are in dram and are larger than the cache
20:46:19 Join stoffel [0] (
20:47:46 Quit fdinel (Ping timeout: 265 seconds)
20:49:46 Quit Topy (Quit: Leaving)
20:49:54 Join Topy44 [0] (
20:50:27***Saving seen data "./dancer.seen"
20:50:40 Quit n17ikh (Quit: ZNC -
20:51:11 Join n17ikh [0] (
20:51:34 Quit n17ikh (Client Quit)
20:52:47Temikpixelma, Now build r27612. Working good
20:53:44CIA-6New commit by mt (r27724): Remove the comment about Coldfire asm. The current code (using fixmul16 macros) is fast enough; even faster than an equivalent CF asm version, as ...
20:53:58CIA-6New commit by kugel (r27725): Revert accidental change to cabbiev2.176x220x16.wps and apply it to cabbiev2.320x480x16.wps as intended.
20:55:22CIA-6r27724 build result: All green
20:56:46 Part watto
20:56:56CIA-6r27725 build result: All green
20:58:02mtsaratoga: Do you think we can merge libwmapro's sine window tables with mdct lookups ?
21:00:48simonrvnsomething broke... fuze v2 isn't playing f*** all.
21:01:41simonrvngonna check if it's pebkac or not
21:05:27simonrvn, ok, not hardware, of plays fine
21:05:38CIA-6New commit by bieber (r27726): Theme Editor: Made progress bars fully movable/savable from the preview panel
21:07:16CIA-6r27726 build result: All green
21:09:29 Quit Dreamxtreme (Quit: Nice Scotty, now beam my clothes up too!)
21:09:57pixelmaTemik: goo to hear
21:10:29pixelmaoh wait, you didn't try the newest
21:10:49simonrvnok, some config issue. ignore.
21:13:17Temiktry r27721 today. Issue present
21:13:50pixelmaanyway. If it was the same build that showed errors for you before, then it shows that there was a problem with your build not the code. That's good to know
21:13:59simonrvnwtf. theme issue prevents playing a file?
21:16:24TemikШ ерштл
21:18:32TemikI think a problem has appeared in r 27613. Now I will try to build r27613
21:24:18 Quit BRi7X (Read error: Connection reset by peer)
21:28:02 Quit stoffel (Remote host closed the connection)
21:28:44pixelmaTemik: do you always do full rebuilds?
21:28:57 Nick Ypsy is now known as YPSY (
21:29:55TemikWith erase dir 'build'
21:35:40Temikr27613 with problem
21:36:13Temikr27612 working good
21:36:57 Quit stripwax (Quit:
21:41:51pixelmaI wonder if it's because the X5 has a greyscale remote. Did you say it's only visible with some themes?
21:43:53TemikYes. Other themes not reading, calling default wps
21:44:47 Join n17ikh [0] (
21:45:33pixelmaare those themes even older - there was a change to the WPS syntax not long ago, did you update them?
21:46:57Temikyes Updated them 16 june
21:49:05TemikAfter updating on June, 16th themes have ceased to work. Has been compelled modify them
22:09:04 Quit Temik (Quit: Leaving.)
22:14:53 Join Corruption [0] (
22:15:17CorruptionIs there any chance of Rockbox working on a 160 GB 6 Gen iPod Classic?
22:17:14 Quit jgarvey (Quit: Leaving)
22:20:56 Quit bieber (Remote host closed the connection)
22:21:00saratogasee front page
22:22:28CorruptionI meant in future versions [I know it isn't working for it as of now]. Sorry for not being very specific. :X
22:23:02gevaertsIt's not impossible, but nobody seems to be working on it
22:23:24 Join bieber [0] (
22:23:29 Join Hillshum [0] (4ba5f419@gateway/web/freenode/ip.
22:24:24 Join angelwolf71885 [0] (
22:24:37CorruptionThanks for letting me know that gevaerts. I guess I'll just have to get along without it.
22:24:39 Quit Corruption (Quit: I-n-v-i-s-i-o-n 3.1.3 (May '10))
22:26:24pixelma'Installing your build in your '' dir' ... don't think I saw that message before. That's when compiling a recent X5 sim in cygwin, clean build configured before
22:26:59pixelmathe build seems to be working correctly though
22:28:33 Quit angelwolf71885 (Client Quit)
22:31:36 Quit milz (Read error: Connection reset by peer)
22:32:03 Quit Hillshum (Quit: Page closed)
22:32:41 Join Hillshum [0] (4ba5f419@gateway/web/freenode/ip.
22:34:11 Quit kugel (Remote host closed the connection)
22:34:39 Join kugel [0] (~kugel@rockbox/developer/kugel)
22:35:29saratogamt: probably but i haven't looked at it carefully
22:35:42saratogabut i think its just a sin window, so it should be possible, at least for the smaller window sizes
22:35:50saratoga4096 won't work, and 2048 may or may not
22:36:38 Join milz [0] (
22:37:05amiconnpixelma: The message used to be "Installing your build in your 'simdisk' dir"
22:37:32pixelmayes, that's how I remember it too
22:38:04Hillshumsaratoga: Do those windows come from Vegas?
22:38:09 Quit mt (Ping timeout: 258 seconds)
22:39:19amiconnSo something is broken, probably due to one of the raaa commits
22:50:31***Saving seen data "./dancer.seen"
22:53:26 Quit sevard (Changing host)
22:53:26 Join sevard [0] (sev@unaffiliated/sevard)
22:58:08 Join fdinel [0] (
23:10:56 Join Buschel [0] (
23:11:29Buschelhmm, somehow "make zip" results in non-installable .zip files...
23:11:59Buschel"rockbox" instead of ".rockbox"-folder, no rockbox.ipod file...
23:16:27kugelI think "%C?" should be an error :/
23:22:00 Quit evilnick_B (Quit: Page closed)
23:22:12 Quit freddyb (Quit: ChatZilla 0.9.86 [Firefox 3.5.9/20100401213457])
23:24:22Buschelok, works now
23:38:16 Quit bertrik (Quit: :tiuQ)
23:40:05 Quit bmbl (Quit: Bye!)
23:42:57 Quit seani (Ping timeout: 258 seconds)
23:43:51CIA-6New commit by Buschel (r27727): Minor of libwmapro on ARM. Swap operands for fixmul31, is 1% faster.
23:43:58 Join Dreamxtreme [0] (~Dreamxtre@
23:45:37CIA-6r27727 build result: All green
23:48:35 Quit AndyI (Ping timeout: 248 seconds)
23:53:59 Join AndyI [0] (~pasha_int@
23:56:35 Join seani [0] (~seani@

Previous day | Next day