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 2009-06-03

00:00:13 Join kachna [0] (
00:02:09 Join PaulJam_ [0] (
00:02:52 Nick fxb__ is now known as fxb (
00:10:21pixelmaUnhelpful: as I thought - the pf crash on Ondio started with it being an overlay plugin: r20863 works, r20865 doesn't (r20864/5 was the overlay change, -64 doesn't build). If I enable "catch mem accesses" I get a "I0C: UserBrk at 090314xx", the last 2 numbers change slightly from revision to revision. Since the crash happens after a few seconds it looks like it has to do with filling the cache.
00:10:29 Join PaulJam__ [0] (
00:11:40pixelmathe rebuild problem when the pictureflow folder does not exist is present on those earlier revisions
00:13:28pixelmatested with SVN builds without backlight, made with cygwin - in case that matters
00:14:43 Quit PaulJam (Read error: 113 (No route to host))
00:15:02 Quit robin0800 (Read error: 60 (Operation timed out))
00:15:32pixelmaoh, and tested with and without voice enabled which made no difference
00:21:27 Quit ender` (" The first 90% of the code accounts for the first 90% of the development time. The remaining 10% of the code accounts for th")
00:22:49 Join togetic [0] (n=ib@unaffiliated/ibuffy)
00:22:56togeticadvantages of using rockbox?
00:23:24funmantogetic: read
00:23:30hillshumlots of themes, audio codecs, games
00:23:51togeticjust bought the ipod nano 1st gen 4gb 20 secs ago
00:25:31Mikachutogetic: it doesn't suck :P
00:26:21 Quit mirak ("Ex-Chat")
00:28:45 Join r0b- [0] (
00:28:52r0b-may i ask what Rockbox is written in?
00:28:57Mikachuyes you mya
00:29:00soapC, some ASM.
00:29:01togetica programming lang
00:29:05togeticyou beat me to it
00:29:10 Join BHSPitMonkey [0] (n=stephen@unaffiliated/bhspitmonkey)
00:29:20Unhelpfulpixelma: interesting. overlay plugins are loaded into the audio buffer, i wonder if plugin_get_audio_buffer returns the entire audio buffer, including the part containing code and static globals.
00:29:33togeticC is so 70s
00:29:43togeticand um 2000s
00:29:46r0b-well is it possible to compile rockbox in Windows without a Emulator for the hardware?
00:29:52togetic..since i'm using an OS written in it...
00:30:05 Quit PaulJam_ (Read error: 113 (No route to host))
00:30:07hillshumyou don's need any emulator to compiler
00:30:31r0b-so it can run nativly on Windows?
00:30:34vedlithr0b-: it probably is possible but it would need recompilation of some target specific compilers like arm gcc and assembler
00:30:54r0b-i was going to try to learn to code for it
00:30:54vedlithrun oO?
00:30:56togetici hope this ipod is worth it since it costs me my entire week's salary :P
00:31:03Mikachur0b-: do you want to compile it for a device or to run it in windows?
00:31:08Unhelpfulthe other thing is not precisely a bug. pictureflow does not *verify* its cache on startup at all, other than to check if the cache version is different from the one in the config file. so far, that is just a mechanism that catches failed cache builds, or forces updates if the format of the cache files changes.
00:31:18*togetic will starve w/ no food, but thrive on music?
00:31:26r0b-lol togetic
00:31:31r0b-what Ipod model?
00:31:50*togetic considers the grass on the lawn, bet it could make a decent green smoothie
00:31:55togetic1st gen ipod nano
00:32:07togeticdamn those pesticides!
00:32:26r0b-when it arrives i hope to put rockbox on my new Sansa e250
00:32:42r0b-i downloaded the Simulator thing for the e250 i love it!!!
00:32:47Unhelpfulcould we please stick to things directly related to rockbox in here? there are guidelines for a good reason. (see topic).
00:32:51togeticseems i'll be blasting some 8bitpeoples tunes and hunting for food the entire week
00:33:05togeticokay okay Unhelpful
00:33:10Mikachutogetic: rockbox plays .nsf and .spc iirc :)
00:33:51r0b-it also plays .gb and .gbc :P
00:34:31 Quit SirFunk__ (
00:34:31 Quit barrywardell (
00:34:31 Quit beta_ (
00:34:31 Quit lucent (
00:34:36pixelmaUnhelpful: should I check other overlay plugins? I'm just done building the most recent revision though and unzipping a complete build... will take a while :/
00:34:37 Quit jgarvey ("Leaving")
00:35:04***Saving seen data "./dancer.seen"
00:35:12Unhelpfulpixelma: i'm working on a patch to both base the cache filenames on album names, and to verify cache files on startup. initially this will probably just mean that files with "old" covers get no-cover, but forcing a rebuild could be another response.
00:35:54Unhelpfulpixelma: i don't know, which others use the audio buffer? we could consider *not* using the audio buffer, now that the plugin buffer only has the overlay stub in it...
00:36:02 Join AndyI [0] (n=pasha_in@
00:36:12funmanI suppose the usberror entry of rbutil.ini isfor detecting mtp mode?
00:36:17 Join HellDragon [0] (
00:36:28kugelUnhelpful: I imagine, for simplicity, the database could place a hook-file in the pf folder foring a rebuilt
00:36:39 Quit togetic (Read error: 54 (Connection reset by peer))
00:37:33NJoinSirFunk__ [0] (
00:37:33NJoinbarrywardell [0] (n=barry@rockbox/developer/barrywardell)
00:37:33NJoinbeta_ [0] (
00:37:33NJoinlucent [0] (i=lucent@unaffiliated/shadows)
00:37:36pixelmaUnhelpful: I thought chess does but am not sure about the "behinds"...
00:37:46 Join togetic [0] (n=ib@unaffiliated/ibuffy)
00:37:58togetickonversation ┬┐krashed?
00:38:05kugelfunman: great success!
00:38:50Unhelpfulkugel: the database doesn't know everything. ideally, i want the cover search itself to be fast enough that i can do it on every PF startup, and then recache covers if a new cover is found, etc.
00:39:12 Quit lucent (Remote closed the connection)
00:39:17 Join lucent [0] (
00:39:22 Quit pat_mulchrone (Remote closed the connection)
00:39:33Unhelpfulpixelma: chessbox uses plugin_get_buffer, which is safe because core knows how much of the plugin buffer the overlay loader used.
00:41:31kugelUnhelpful: the database does know when it gets updated
00:41:56Unhelpfulkugel: it doesn't know when non-music files are changed, though.
00:41:59pixelmaok, well seems like some grep would be in order. Btw. does your rebuild cache patch also solve the problem with no "pictureflow" folder present? ;)
00:42:22obofunman: sorry I missed you earlier - I'm using arm-elf-objdump and an IDA demo (after wrapping the binary into an elf)
00:42:27Unhelpfulpixelma: it ought to, as anything that triggers a cache build will create that directory.
00:42:54bertrikkugel, the DBOP FIFO fix now makes a lot of difference on my clip, 38% faster unboosted
00:43:21pixelmawell the directory was created but only filled with emptyslide after entering pf first...
00:43:28kugelbertrik: I can give it another try. Can you upload it once more please?
00:43:43funmanobo: ok, because if you use IDA only you won't see cross references to some PP registers if they are loaded in registers with mov/shifts
00:45:06CIA-38New commit by funman (r21171): Sansa AMS : don't use the SHAREDBSS_ATTR since we don't have a coprocessor
00:46:44Unhelpfulpixelma: can you give this a try? it uses plugin_get_buffer on all targets, instead of plugin_get_audio_buffer when the plugin buffer is tiny. with only the overlay loader in the plugin buffer on archos, there should be room for about 19 slides in the plugin buffer, so that should be fine.
00:46:46 Quit AndyIL (Success)
00:47:22kugelbertrik: is this with 64MHz unboosted?
00:47:49kugel62 ven
00:47:58funmankugel: yep, I noticed again this CCU_SCON register when looking for USB init code in the OF ;)
00:48:28pixelmaUnhelpful: yes, trying
00:48:57obofunman: do you use anything to post-process the objdump output? as in getting it to place the values next to the operators (like IDA), rather than having to scroll around to find them?
00:49:10funmanbertrik: did you notice the forum post : as3514's DCDC is needed for some Clip screens
00:49:16kugelUnhelpful: I fear that doesn't apply cleanly to SVN :)
00:49:20pixelmaUnhelpful: if the hunk didn't fail though it would be easier
00:49:30pixelmaexactly ;)
00:49:39bertrikfunman, yes I did. I wondered if there's some way to detect whether DCDC is needed or not
00:49:46*Unhelpful will sync it
00:49:55funmanobo: i usually don't use objdump, but vitja wrote some code to enhance objdump output, let me find it
00:49:56bertrikI vaguely remember seeing it in the OF
00:50:35kugelfunman: that's sad. either we support all displays and waste a good deal battery life on it, or we support some not with more battery life for the others. or we supply a setting that most people don't undersand :(
00:50:41obofunman: you use IDA or something else?
00:51:15funmanobo: - yes i use IDA
00:51:52*kugel wonders why changes to clock-target.h force a full rebuild :(
00:51:55funmankugel: checking how much battery this drains would be interestin indeed
00:52:26kugelafter your formula, it would be 12.5mA
00:52:33funmankugel: I think every file including system-target.h depends on it
00:52:39 Quit dfkt ("-= SysReset 2.53=- Ph'nglui mglw'nafh Cthulhu R'lyeh wgah'nagl fhtagn.")
00:52:59Unhelpfulkugel: assuming we don't get a supported swcodec target with a tiny plugin buffer, you can amend PF_PLAYBACK_CAPABLE to just checking for swcodec
00:53:10 Quit Thundercloud (Remote closed the connection)
00:53:11Unhelpfulpixelma: resynced:
00:53:26funmankugel: hum right, but how much more % is it related to the actual consumption?
00:53:39Unhelpfulusing the plugin buffer won't make it playback-capable on hwcodec, though, because there is still code loaded in the audio buffer.
00:53:51kugelUnhelpful: I made the #define so that it can be replaced easily (even with 1 :) )
00:54:01kugelwasn't a hard job actually :P
00:54:23bertrikif there's no current drawn from the DCDC in a clip that doesn't need it, then the impact can be limited I think
00:54:40kugelfunman: no idea, but the v1 have shown that even 4mA can make a noticeable difference
00:54:54 Quit froggyman (" ajax IRC Client")
00:55:08kugelhowever, I assume we can "power down" the dcdc anyway in backlight_off(), so it probably won
00:55:16kugelwon't make such a huge difference
00:55:55 Nick fxb is now known as fxb__ (
00:56:09bertrikand we shouldn't give up yet on auto-detecting the need for the dcdc
00:56:10JdGordon|Unhelpful: if you are alooking for a way to make it work.. you should be able to create the playlist while the plugin is runnign, then just splash "press play to start the playback"
00:56:28JdGordon|or use some new thread messages to start when the plugin unloads
00:56:56Unhelpfulafter i take a look at Horscht's jpeg decoder bug report, and tackle the hash-based cache keys for PF, i think i want to take a look at making the plugin buffer entirely dynamic, as amiconn and i talked about a couple weeks ago. no wasted space without a plugin running would be a nice side benefit. as would support for "oversized" plugins without an overlay loader.
00:57:25kugelbertrik: from 45 to 65 fps it seems
00:57:44*JdGordon| 's curiosit pricks up
00:58:12kugelbertrik: 100fps while boosted :O
00:59:03*kugel still thinks we are in need for a splash that disappears upon button press, like a splash-based yesno screen
00:59:25JdGordon|so do it :)
00:59:31JdGordon|we already got an ok for it didnt we?
01:00:05Unhelpfulsplash (-1,...) ;)
01:00:28 Quit funman (Read error: 54 (Connection reset by peer))
01:00:36amiconn[00:29:20] <Unhelpful> pixelma: interesting. overlay plugins are loaded into the audio buffer, i wonder if plugin_get_audio_buffer returns the entire audio buffer, including the part containing code and static globals. <== It sure does; the plugin needs to make sure to not overwrite itself
01:00:43JdGordon|no, all splashes are either no block, or some standard timeout value, pressing any button would break the timeout
01:00:48kugelJdGordon|: yea, I haven't really found a place where it would be really usefull yet
01:00:57kugelexcept some plugins
01:01:00amiconnThis is possible; one of the first overlay plugins (*the* first iirc) - rockboy - does that
01:01:03 Quit bmbl ("Woah!")
01:01:07 Join funman [0] (n=fun@rockbox/developer/funman)
01:01:30pixelmaUnhelpful: yes, that fixes it (and the keymap issue is really fixed as well)
01:01:36JdGordon|kugel: the "nothing to resume" splash :p
01:01:47kugela right :)
01:02:01kugelbertrik: lcd updates show weird blue bars when boosted
01:02:28kugelmaybe almost full would be better?
01:02:39amiconnUnhelpful: And we do have a lowmem swcodec target - the clip
01:02:41Unhelpfulamiconn: it seems the easiest way might be to check the address of __header?
01:02:54 Quit bertrik ("sleep")
01:03:08funmanamiconn: the plugin buffer is not that small
01:03:18pixelmaUnhelpful: seems to be a bit snappier too (you notice a slowdown in fps if the plugin is busy filling the cache)
01:04:00*pixelma mentions the Logik Dax ;)
01:04:26Unhelpfulamiconn: the clip still has a rather large plugin buffer ;)
01:04:44amiconnapps/plugins/rockboy/rockboy.c lines 356ff
01:06:09amiconnkugel: Iirc the ipod mini also achieves 100fps in PF
01:06:30kugelamiconn: I was talking about 100fps in test_fps :/
01:07:06kugelwhich is by far not a great value, but the best I've seen so far by the fuze
01:07:35*amiconn somehow didn't see a reference to test_fps
01:07:53funmanpff, Clip shows 2700fps
01:08:29kugelamiconn: sorry if it wasn't clear. bertrik was playing with the fifo to get better fps in general
01:08:40Unhelpfulthe clip, being mono, should do fine with the 384KiB plugin buffer. i'm guessing PF will take about 40-80KiB of that itself... there will be plenty left for covers, as they should cost a bit under 2KiB on average.
01:08:47kugelfunman: mono, tiny display.........:)
01:09:17Unhelpfulkugel: which is why i can't hit the high mark of something like 270fps in PF on beast :/
01:09:28amiconnIirc the clip used to have 64KB plugin buffer. Someone must have enlared that
01:10:32obofunman: thanks. that thread also reminded me of the disassembler under /utils
01:10:48*Unhelpful would still like to see no-fail bufopen, and buf_clear_end(size), or the like, and equivalents for hwcodec. it would be the end of buflib, though.
01:11:14funmanobo: there is also - but it's still in early development -
01:12:25 Quit tessarakt ("Client exiting")
01:12:30funmanhum .. apps/main.o is 0 bytes
01:13:29obofunman: Doesn't currently compile, but I'll keep an eye on it. Cheers!
01:14:17funmanobo: check funman/eda-2/tree/master"> - that one should compile ;) -
01:14:51 Join pixelma_ [50] (n=pixelma@rockbox/staff/pixelma)
01:14:51 Quit pixelma (Nick collision from services.)
01:14:59 Quit togetic (Read error: 110 (Connection timed out))
01:15:10 Nick pixelma_ is now known as pixelma (n=pixelma@rockbox/staff/pixelma)
01:15:25 Quit amiconn (Nick collision from services.)
01:15:28 Join amiconn_ [50] (n=jens@rockbox/developer/amiconn)
01:15:47 Nick amiconn_ is now known as amiconn (n=jens@rockbox/developer/amiconn)
01:16:05 Join onlysoaa [0] (i=42248748@gateway/web/ajax/
01:16:07kugelJdGordon|: how useful would get_action() be for that splash? aren't there a few actions which don't come from buttons? (e.g. ACTION_REDRAW?)
01:18:49JdGordon|_REDRAW isnt sent from the button driver i tihnk... its more of a hack to fix draw issues...
01:18:59JdGordon|_NONE and SYS_* are the ones to watch for
01:19:12 Join funman_ [0] (
01:19:14 Quit funman (Read error: 113 (No route to host))
01:19:50 Nick funman_ is now known as funman (
01:21:42pixelmaredraw issues... suddenly reminds me of the colour choser statusbar problem...
01:22:06kugelnot that a ACTION_REDRAW suddently removes the splash though
01:22:37kugelon the other hand, SYS_* should probably remove it, like SYS_USB_CONNECTED
01:22:46JdGordon|kugel: my point was.. you would never get that action while a blocking splash is open
01:23:10kugelhm yea, it can only come from the main thread
01:23:29JdGordon|iirc just about only in a list control
01:26:13 Quit barrywardell (Remote closed the connection)
01:26:25obofunman: that does indeed build, and runs. Are there any instructions anywhere?
01:26:49funmanno ^^
01:27:31 Quit BHSPitMonkey (Read error: 113 (No route to host))
01:28:00funmanyou need to modify for file / loading address, and then you can connect to , what it does (even the only useful thing) is simulate functions so you can see effects on registers
01:28:10mtWhat header defines file functions like read(), lseek .. etc. ? (equivalent to unistd.h)
01:28:35 Join FlynDice [0] (n=FlynDice@
01:28:47funmanmt: include/file.h it seems
01:29:15funmanFlynDice: hi! rolling out my last patch - to get mpegplayer working - before I leave
01:29:20mtI included that but keep getting undefined reference to lseek, read ..etc.
01:29:28mtI'll recheck. :/
01:29:44funmanmt: undefined references mean that the functions are not present, not that you didn't include the right header
01:29:49FlynDiceGreat! sorry to say my day has been mostly wasted
01:30:25funmanthey should be in firmware/common/file.o (and firmware/libfirmware.a)
01:31:18mtfunman : yes, but when I include the right header the functions should be available.
01:31:31mtDo you mean I have to link file.o ?
01:31:52FlynDiceI svn upped on my laptop before I left this morning and got a bad update :(
01:31:53kugelfunman: read my edit too :/
01:32:00kugel(re: FS #10048)
01:32:36kugelFlynDice: that's why you have a plain SVN checkout, and use rsync to checkout from that
01:32:50kugelThat's how I do it, at least :)
01:33:44 Quit hillshum (Remote closed the connection)
01:34:38funmanI use git & git stash :)
01:35:17FlynDiceI have my one svn copy that I actually update and then I make copies from that to start from, but I'm all ears this is mostly still new stuff for me ;)
01:35:53FlynDiceI thought I would get a lot done today but that sunk me...
01:37:54kugelfunman: what kind of wps corruption were you getting?
01:38:18*amiconn just uses two svn working copies, one for actual work and a clean one for bisects, and makes no further copies
01:38:34FlynDiceI've been having problems today with last night's FS #10048 and couldn't roll a new build. Couldn't get anything to complete compiling and no internet access for most of the day
01:38:35funmanshort (~8px) horizontal bars, vertically separated (~8px) and offseted (~8px)
01:39:10kugelI'm getting blue bars moving on the screen with bertriks fifo patch
01:39:40funmanhum now I can't boot rockbox oO
01:40:15*JdGordon| has i tihnk 4 svn copies.. ranging from >1yr old to the one i actually use... im too lazy to delete them :p
01:41:20kugelfunman: have you looked at implementing a udelay()? It doesn't seem very complicated even if we're reusing the tick task timer
01:41:30kugelI think I have 8 or 9
01:42:33funmanno i've not but i have an idea on how to do it. how would you use it?
01:43:21kugelthe delays I'm messing with are clock dependant, so I wouldn't even have a use :)
01:44:00kugelbut bertrik wants one. And I think for FM, a fixed delay would be very usefull (e.g. udelay() made FM radio actually possible on the D2 port)
01:44:24 Quit kachna (Read error: 60 (Operation timed out))
01:45:58funmankugel: FlynDice does mpegplayer also deadlocks with FS #10048?
01:47:18 Quit Res1 (Read error: 60 (Operation timed out))
01:47:19 Quit bubsy (Read error: 54 (Connection reset by peer))
01:47:22kugelfunman: I'm fairly sure it worked yesterday (now it deadlocks too) even though I didn't change the patch, I'm still on the one before the most recent
01:47:52FlynDiceDon't know, didn't try it. Might just be the copy on the player right now, I couldn't replace it because I couldn't compile a new one. I'm on my laptop now away from home
01:48:24funmanok, i'll look into it tomorrow
01:49:07funmanFlynDice: please have a look at what kugel said earlier (around 7PM on yesterday log - UTC or CEST, whatever say the logs -)
01:49:25 Quit funman ("leaving")
01:49:46FlynDiceI'm headed home now though and I'll try to check it out. I was getting poor performance numbers at 248/31/31 also boosted 30% of the time and 86MHz for mp3
01:50:00kugelI already needed to higher the delay in the dbop reading again :S
01:50:29FlynDicekugel: Ok I'll take a look
01:50:42Unhelpfulamiconn: is audio buffer *always* adjacent to, and before, plugin buffer? if that's the case, couldn't one just unconditionally set the audio buffer size to plugin_start_addr - audio_bufferbase?
01:50:55 Quit FlynDice (Remote closed the connection)
01:52:00 Join Res1 [0] (
01:52:19*amiconn points Unhelpful 48 minutes back
01:53:48Unhelpfulamiconn: at how rockboy does it? i looked, i'm just not sure if the test is necessary. it seems to me that if conditions are as i understand them to be, the assignment can be made unconditionally.
01:54:32amiconnThe test might be unnecessary
01:54:35 Join bubsy [0] (i=Bubsy@
01:55:10amiconnIt also seems that a bug slipped into rockboy (or rather, unnecessary code)
01:55:51amiconnAbove the adjustment for not overwriting itself, rockboy checks whether audio is stopped or not, in order to decide whether to use the audio buffer
01:55:54 Quit toffe82 (Read error: 110 (Connection timed out))
01:56:12amiconnHowever, on lowmem audio will always be stopped, because the overlay loader has to do that
01:56:59*amiconn also wonders whether this behaviour is desirable on swcodec
01:57:01 Join PaulJam [0] (
01:57:47amiconnIt means that some ROMs won't work when music is playing, but will work when music is stopped. Other ROMs will always work.
01:57:56amiconnMight confuse the user...
02:05:13 Quit JdGordon| (" ajax IRC Client")
02:07:58mtdoes adding a rule in codecs.make link to common files like file.c ?
02:11:33amiconnRockbox codecs must not link core objects.
02:12:09amiconnIf you need access to core functions, you need to use the codec api
02:14:05pixelmadoesn't Rockboy stop playback anyways (except on the gigabeat)?
02:14:28evilnick_home1It should really give the user the choice, as in Jpegviewer
02:14:49 Quit PaulJam__ (Read error: 113 (No route to host))
02:16:01mtamiconn : So functions like lseek and read should be accessed through the codec api ?
02:16:56 Join JdGordon| [0] (i=4607ea54@rockbox/developer/JdGordon)
02:18:16JdGordon|kugel: last week you stopped me from working on the latest custom vp patch?
02:18:45pixelmasome forum admin around?
02:23:48 Quit efyx_ (Remote closed the connection)
02:25:54 Quit mt ("ChatZilla 0.9.84 [Firefox 3.0.10/2009042316]")
02:28:19kkurbjunamiconn, if rockbox's behavior is confusing for users, which I havn't seen much of, I would suggest that just changing the exit message would be enough
02:28:27kkurbjunrockboys that is
02:29:08kkurbjunon the gigabeat's rockboy is very playable with music running and why not take advantage of that if possible
02:33:02 Join cool_walking_ [0] (i=cb3b81c3@gateway/web/ajax/
02:35:06***Saving seen data "./dancer.seen"
02:36:24Unhelpfulamiconn: *is* the plugin buffer right now always immediately after the audio buffer?
02:38:09JdGordon|there is a wiki page which has the mem layout
02:38:13kkurbjunUnhelpful: that is up to the individual targets to define, on the M:robe 500 it is not
02:38:39JdGordon|remember audio buffer has a 32k buffer tacked onto the end of it (which may be considered a sepearate buffer... i dunno)
02:38:44kkurbjunI'm not sure about other targets offhand
02:38:51Unhelpfulhrm... are there macros defined that i coulduse to check the layout?
02:39:28kkurbjunthere's variables that are defined in the linker
02:39:44kkurbjunI think most targets have them setup
02:40:32Unhelpfuli was looking to eliminate a run-time check. it won't be that much code, though, so there might as well be one.
02:40:57kkurbjunthings like _audiobuffer _codecbuf
02:41:30kkurbjunyou could change the variables around too to include the beginning and end of each
02:42:17Unhelpfulthat would still be a runtime check... so i may as well have a runtime check :)
02:42:21kkurbjunor you could make sure that all the linker's are the same, but I think it would be better to check it in code than make the linker script have an arbitrary requirement
02:42:47kkurbjunlinker scripts I mean
02:43:25 Quit JdGordon| (" ajax IRC Client")
02:43:35kkurbjunit would be trivial to change on the m:robe, but I don't think it would be good to expect them to remain that way unless you had a #error that could make sure that if someone changes it they know there will be problems
02:43:57kkurbjunwell, I say m:robe, but it's any dm320 based target right now
02:44:39kkurbjunthe gigabeatF/X doesn't have the plugin buffer immediately after the audio buffer either
02:45:24Unhelpfulshould i check for the possibility that the plugin buffer is *before* the audio buffer?
02:45:28kkurbjunnor the as3525 based devices.. they all have the codec buffer between the audio buffer and the plugin buffer
02:45:38kkurbjunI think that would be reasonable too
02:46:06kkurbjunI was looking at changing it that way on the dm320 at one point, I don't remember why offhand, but I never did it
02:48:42saratogaunboosted on the AMS targets is 62 MHz with the latest MMU work right?
02:54:58CIA-38New commit by unhelpful (r21172): PictureFlow: subtract space used to load plugin from available audio buffer space if needed.
02:57:42kugelJdGordon|: I stopped you?
02:59:15kugelanyway, the recent progress of the AMSes carries me a bit a away from custom vp :S
03:00:13kugelsaratoga: without too. 62 is svn
03:02:21JdGordonkugel: not really stopped... you asked if i had started fiddling.. then left...
03:02:35JdGordoni took that to mean youd done something bad/diffrent?
03:02:54 Quit faemir ("Lost terminal")
03:03:35kugelI don't think so (but I can't really remember too)
03:03:58kugelI always welcome if people take a look (and shot) and my patches :)
03:05:12JdGordonis there much to do on it still? or still the same?
03:05:57kugeljust hunting down screens which don't clear properly, basically
03:06:20kugellocally, I've changed plugin behavior to default to fullscreen (for now)
03:06:29kugeland of course, viewport.c needs a cleanup as well
03:08:43*JdGordon wants to say do that clean up later
03:09:19 Join barrywardell [0] (n=barrywar@
03:12:21 Join BHSPitMonkey [0] (n=stephen@unaffiliated/bhspitmonkey)
03:19:22 Join maxio [0] (n=c3f11d8d@gateway/web/cgi-irc/
03:20:18 Quit kugel ("ChatZilla 0.9.84 [Firefox 3.0.10/2009042316]")
03:22:01maxioi thought i should join this place now. getting a ipod video tomorrow and gonna rockbox it. and if something goes wrong or im confused i guessed someone would know here.
03:22:41cool_walking_Make sure you actually get an iPod Video and not an iPod Classic (which Rockbox doesn't run on).
03:22:48kkurbjunyep, but if you use rbutil it should be pretty painless to install
03:24:04maxioahh dont worry ive researched my buy well.
03:24:14maxiopretty lucky find
03:24:35maxiovideo 30gb in box with all the stickers on it never used
03:24:47maxionot 5.5g tho
03:25:37maxiojust have to find some kind of case which is best for it
03:26:44cool_walking_I like the silicone skins.
03:27:07maxiomy friend has the silicone kind
03:27:35maxiothe only thing i dont like is how the wheel feels sticky
03:28:31cool_walking_They don't add too much size, they absorb impact rather than breaking themselves when they drop, and they don't get in the way of buttons/ports (and can easily be cut if they do).
03:28:47*scorche notes that we have an offtopic channel for this sort of discussion
03:29:14 Quit AndyI (Read error: 60 (Operation timed out))
03:29:33 Quit barrywardell ()
03:29:53maxiooeps sorry didnt know there was such a channel
03:30:37cool_walking_From this channel's topic: Please direct offtopic/social chat to #rockbox-community
03:31:34 Join AndyI [0] (n=pasha_in@
03:40:09Unhelpfulsaratoga: ok, the area scaler costs 3 * dw * sh + 3 * dw * dh multiplies per component. the linear scaler costs 3 * sw * sh + 3 * dw * sh. the linear scaler is designed for upscaling, and would probably break on a downscaling case. in general, the cost is 3 * min(sw, dw) * sh + 3 * dw * min(dh, sh)
03:41:43Unhelpfulif we assume that the area scaler is never scaling down by more than 2x, so that every output pixel crosses an input pixel boundary, i believe we can reduce the constant multiplier from 3 to 2. the same restriction also means that it's not doing many additions.
03:41:43 Quit maxio ("CGI:IRC (EOF)")
03:47:19 Join Llorean1 [0] (
03:49:22 Quit Llorean (Nick collision from services.)
03:49:24 Nick Llorean1 is now known as Llorean (
04:11:13 Join Tichodroma [0] (n=8cb40245@gateway/web/cgi-irc/
04:12:38onlysoaaHello all; how do I disable compiling plugins?
04:13:23onlysoaaI'll likely run a bunch of tests with the binary alone, so I'd rather not rebuild all the plugins too every time I need to test some changes.
04:13:48krazykityou could just make bin
04:13:49cool_walking_remove them from the SOURCES/SUBDIRS files.
04:13:51TichodromaI'm running svn on an ipod video. I tried pictureflow but only a small number of my covers appeared; the rest were '?'. I couldn't see any difference between the images that did and didin't work. Is pictureflow a work-in-progress so I should shut up and wait, or should it be working?
04:14:08onlysoaamake bin? sounds good. Thanks.
04:18:36saratogausually it won't rebuild plugins unless you're changing them
04:19:31onlysoaaCool, so I just rerun make?
04:21:05saratogamake -j is probably faster though
04:21:09onlysoaaCool, thanks.
04:27:05 Quit HellDragon (Read error: 104 (Connection reset by peer))
04:27:19 Join HellDragon [0] (
04:31:31 Join FlynDice [0] (
04:35:07***Saving seen data "./dancer.seen"
04:37:06JdGordonare we happy with the custom list/ui viewport setting being named "list viewport" in the .cfg?
04:44:30 Join hd [0] (
04:44:53 Quit HellDragon (Read error: 104 (Connection reset by peer))
05:00:31 Quit chandoo ("Leaving")
05:02:42onlysoaaHey everyone, I don't think there's a power on/off sequence in the Samsung P2's LCD datasheet, right?
05:03:41onlysoaaThe sequence is there, but it's a graph, and it doesn't show which registers should be set.
05:08:12 Quit Tichodroma ("CGI:IRC (Ping timeout)")
05:21:07onlysoaaOh, nevermind, I think I found it.
05:21:58LloreanJdGordon: Sounds okay to me. Is there some reason that isn't appropriate that I don't know of?
05:23:04JdGordonim not saying its inappropriate.. just wondering if we cant come up with something better?
05:23:30Lloreanlist dimensions?
05:24:24JdGordonwell... i dont like the list part of it... its for the whole ui...
05:27:23 Quit jordan` (Read error: 104 (Connection reset by peer))
05:27:37 Join jordan` [0] (i=gromit@
05:29:28 Quit PaulJam (".")
05:36:21 Quit relentless (Remote closed the connection)
05:46:36 Quit Horscht ("Verlassend")
06:13:56 Join ender` [0] (
06:17:20 Join VytenisS [0] (
06:35:08***Saving seen data "./dancer.seen"
06:45:46 Join patmulchrone [0] (
06:57:43onlysoaaHelp, when I make bin, it still goes through the plugins and throws errors because the P2 uses a 480x272 screen and they don't support that resolution yet.
06:58:10onlysoaaIs there a way I can just make a rockbox binary to test using tcctool?
06:58:19JdGordonyou can disable plugins and codecs in configure
06:58:34JdGordonbut, make bin should ignore them also...
06:58:39onlysoaaAlright, I'll check it out.
06:58:44onlysoaaIt's strange, it didn't.
06:59:20 Join boobaloo [0] (n=chatzill@
07:01:23onlysoaaHuh! Even after disabling plugins in configure it still went through the plugins and threw errors.
07:01:25 Quit kkurbjun (Read error: 110 (Connection timed out))
07:01:59 Quit crashd (Read error: 110 (Connection timed out))
07:02:10cool_walking_try removing from apps/plugins/SOURCES and apps/plugins/SUBDIRS
07:09:03onlysoaaAh hah! I just wiped out the contents of those two files and it worked.
07:09:14onlysoaaI guess I could svn revert when I'm ready to work on the plugins.
07:17:33*Unhelpful suspects he has had a useful insight regarding pictureflow... by adding a few fields to struct album_data, it can also track the state of a background cache build and of cache loads... and will probably make the memory cache code a good deal simpler, which is nice as it may be the cause of some hangs on target that have been hard to debug :/
07:22:57 Quit patmulchrone (Read error: 110 (Connection timed out))
07:29:35 Quit jmillikin (Read error: 60 (Operation timed out))
07:33:44 Join matsl [0] (
07:34:16onlysoaaQuestion: does anyone know how the LCD panel GPIO A pins were determined for the D2?
07:35:05JdGordonprobably from the disassembled OF
07:35:29onlysoaaI can't seem to find where the LCD code is in the P2 firmware.
07:35:46JdGordonit has to be there somewhere :D
07:36:16onlysoaaI know, but 3.7MB if binary is tedious to look through. :P
07:36:39onlysoaaAlso, bear in mind I'm new to all this.
07:36:52 Quit Llorean (Read error: 104 (Connection reset by peer))
07:40:52tmztyou wouldn't be looking through "binary" you have to read assembler
07:45:20onlysoaaOh, of course, I'm going through the disassembly.
07:45:37onlysoaaNo sense in reading actual binary. :P
07:45:43onlysoaaI think I found something though.
07:49:00 Join bertrik [0] (
08:06:15CIA-38New commit by unhelpful (r21173): Fix red.
08:06:47 Join timc`` [0] (n=aoeu@
08:13:27 Quit timc (Connection timed out)
08:16:46 Join Zagor [242] (n=bjorn@rockbox/developer/Zagor)
08:16:48 Join tvelocity [0] (
08:16:48 Quit Zarggg ()
08:23:19 Quit safetydan ("Leaving.")
08:26:41 Join Rob2222 [0] (
08:26:56onlysoaaAh shoot, I'm stuck.
08:27:15onlysoaaWell I'm going to sleep now. (: I'll keep hacking tomorrow.
08:30:43 Join kachna [0] (
08:35:10***Saving seen data "./dancer.seen"
08:35:14CIA-38New commit by unhelpful (r21174): Don't perform YUV->RGB conversion before row output for unscaled loads of greyscale JPEG, as store_row_jpeg already does this.
08:36:39 Join Trid [0] (
08:38:10 Part Trid
08:38:55 Join JdGordon| [0] (i=43a02c5a@rockbox/developer/JdGordon)
08:39:26 Quit bertrik (Read error: 113 (No route to host))
08:43:39 Quit Rob2223 (Read error: 110 (Connection timed out))
08:45:18 Join flydutch [0] (
08:46:11 Part boobaloo
08:48:34 Join andrewbeveridge [0] (i=586d6058@gateway/web/ajax/
08:48:51 Join pondlife [50] (n=Steve@rockbox/developer/pondlife)
08:49:02 Quit JdGordon| (" ajax IRC Client")
08:49:07 Part andrewbeveridge
08:49:50pondlifeDoes anyone here understand ASM parameter passing?
08:51:19pondlifeIn FS #8894, I added const to various routines (typically changing "int32_t *src[]" to "const int32_t *src[]".
08:51:53pondlifeThere are some assembly implementations used - I wonder what the impact might be.
08:55:29 Join Zarggg [0] (
08:57:24 Join n1s [0] (n=n1s@rockbox/developer/n1s)
08:58:53 Join einhirn [0] (
08:59:08Unhelpfulpondlife: i don't think const changes ASM ... just what gcc expects C code to do to the data.
08:59:17 Quit BHSPitMonkey (Connection timed out)
08:59:20pondlifeThat was my guess, thanks.
09:13:42 Join places [0] (
09:18:33 Join Thundercloud [0] (i=thunderc@
09:21:25 Quit kachna (Remote closed the connection)
09:31:12 Join jon-kha [0] (
09:32:40 Quit Unhelpful (Read error: 60 (Operation timed out))
09:32:57 Join Unhelpful [0] (n=Militant@rockbox/developer/Unhelpful)
09:34:25 Quit Thundercloud (Remote closed the connection)
09:40:20 Join barrywardell [0] (n=barrywar@
09:47:06 Quit Zarggg (Read error: 110 (Connection timed out))
09:51:19 Quit perrikwp (" ajax IRC Client")
09:53:43GodEatermmmm, cofffeeeeeee
09:54:29*FrankTM_ purrs
10:06:13pixelmaUnhelpful: the committed fix for overlay pf is different to the one I tested yesterday, right?
10:06:16 Quit Unhelpful (
10:06:20 Join Unhelpfu- [0] (
10:06:49*pixelma wonders whether Unhelpfu- still got the question
10:07:03 Quit cool_walking_ (Read error: 104 (Connection reset by peer))
10:07:03 Quit onlysoaa (Remote closed the connection)
10:07:03 Quit saratoga (Read error: 54 (Connection reset by peer))
10:07:24 Nick Unhelpfu- is now known as Unhelpful (
10:07:32FrankTM_pixelma: doubtful :p
10:07:41Unhelpfulpixelma: i don't recall seeing a question, no.
10:07:52FrankTM_10:06 < pixelma> Unhelpful: the committed fix for overlay pf is different to the one I tested yesterday, right?
10:08:03pondlifegevaerts: If/when you get time - there's a new patch on FS #8894...
10:08:10pixelmaFrankTM_: the channel is logged (if server is up)
10:08:20FrankTM_oh well
10:08:28FrankTM_this is quicker :p
10:09:17Unhelpfulpixelma: yes, it is. i based it on the rockboy solution, since the audio buffer will allow a larger number of slides in the in-memory cache.
10:10:13pixelmaok, I'll update and test then
10:11:29Unhelpfulthanks, although after the first one worked, i'm pretty sure that the problem has been identified, and that this will work also.
10:12:57 Join cool_walking_ [0] (i=cb3b81c3@gateway/web/ajax/
10:13:50 Join perrikwp [0] (i=4aa794a0@gateway/web/ajax/
10:15:46 Join pyro_maniac [0] (
10:19:36CIA-38New commit by unhelpful (r21175): Replace arrays of tags that are numeric/sorted/uniqued with bitfields flagging each tag that is a member of the set, and replace the membership tests ...
10:31:36 Join efyx_ [0] (
10:31:42 Quit cool_walking_ (" ajax IRC Client")
10:35:12***Saving seen data "./dancer.seen"
10:46:58pondlifeHmm, if you change a setting and then ROLO, the setting change isn't saved to disk...
10:48:13GodEatergood to see you're back working with as again pondlife ;)
10:48:55pondlifeJust coming up for air really...
10:53:50pondlifeGodEater: Do you still use the iPod Video? If so, a test of the latest FS #8894 would be useful...
10:54:03GodEaterI do
10:54:08GodEaterlet me get on that for you
10:54:41amiconnUnhelpful: For SH you could use a small lookup table for 1<<n. That'd be just 32 entries...
10:55:24Unhelpfulamiconn: i'd though of a lookup table with flags, but that's an even better idea :)
10:59:00Unhelpfulhrm, the 1<<n LUT would be about 128B. if i use a byte LUT per set, it would actually be 3/4 the size of the 1<<n LUT.
11:01:17GodEaterpondlife: building now
11:02:11 Join pyro_maniac1 [0] (
11:02:19*GodEater watches his ipod carefully incase of magic smoke incidents
11:02:53 Quit pyro_maniac (Read error: 110 (Connection timed out))
11:05:09 Join LinusN [0] (n=linus@gateway/web/cgi-irc/
11:10:41 Join kachna [0] (
11:17:30GodEatergah stupid ipod
11:17:35GodEater"Very low battery" my arse
11:17:54*GodEater performs the "one thumb salute"
11:20:04GodEaterpondlife: why does it require a reboot ?
11:20:23 Quit barrywardell ()
11:22:22pondlifeGodEater: To allocate a bit of memory
11:22:26 Join stripwax [0] (n=cc04828c@gateway/web/cgi-irc/
11:22:28GodEateralso, once I've turned it on, how do I use it ? :)
11:22:38pondlifeGo into the pitch screen
11:22:53pondlifePress the mode key (whatever that is).
11:23:06pondlifeIt will go to semitone mode
11:23:08stripwaxr21175 - tagcache.c has a bug, tagcache_is_sorted is looking at TAGCACHE_UNIQUE rather than TAGCACHE_SORTED
11:23:32pondlifePress the mode key again - now you should see Speed and Pitch
11:23:49pondlifeLeft/right controls speed, up/down controls pitch
11:23:52GodEaterI can't work out what the mode key is
11:23:55stripwaxcan't fix from here (would if I could)
11:24:07stripwaxcan someone fix?
11:24:12stripwax(gotta go)
11:24:14 Quit stripwax (Client Quit)
11:24:24*pondlife calls Mr Someone
11:24:33GodEaterah, got it
11:24:51GodEaterit's a bit choppy
11:25:05GodEaterinstant artifacts
11:25:08GodEatereven at only 101%
11:25:11pondlifeYes, and mono
11:25:49GodEater"not quite ready" imo
11:26:35pondlifeDoes it crash?
11:26:43GodEaternot so far
11:26:50pondlifeIt's meant for spoken word really
11:27:00GodEaterI can try that
11:27:05GodEaterI have HHGTTG
11:27:07pondlifeThe quality won't be improved pre commit
11:27:15 Quit _lifeless (Remote closed the connection)
11:27:18GodEaterit's all .spx
11:27:32 Join _lifeless [0] (n=lifeless@
11:28:16GodEaterwell the quality doesn't get any worse
11:28:40GodEaternow up to 200%
11:28:54GodEaterand it's fine for speech
11:28:59GodEaterquality isn't nearly as bad
11:29:04GodEater(or at least, not as noticeable)
11:30:00GodEatercutting out at 238%
11:30:03GodEaterno crash tho
11:31:20GodEaterbit dire at slower speeds?
11:31:29GodEaterat 50% it's very crackly
11:33:15 Join cool_walking_ [0] (
11:36:57AlexPGodEater: There was the start of a conversation at some point regarding where in the manual to put HID
11:37:05AlexPBut I don't think anything was decided
11:37:14AlexPSo if anyone has any bright plans...
11:38:09CIA-38New commit by unhelpful (r21176): Replace bitfield for tagcache tag sets with array of char on SH.
11:38:29GodEaterseems a bit odd
11:39:26GodEaterah ok
11:39:29GodEaterit's working now
11:39:36GodEaterseems winamp took a few moments to get the hang of it
11:39:54GodEaterdon't have winamp AND spotify open at the same time
11:45:18 Join snill [0] (n=chatzill@linuxc41.rz.RWTH-Aachen.DE)
11:47:25pixelmaUnhelpful: seen stripwax's report about r21075 about 50 minutes ago?
11:47:32GodEaterpondlife: is the quality any better on a more powerful target than the ipod ?
11:47:59pixelmaUnhelpful: err.. 21175
11:48:00*GodEater is amazed stripwax is finding the time to do bug reports whilst he's at work
11:48:05GodEaterthey're working him like a slave over there
11:49:08pixelmaand 50 is wrong too, wonder what I thought there - I meant 20
11:49:47GodEaterwhere is his report ?
11:49:53GodEaternot in FS in take it ?
11:50:14pixelmano, here in the channel
11:50:25*GodEater adjusts his non-existent glasses
11:53:08pondlifeGodEater: Sorry, work called. If you're getting choppiness/breakup then that's probably CPU limited.
11:53:30pondlifeI don't get that on H300, jus tslightly garbly artifacts.
11:53:34GodEaterpondlife: that's what I thought
11:53:44GodEaterI don't have my gigabeat to test it on
11:53:49GodEaterwill have to try it at home
11:54:01GodEater(assuming no-one's tested it on that yet)
11:55:52pondlifeI'd be interested, especially in seeing if it Data Aborts..
11:56:25pondlifeAlso, an e200 test would be good if anyone out there has one
11:59:07GodEaterI have a c200
11:59:11GodEatercould try it on that I suppose
12:00:24pondlifeYup - I know that e200 owners have reported Data Aborts
12:04:03Unhelpfulpixelma: yes, saw that, the function in question is gone now, but i just double-checked to make sure its replacement is correct
12:35:16***Saving seen data "./dancer.seen"
12:37:10 Quit ender` (Read error: 54 (Connection reset by peer))
12:37:59 Join funman [0] (n=fun@rockbox/developer/funman)
12:38:59 Join dionoea [0] (
12:39:10funmankugel bertrik: usleep() for Sansa AMS using TIMER2 :
12:49:37linuxstbfunman: Why the ">> 1" for HAVE_SCROLLWHEEL?
12:50:47funmanlinuxstb: it comes from kernel-as3525.c tick_start() : the scrollwheel is read twice per tick, so the timer interrut frequency is doubled
12:51:30linuxstbThen shouldn't TIMER_FREQ be different on those targets?
12:51:51funmanhum and then I just notice this code is wrong and the scrollwheel is read once per tick (on one timer interrupt call_tick_tasks is called, then on the following the scrollwheel is read)
12:53:53funmanlinuxstb: yes this could work this way. but I will ask kugel to just remove this tweak
12:54:06linuxstbAlso, how many timers does the AS3525 have? Some parts of Rockbox use a second timer (i.e. not the one for the 100Hz tick) for things (it's called the user timer). So that could conflict.
12:54:49funmanthere is 2 hardware timers : one is used with a fixed interval for the kernel tick (the one I use), the other one is user configurable
12:55:14linuxstbHmm, then I think your usleep may cause problems.
12:55:26funmanhow ?
12:55:39linuxstbWhat I just said - parts of Rockbox use a second timer.
12:55:39FrankTM_is there stuff that needs testing on fuzes?
12:56:04linuxstbSo it's not free for you to use in usleep.
12:56:23funmanwell this usleep doesn't use the user timer
12:56:41funmanFrankTM_: no
12:57:04FrankTM_funman: alrighty
12:57:35FrankTM_seemed very stable so far anyway :)
12:57:38linuxstbfunman: Ah, so TIMER2 is the one already being used for the kernel tick?
12:58:08funmanyes, (I wrote this function in kernel-as3525.c to not forget this fact)
12:58:16linuxstbOK, then ignore me ;)
13:00:22funmanlinuxstb: there is a lot of ways to make errors in rockbox, so thanks for checking. Don't worry I won't ignore you ;)
13:00:29*GodEater starts trying to put together an iphone toolchain again
13:01:08GodEaterlinuxstb: you don't happen to have the OSX 10.5 SDK anywhere do you? Or know where I can get it from. Apple's developer site is a nightmare.
13:01:17funmanGodEater: are you porting rockbox to iphone?
13:01:34GodEaterfunman: I might be ;)
13:02:09 Join kugel [0] (n=kugel@rockbox/developer/kugel)
13:02:24 Quit snill ("ChatZilla 0.9.84 [Firefox 3.0.10/2009042708]")
13:02:24FrankTM_GodEater: i might be able to supply that
13:03:00GodEaterideally a link to it on the apple site would be the best
13:03:14FrankTM_do you have an apple developer account?
13:03:15GodEaterat the moment I'm downloading Xcode in the hope it's in there somewhere
13:03:26FrankTM_i think it is
13:03:26GodEaterI have some sort of apple dev account
13:03:33GodEaterI've no idea where it allows me to go though
13:03:57FrankTM_apple isn't so open, sadly :(
13:04:48funmanGodEater: do you plan to build an iPhone OSX app, or a complete firmware?
13:06:19FrankTM_GodEater: is OSX SDK the same as iPhone SDK?
13:06:43linuxstbGodEater: No, but I can't remember needing to download it. I think it either comes on the OSX CD/DVDs or may even already be on your hard disk as a .dmg file somewhere.
13:06:57FrankTM_should be on the DVD
13:07:19GodEateriphone OSX app
13:07:20funmanyou need to download updates at least (iphone SDK wasn't on my 10.5 DVD)
13:07:23GodEaternot a complete firmware
13:07:28GodEaterthat would be pointless imo
13:07:32FrankTM_but the internet onces might be more up to date
13:08:12 Nick fxb__ is now known as fxb (
13:10:37GodEaterI have the iphone sdk already
13:10:42GodEaterit's the osx sdk that I appear to need
13:11:02 Quit funman ("leaving")
13:14:46kugelDon't we just need SDL on it? :)
13:15:00 Join wark [0] (
13:15:05 Join lennyf_ [0] (
13:16:01GodEaterwe need a way to compile for it
13:16:05GodEaterwhether it has SDL or not
13:16:30kugellinuxstb: the >>1 is to alter the TIMER_FREQ actually..
13:17:43kugelso that it's halfed for scrollweel, which makes scrollwheel reading more often possible
13:19:01linuxstbkugel: Wouldn't it be simpler to have TIMER_FREQ contain the correct value?
13:19:21kugelI don't know
13:19:37kugelI thought it would be better for TIMER_FREQ to translate into the tick task frequency
13:19:40linuxstbOr at least, it would seem clearer to me.
13:19:56kugelfor me, this was clearer
13:20:59linuxstbBut doesn't it mean using #ifdef HAVE_SCROLLWHEEL anywhere you're using that value?
13:22:23kugelthe value isn't used elsewhere
13:22:42kugelbut yea, I'm not opposed to changing that
13:26:17 Quit fyrestorm (Read error: 104 (Connection reset by peer))
13:26:33kugellinuxstb: funman said "the code is wrong", was he refering to my or his code?
13:26:45kugeltoo bad he isn't here anymore to ask him directly
13:28:07 Join fyrestorm [0] (
13:29:42 Join Llorean [0] (
13:30:19 Join barrywardell [0] (
13:32:07linuxstbkugel: I understood that he was talking about his code, but I'm not sure...
13:36:01CIA-38New commit by kugel (r21177): Change the timer interrupt setup so that TIMER_FREQ is changed for HAVE_SCROLLWHEEL (which read the scrollwheel between tick tasks too) instead of the ...
13:38:33kugelI'm not sure if his code is working
13:39:13 Quit gevaerts (Nick collision from services.)
13:39:22 Join gevaerts [0] (n=fg@rockbox/developer/gevaerts)
13:40:11 Join Tomers [0] (
13:55:55 Quit tvelocity (Read error: 60 (Operation timed out))
13:57:26 Join tvelocity [0] (
13:57:49 Quit tvelocity (Client Quit)
14:04:55 Join PaulJam [0] (
14:17:58 Quit places ()
14:25:58 Join nibbler [0] (
14:31:26CIA-38New commit by zagor (r21178): Branching for the new www design.
14:33:31Zagormeh, my svn fu is weak. wrong module...
14:35:01CIA-38New commit by zagor (r21179): Incorrect branch...
14:35:19***Saving seen data "./dancer.seen"
14:35:24 Join mcuelenaere [0] (
14:35:45 Join LambdaCalculus37 [0] (i=44a0430d@rockbox/staff/LambdaCalculus37)
14:36:07mcuelenaerelinuxstb (or someone else): you said parts of Rockbox use the second hardware timer; can you tell me which parts? (because the Onda doesn't have it (properly) implemented)
14:37:11linuxstbmcuelenaere: I don't think it's anything vital. I think the backlight fading code does, plus some plugins such as Doom and maybe metronome. Plus the greyscale lib.
14:37:15***Alert Mode level 1
14:37:15CIA-38New commit by zagor (r21180): Branching for the new www design.
14:38:49mcuelenaerebacklight fading code as in HW PWM backlight fading code? (or just the generic backlight fading code?)
14:39:28linuxstbOne of them... I've lost track.
14:40:08 Join AndrewRB [0] (
14:41:05linuxstbmcuelenaere: Search the source for "timer_register" - that registers a user timer.
14:41:40kugelso that should be usable for udelay()?
14:42:04 Part lennyf_ ("Ex-Chat")
14:43:44***Alert Mode level 2
14:43:44CIA-38New commit by zagor (r21181): as-is. Much (re)work remains.
14:46:13FrankTM_Zagor: wow
14:46:35FrankTM_nice design
14:47:05EtuZagor: Looks great
14:49:14Zagoryeah it's pretty but I'm not happy with the "busyness" of it. I'm (slowly) working on splitting it into a "user" and "developers" page (or abstract / detailed)
14:49:34LambdaCalculus37Zagor: /branches/www-redesign/media/images/players/toshiba_gigabeat_x.jpg is actually an S. ;)
14:49:47ZagorLambdaCalculus37: and now you have the chance to fix it! :-)
14:50:50 Join kkurbjun [0] (n=kkurbjun@rockbox/developer/kkurbjun)
14:51:52LambdaCalculus37Zagor: Maybe toffe82 can supply you with a picture of an X. I don't have one myself.
14:53:45***Alert Mode OFF
14:55:16 Quit r0b- (Read error: 104 (Connection reset by peer))
14:59:04 Join r0b- [0] (
15:01:59 Join jgarvey [0] (
15:02:42 Quit kugel (Read error: 110 (Connection timed out))
15:04:08AndrewRBis the an ACTION to switch the backlight on?
15:04:31AndrewRBor perhaps toggle
15:07:52mcuelenaerehmm anyone ever tried git bisect with Rockbox?
15:10:50linuxstbAndrewRB: Any action would normally turn on the backlight, so you could probably just define anything. But no, there's nothing to specifically toggle it - no targets have any consistently free buttons for that.
15:11:26CIA-38New commit by FlynDice (r21182):
15:12:29AndrewRBlinuxstb: yeah, i've defined the power switch on my D2 as ACTION_REDRAW. No way to switch it off by choice though, presumably, without modifying something other than just the keymap for my device.
15:12:42 Quit antil33t (Read error: 54 (Connection reset by peer))
15:12:56 Join antil33t [0] (n=Mudkips@
15:14:02FlynDicewhoops, any way to get the comments into that latest commit message? My fingers were a bit too fast :(
15:14:33FrankTM_FlynDice: it's possible, if configured
15:16:24FlynDiceFrankTM_: what do you mean?
15:17:14FrankTM_the server must allow it, and it's not on by default
15:18:10FrankTM_it's very simple to enable though
15:19:15 Quit dmb (Read error: 113 (No route to host))
15:27:09 Join gregzx [0] (
15:28:10 Join evilnick [0] (i=0c140464@gateway/web/ajax/
15:36:52 Join bmbl [0] (n=Miranda@unaffiliated/bmbl)
15:42:09FrankTM_FlynDice: any luck?
15:45:29 Join killan_ [0] (
15:47:54linuxstbFrankTM_: I don't think it's enabled for Rockbox. I could be wrong though...
15:48:38FrankTM_not likely :p
15:49:03 Quit jgarvey (
15:49:03 Quit barrywardell (
15:49:03 Quit Zagor (
15:49:03 Quit goffa__ (
15:49:03 Quit killan (
15:49:03 Quit pabs (
15:49:03 Quit avacore^ (
15:49:03 Quit soap (
15:49:18linuxstbFrankTM_: Would there be a downside to enabling it?
15:49:41NJoinjgarvey [0] (
15:49:41 Join barrywardell [0] (n=barry@rockbox/developer/barrywardell)
15:49:41NJoinZagor [242] (n=bjorn@rockbox/developer/Zagor)
15:49:41NJoingoffa__ [0] (n=goffa@
15:49:41NJoinkillan [0] (
15:49:41NJoinavacore^ [0] (
15:49:41NJoinsoap [50] (n=soap@rockbox/staff/soap)
15:49:41NJoinpabs [0] (
15:49:59FrankTM_i'm not 100% sure actually
15:50:03FrankTM_we have it enabled here
15:50:40FlynDiceFrankTM_: still wading through... sorry my notifications are all messed up since the konversation "enhancement" the other day
15:51:06FrankTM_using kde was a mistake after all ;)
15:51:07*FrankTM_ runs
15:53:07FrankTM_linuxstb: i think comments are not under version control
15:53:14FrankTM_so once updated. there's no way back
15:53:28 Quit goffa__ (
15:53:28 Quit barrywardell (
15:53:28 Quit avacore^ (
15:53:28 Quit jgarvey (
15:53:28 Quit pabs (
15:53:28 Quit Zagor (
15:53:28 Quit soap (
15:53:28 Quit killan (
15:55:44NJoinjgarvey [0] (
15:55:44NJoinbarrywardell [0] (n=barry@rockbox/developer/barrywardell)
15:55:44NJoinZagor [242] (n=bjorn@rockbox/developer/Zagor)
15:55:44NJoingoffa__ [0] (n=goffa@
15:55:44NJoinkillan [0] (
15:55:44NJoinavacore^ [0] (
15:55:44NJoinsoap [50] (n=soap@rockbox/staff/soap)
15:55:44NJoinpabs [0] (
15:56:19AndrewRBsorry for the seemingly-helpless questions, but I'm trying to find my way around the rockbox source and getting confused. Where could I add code for a custom action? I would like to create "ACTION_TOGGLE_BACKLIGHT" for personal use, yet I am lost.
16:00:16 Join dfkt [0] (i=dfkt@unaffiliated/dfkt)
16:00:42FrankTM_linuxstb: is rockbox on dedicated servers?
16:02:27 Quit killan (Connection timed out)
16:03:09 Join pabs_ [0] (n=pabs@
16:09:21 Join Zagor_ [0] (n=bjorn@
16:09:51 Quit pabs (Read error: 113 (No route to host))
16:10:01 Part LinusN
16:10:32 Quit Zagor (Nick collision from services.)
16:10:37 Nick Zagor_ is now known as Zagor (n=bjorn@
16:13:11 Join andrewbeveridge [0] (
16:13:32 Quit andrewbeveridge (Remote closed the connection)
16:14:49linuxstbFrankTM_: In what way? The admins own the boxes, but I think they're used for other things as well.
16:17:30FrankTM_ah. i would like to contribute to the project :P
16:17:38FrankTM_but i don't think my C will help a lot
16:19:01linuxstbYou want to contribute hosting?
16:19:38FrankTM_i'm a sys-admin
16:19:43FrankTM_dunno if i can help :p
16:21:44n1sFrankTM_: donating cpu/bandwith by hosting a build server is appreciated
16:21:58linuxstbThere are lots of scripts (mainly Perl) behind the Rockbox website. I'm sure contributions to hacking those would be welcome if you have time. But I don't think general sysadmin is needed.
16:22:29 Join dmb [0] (n=dmb@unaffiliated/dmb)
16:23:18 Join funman [0] (n=fun@rockbox/developer/funman)
16:23:51FrankTM_linuxstb: i could have a look
16:24:50FrankTM_n1s: i could do builds aswell
16:25:21FrankTM_although i might not have the cpu-power
16:25:23Zagorstart with adding a build server. we don't have any particular perl needs right now.
16:26:05FrankTM_are there any specs on the current build servers?
16:26:46linuxstbZagor: I'm just thinking that there are always changes people ask about, but no-one does them. I can't remember any examples though...
16:27:38FrankTM_hehe.. hacking perl isnt what most ppl like :p
16:28:43linuxstbFrankTM_: Here are some general things for "someone" to do -
16:29:35Zagora toplist of build server speeds is a vital missing feature! ;)
16:30:09FrankTM_Zagor: hm?
16:30:35 Quit AndrewRB (Read error: 110 (Connection timed out))
16:32:13ZagorFrankTM_: on the far right of you have a link to build server stats, which in turn has link to the buildmaster log. it could be parsed to calculate the time used for each build on each server.
16:32:39Zagorif we compare those times for different build batches, we can construct a speed toplist.
16:33:03Zagorjust for fun. but oh, how fun! :-)
16:35:12FrankTM_stats are always good
16:35:21 Quit gregzx ("ChatZilla 0.9.84 [Firefox 3.0.10/2009042316]")
16:35:22FrankTM_preferbly supported by graphs :P
16:35:23***Saving seen data "./dancer.seen"
16:35:30Zagorthat's the spirit!
16:35:59evilnickZagor: Didn't rasher do something similar recently?
16:36:19FrankTM_Zagor: red <td>'s are errors/warnings i suppose?
16:36:20Zagordid he? if so, I haven't seen it.
16:36:28ZagorFrankTM_: yes
16:38:44evilnickI'm not sure precisely what it was but it definitely involved a lot of graphs
16:39:48pixelmaI think it was binsize (maybe RAM too) graphs
16:39:52 Join toffe82 [0] (n=chatzill@
16:40:30evilnickThat does ring a bell"> <−− These?
16:43:02ZagorLlorean: well those are afaik not measuring build speeds, and not for the buildmaster builds
16:43:21LloreanZagor: I was responding to evlinick and pixelma, really
16:43:22 Quit cool_walking_ (Remote closed the connection)
16:43:50LloreanThere seemed to be some uncertainty as to what Rasher had graphed is all.
16:44:21*Llorean would love to see build-related graphs over time like that, just for curiosity's sake.
16:44:31evilnickLlorean: Thanks for the explanation, I simply hadn't followed what all the discussion was about while it was happening
16:46:02LloreanZagor: How are builds handed out? Does the first build always go to the first build server in the list?
16:46:31 Join PaulJam_ [0] (
16:47:19ZagorLlorean: it looks like that, from the logs.
16:47:28Zagorbagder knows
16:47:39FrankTM_badger badger badger
16:47:46*FrankTM_ goes back to work
16:48:03LloreanI wonder if times could be optimized a bit just by being more selective about which server gets which builds.
16:48:17Zagoractually, no
16:48:55FrankTM_don't they all take pretty much the same time to build?
16:48:59Zagorthe biggest gains I think would be from ensuring that a slow server doesn't end up with a big build in the end
16:49:04ZagorFrankTM_: no, _very_ different
16:49:27Zagorthe values in parenthesis are seconds/build
16:49:46Zagorbetween 4 and 228 seconds in that log...
16:49:57FrankTM_oh wow :P
16:49:57LloreanZagor: Also pass out the big builds earliest, so that when they're finished we can follow them up with smaller ones, to avoid ever stacking servers with more than one big one in a row, too?
16:50:08FrankTM_just out of curiousity: is fuze a big build?
16:50:28n1sFrankTM_: bootloaders are the very fast ones (in only a few seconds)
16:50:41ZagorFrankTM_: parse the logs and find out ;-)
16:50:46funmanFrankTM_: 7.9MB here
16:51:01FrankTM_i meant in time :P
16:51:13n1sfull target builds are a lot slower to build and swcodec take a lot longer than hwcodec because of the codecs and usually more plugins
16:51:58FrankTM_i'm quite new to this project. so that explains some of these questions ;)
16:55:35 Join ender` [0] (
16:56:01EtuHmm, Does rockbox work on Sandisk Sansa Clip?
16:56:27LloreanIt's in development on it.
16:56:27funmanEtu: see
16:56:40Etu(I have seen it on the wikipage with non released players)
16:57:56EtuOk, if I want a Clip I want a ClipV1
17:02:26amiconnlinuxstb, Zagor: speculative parallel builds and uploads in the background are examples
17:02:40 Join cmwslw [0] (
17:02:53 Part cmwslw ("Ex-Chat")
17:03:06 Join patmulchrone [0] (
17:03:12Zagoramiconn: yeah, but perhaps a little tough to start with as a first introduction to rockbox...
17:04:12 Quit Zagor ("Don't panic")
17:08:53 Quit PaulJam (No route to host)
17:12:53 Join onlysoaa [0] (i=42248748@gateway/web/ajax/
17:30:02 Join robin0800 [0] (
17:31:26 Join jmillikin [0] (
17:31:43 Quit funman ("leaving")
17:33:04 Quit perrikwp (" ajax IRC Client")
17:39:24 Quit einhirn ("Miranda IM! Smaller, Faster, Easier.")
17:44:38 Join Zarggg [0] (
17:46:29 Quit timc`` (Read error: 110 (Connection timed out))
17:49:59 Join yomamathecablegu [0] (n=438d8f12@gateway/web/cgi-irc/
17:50:55 Quit barrywardell (Remote closed the connection)
17:51:16r0b-what transfer mode do i have to put my e250 in to install rockbox
17:52:00yomamathecableguHey everyone ^_^ I'm working on a site, and I just want to see how much support it gets from the rockbox community - just to get others' opinions. I'm working on a site that provides many, many downloads that are compatible with rockbox. For example, full length shows fit to your mp3 player's screen, a backdrop gallery, etc. Tell me what you think, and ask if you want more info ^_~
17:53:04 Quit at0m (Read error: 110 (Connection timed out))
17:53:15 Join at0m [0] (
17:54:33n1sr0b-: ums/msc
17:55:10r0b-installing :)
17:55:33Lloreanyomamathecablegu: There's already a theme gallery site. There's not really much need for a separate backdrop gallery - in the direction themes are involving backdrops are going to probably become less "fullscreen image" and more borders and layout imagery.
17:56:32LloreanAs for shows formatted, that's up to you but we don't really endorse or involve ourselves with media. People can get their videos or audio from wherever they like. We do, however, suggest you be very careful about copyright, and be aware there's an ever increasing number of screen dimensions so it will be a lot of work to keep such a site up to date with compatibility.
17:56:41 Join evilnick_230 [0] (i=0c140464@gateway/web/ajax/
17:58:05r0b-ROCKBOX IS LOVELY!!!!!!!!!!!
17:58:38evilnick_230It certainly is.
17:58:40yomamathecableguLlorean: it wouldn't just be a backdrop gallery, that's just to make it a bit more convenient - as to have everything in one place. there would be many types of files. it's really just in the scratchwork stage right now. and thanks for your suggestions about the shows. it'll take a while, but I think I can do it ;)
17:58:53r0b-i got it on this sansa e250 :)
17:59:44r0b-on the back it shows its v2 but the firmware it has is v1 something
17:59:53evilnick_230r0b-: Make sure that you read our fine manual so that you get to know exactly what your Sansa is now capable of doing
18:00:04r0b-i know some of the stuff
18:01:27 Quit robin0800 (Remote closed the connection)
18:01:50*pondlife points out the broken 'mipper' theme at again...
18:03:11yomamathecableguSo does anyone else have any opinions about it? ^_^
18:03:17pixelmaI still think it's not the theme that's broken
18:03:38pondlifepixelma: Well, I meant only one theme shows a problem...
18:03:49Lloreanyomamathecablegu: You haven't really said anything to give opinions on.
18:03:57 Join JdGordon| [0] (i=ad75fd4b@rockbox/developer/JdGordon)
18:05:06yomamathecablegupondlife: it's not a problem with the theme. it's one of two things: 1) there's an error in the pages php script, or 2) the image for the theme is in a different location than they told the web page
18:06:04pondlifeI'm more thinking it should be removed from the list
18:06:15Lloreanrather than fixed?
18:06:32 Quit onlysoaa (" ajax IRC Client")
18:06:55pondlifeWell, if the image is available then yes, fixed.
18:07:07yomamathecableguLlorean: true I guess. lol I'm sorry. I'm going off of like 2 hours of sleep. T.T Anyways, I guess I'm asking for suggestions mainly. 1) suggestions on the design of the site (can be seen here: 2) suggestions for what kinds of files i should include 3) what shows/movies you owuld want to see available for download etc. Just things like that ^_^ That would be very helpful.
18:07:37pixelmathe theme site needs to be fixed IMO - the theme is actually a 160x128 one and shows up correctly there and is not on the other 220x176x16 targets' theme pages (Ipod Photo/Colour etc.)
18:07:40pondlifeNon-copyright shows only, I guess.
18:08:14evilnick_2303) would be leaping face-first into copyright issues unless you stick to CC (or similar) shows
18:08:15yomamathecablegupondlife: if they don't have an image, they shouldn't completely get rid of an entire theme. i think they should just put a normal text link to the theme.
18:08:16pixelmasomething goes wrong when excluding it for the H300 (maybe because of the RWPS relations or so)
18:09:27pondlifeyomamathecablegu: At point (2), there's no need to include themes, we already do those "in house".
18:09:29yomamathecableguSome examples of some shows I'm already doing: Code Geass, Death Note, Ben 10.. just anything like that. I'll worry about the copyright issues ^_^
18:09:47r0b-thats pretty cool
18:10:19yomamathecablegupondlife: I won't include whole themes, just backdrops so that you can customize your theme. Nothing to fancy, just bitmap images sized right for your mp3.
18:10:25pondlifeyomamathecablegu: Well if you plan to violate copyright, don't expect any help from us.
18:10:52yomamathecablegupondlife: I don't plan to violate copyright.
18:10:58Lloreanyomamathecablegu: Few themes work well with interchangeable backdrops, and that's going to become even more true as soon as customizable list viewports are included.
18:11:19evilnick_230yomamathecablegu: And also, don't use this channel to discuss copyright violations either
18:11:26yomamathecableguLlorean: I see.. maybe I won't include that then ^_^;;
18:12:24yomamathecableguevilnick_230: I'm not trying to talk about copyright violations. I said 3 posts up that I would worry about that. That was meant to stop the speculation of copyright issues.
18:13:28 Quit kachna (Connection timed out)
18:13:35Lloreanyomamathecablegu: I'd suggest figuring out exactly what you want your site to be, before worrying specifically about things like what shows will be on it.
18:13:39LambdaCalculus37yomamathecablegu: You are talking about copyright violations just by mentioning it.
18:13:57LloreanAmong other things, if you do license them you're going to have to come up with a plan to prevent other people violating copyright, since Rockbox doesn't have any DRM measures.
18:14:18yomamathecableguLlorean: I will think about it more. Thanks :)
18:14:20evilnick_230Doesn't, and will not ever.
18:14:59yomamathecableguLambdaCalculus37: I wasn't the one that brought it up. Someone else brought it up, and I tried to end the talk about it.
18:15:51 Join domonoky [0] (n=Domonoky@rockbox/developer/domonoky)
18:19:32 Quit FlynDice (Remote closed the connection)
18:20:04 Part pondlife
18:21:01yomamathecableguWell, this has turned into an overall negative discussion. I guess I'll see you guys later.
18:21:37evilnick_230yomamathecablegu: You haven't really given us any ideas that are new
18:21:53evilnick_230I don't want to sound negative here, just stating facts
18:23:02evilnick_230I mean, backdrops and the like would be better served from the themes site (as it's centralised) rather than having "Nano backdrop site" and "iPod Video backdrop site" etc.
18:23:36 Join perrikwp [0] (i=18ac0c41@gateway/web/ajax/
18:23:42yomamathecableguI understand. I will come back later on with some more ideas. Just gotta sit down and brainstorm for a while. ^_^ Also, it is all just one site. The is just a dummy site. All of the mp3 players will be included in one main site.
18:23:59 Join PaulJam [0] (
18:25:58CIA-38New commit by mcuelenaere (r21183): Ingenic Jz4740: add extended exception debug reporting option
18:26:10JdGordon|yomamathecablegu: I missed the entire convo... but wouldnt it be better to just help out and make the site better?
18:26:17 Join Jaykay [0] (
18:28:43r0b-can i watch fullsize movies with rockbox or do they have to be the size of my players lcd
18:28:48evilnick_230JdGordon|: Darned good point
18:28:49Jaykayi get when i try to enter the history of the ipod5g wps gallery..
18:29:17evilnick_230r0b-: You *can* watch full-size movies, but the fps will be a lot better if you convert them to the size of the screen
18:29:38Lloreanr0b-: Size of your player's LCD.
18:29:46evilnick_230r0b-: That's "can" as in "yes it is possible, but not advisable"
18:29:47LloreanThese players don't have much in the way of CPU power.
18:30:10r0b-well i wanna watch like 2 fast 2 furious on my MP3 player :P
18:31:07 Quit Xerion (Read error: 104 (Connection reset by peer))
18:31:12gevaertsr0b-: have a look at
18:31:30 Join Xerion [0] (
18:31:58yomamathecableguJdGordonl: Good point. How would I go about doing this though? I'm sure they don't want all of those videos clogging up their server, and I'm just now getting into C programming, so I can't offer too much in the area of plugin development yet.
18:32:50JdGordon|c wont help you in web design anyway..
18:33:33 Quit Xerion (Read error: 54 (Connection reset by peer))
18:33:57 Join Xerion [0] (
18:34:02yomamathecableguI understand that. I mean plugin development. I'm very fluent in HTML, JavaScript, and CSS. Why would my web design skills help?
18:34:10yomamathecableguSorry if that seems like a stupid question.
18:34:48*JdGordon| never mentioened anything about plugins
18:35:24***Saving seen data "./dancer.seen"
18:37:08yomamathecableguNo no I mentioned plugin development. Saying that I'm new to C, so I wouldn't be able to help the plugin area. I didn't realize that you meant web design when you said that I could help make the site better.
18:38:29gevaertswhat site did you think he meant?
18:38:47 Join robin0800 [0] (
18:39:39yomamathecableguUgh this is so confusing. XP I know he meant, but I didn't know that he meant I could help with web design. I thought he meant I could help with plugins and adding things to the site and such.
18:41:33 Quit PaulJam_ (Read error: 113 (No route to host))
18:41:57 Quit JdGordon| (" ajax IRC Client")
18:42:18 Join kugel [0] (i=kugel@rockbox/developer/kugel)
18:44:10 Quit matsl (Read error: 110 (Connection timed out))
18:44:12evilnick_230yomamathecablegu: By all means come up with new ideas, but please don't take a "this would be better on the site" as a personal attack or being negative
18:45:02 Part pyro_maniac1 ("Leaving.")
18:46:56yomamathecableguevilnick_230: nonono I don't take it that way at all. In my experiences making websites, I've always wanted to have my OWN site. However, I don't think I realized until now that it would probably be more beneficial if I actually helped add on to the main site. I see it as a positive thing. ^_^ And hopefully soon I'll pick up C, and be able to start messing around with plugins. I'm gonna brainstorm some
18:47:23 Join FlynDice [0] (
18:47:31 Quit FlynDice (Client Quit)
18:48:02yomamathecableguI do have one question though - a while back I asked and got a good answer, but I've forgotten. I have on my mac (which I'm on right now). How do I run that to build a compiler?
18:48:35LloreanThat question's kinda confusing.
18:48:41LloreanYou just run it.
18:49:12FrankTM_yomamathecablegu: you run it from a terminal
18:49:17FrankTM_using sudo
18:49:53yomamathecableguMy bad. What would the command be?
18:50:03FrankTM_i suppose you need to have xcode installed too
18:50:12FrankTM_sudo ./
18:50:19yomamathecableguI have xcode installed
18:50:26yomamathecableguAnd thanks I'll try that out ^_^
18:51:14 Join PaulJam_ [0] (
18:51:35 Join FlynDice_ [0] (
18:51:52yomamathecableguit says "command not found" :/
18:52:20n1sare you in the right dir?
18:52:52Lloreanyomamathecablegu: usually we suggest people not try to compile until they're familiar with things like this that might be considered basic computer use. Googling how to use the terminal on your mac may help
18:53:49yomamathecableguI've used the Terminal before, just not with anything like this. I'm more familiar with the DOS Command Prompt, but the mac is all I have right now :/
18:54:45 Quit FlynDice_ (Client Quit)
18:55:34VytenisSHello, can anyone have a look at
18:55:52VytenisSAnd commit if everything is ok
18:55:57LloreanVytenisS: We usually look at translations more as we get closer to the release.
18:56:38 Join JdGordon| [0] (i=836b0070@rockbox/developer/JdGordon)
18:58:00 Join {phoenix} [0] (
18:59:05kugelVytenisS: it's assigned to LambdaCalculus37, you might want to talk to him
19:02:01r0b-why does rockbox make the music folder hidden?
19:02:18r0b-meaning the player itself cant see it
19:02:33pixelmaRockbox doesn't - the Sansa OF does
19:02:45pixelmaand you can see it
19:03:17LambdaCalculus37VytenisS: I haven't forgotten about your patch.
19:03:19 Join bertrik [0] (
19:03:26r0b-only if i tell xp to unhide it
19:03:38kugelbertrik: hey
19:04:06kugelI have put some more work into the lcd_write_data function, using an interrupt for FIFO FULL
19:04:25kugelthat gives even more fps (90fps unboosted)
19:04:58Lloreanr0b-: Or if you set Rockbox to show all files, rather than just supported
19:05:03bertrikoh nice, I wouldn't have expected to get more fps from using an interrupt though
19:05:34bertrikisn't it a lot more complex than simply polling the fifo status?
19:05:49kugelpolling hardware is a bit slower than polling a volatile variable, I guess
19:06:17kugelit's more complex yes, but the extra fps are worth it
19:06:49bertrikand you talked about some bits or something last night, were you able to get rid of those?
19:06:51kugelwhat I find weird though, I can't get over 100.0 fps, using either your or my patch, when boosted
19:06:59kugelis this a test_fps restriction?
19:07:46kugelor we just hit the FIFO's maximum(?)
19:07:50bertrikmy clip gets way more than 100 fps :P
19:08:24 Join saratoga [0] (i=98039dd8@rockbox/developer/saratoga)
19:08:29 Quit JdGordon| (" ajax IRC Client")
19:08:34kugel100.0 is a pretty suspicious number, especially if you can't manage to get above
19:08:41bertrikkugel, maybe that's just the maximum throughput indeed.
19:09:04bertrikhow much data is a full frame on the fuze?
19:09:15 Join SirFunk_ [0] (
19:09:26kugel220x176x2 Bytes
19:09:49bertrikI meant to say "blue bits" at 19:06
19:10:27kugelI don't understand
19:10:48kugelblue is 5bits, so 220x176x5 bits if you mean that
19:10:57bertrik01:02 last night : "bertrik: lcd updates show weird blue bars when boosted"
19:11:04n1skugel: that still seems a little slow for such a fast processor unless the lcd hookup is seriously crippeling the thouroughput, see LcdFrameRate for reference
19:11:09r0b-whats the easiest way to get talking menues
19:11:27n1sr0b-: install a voice file
19:11:36kugeln1s: yes, I looked at that page. I was wondering about the general slowness too
19:12:03r0b-i tried using the autoinstaller for rb but it didnt work right'
19:12:20n1sr0b-: please be more descriptive
19:12:31r0b-it cant find the sapi scripts
19:12:39bertrikkugel, as far as I understand, the FIFO just reduces the slack in between transfers to the display. I can imagine that at some point there is zero slack, giving a max theoretical fps
19:12:41r0b-i used the sapi speak engine
19:13:37n1sthe *easiest* way is of course to use a pregenerated voice file, i dont' know if rbutil ccan download those but they can be found here made with festival and English only)
19:14:00kugelbertrik: I'd like to see e200v2 reference messurements. It has a more complex initialisation, we might miss some bits
19:14:08 Join PaulJam__ [0] (
19:15:38kugelbertrik: re: blue bars, I was thinking if they have to do with caching. funman got weirdnesses with having buffers not cache aligned. I think the framebuffer isn't cache alligned too
19:15:56bertrikkugel, either me or you should create a fs patch to enable the dbop fifo, to make it easier for other people to test too
19:16:09kugelyep, I'll do
19:16:19 Quit at0m (Read error: 110 (Connection timed out))
19:16:26bertrikok thanks, I'll attach my non-interrupt patch to it too, OK?
19:16:32 Join BryanJacobs [0] (i=80974574@gateway/web/ajax/
19:18:00r0b-oh well i dont need a talking mp3 player
19:18:02r0b-i can see fine :P
19:18:58 Quit PaulJam (No route to host)
19:20:02 Join Horscht [0] (n=Horscht@xbmc/user/horscht)
19:22:12 Quit SirFunk__ (Read error: 110 (Connection timed out))
19:22:40kugelbertrik: fs#10272
19:23:32bertrikkugel, as far as I understand DBOP, it's basically a cycle generator to control a bunch of data and control lines towards a display. Suppose one write cycle takes 8 DBOP cycles at 62 MHz, then it can move 7.75 MB/s, a full frame is 77440 bytes which would be consistent with a max of 100 fps
19:24:03 Join JdGordon| [0] (i=836b0070@rockbox/developer/JdGordon)
19:24:04bertrikAlthough 8 cycles for one write cycle seems a little high
19:24:06 Quit PaulJam_ (Read error: 113 (No route to host))
19:24:42bertrikI could look that up in the source of course ...
19:25:13kugelinteresting calculation
19:25:21 Quit Jaykay ("ChatZilla 0.9.84 [Firefox 3.0.10/2009042316]")
19:25:46kugelis the 8 documented somewhere or did you guess to make the equation work?
19:25:49 Quit yomamathecablegu ("CGI:IRC (Ping timeout)")
19:25:53bertrikI guessed it
19:27:50 Join mt [0] (n=MTee@rockbox/developer/mt)
19:29:40kugelbertrik: I'm not really suprised that the interrupt is faster. But what I find suprising is that we can insert a free yield()
19:30:46kugelon the other hand, the exit part of that function is faster by polling the status register again (but only by 0.5fps or so)
19:32:21kugeln1s: 7.75MB/s is not a that bad value it seems
19:32:57kugelonly very few targets beat that (but those who do do vastly better....32MB/s @ e200v1)
19:33:48 Join FlynDice [0] (
19:34:39n1skugel: my comment was in relation to the speed of the cpu, it's quite a lot faster than pp
19:35:09n1sbut it may of course be limited by the hookup or the lcd controller
19:35:47kugeln1s: yes, I agree. But if you look at c200 vs e200, the lcd hookup makes a huge difference
19:36:36kugelI think the lcd didn't change in the e200v2, I'm very interested in the numbers of that one
19:38:18kugelalso, 90fps when unboosted vs 100fps when boosted (while the cpu clock is 4x) makes me think there's some hardware limitation (or missing bits in the init)
19:44:50gevaertsI've looked a bit at upload speeds of build hosts, and I wonder if the slower ones shouldn't be restricted to bootloaders and sims a bit more
19:45:13gevaertsNumbers are at
19:45:14kugelUnhelpful: wouldn't it make more sense to pre-shift the tags?
19:45:24bertrikkugel, we should be able to determine the numbers of DBOP clock ticks per write cycle from the DBOP_TIMPOL_XX registers I think
19:45:34kugellike #define tag_artist (1<<X)
19:45:37Bagdergevaerts: I don't think it's that simple
19:45:54Bagdersee how my host is #8 slowest, but often #1 in number of builds
19:46:23mcuelenaerekugel: regarding that 'free yield()', are you doing something like func() { while(wait_for_var_to_change) yield(); } irq_handler() { change_var; } ?
19:46:33BryanJacobskugel: with a precompiler define you won't get any performance gain - it's a direct substitution at the time of usage
19:46:41kugelmcuelenaere: no
19:47:02kugelor well, maybe yes
19:47:05 Join stoffel [0] (
19:47:20kugelhm, no (needed to figure out your example :P )
19:47:22mcuelenaereif so, try using wakeup_* calls, they will probably be faster
19:47:33mcuelenaere(they were on my target at least)
19:48:07gevaertsBagder: true. Maybe we should consider build time + upload time instead. I'll see if I can cook up some numbers
19:48:28Bagderthat could work
19:48:56Bagderand to make it really swell: do that prioritization a run-time decision/measure
19:49:15kugelmcuelenaere: this is the snippet:
19:49:24gevaertsthat would be perfect, yes, but also slightly more work :)
19:49:49kugelsee where I commented yield() out. The IRQ makes the FIFO filling loop stop (FIFO is full).
19:50:45kugelthe fifo needs to empty again. And as long as we fill faster than the fifo empties, we can put code there theoretically for free
19:51:25kugelwithout the yield, it will refill straight away, but we actually have time until the fifo is almost empty again without loosing fps
19:53:10kugela yield takes less than a tick, right? if so, and we can't get over 100fps (1 frame per tick), the yield should really be free
19:53:23kugelalthough it would be nice if someone understands and confirms my logic :P
19:55:20kugelFlynDice: a) I think the delay you added in the button driver is slightly wrong b) which delay makes you unable too boot (re your post to FS #10048)
19:55:41bertrikassuming 8 DBOP clocks per write cycle again, the FIFO would empty in 8 cycles * 128 entries / 62 MHz = just 16.5 us
19:56:28FlynDicekugel: If I don't touch the button driver but simply comment out the lcd delay it boots fine though...
19:56:54kugelthe fuze's lcd delay is only 4 nops, not 8. Have you tried that?
19:57:52kugeland for the button_dbop delay; the data sheet says that the DBOP_DIN read is only valid if it's read immediately after the read_valid status is set ("next cycle" it says, IIRC)
19:57:54mcuelenaerekugel: hmm I can't really see where interrupts come into the picture here.., could you post the interrupt handling function too?
19:58:01FlynDiceI've just been using delay−− What's the difference with using the nops, I haven't made it to that chapter yet... ;)
19:58:05kugelthe fuze has the delay above the /* start read */ line
19:59:14kugelFlynDice: in theory it's the same with the exception of 1 extra nop instruction. But I haven't found a actual difference
19:59:55kugeli.e. when I tried 16−− it wasn't working with or without nop, when I tried 24−− it was working for both
20:00:20mcuelenaereah irq sets full_empty?
20:00:24kugelmcuelenaere: the ISR just does full_empty = 1;
20:00:41kugel(and resets the interrupts, of course)
20:01:01FlynDiceI tried a delay value of 3 that did not work, 4 worked intermittantly 5 worked every time for me so I padded it for now(for the lcd)
20:01:39 Quit bmbl ("Woah!")
20:01:50kugelalso, I think you can stop your 31MHz experiments. From what I've seen it will be too slow in terms of UI resonsiveness even with caches enabled
20:02:37 Quit nibbler ("Ex-Chat")
20:03:09 Join Thundercloud [0] (
20:03:23FlynDiceYes I beleive that is correct but I was wondering about possibilities of taking it down there when the backlight was not on
20:03:27 Quit stoffel ("leaving")
20:03:30 Join pyro_maniac [0] (
20:04:21tmztcan lcd be fully disabled or will it take to long to reinit?
20:04:33tmztor could updates just be stopped with the clocks on
20:04:38AlexPFlynDice: Keeo in mind people that use their player without backlight
20:04:58AlexPFlynDice: Blind people that use voicing for instance might as well disable the backlight to save power
20:05:22mtIs there any debugging howto for the sansa similar to this :
20:06:16FlynDiceAlexP: That's fine, this is all just experimenting at this point, Real life calls, back later
20:07:04AlexPFlynDice: Sure, I was just pointing out that backlight off doesn't always mean the UI performance isn't important :)
20:07:42kugelFlynDice: it slows down the other peripherals too though. We actually lose decoding speed then which results in more boosts
20:07:49bertrikI guess the current TIMPOL timing values in SVN were designed for 62 MHz DBOP clock and can then indeed be expected to be rather slow at 31 MHz DBOP
20:08:40kugelthey weren't designed for 62MHz (unless Sandisk did that). The values are purely based on the disassembly
20:10:59bertrikwe should know what DBOP clock Sandisk designed them for, or else it's all just pure guesswork and tinkering
20:13:02kugelI don't :)
20:13:41kugelAnd I'd lie if I said I understand the TIMEPOL registers
20:13:48 Quit FOAD (Read error: 104 (Connection reset by peer))
20:14:57 Join FOAD [0] (
20:15:14 Join funman [0] (n=fun@rockbox/developer/funman)
20:20:49funmanAll the AMS OF use 32MHz for DBOP
20:22:58bertrikkugel, it looks like a DBOP read/write cycle is either 16 or 32 clocks long. The timpol registers indicate at what point within this read/write cycle the control signals change state
20:23:24bertrikfunman, hm, so perhaps we were overclocking it already?
20:23:35mcuelenaeremt: I think there's an ARM JTAG patch in the tracker (and perhaps an ARM GDB stub in SVN), but that's it AFAIK
20:26:56funmanbertrik: which problem could it cause : LCD not functioning properly if the times in DBOP_TIMPOL_* (i suppose they are in cycles) are too short?
20:27:17 Join perfectdrug [0] (
20:27:24gevaertsBagder: new numbers at (total time from build start to end copy, slowest at the top)
20:28:41*JdGordon| smiles at his #4 spot
20:28:47 Quit Thundercloud (Remote closed the connection)
20:29:26 Quit hd (Client Quit)
20:31:45bertrikfunman, no idea exactly, with a mild overclock I would expect to see artifacts, maybe hangs with a heavy overclock
20:32:38gevaerts has all data
20:33:37 Join HellDragon [0] (
20:33:41kugelbertrik: what do you mean by overclocking? the dbop is specified for upto 65MHz
20:35:26***Saving seen data "./dancer.seen"
20:35:38bertrikI looked at the DBOP control register setting and it's configured for a 16-clock write cycle with 16-bits per write, so this means indeed a max throughput of 7.75 Mb/s as guessed earlier
20:37:21bertrikkugel, I mean a write cycle at 62 MHz takes only half as long a write cycle at 31 MHz, which could be too fast for the display
20:38:38kugelwhere did you see it's configured for the 16-clock write cycle?
20:38:50kugeland, can we get faster?
20:39:04kugelwe can write 32bits IIUC
20:39:57kugelI tried setting sdc (1<<18) with..uhm funny results
20:39:58 Join mks [0] (n=a3b2a233@gateway/web/cgi-irc/
20:40:33mksI have a questions
20:40:56amiconnkugel: What LCD controller does the fuze use?
20:41:06kugelno idea
20:41:13Bagdergevaerts: that looks like a pretty good table
20:41:28mksit's posible install rockbox on a mp4 Titan
20:41:38amiconnI suspect you just hit the maximum transfer rate of it. Going higher would cause garbled display, and my guess is that the fifo is programmed so that this doesn't happen
20:42:00kugelthere's a comment about similar registers as for the HD66789R
20:42:17gevaertsmks: no
20:42:34mksah ok
20:42:37kugeland in fact, the contrast registers which the e200v1 and v2 use work on the fuze too
20:43:07kugelthough other registers aren't exactly the same
20:44:16gevaertsBagder: now the question is how to make use of it...
20:44:32bertrikkugel, yes I meant the sdc bit in the control register and as far as I can see it's already set during initialisation
20:44:54mksand just for to know, why it isn't posible?c
20:44:57Bagdergevaerts: primarily I think I want to keep the table in mind when thinking about an upcoming remake of the system
20:45:09AlexPmks: Nobody has done the work
20:45:28AlexPmks: Ports are down by interested owners - see
20:47:05JdGordon|Bagder: maybe grab rasher's distrubted build script and use that as a base?
20:47:16 Quit SirFunk_ (Read error: 110 (Connection timed out))
20:47:44BagderI think we have slightly different goals and ideas than his script(s) too
20:47:58 Join SirFunk_ [0] (
20:48:03Bagdermy plan is to get some discussion/brainstorming on this at devcon
20:48:10Bagderand then start on actually working on it
20:48:47kugelbertrik: I can't see that :(
20:49:10JdGordon|somehing to maybe think about... have a way so others can add/remove servers without havin too bug you?
20:49:13 Quit mks ("CGI:IRC (EOF)")
20:49:24gevaertsThere are basically two different goals in the remake. I think we want it to be more dynamic to avoid a slow host holding things up at the end, and we want to reduce and/or reverse the connections. I think those two are pretty separate
20:49:37BagderI should start a wiki page for ideas/notes about a remake
20:49:46kugelbertrik: the 3rd nibble is 0, no bit set
20:50:12Bagdergevaerts: the prevention of slow hosts holding up will be fixed by doing (speculative) parallell builds
20:50:13bertrikbit 18 is not in the 3rd nibble :)
20:50:32Bagdergevaerts: we just make sure all build clients are building until all builds are done
20:50:54gevaertsBagder: that's one way, yes. I'm not convinced that there aren't more (additional) things we can do though
20:51:09BagderI'm sure there's more!
20:51:21BagderI'll start a wiki page
20:51:52gevaertse.g. if at the end you have a sim build and a bootloader build to allocate, and you have one slow host and one fast host, you want the fast host to take the sim
20:52:21kugelbertrik: uhm, yea...sorry
20:52:34gevaertsOf course, if you add more remaining builds and more servers, it gets less easy
20:53:44gevaertsspeculative builds kick in as soon as you have more servers than remaining builds
20:55:53 Quit BryanJacobs (" ajax IRC Client")
20:57:00n1swouldn't saving the bootloaders for last help a bit too?
20:57:01 Quit bagawk (Read error: 104 (Connection reset by peer))
20:57:17Bagderit already does that
20:57:35Bagderthe ordering of the handing out jobs isn't a big factor
20:58:32gevaertsthe problem is you can get all bootloaders built on two or three fast servers in much less time than one normal build on a slow server
20:58:51n1swe need more bootloaders then!
20:59:16JdGordon|it would also be good to try keeping the same builds on the same servers for ccache
20:59:36Bagderyeah, but that's really shaky
21:00:17BagderI mean, clients come and go, the load comes and goes etc
21:00:33n1sif the connections are reversed and the clients request builds they could have a list of what they built most recently and request that first
21:00:38BagderI'll add that as an idea though
21:00:49JdGordon|ranking builds and servers so chunky builds are given only to faster servers (unless everyone else is busy) could be done...
21:00:50Bagdern1s: good thinking
21:02:19JdGordon|and with a bit of magic, the rankings could even be updated at the end of each round
21:02:36 Join bagawk [0] (
21:02:47gevaertsSlow network vs slow CPU makes things a bit more complicated of course
21:03:26kugelwhat's speaks against doing the uploads in the background?
21:03:26JdGordon|yes, but not all builds are uploaded...
21:03:43kugelJdGordon|: even more complicated
21:04:38 Join Thundercloud [0] (
21:04:48gevaertskugel: nothing much, except of course complexity :)
21:05:06gevaertsbut yes, that can be done as well
21:05:21Bagdererror handling gets trickier
21:05:30Bagderbut still, it's good use of threads
21:05:57gevaertsBagder: I'd expect paralel builds to magically handle part of that problem
21:06:19JdGordon|we cant have it hide server problems though...
21:06:26*gevaerts would do this as a makefile :)
21:07:08*JdGordon| spots too many l's :)
21:08:02*Bagder removed some read underlines
21:08:49*Bagder denies that stout having an impact on his typing
21:09:09gevaerts adds sim and bootloader times
21:09:14*JdGordon| doesnt think E and C will conflict much
21:09:44kugelCould we, say, calculate avg build time*avarage upload time, and use this for ranking (highest ranking will get the "heavy builds", most featured targets with big screen resolution)
21:10:21gevaertskilling slow builds if a fast client has overtaken them may not be needed. They can just get on with it and be ignored
21:10:22Bagderactually, we should count upload speeds separately and use that for the zip files we host
21:10:39Bagdergevaerts: not if another build is triggered
21:11:09gevaertsBagder: if it's near the end, that likely won't happen, and if it does we want a fast host to get it
21:11:24gevaertsah no, I misread
21:12:00BagderI meant another buildround (or whatever name we should have for all the 100+ builds in bundle)
21:12:35kugel"tick" :)
21:13:04kugeltock when the builds are done, respectively :p
21:14:15bertrikkugel, I guess this means that 31 MHz DBOP clock results in a max fps of 50
21:14:56kugelbertrik: have you in the meantime tried my patch on yor clip?
21:15:03*kugel could also do it himself :/
21:15:03amiconnSpeculative concurrent builds are supposed to solve the slow server problem automatically
21:15:35bertrikkugel, sorry not yet
21:15:43*kugel should probably get at the unified driver patch again
21:18:27gevaertsStatistics on which hosts usually end last at It counts the "1 builds left" cases, so this is not exactly the number of times they end up last
21:18:45gevaertsMaybe I'd better stop that one server...
21:19:00JdGordon|are you going by previous logs? or actually running new builds?
21:19:13gevaertscurrently available logs
21:19:22gevaertsi.e. the last 20 builds
21:19:37JdGordon|in that case.. why isnt mine getting sim buidls?
21:20:07kugelno sdl?
21:20:18kugelI think there was a problem with yours some weeks ago
21:20:19Bagderit has sdl tagged for it
21:20:35JdGordon|must be just good luck then
21:21:19gevaertsSome numbers may be slightly off. The logs are written by several threads without much locking, and now and then lines get garbled by that
21:21:59Bagderand a host ending up last is not necessarily a bad thing
21:22:04JdGordon|sdl works fine, so yeah, just luck
21:22:43gevaertsBagder: of course not, but if it always does, and it's near the bottom of all other rankings, I think it's not unlikely to hold things back
21:23:16funmando you need/welcome new build servers?
21:23:16Bagderyeah, it's one piece of the jigsaw
21:24:07Bagderbut we need them to be pretty fast to be worth it, no old gits
21:25:07funmanfast as in CPU or bandwidth?
21:25:21BagderCPU primarily
21:25:36Bagderthere's only like 20 big uploads out of the 106 builds
21:25:55Bagderor maybe 30
21:26:09n1swould the reversed connection system be more suited for not-always-on hosts?
21:26:19JdGordon|oh, something to remember.. we wanted to add the tools/ stuff in the compile run
21:26:25Bagdern1s: I think so, yes
21:26:41 Quit Thundercloud (Read error: 113 (No route to host))
21:26:54BagderJdGordon: I've hesitated with that since it'll slow them down
21:27:40JdGordon|? not do tools in every build.. as a seperate target
21:27:52 Join Thundercloud [0] (
21:28:21Bagderright, that's a good idea
21:28:26Bagderand cross-compiled sims is another
21:28:41JdGordon|and rbutil...
21:28:53gevaertsand the manual
21:29:00Bagderrbutil and manuals are slightly different though
21:29:01JdGordon|windows sims with the uploaded zip would be good (i tinhk)
21:29:07Bagderwe can't easily mix them in
21:29:22tmztcan mingw32 build it on linux?
21:29:39JdGordon|sure we can... if make can be used to play sudoku we can do this!
21:29:56Bagder"if you can land a man on the moon"...
21:30:14gevaerts*we* can't :)
21:30:24Bagderhow hard can it be? B)
21:30:32 Quit flydutch ("/* empty */")
21:30:39Bagder"The Rockbox Man On The Moon Project"
21:30:45Bagderstarting right after devcon
21:30:50JdGordon|well... if the individual builds are just "run some command on a system with X dependancies" why cant we do it?
21:31:02funman1st step: "The Rockbox Man Project" ?
21:31:18BagderJdGordon: well, first the plain simple reasons like the builds are triggered on non-rbutil and non-manual changes
21:31:41BagderI mean, we'd do a whole lot of builds in vain
21:32:20JdGordon|sure, but it its already smart enough to not run a build if its a manual commit, why cant we block rbutil builds if its not touched?
21:32:21 Join merbanan [0] (
21:32:23kugelI think we can leave manual and rbutil on a daily basis, but show them on the build tabe
21:32:51Bagderkugel: I think I'm in that boat too
21:33:37kugelor trigger rbutil on rbutil/ changes and manual builds on manual/ changes :)
21:33:55 Quit wark ("Leaving... for now. But I'll be back.")
21:34:02JdGordon|thats what I was trying to say :p
21:38:18 Quit perfectdrug (Read error: 104 (Connection reset by peer))
21:40:28saratogaanyone mind if I add "few ssh connections per rebuild" to the build server improvements?
21:43:57n1ssaratoga: no, but i'm curious why you want that?
21:44:52saratogan1s: IT keeps blocking the build server
21:45:23saratogawe open and close a lot of connections which i guess makes their firewall angry
21:47:00 Join gartral [0] (
21:47:44gartralhi all, whats with r21182?
21:49:15bertrikgartral, anything wrong with it in your opinion?
21:49:33JdGordon|that rev easts your cats and burns your house down
21:49:59LloreanI guess it's the missing commit message?
21:50:44funmankugel: about 21177, I think TIMER_FREQ represents the highest resolution possible (number of timer cycles per second), and is used by plugins (greylib, doom, metronome, scanrate..)
21:51:13kugelso the change was correct?
21:51:13 Join PaulJam [0] (
21:51:18funmannope :/
21:51:49funmanTIMER_FREQ should stay the same, because the other timer has the same resolution and is used by plugins and such with user configurable delays
21:52:04 Join Juice^ [0] (
21:52:10kugelbut the TIMER is actually lower
21:52:18kugelTIMER_FREQ was a useless value before
21:52:21funmani.e. we can have TIMER_FREQ = 24000000/16 timer interrupts per second
21:52:57 Quit pyro_maniac ("Leaving.")
21:52:58kugeland on scrollwheel we have more
21:53:02kugel /32
21:53:12funmanTIMER_FREQ is used by Sansa AMS' Timer2 (to generate kernel tick) and Timer 1 (user configurable timer)
21:53:51funmanthe presence of scrollwheel only affect Timer2 resolution, not Timer1
21:55:08funmanThe whole timer api could benefit from some work also
21:55:30funmanfirmware/timer.c could be split in the target tree
21:56:02funmanand tick_start could be tick_start(void), and the implementations use HZ directly
21:56:32kugeloh, it's used for both
21:56:37kugelI wasn't aware of that, sorry
21:56:47bertrikfunman, totally agree about timer.c
21:56:54funmanis that something MrSomeone could do ?
21:57:16funmankugel: eh i just figured it out
21:57:58funmankugel: in fact I would like to make the number of timer cycles for a tick static const, so it could be reused by usleep() if it becomes used
21:58:24 Quit LambdaCalculus37 (" ajax IRC Client")
22:00:12funmanhow will my reputation be affected if I work on some code not specific to Sansa AMS? :/
22:01:27gartralbertrik: no, but Ide like to know what it doesm the entry is empty
22:01:28bertrikyou will be greatly respected and rewarded with special beers should you decide to come to devcon
22:01:34domonokyfunman: +42 :-)
22:02:29Lloreangartral: Looking at the code, it's just the addition of a couple delays in two places.
22:06:09 Quit Juice^ ("- nbs-irc 2.0 - -")
22:06:19 Quit JdGordon| (" ajax IRC Client")
22:07:52 Join froggyman [0] (n=47ba06b6@gateway/web/cgi-irc/
22:08:07r0b-ok sthe talking menus wont work on this sandisk e250
22:08:16 Quit PaulJam__ (Read error: 110 (Connection timed out))
22:08:23r0b-i wonder if i need a speech engine
22:08:43Lloreanr0b-: You need one on your computer to generate the voice files, yes.
22:08:48LloreanUnless you've downloaded premade ones
22:09:28r0b-well i got the premade ones
22:11:20LloreanSo, what exactly have you done?
22:11:28LloreanAnd where did you get the voice file from?
22:11:50r0b-i forgot who posted it
22:12:34GodEater"posted" it ?
22:12:39 Quit {phoenix} (Read error: 54 (Connection reset by peer))
22:12:43LloreanIf you didn't generate it yourself, and it's not one from our voice files page, we don't provide support for it.
22:12:44GodEateryou didn't download it from our website ?
22:13:07r0b-i just downloaded the one from your site
22:13:19LloreanWhich one?
22:13:37gartralLlorean: thanks
22:13:45Lloreanr0b-: That's not how ours are named.
22:13:55LloreanWhat site / link did you follow?
22:13:57r0b-could you point me to yours
22:14:18*Llorean wishes people wouldn't attach such things to the wiki
22:14:23Lloreanr0b-: Click the "Extras" link on the left, and follow that
22:14:42LloreanThose voice files at the page you linked are 1.5 years old.
22:14:54LloreanWell, 1 year old
22:15:27n1sr0b-: i gave you the link to the release voices earlier...
22:15:40r0b-i just downloaded the latest version of rockbox
22:15:44r0b-we will see what happens
22:17:01r0b-still no voice
22:17:09LloreanNobody told you to download a new version of Rockbox...
22:17:19LloreanIf you have no interest in actually following instructions, why ask for help?
22:17:39r0b-i also downloaded the release voices
22:17:58LloreanWhat is the filename, and where *exactly* did you put it on your player?
22:21:59r0b-i extracted it to the wrong directory
22:22:06r0b-i just moved the file to the proper area now it works
22:33:44 Quit merbanan (Read error: 110 (Connection timed out))
22:35:27***Saving seen data "./dancer.seen"
22:37:14 Join kachna [0] (
22:44:29 Join Russel-Athletic [0] (
22:46:20GodEaterit's off to work we go
22:46:26Russel-Athleticit seems the developers have made some major changes in apps/tagcache.c which gives me some headaches because I want to have this patch in my rockbox version:
22:47:04Russel-Athleticit would be nice if somebody could help me port this
22:48:06LloreanThat patch was rejected some months ago.
22:48:22Russel-Athletici know
22:48:24GodEatertherefore it's not going to require a "port"
22:48:30GodEaterit's going to require a complete rewrite
22:48:36GodEaterwhich is a substantial amount of work
22:48:50Russel-Athleticis no quick fix for this possible?
22:49:10LloreanPlease don't ask in this channel for help getting rejected patches working. They were rejected because they were decided no acceptable for the official version of Rockbox, which is what this channel is in place for.
22:49:22LloreanWhile it may be possible to get them working for personal use, it's not something that's on-topic here.
22:49:32Russel-AthleticI know that this patch is rejected but I only want to have it in my version
22:49:44LloreanAs I just said, it's not on-topic here.
22:49:59Russel-Athleticoh ok
22:55:06GodEaterit actually looks like you sync'd it yourself a few times
22:55:21GodEaterso you seem up to the job of doing it yourself surely?
22:56:20Russel-Athleticthe syncing was only minor changes
22:56:32Russel-Athleticlike some tag was added
22:56:39Russel-Athleticbut this bit stuff is a bit too much for me
22:57:10Russel-Athleticand i know it is your right to not support this patch but i just wanted to try asking
22:57:42GodEaterhas anyone ever looked at replacing tagcache with sqllite or something ?
22:57:52 Join Tomers_ [0] (
22:58:01LloreanGodEater: There's an sqlite patch in the tracker, I believe
22:58:15evilnick_230Llorean: Do changes to the database affect users who don't use the database (i.e. not even Initialize it)?
22:58:17LloreanThough it hasn't been adapted to the database, I think it's just a plugin of some sort
22:58:21GodEaterthere you go Russel-Athletic get cracking on that :D
22:58:31Lloreanevilnick_230: Most shouldn't, no.
22:59:06Lloreanevilnick_230: The problem with that patch is that it'd affect all users of the database for a feature that could be implemented in a lot lower impact manner, if I recall
22:59:20Russel-Athletici know the argument
23:00:33evilnick_230Llorean: Sure. That's what I'd expect, but it's always best to make sure rather than making assumptions.
23:00:48gevaertsGodEater sqlite would increase memory usage by a few hundred kilobytes
23:01:11Lloreangevaerts: Is that "ever" or "on top of any increases in database size"?
23:01:22GodEatergevaerts: hence my "or something" too :D
23:01:48*Llorean would argue in favor of sacrificing those few hundred KB if it meant no more bugs relating to duplicate entries in the database, etc.
23:04:53bertrikfunman, in the clip OF 1.1.30 I see that DCDC15 is written with 1 without any condition when turning on the OLED
23:05:09gevaertsLlorean: I don't remember the exact numbers. Also, I somehow doubt that most of the bugs we have are due to the actual storage engine. They sound more like update logic gone wrong
23:05:11funmanmcuelenaere: cf timer-jz4740.c : __timer_set(), the cycles argument means the number of timer cycles between each interrupt (calculated from TIMER_FREQ, number of cycles per second)
23:05:23 Quit Tomers_ ("ChatZilla 0.9.84 [Firefox 3.0.10/2009042316]")
23:05:26funmanbertrik: I check 1.1.29 and 1.1.32 they are the same
23:05:47funmanI didn't test what was the effect on my Clip however
23:05:50mcuelenaerefunman: thanks
23:06:03funmanmcuelenaere: thanks for not adding your code to timer.c mess ;)
23:07:11bertrikfunman, so if we do it the same way (as in matsch' patch) we're at least not doing worse than the OF :/
23:08:04funmanI don't know how much extra power that represents (1.25mA)
23:08:24bertrikI was hoping for some way to autodetect it
23:08:44 Join PsyCl0ne [0] (n=robbie@
23:09:11PsyCl0neHi would anyone be able to help me get rockbox installed on my iPod
23:09:21bertrikMy guess is that there's nothing to draw any current for the clips that don't need this, so there's hardly any power wasted
23:09:22LloreanPsyCl0ne: Where are you getting stuck in the instructions?
23:09:45funmanbertrik: would the difference be noticeable with battery_bench if it wasn't the case?
23:10:03 Join matsl [0] (
23:10:04PsyCl0neLlorean: Actually all that I have done is check to see if rockbox is compatible with my ipod, and got the visual installer
23:10:13mcuelenaerefunman: so __timer_set(long cycles, bool start)
23:10:23LloreanPsyCl0ne: Why not try following the instructions then before asking for help? ;)
23:10:30PsyCl0nellorean: but for some reason the app is telling me that my ipod is noncompatible
23:10:41LloreanThen you mis-identified your iPod
23:10:41evilnick_230PsyCl0ne: What generation iPod is it?
23:10:43bertrikfunman, hard to say, I'll have to take a guess at the normal current consumption
23:10:59mcuelenaerefunman: so __timer_set(TIMER_FREQ/2, true) means setting a timer for every half a second?
23:11:10saratoga1ma comapred to teh entire screen consumption is probably really small
23:11:26PsyCl0neLlorean: its an iPod classic 80gb serial does end in YMV
23:11:52saratogaPsyCl0ne: check the apple docs and confirm what you have
23:11:59saratogayou're probably wrong about which ipod you own
23:12:01LloreanPsyCl0ne: The "classic" isn't supported
23:12:06bertrikfunman, the clip battery is 350 mAh, how long does it last now with SVN?
23:12:19LloreanPsyCl0ne: It says "(not the Shuffle, 2nd/3rd/4th gen Nano, Classic or Touch) " on the front page of the site.
23:12:39evilnick_230iPod Classics are not supported, but to make things more confusing, some people refer to all full-size iPods as Classic, when they are not
23:12:47funmanmcuelenaere: right, the 2nd argument set to true means that it will unregister previous user of the timer (afaiu)
23:12:53 Quit Tomers (Read error: 113 (No route to host))
23:13:30mcuelenaereand if set to false it just changes the period of the timer?
23:13:48PsyCl0nekk, my bad I went with the guide at, but going by my model number its a 6th gen
23:14:17LloreanAnd you'll notice the 6th gen isn't supported.
23:14:26LloreanWe go up to the 5.5th gen.
23:14:30funmanbertrik: i haven't run battery_bench since;topic=14064.0;attach=3692;image so I suppose around 6 or 7 hours
23:14:30GodEaterthe guide at that page is perfectly correct
23:14:32PsyCl0neyea, I thought it was a 5th gen....
23:14:35PsyCl0nemy bad
23:14:38PsyCl0nethanks though
23:14:42 Part PsyCl0ne
23:14:56funmanmcuelenaere: yes
23:16:11bertrikfunman, so that's about 50 mA on average. Then I wouldn't worry about 1.25 mA and I doubt this 1.25 mA even runs for clips that don't need it. Also we can turn it off when the display is off.
23:16:20r0b-where can i find movie trailers
23:16:33saratoganot here
23:16:34r0b-i want some clips to put on my sansa
23:16:34n1sgoogle maybe
23:16:47funmanbertrik: ok tha,ks for the explanation
23:16:49GodEaterwe definitely do not provide media
23:17:01GodEaterapart from the ONE we have on the pluginmpegplayer page in the wiki
23:17:02n1sr0b-: this channel is about rockbox anything else isd off topic
23:21:48 Quit evilnick_230 (" ajax IRC Client")
23:22:56 Quit gartral ("leaving")
23:23:28 Quit bubsy ("Mrrrrreow!")
23:25:39funmanin fact timer.c / timer-target.h is alreayd made for target tree, but only some targets use the target tree (tcc, gigabeat-fx, lyre proto1, dm320 and ingenic jz47xx)
23:25:52funmani'll just move targets one by one
23:26:06 Quit VytenisS (Remote closed the connection)
23:27:58 Join onlysoaa [0] (i=40e56ec7@gateway/web/ajax/
23:30:35onlysoaaHello all! Anyone know if I can get a hold of shotofadds?
23:30:56onlysoaaHe's the one who did most of the D2 port, so perhaps he could help me out with my P2 port. ;)
23:31:39 Quit bertrik (Remote closed the connection)
23:32:12saratogahe posted on the forums about 10 minutes ago, so maybe you could just PM him and ask to meet him in IRC
23:32:16n1sonlysoaa: quite a few devs read the logs of this channel so if you have any specific questions maybe you should just ask here, also maybe someone else can answer
23:32:40 Join bertrik [0] (
23:37:53 Nick fxb is now known as fxb__ (
23:38:10onlysoaaAlrighty, thanks. I noticed he posted on the forums not long ago. :P
23:39:15 Join safetydan [0] (n=deverton@rockbox/developer/safetydan)
23:52:13onlysoaaI just updated the Samsung P2 port thread. (:
23:52:13 Join JdGordon| [0] (i=836b0070@rockbox/developer/JdGordon)
23:55:47 Quit Thundercloud (Remote closed the connection)
23:56:23kugelonlysoaa: blueish display?
23:57:16kugelonlysoaa: I too had that when I first saw someting on my fuze's display. linuxstb was able to fix it with a single line IIRC, maybe you ask him
23:57:49onlysoaakugel: it's not only blueish; it's also stretched horizontally three-fold.
23:58:13onlysoaakugel: It's like it's displaying data from RGB in the B channel only.
23:58:24onlysoaaTimings should be set correctly.
23:58:31kugelonlysoaa: as I said, yo need to replace the resoltion at some places
23:58:42kugelthe d2 lcd driver is far from multi target friendly

Previous day | Next day