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 2009-11-21

00:00:06LambdaCalculus37toffe82: For the SA9200, plugin keymaps.
00:00:13LambdaCalculus37For the HDD6330, the touchpad driver.
00:00:20LambdaCalculus37For both, an easy install method.
00:00:29toffe82and the 1830 ?
00:03:04 Quit n1s ("Lämnar")
00:03:28LambdaCalculus37toffe82: Install method as well IIRC.
00:03:34 Join martian67 [0] (
00:03:40LambdaCalculus37The HDD1x30 works well for the most part.
00:04:35toffe82it would be nis to have them going in the unstable port
00:04:47toffe82nice ... :)
00:07:11LambdaCalculus37toffe82: Yes, I agree.
00:07:47funmanregarding Sansa AMS, is there anything preventing Fuze & e200v2 at least (and Clip/m200v4/c200v2 when fs#10605 gets committed) to be part of the next release?
00:07:49LambdaCalculus37Would anyone object to bumping the GoGear HDD1x30 series up from "Unusable" to "Unstable"?
00:08:02LambdaCalculus37funman: No USB stack.
00:08:51funmanthat didn't prevent ipod & e200v1/c200v1 to be supported?
00:09:08LambdaCalculus37That's true.
00:09:43LambdaCalculus37toffe82: There's no install method in rbutil yet for the HDD1x30 models, either.
00:10:26 Quit p3tur ("Zzzzz")
00:10:44funmanand while USB support in rockbox would be nice, that doesn't prevent daily use (i use my fuze daily)
00:10:44gevaertspamaury: what exactly should I test on the gigabeat S OF?
00:13:44toffe82LambdaCalculus37: the install procedure is on the wiki for all the gogear ? all uptodate ?
00:14:11gevaertsfunman: "daily use" is what gets you into unstable :)
00:14:55LambdaCalculus37toffe82: For the SA9200, that's the current install procedure.
00:15:08LambdaCalculus37For the HDD1x30 and the HDD6330, we need to add it.
00:15:18funmanno, ports in unstable "run, but are incomplete, less usable or have problems that limit them to advanced users"
00:15:39gevaertsoh certainly, but they are usable for daily use
00:15:54funmani believe rockbox "runs well", "is supported by the installer", and the manual should be complete though i didn't check in details
00:15:58gevaertsI don't object to having them in the release though
00:16:55LambdaCalculus37toffe82: Did you ever send me an extracted firmware update fileset for the MEG50JS or the MEG201?
00:17:07 Part ercani
00:17:47gevaertsfunman: I'm not sure about the clip though. The release should be in about a month, is that enough time to be confident that it works well?
00:18:13funmanno idea
00:18:34 Quit petur (Remote closed the connection)
00:19:17gevaertswe'll see
00:21:57toffe82LambdaCalculus37: I don't remember
00:34:01*LambdaCalculus37 extracts the firmware updates
00:34:50saratogafunman: I'm going to make the clip unstable tomorrow
00:35:05saratogaIMO the Fuze and e200v2 can become stable too
00:35:12saratogashould I change all of them tomorrow?
00:40:24funmanshoudln't we wait for the release ?
00:41:30LambdaCalculus37toffe82: It looks like I may be able to get some very useful information out of the Gigabeat MEG201 and MEG50JS in terms of flash ROM if I can just cat /dev/from into an external file and read it.
00:45:06 Quit bertrik ("De groeten")
00:45:12*toffe82 lets LambdaCalculus37 do this....
00:46:36kugelsaratoga: the clip has still a mostly unusable keymap IMO
00:48:23JdGordonno it doesnt...
00:49:58kugelit does
00:50:09 Join MethoS- [0] (n=clemens@
00:50:15kugelI think you can't even exit some plugins
00:57:12 Quit funman ("free(random());")
00:57:42kugelI think the clip shouldn't be stable already, it wasn't even unstable yet
00:58:07kugelnobody can really tell if it's sufficient for stable because it was basically impossible to use it on a every day basis
00:59:12LambdaCalculus37HA... minix filesystem is used on the MEG201. :)
01:00:25JdGordonits not useable becasue its only 2GB :)
01:02:30 Join CaptainKwel [0] (
01:05:56 Quit dmb (Read error: 113 (No route to host))
01:18:13***Saving seen data "./dancer.seen"
01:22:52 Quit sbhsu (Read error: 60 (Operation timed out))
01:24:47Unhelpfulhrm, what's the "right" way with strict aliasing to address alias an array of char w/ an array of some other type? :/
01:25:11Torneisn't the point that you don't do that?
01:27:37Tornealso char* is allowed to alias points of any size, no?
01:28:01UnhelpfulTorne: the point is that you make sure the compiler *knows* when you do such things... usually via unions and the like. or that's what i though? the example here is that the bmp reader reads raw file data into a buffer, and then decodes it a pixel at a time into the same buffer
01:28:27Torneif you use a union that's not aliasing :)
01:28:32Tornebut, yes, i see wht you mean
01:28:40Tornechar* is allowed to alias other pointer types, though
01:28:49Torneso it shouldn't be a problem?
01:31:45Unhelpfuland yet it seems to be... the read buffer is cast to uint8_rgb* and then bmp.c:825 throws an aliasing warning
01:32:35Torneuint8_rgb is presumably a typedef for unsigned char though.. i'm not sure if that counts
01:32:51Tornei'm not sure tbh :)
01:33:13Unhelpfulum, no, it's a struct. the issue might be that the dereferenced value is "passed" to an inline function. i'm not sure.
01:35:51 Quit toffe82 ("ChatZilla 0.9.85 [Firefox 3.5.5/20091102152451]")
01:36:00 Quit parafin (Read error: 60 (Operation timed out))
01:36:04 Join Rob2223 [0] (
01:37:56 Join parafin [0] (
01:38:24Unhelpfulthe color case passes the buffer to output_row_8_native, which seems to have no problem casting the received char * to struct uint8_rgb * and addressing it. i'm guessing either 1) a distinction is being made between array and char 2) the inline function is an issue
01:39:45Tornewell try uninlining it?
01:39:49Tornejust to see?
01:41:53Unhelpfulwell, i was thinking not so much "it's inline" as "struct-as-argument"
01:42:10 Join rodion09 [0] (
01:42:29Unhelpfulwhich is probably not nice to do even for a <int-sized struct
01:43:09 Quit Rob2222 (Read error: 60 (Operation timed out))
01:45:34 Quit liar (Read error: 113 (No route to host))
01:45:52Unhelpfulvery strange. the compiler does *not* complain about using the aliased pointer to access the struct members, so that they can be packed to native format. it *does* complain about passing the struct, *or* the aliased pointer to it, as an argument to brightness(). perhaps rewriting it as a macro?
01:47:16Unhelpfulif i pass it as a pointer, the alias warnings trigger on bmp.h for the access within brightness
01:52:04*Unhelpful will try putting an anon union into bmp_args of a char array or a struct uint8_rgb array... surely if they're in a union then gcc "knows" they alias each other?
01:55:57 Join GeekShadow [0] (n=Antoine@reactos/tester/GeekShadow)
01:57:32Unhelpfulthe unionized buffer works even if i don't ever *use* the uint8_rgb[] member. a bit weird.
01:58:10Torneyou know -Wstrict-aliasing isn't perfect, right? :)
01:58:26Torneit cannot detect all possible cases of unacceptable aliasing
01:58:30Unhelpfulalthough i *could* replace all of the (uint8_rgb*)ba.buf now with ba.pbuf. i'm not sure if that would be more or less readable than it is now.
01:58:34Torneotherwise there would be no need to have a warning
01:58:47Tornegcc could just not optimise around things it knows are aliased badly :)
01:59:09Tornesadly the gcc authors expect you to actually be sure whether what you are doign is okay :)
01:59:13 Quit rodion09 ("CGI:IRC (EOF)")
01:59:18Tornethe warnings are just a partial crutch
01:59:40UnhelpfulTorne: of course it's not perfect. but n1s saw some things improve in speed on 4.4 with -fstrict-aliasing, and so i thought i might have a go at cleaning up the warnings that causes
01:59:58TorneIndeed, the strict aliasing optimisations do help
02:00:13Tornei'm just saying that you probably should be sure ;)
02:01:14Unhelpfulit's in-place conversion of a buffer from one format to another... is there a more correct way to do it?
02:01:31Unhelpfulalso, i think the union is pretty much guaranteed to be safe, isn't it?
02:02:01kugelis that the only aliasing problem in the core?
02:02:40Unhelpfulit's not the only one gcc warns for :)
02:03:40Unhelpfulsince i have a pbuf member now that is explicitly "buf, but as struct uint8_rgb[]", should i be using that? the structure definition makes it quite obvious what types buf and pbuf are, and that they alias each other.
02:12:56 Quit GeekShado_ (Read error: 110 (Connection timed out))
02:22:32*Unhelpful has no idea how to fix the ones in kernel.c :/
02:22:34 Quit HBK (Read error: 54 (Connection reset by peer))
02:23:43 Quit kugel ("exit(0);")
02:25:11 Join phanboy4 [0] (
02:26:07 Join HBK [0] (
02:27:18 Quit Omlet ("( :: NoNameScript 4.22 :: )")
02:34:57 Join kugel [0] (n=kugel@rockbox/developer/kugel)
02:40:40 Join bughunter21 [0] (
02:40:40 Quit bughunter2 (Nick collision from services.)
02:40:51 Nick bughunter21 is now known as bughunter2 (
02:41:35 Join darkhamm [0] (
02:42:34 Join pamaury [0] (
02:43:07 Join toffe82 [0] (
02:46:14 Quit hebz0rl ()
02:48:33darkhammwill rockbox play more lossless audio codecs'
02:48:37darkhammlike tta?
02:50:09Torneif someone implements it
02:51:50Unhelpfulit already supports wavpack, monkey's audio (but only on the fastest hardware), flac, and apple lossless. do you *really* need tta?
02:52:34 Quit pamaury ("exit(*(int *)0 / 0);")
02:53:54 Join webguest30 [0] (n=62524b71@
02:53:57 Join pamaury [0] (
02:54:02 Quit webguest30 (Client Quit)
02:54:19 Quit pamaury (Client Quit)
02:55:16darkhammUnhelpful, not, i'm only interesting to lossless support
02:55:35darkhammi only look that tta have very little encoding times
02:56:01Unhelpfulyou could work on supporting it. for lossless generally decoding time and compression ratio are more important than encoding times.
02:57:49LambdaCalculus37Unhelpful: There's always some nut out there who has a weird, obscure format. ;)
02:57:57darkhammUnhelpful, yes i became curious browsing
02:59:09darkhammand at highest level , flac (my favorite) have the longest time
02:59:19UnhelpfulLambdaCalculus37: tbh the only one we *don't* support that i'd like to see is tak, and that only if the author ever gets around to releasing source. winning on encode time *and* compression ratio vs many common codecs, with decode times competitive with flac? sounds good.
03:00:43Unhelpfuldarkhamm: encode time is not really where you should be looking, unless you're comparing codecs that are on fairly even footing with regard to decode time and compression ratio.
03:01:45Unhelpfulalso, high-compression encode in flac has gotten *much* faster since 1.1.0.
03:03:08 Join fyre^OS [0] (
03:03:37 Join HellDragon_ [0] (
03:04:13 Join toffe82_ [0] (n=chatzill@
03:04:56Unhelpfuli believe flac-1.1.4 incorporated the work done on flake for estimating/solving for parameters that used to be brute-force-searched in high modes
03:05:45 Join pixelma_ [0] (i=quassel@rockbox/staff/pixelma)
03:05:45 Quit pixelma (Nick collision from services.)
03:05:51 Quit toffe82 (
03:05:51 Quit hd (
03:05:51 Quit fyrestorm (
03:05:51 Quit AaronM (
03:05:51 Quit freqmod_qu (
03:05:51 Quit jasio_ (
03:05:51 Quit rasher (
03:05:51 Quit Utchybann (
03:05:56 Nick toffe82_ is now known as toffe82 (n=chatzill@
03:06:04NJoinfreqmod_qu [0] (
03:06:04 Nick pixelma_ is now known as pixelma (i=quassel@rockbox/staff/pixelma)
03:08:21 Join rasher [0] (
03:08:29 Quit toffe82 (Killed by (Nick collision))
03:08:31 Join jasio [0] (n=yann@
03:08:31NJointoffe82 [0] (
03:08:31 Join hd [0] (n=jd@Wikipedia/HellDragon)
03:08:31NJoinfyrestorm [0] (
03:08:31NJoinAaronM [0] (
03:08:31NJoinjasio_ [0] (
03:08:31NJoinUtchybann [0] (
03:09:03 Quit jasio_ (Read error: 104 (Connection reset by peer))
03:09:13 Quit hd (SendQ exceeded)
03:09:13 Join toffe82_ [0] (
03:12:27LambdaCalculus37Unhelpful: And there's OptimFROG, although that's proprietary.
03:13:23Unhelpfulso's tak. "i need to clean it up, fix some issues, and rewrite in c or c++ before i open the source", as though a community effort couldn't make all of that *easier*
03:16:36 Quit CaptainKwel (Read error: 54 (Connection reset by peer))
03:18:16***Saving seen data "./dancer.seen"
03:18:40 Join sbhsu [0] (
03:23:03 Quit toffe82 (Connection timed out)
03:23:30 Quit fyrestorm (Connection timed out)
03:25:31LambdaCalculus37Unhelpful: Hey, community effort helped us out a lot. :)
03:25:41 Quit Thundercloud (Remote closed the connection)
03:28:34 Join CaptainKewl [0] (
03:28:34 Quit CaptainKewl (Client Quit)
03:28:55 Join CaptainKewl [0] (i=jds@
03:32:44 Quit darkhamm ("Sto andando via")
03:35:21 Quit LambdaCalculus37 ("cd ~")
03:56:51 Quit Res1 (Read error: 60 (Operation timed out))
04:04:15 Part Llorean
04:06:32 Quit tchan ("WeeChat 0.3.1-dev")
04:08:09 Quit Rondom (Nick collision from services.)
04:08:19 Join Rondom [0] (
04:10:35 Join Res1 [0] (
04:10:41 Quit TheSeven (Nick collision from services.)
04:10:48 Join The_Seven [0] (n=theseven@rockbox/developer/TheSeven)
04:11:00 Nick The_Seven is now known as TheSeven (n=theseven@rockbox/developer/TheSeven)
04:19:27 Join faemir [0] (n=faemir@
04:22:38 Join Llorean [0] (n=DarkkOne@
04:31:26 Quit jasio (
04:31:26 Quit Utchybann (
04:31:26 Quit AaronM (
04:32:29 Join lovingHDTV [0] (
04:32:51lovingHDTVAnyone know if there is a hearing aid type plugin?
04:35:28NJoinjasio [0] (n=yann@
04:35:28NJoinAaronM [0] (
04:35:28NJoinUtchybann [0] (
04:35:35 Join jasio_ [0] (n=yann@
04:35:56 Quit GeekShadow ("The cake is a lie !")
04:40:59 Quit kugel (Remote closed the connection)
04:43:29 Join KBH [0] (
04:45:48 Quit HBK (Read error: 60 (Operation timed out))
04:45:48 Nick KBH is now known as HBK (
04:46:33 Quit jasio (Read error: 110 (Connection timed out))
04:54:12 Join BiptoN [0] (n=4cad70c0@
05:00:54 Quit BiptoN ("CGI:IRC (EOF)")
05:01:06 Join bipton [0] (
05:01:52biptonanyone here use the svn ipodpatcher today?
05:05:37 Join tmzt_ [0] (
05:06:41 Join tchan [0] (n=tchan@lunar-linux/developer/tchan)
05:10:41 Join tmzt__ [0] (
05:16:00 Quit tmzt (Read error: 110 (Connection timed out))
05:18:19***Saving seen data "./dancer.seen"
05:22:47 Quit tmzt_ (Read error: 110 (Connection timed out))
05:38:23biptonno one...?
05:45:56*Unhelpful doesn't even own an ipod :)
05:48:43Unhelpfuln1s: i've got my beast results from running different -O levels. the tuned-level times for beast are probably the best for any target i've looked at so far. ape is ~4% slower at best, vorbis as much as 1.7% slower, but in the .1-.5% range for most bitrates, and everything else is faster after tuning
05:49:07 Join webguest43 [0] (n=dee1773c@
05:50:00Unhelpfulmpc by ~5.5%, flac by ~16%, wma by ~8%, and aac-he by ~12% are the most impressive numbers. clearly 4.4.2 knows how to work armv6 better than 4.0.3 :)
05:52:04Unhelpfulvorbis and ape were the losers on beaft also, though neither by more than 1.25%. and the gains were not as large, either, except the m4a samples
05:53:20webguest43Does anyone know whether bockbox bootloader supports eabi?
05:54:56Unhelpfuli doubt anybody has tried
05:55:10 Quit mt (Read error: 104 (Connection reset by peer))
05:55:38 Join mt [0] (n=mtee@
05:55:58Unhelpfulbut the bootloader is essentially a stripped-down version of rockbox, so it likely works properly
06:06:47 Quit tchan ("WeeChat 0.3.1-dev")
06:07:33 Quit bipton ("CGI:IRC")
06:07:44webguest43i see.
06:08:48 Quit webguest43 ("CGI:IRC (EOF)")
06:09:19Unhelpfulresults from tuning work so far are here:
06:09:29Unhelpfulthe lesson: arm7tdim sucks :)
06:18:28 Join tchan [0] (n=tchan@lunar-linux/developer/tchan)
06:26:02 Quit panni_ (Read error: 131 (Connection reset by peer))
06:49:32 Join FOAD_ [0] (n=dok@
06:49:44 Quit bluebrother (Nick collision from services.)
06:49:47 Join bluebroth3r [0] (n=dom@rockbox/developer/bluebrother)
06:52:04 Quit jasio_ (
06:52:04 Quit Utchybann (
06:52:04 Quit AaronM (
06:57:36 Quit FOAD (Read error: 145 (Connection timed out))
06:57:36 Nick FOAD_ is now known as FOAD (n=dok@
07:05:16NJoinjasio_ [0] (n=yann@
07:05:16NJoinAaronM [0] (
07:05:16NJoinUtchybann [0] (
07:05:55 Quit Utchybann (Read error: 104 (Connection reset by peer))
07:06:04 Join AaronMcD [0] (
07:06:05 Join Utchybann [0] (
07:06:18 Quit AaronM (Connection reset by peer)
07:06:45 Quit jasio_ (Read error: 110 (Connection timed out))
07:10:24 Join jasio [0] (
07:13:40 Join AlexC [0] (n=AlexC@unaffiliated/alexc)
07:13:57 Join shai [0] (
07:15:21AlexCRecently nearly all of the new music i put onto my ipod doesn't play on rockbox and causes rockbox to stop playing all music until reboot
07:15:33AlexCalso my apple firmware doesn't work
07:15:51AlexCIf i format it will that fix anything?
07:17:05sintheteki am no expert, AlexC, but i would think it's at least possible some sort of fs errors could cause problems so fs repair or format may help
07:17:30sinthetekno idea on apple firmware (i use rockbox on a different dap myself)
07:18:22***Saving seen data "./dancer.seen"
07:21:20*AlexC nods
07:21:28AlexCyeah it's a bit of a pain
07:25:40 Quit CaptainKewl (Read error: 145 (Connection timed out))
07:48:02 Join Horschti [0] (n=Horscht2@xbmc/user/horscht)
07:56:29 Quit Horscht (Read error: 145 (Connection timed out))
08:03:26 Join dmb [0] (n=Dmb@unaffiliated/dmb)
08:14:04 Part toffe82_
08:15:22 Quit AaronMcD ("Emo Time In My Corner... //_-")
08:16:21 Quit MethoS- (Remote closed the connection)
08:27:52 Quit Tomis ()
08:37:54 Quit phanboy4 ("Leaving")
08:47:49 Join ecognito [0] (
08:54:07 Join Rob2222 [0] (
09:00:08 Join pondlife [50] (n=Steve@rockbox/developer/pondlife)
09:06:23 Join bertrik [0] (
09:11:52 Quit Rob2223 (Read error: 110 (Connection timed out))
09:15:13 Quit pondlife ("Leaving.")
09:17:51 Quit AlexC (Remote closed the connection)
09:18:25***Saving seen data "./dancer.seen"
09:20:46 Join n1s [0] (n=n1s@rockbox/developer/n1s)
09:22:02 Join flydutch [0] (
09:26:43 Join bmbl [0] (n=Miranda@unaffiliated/bmbl)
09:42:24 Join ender` [0] (i=krneki@
09:49:13n1sUnhelpful: nice testing, seems to be about the same results i got, i didn't test all the O levels though, it looka like a clear win for both the beast and the gigabeat f and while the sansa has more losses they are not very big
09:51:26*amiconn thinks that the value of a speedup/slowdown depends on how fast that codec was to begin with
09:52:38amiconnA speedup in a codec with is already fast (like flac) is worth nothing if it means a slowdown in a slow codec (like e.g. aac)
09:55:05n1samiconn: true, but more than flac are faster, on sansa flac, alac, aac, wv and mp3 are faster
09:55:49n1swhile vorbis, wma, mpc, ac3 and ape are slower
09:56:35n1sUnhelpful: could you post your patch to easily swithc the codec compilation options? it might come in handy in the future
10:00:30n1sUnhelpful: btw, i only get aliasing warnings in the zxbox plugin
10:01:33n1sif it turns out fixing those warnings is tricky maybe we should just turn it on for codes as you suggested
10:02:46 Join BHSPitLappy [0] (n=BHSPitLa@unaffiliated/bhspitmonkey)
10:03:37n1samiconn: although the comparison is a bit unfair since i don't think anyone has set the optimal compiler optimization level for most of our codecs in svn
10:04:03n1sso a bunch of them use the core optimization settings
10:04:20amiconnIt seems that especially wavpack optimisation is hosed in svn
10:05:15n1sflac, faad and cook also build with the core optimization setting
10:06:17 Join Enok [0] (
10:06:32n1swavpack is using O2 so that might be a real compiler improvement
10:06:56 Quit Enok (Client Quit)
10:08:08n1sand it only sees the big improvement for arm7
10:20:51Unhelpfuln1s: when compiling with the new eabi compiler, and using -fstrict-aliasing in the makefile? interesting, i get several issues in core and a few of the plugins are *covered* in them
10:21:43Unhelpfulmost codecs and/or decoder libraries they use don't really touch -O
10:22:35n1sUnhelpful: i just removed the -fnostrict-aliasing from the makefile but yeah, only zxbox, which target are you building for?
10:25:01Unhelpfulbeast. i replaced it with a -fstrict-aliasing... it looks like that changes the default warning settings for aliasing as well
10:27:35Unhelpfulit looks as if -fstrict-aliasing sets -Wstrict-aliasing, which is equivalent to setting -Wstrict-aliasing=3 ; this is the highest level, which does more detailed aliasing analysis and should, per the manual, have "very few false positives and few false negatives"
10:29:51Unhelpfulprobably we should at least *try* to resolve warnings regarding aliasing. as you saw with test_codec, strict aliasing can allow the compiler to be a bit more successful in optimization.
10:32:50Unhelpfulit would be nice to have a "checksum directory" option in test_codec that could be used to verify that optimization changes and such don't impact codec output... obviously it's no help when somebody makes algorithmic changes, but compiler flags *shouldn't* change results, at least not in fixed-point codecs...
10:35:42 Join stoffel [0] (
10:36:00n1sshould be easy to implement really
10:36:23n1swe have the write wav functionality and can md5sum the wavs
10:36:36 Quit ender` (" The most important lesson we can learn from history is the inability of humans to learn lessons from history. -- Aldous Hux")
10:38:42 Join DerPapst [0] (n=DerPapst@
10:39:09Unhelpfuli don't think that write wav can process a whole directory... aside from that it would also be more efficient to checksum the decoded data in memory. and if we're checking for errors rather than guarding against malicious alteration of data, md5 might be a fair bit of overkill. the jhash algorithm that i added to pluginlib and never got to using would probably be fine :)
10:49:05 Quit Utchybann ("I like core dumps")
10:57:44amiconnUnhelpful: Rockbox already has crc32. Maybe that is sufficient
10:58:13Unhelpfulthat probably is, too...
11:04:32 Quit ecognito ("I read it on the Internet so it MUST be true!")
11:18:29***Saving seen data "./dancer.seen"
11:26:47 Join hebz0rl [0] (
11:37:08 Join ender [0] (i=krneki@
11:38:42 Quit shai (
11:38:42 Quit jasio (
11:38:46 Join shai_ [0] (
11:43:39 Quit Kopfgeldjaeger (Read error: 104 (Connection reset by peer))
11:43:41Unhelpfuln1s: patch to force -O level on *all* codecs. this includes anything that has a tuned -O level already - those are gone w/ this patch. :)
11:46:56 Join Kopfgeldjaeger [0] (
11:49:40n1sUnhelpful: aha, still usefull though (can it set other switches than O levels?)
11:51:12Unhelpfulyou could add some other switch in codecs.make, *but* i didn't remove overrides for anything else from the sub-makefiles
11:52:46n1sso you just added the switches to codecs.make and removed the overrides?
11:54:14 Join pamaury [0] (
11:54:43Unhelpfulthat's all, yes. also, i can see why we do this where there's more than one -O we might add based on target etc, but why do we *always* seem to use filter-flags when in some cases subst would do?
11:57:45 Quit hebz0rl ()
11:58:14n1swhat do you think about fixing the aliasing warnings? i'm not very sure how to fix the trickier ones but won't work on it if you are already or think it isn't worth it
11:59:43Unhelpfuli'm not sure *either*. i think it would be best to start with enabling it for codecs and tackling any there.
12:00:53n1sbtw, casting the pointers to (void*) "fixes" the warnings in kernel.c, not sure about that though
12:01:31n1s(void*)arr == NULL
12:03:07n1swell, any explicit cast seems to do it which makes me wonder even more
12:04:36Unhelpfuldon' t know. the solution i used for bmp.c can't work everywhere... in fact, it can only work for aliased members of structs :/
12:05:18Unhelpfulas nice global anonymous unions for creating variables that explicitly alias each other
12:06:05NJoinjasio [0] (
12:09:38n1safaik void pointers alias everything so this should be legal, but maybe casting to struct timeout* is nicer
12:12:18Unhelpfuli didn't look very closely at those... i was honestly a bit confused as to what was going on, it looked like the array of structs, including their various members, was addressed as an array of void*...
12:12:55n1syes, i think it's doing that
12:15:25*Unhelpful couldn't figure out how manipulating it in that way could be valid, since the code doesn't even know which struct member it's touching...
12:15:32 Quit BHSPitLappy ("Ex-Chat")
12:17:25n1safaict it only messes with the array of struct pointers and is addressing those as void
12:20:17Unhelpful...*oh*. it's an array of struct *, not of struct? :)
12:27:22n1sin fact uisng struct timeout** pointers and casting to void** in the function calls fixes the warnings and seems nicer to me
12:27:45n1si wonder of a good way of testing this
12:28:11Unhelpfulyou mean, testing whether it's valid?
12:29:14Unhelpfulapparently for types you can use the may_alias attribute
12:30:15Unhelpfulin fact, i wonder if that might be a cleaner fix for bmp.c
12:30:27 Join liar [0] (n=liar@
12:31:07 Quit mt (Read error: 104 (Connection reset by peer))
12:31:37 Join mt [0] (n=mtee@rockbox/developer/mt)
12:32:10 Join GeekShadow [0] (n=Antoine@reactos/tester/GeekShadow)
12:38:15 Join FOAD_ [0] (
12:40:00 Quit yosafbridge ("Coyote finally caught me")
12:40:16 Join yosafbridge [0] (n=yosafbri@
12:47:15 Quit FOAD (Read error: 145 (Connection timed out))
12:47:17 Nick FOAD_ is now known as FOAD (
12:52:17 Join Grahack [0] (
13:07:17n1syeah i mean teting if it actually works
13:14:40 Quit Kopfgeldjaeger (Read error: 60 (Operation timed out))
13:15:28 Join petur [0] (n=petur@
13:18:29 Join Kopfgeldjaeger [0] (n=nicolai@
13:18:31 Join FOAD_ [0] (
13:18:32***Saving seen data "./dancer.seen"
13:30:25 Join hebz0rl [0] (
13:32:31 Quit stoffel (Remote closed the connection)
13:35:29 Quit FOAD (Read error: 110 (Connection timed out))
13:35:29 Nick FOAD_ is now known as FOAD (
13:45:06 Quit n1s ("Lämnar")
13:47:13 Join n1s [0] (n=n1s@rockbox/developer/n1s)
13:48:04n1sof the codecs, mod, spc and the mp3 encoder gives aliasing warnings
13:48:14 Join bimbel [0] (n=Miranda@unaffiliated/bmbl)
13:48:43*petur wonders if JdGordon is looking into the recording statusbar issue or was merely hinting somebody should fix it
13:49:58 Join fdinel [0] (
13:50:18 Quit seani_ ("Leaving")
13:55:28 Nick Tuplanolla is now known as Tuplis (n=jani@unaffiliated/tuplanolla)
13:55:31n1shmm, spc is accessing a char array as uint16 so that's no surprise, just need to dig up a test file and this should be easy to fix
14:03:40 Join Omlet [0] (
14:05:23 Quit bmbl (Connection timed out)
14:10:07 Join Thundercloud [0] (
14:11:29 Quit lovingHDTV (Read error: 110 (Connection timed out))
14:11:57 Quit Bob_C (Remote closed the connection)
14:12:08 Quit flydutch ("/* empty */")
14:40:28 Quit Thundercloud (Remote closed the connection)
14:44:43 Join pondlife [50] (n=Steve@rockbox/developer/pondlife)
14:52:01 Part pondlife
15:03:05pamaurygevaerts: here ?
15:05:16CtcpPing from gevaerts!n=fg@rockbox/developer/gevaerts
15:05:36*gevaerts hears beeping
15:06:44 Join CaptainKewl [0] (
15:07:45gevaertspamaury: yes
15:10:25pamaurygevaerts: I really don't have time now but I saw on the logs that you asked me something after I quit irc. This was about what you should test on your device: you should try to get the string descriptor at 0xee and then send a get compat id and get extended property reequest and dump the answer
15:10:34 Join oblib [0] (
15:10:53gevaertsok. I'll try to look for that
15:11:06gevaertscan any of the mtp tools help with it?
15:11:53oblibHi, I've finally gotten around to updating my rockbox build (it's been at least a year), and my make is failing. It says "make: *** No rule to make target `/home/mythtv/stuff/sources/rockbox/rockbox/build/id3.h', needed by `/home/mythtv/stuff/sources/rockbox/rockbox/build/apps/bookmark.o'. Stop." Anyone know how to fix this?
15:12:57pamauryyes and no, the mtp-detect tools does it but if your device if know, it won't even probe the descriptor. So I advise you to pick the code of "probe_descriptor" in "usb_glue.c" of libmtp, that just a few lines and it does the job
15:13:02pamauryI have to leave
15:15:09 Quit Horschti ("Verlassend")
15:18:04 Join cowgarden [0] (n=qgar10@
15:18:32 Join Horscht [0] (n=Horscht2@xbmc/user/horscht)
15:18:36***Saving seen data "./dancer.seen"
15:19:08 Quit antil33t (Read error: 131 (Connection reset by peer))
15:19:21 Join antil33t [0] (n=Mudkips@
15:23:38gevaertsoblib: is the build directory empty before starting?
15:23:54 Quit parafin (Read error: 60 (Operation timed out))
15:25:55gevaertsand the tree is clean?
15:26:44oblibby tree to do you mean the source directories? No, I've just done an svn update and I've got a couple of custom changes
15:27:20 Join parafin [0] (
15:27:23oblibIf I can't figure it out, I'll do a clean svn directory and add my changes back in, but that's last resort
15:27:50gevaertsthere have been a lot of changes during the past year, including a complete overhaul of the makefiles
15:32:24 Join Topy [0] (
15:34:23 Join dfkt [0] (i=dfkt@unaffiliated/dfkt)
15:34:31oblibWould there be an easy way to just delete all the makefiles. It looks like they mostly end it .make ...
15:37:55 Join SchwarzerRiese [0] (
15:39:18oblibIf I got rid of them and then did 'svn update' that could fix the problem without redoing the whole tree
15:39:59gevaertsjust run svn revert on them then
15:40:14oblibon just the make files?
15:40:35oblibsvn diff doesn't think there are any diffs besides my source changes
15:41:18gevaertswell there most be something relevant that changed
15:42:47gevaertscheck if svn stat shows files that shouldn't be there
15:44:42 Quit fyre^OS ("lamers envy me like they envy bill g -- main boot xp, just the way it should be!")
15:44:58 Join fyrestorm [0] (
15:46:01 Part SchwarzerRiese
15:48:02gevaertspamaury: usb_control_msg() returns -32 when asking for the extended property descriptor
15:50:40 Quit T44 (Read error: 110 (Connection timed out))
15:51:05 Quit hebz0rl (Read error: 104 (Connection reset by peer))
15:52:10 Join pondlife [50] (n=Steve@rockbox/developer/pondlife)
15:53:38 Part pondlife
15:55:58 Join watto [0] (
16:01:48oblibsvn had skipped over apps/codecs.c codecs.h and playback.c
16:01:59oblibturns out those are sort of important...
16:02:12oblibThanks for your help gevaerts
16:06:11oblibbuild completed successfully
16:06:13 Quit oblib ("Leaving")
16:09:13 Quit bimbel ("Bye!")
16:11:06 Join GeekShado_ [0] (
16:14:15 Quit GeekShadow (Read error: 60 (Operation timed out))
16:14:35 Join Thundercloud [0] (i=thunderc@
16:18:24 Join GeekShadow [0] (n=Antoine@reactos/tester/GeekShadow)
16:24:03 Quit GeekShado_ (Read error: 60 (Operation timed out))
16:39:54 Join solexx_ [0] (
16:41:33 Quit solexx (Read error: 60 (Operation timed out))
16:46:13 Quit mc2739 (Read error: 104 (Connection reset by peer))
16:46:25 Join mc2739 [0] (n=mc2739@rockbox/developer/mc2739)
17:12:20n1shmm, no, the trickery in spc is much worse than i thought
17:13:56n1scasting the address of a member of an array to a struct pointer and then doing indexed addressing on that pointer...
17:14:58 Quit shai_ ("Leaving")
17:15:25 Join shai [0] (
17:18:38***Saving seen data "./dancer.seen"
17:21:27 Join flydutch [0] (
17:29:39n1sugh, it's even casting a pointer to an array inside a struct to a struct pointer and does indexedaddressing on that struct pointer. And then accesses this char array as a uin16_t so double aliasing breakages in a single line :)
17:30:06n1ss/dadd/d add/
17:30:22*gevaerts doesn't loke horror stories at this time of day
17:32:13gevaertsnot your fault :)
17:34:05 Join stoffel [0] (
17:37:44 Join toffe82 [0] (
17:42:17saratogaso make the ClipV1 unstable and the fuze/e200v2 stable today?
17:43:14saratogai wonder if the c200v2 now works?
17:43:28 Join kugel [0] (n=kugel@rockbox/developer/kugel)
17:43:55gevaertssaratoga: how about committing your work first? ;)
17:44:00kugelsaratoga: can the fuze wait until the display glitches are sorted out?
17:44:12kugelit just needs bertrik's patch committed
17:45:19saratogakugel: and the e200v2?
17:45:35kugelI don't know if that one has the glitches too
17:46:12saratogaok i'll leave both alone for today
17:47:05pixelmasaratoga: I tested the 1a patch a bit on my M5 which is coldfire and has an HD (but only playing a single album). Seemed to work fine as far as I could see - I don't use feature like "gather runtime data" which I think caused trouble when there were small bugs in buffering causing disk spinup at every track change (haven't tested for that yet)
17:47:37saratogapixelma: I've also been running in on my gigabeat F without issue
17:48:05saratogathe only change i'm not sure about is the one in playback.c and blue dude and matsch both seemed to think it was reasonably safe
17:48:14gevaertsIf there are issues left, committing now will increase the chances to find them before the next release
17:56:38 Join Utchybann [0] (n=lolo@
17:57:12n1shmm, using memcpy seems to be the only way to avoid these warnings in spc without some pretty intrusive changes
17:57:36n1sor maybe creating a big ass union...
17:59:30 Quit Grahack (Success)
18:00:41CIA-80New commit by saratoga (r23680): Commit FS #10605 - stable playback on low memory swcodec targets by Matthias Schneider. Should allow stable playback on targets with less then 4MB of ...
18:01:29 Join Grahack [0] (
18:02:56saratogadoes anyone have a c200v2?
18:03:48CIA-80New commit by saratoga (r23681): Update the website to promote clipv1 from unusable to unstable.
18:05:01 Join AlacerFox [0] (
18:05:22saratogaBagder, Zagor: would either of you mind uploading this bootloader for the Clip: and then updating the front page?
18:06:28saratogawow all that and no binsize increase
18:06:40 Join Sajber^ [0] (
18:07:12gevaertsnot even a yellow!
18:07:53saratogai tested it on the sims and a lot of targets first
18:08:15saratogai kept expecting it to fail
18:09:29 Join funman [0] (n=fun@rockbox/developer/funman)
18:09:53saratogafunman: congrats
18:11:17 Join bluebrother^ [0] (n=dom@rockbox/developer/bluebrother)
18:11:25kugelwhat for?
18:11:54saratogahe basically started the clip port
18:12:16saratogaand now 1.5 years later its basically working
18:13:07funmani think it really started on the clip 1 year before i first browsed the forum
18:13:29 Quit Thundercloud (Remote closed the connection)
18:13:38funmani'm just a drop in the ocean, which happened to have a lot of free time at the right point in time
18:13:56funmanbut i'm very happy it reaches an usable point ;) (just downloaded the current build)
18:14:07 Join Blue_Dude [0] (n=chatzill@rockbox/developer/Blue-Dude)
18:15:17kugelah yea of course, good work! :)
18:16:09funmanah it reminds me FM radio still doesn't work on the Clip
18:16:19gevaertsfunman: you *downloaded* a build?
18:16:36 Quit dmb ("Leaving")
18:17:31funmansure, why ?
18:18:08funmani don't work much on rockbox these days, downloading is faster than building from scratch
18:18:18gevaertsI know. Just teasing
18:20:25n1syay, got rid of the warnings with some amazing union hack :)
18:20:36n1sno idea if it works though :)
18:20:58 Quit Blue_Dude ("ChatZilla 0.9.85 [Firefox 3.5.5/20091102152451]")
18:21:10funmanby the way it seems there is some similarities between the USB chip used in sansa AMS and the chip used in nano2g
18:21:44saratogafunman: any interest in cleaning up and committing your clipv2 work?
18:22:49pamaurygevaerts: re, on which target(s) did you try it ?
18:23:01gevaertspamaury: gigabeat S, OF
18:23:18funmansaratoga: not sure if it will help someone figure out how to transfer data from SD
18:23:34gevaertsso I guess not implementing those descriptors is fine
18:24:16pamauryyes, I guess the guys of libmtp and gPhoto2 implement it without really testing it.
18:24:21saratogafunman: i realize that but getting the code in svn would still be nice so that it doesn't rot
18:24:38saratogasooner or later we'll get someone interested
18:24:53saratogaanyway, time for lunch
18:25:23Lloreansaratoga: Does the clip require the new bootloader for fixed playback, or just a current build?
18:25:45funmanLlorean: jut current build
18:25:50AlexPI'm fixing the manual for the clip install instructions - for the difference between clip and clipv2, is it a case of OF version number to tell the difference? And if so, what are the versions?
18:26:02funmanmy clip bootloader is r21285
18:26:34funmanAlexP: v1 begin with 01. and v2 with 02.
18:26:47AlexPfunman: Thanks
18:28:15Lloreanfunman: My bootloader is 19385. So a bit older.
18:28:49 Quit DerPapst ("Leaving.")
18:28:49funmanLlorean: at some point it was needed to update the bootloader (perhaps when data cache was enabled, not sure), else it would lock up
18:28:51Llorean19385 doesn't seem to be able to boot a current build.
18:29:04LloreanI guess mine's from before that point.
18:29:16funmanoh we'll need a released bootloader and activate rbutil support
18:32:08Lloreansaratoga's build a bootloader, I think.
18:32:09 Quit cowgarden (Read error: 131 (Connection reset by peer))
18:32:50AlexPfunman: For the AMS Sansas, is the bootloader on "disk" or "flash" (ignoring that the disk is flash)
18:33:36funmanit's on flash (not visible on the disk)
18:36:38LloreanDo clip+ work too, or just the clip for now?
18:36:56funmanonly clipv1 (clipv2 has lcd/buttons, clip+ nothing)
18:37:39LloreanOh. I thought I remembered someone having something running on the +. Thanks.
18:37:40funmanthe sansa ams forum thread will be 2 years old in 8 days :o)
18:38:02kugelit is on flash? I thought it was on the hidden firmware area of the "disk"
18:38:31funmankugel: well the disk is flash :)
18:38:46kugel"ignoring that the disk is flash" still applies
18:39:04*kugel was just following AlexP's convention
18:39:26AlexPDisk in that case then I think
18:39:48AlexPThis is for the manual incidently
18:40:00AlexPI assume that the bootloader download location will be bootloader/sandisk-sansa/clip/bootloader-clip.sansa ?
18:40:21kugelwe hope so :p
18:40:48AlexPand that mkamsboot on the download server will be updated to deal with it (if it doesn't already)?
18:40:59gevaertskugel: I'd assume that flash means "an area you can't reach directly from rbutil" in this case
18:41:06funmanAlexP: mkamsboot already works
18:41:29funmanor from a formatting tool / file browser
18:43:04AlexPWhat is the clip OF .bin called?
18:43:11 Join Lss [0] (n=Lss@
18:43:48 Quit stoffel (Remote closed the connection)
18:44:12 Quit kugel (Read error: 131 (Connection reset by peer))
18:45:30 Quit Omlet (Read error: 113 (No route to host))
18:47:18 Quit Lss (Read error: 131 (Connection reset by peer))
18:47:27 Join kugel [0] (n=kugel@rockbox/developer/kugel)
18:48:31LloreanShould be an actual current build of Rockbox for clip?
18:52:02AlexPfunman: Does rbutil support the clip yet?
18:52:43gevaertsLlorean: is the one you want I think
18:52:56LloreanThe -clip one should probably be deleted then
18:53:20bluebrother^AlexP: rbutil has preliminary support for the clip in svn, but it's disabled for releases
18:53:32AlexPOK, cheers
18:53:53bluebrother^no idea what state that support is in, though. But if mkamsboot is doing the job I'd guess it's pretty much done
18:54:13*bluebrother^ thinks that a target support table on the rbutil wiki page might make sense
18:54:17funmanbluebrother^: just like fuze/e200v2, i think only the "disabled" line needs to be removed from rbutil.ini
18:56:10funmanand add "(Unstable)" to the model name
18:56:28funmanby the way, m200v4 and c200v2 could go into unstable as well (if they are tested)
18:56:29LloreanAm I to understand that in SVN Rockbox you actually have to edit a text file to get numerical volume now because of the SBS stuff?
18:58:03kugelLlorean: no
18:58:10CIA-80New commit by alex (r23682): Update Sansa Clip install instructions in the manual.
18:58:20Lloreankugel: Where's the setting for it then?
18:58:27kugeltheme settings
18:58:53kugelin status/scroll bar settings
18:59:12LloreanAh, that's where status/scroll bar settings are now. okay.
19:00:44 Quit GeekShadow (Read error: 131 (Connection reset by peer))
19:02:01 Join GeekShadow [0] (n=Antoine@reactos/tester/GeekShadow)
19:02:07 Join Omlet [0] (i=omlet05@
19:02:38 Join kugel__ [0] (
19:02:52 Quit kugel (Nick collision from services.)
19:02:58 Nick kugel__ is now known as kugel (
19:04:31AlexPer, the clip doesn't have a recovery mode does it?
19:04:32funmanAlexP: well the option HAVE_RB_BL_ON_DISK shouldn't not be used on Sansa AMS
19:04:38funmanno it doesn't
19:04:44AlexPshouldn't not?
19:04:48funmanthe rockbox bootloader is not on disk
19:05:07AlexPdisk here means the drive, not hard disk vs flash disk
19:05:13funmanwhere disk means the mass storage visible over usb
19:05:21AlexPit is on a flash chip then?
19:05:39funmanit is on the same flash chip than the mass storage, but it's not visible from USB
19:05:56kugeldid someone try if the buffering fix helped on the c200v2?
19:06:25CIA-80New commit by alex (r23683): The Clip has no recovery mode.
19:06:33 Join dmb [0] (n=Dmb@unaffiliated/dmb)
19:07:11AlexPfunman: OK, I'll swap it
19:07:21funmanhum this option is not clear
19:07:32AlexPIt seems a semantic argument to me
19:07:57AlexPfunman: If you read the actual text of the manual, is it incorrect? (ignoring what the option is called)
19:07:58funmanwhat's the difference between _ON_DISK and _ON_FLASH ?
19:09:25funmani believe it's incorrect
19:09:55funmanit says there is the OF bootloader, which loads the rockbox bootloader, which loads rockbox
19:10:34kugelisn't that technically true?
19:10:47funmanthe rockbox bootloader loads the OF
19:10:57funmanthere's no "OF bootloader"
19:11:14AlexPSo the very first thing that runs is the RB bootloader?
19:11:20AlexPThere is nothing before that?
19:11:23kugeldon't we patch the OF to load our bootloader?
19:11:51AlexPI think it should be IN_FLASH anyway
19:12:10funmani think it's closer to reality yes
19:12:34AlexPOK, I'll change that
19:12:36funman \opt{HAVE_RB_BL_IN_FLASH}{Unlike the \playerman{} firmware, which runs
19:12:36funman entirely from flash memory,}
19:12:47funmanhm that's true
19:13:07kugelIIRC that's the same procedure as on h100, so it should read what the h100 manual read
19:13:29 Join Lss [0] (
19:13:32funmanAlexP: hum actually your last commit is wrong
19:13:51AlexPI asked and you said there was no recovery mode
19:13:51funmanthe *n*opt should include e200v2/clip/fuze because they don't have a recovery mode like e200v1
19:14:03AlexPoh, woops
19:14:06 Join AaronM [0] (
19:14:06funmanyes there is no recovery mode, so clip should be in nopt, like fuze
19:14:14 Quit amiconn (Nick collision from services.)
19:14:17 Join amiconn_ [0] (i=quassel@rockbox/developer/amiconn)
19:14:20 Join pixelma_ [0] (i=quassel@rockbox/staff/pixelma)
19:14:20 Quit pixelma (Nick collision from services.)
19:14:36 Nick amiconn_ is now known as amiconn (i=quassel@rockbox/developer/amiconn)
19:14:37AlexPSansaAMS in fact
19:14:37 Nick pixelma_ is now known as pixelma (i=quassel@rockbox/staff/pixelma)
19:15:43CIA-80New commit by alex (r23684): No SansaAMS have a recovery mode, and set the Rockbox bootloader to be in flash.
19:15:43saratogai bet the c200v2 works now, but who has one?
19:16:05gevaertsbertrik has one IIRC
19:17:22 Join panni_ [0] (
19:18:41***Saving seen data "./dancer.seen"
19:20:54kugelclip playback works great it seems :)
19:21:05saratogayeah my playback counter is at 70 hours
19:24:10 Quit Lss (Read error: 104 (Connection reset by peer))
19:31:06 Quit funman ("free(random());")
19:34:02 Quit flydutch ("/* empty */")
19:41:28 Quit toffe82 (Read error: 104 (Connection reset by peer))
19:41:47saratogawe haven't been updating MajorChanges with recent improvements to the WPS, statusbar, etc
19:42:39kugelbrowsing and playing music in/from pictureflow on the clip is ... strange :P
19:43:15saratogashould pf be enabled?
19:43:20kugelit is
19:43:38gevaertswhy not?
19:43:38saratogamaybe it should be disabled
19:43:40bertrikit looks a bit weird because of the yellow/cyan split, but it does work
19:43:47kugelthere was no major improvement in the wps/area, mostly introducing and fixing of small bugs
19:43:55gevaertssaratoga: it's enabled on the archoses...
19:43:58saratogai thought we had translated strings now
19:44:15kugelthat's older an in MajorChanges already
19:44:30saratogaoh must have missed it
19:45:47pixelmait even looks better on the Archoses (because of the clip's flickery OLED)
19:46:09pixelmaat least that's what I remember from DevCon ;)
19:47:00kugelI seem to remember the same ;)
19:47:53 Quit Torne (Read error: 60 (Operation timed out))
19:48:03 Join Torne [0] (i=torne@
19:48:58saratogabertrik: do you have a c200v2?
19:49:13bertriksaratoga, yes
19:49:25saratogacould you try current svn on it and tell me if it works?
19:50:24 Join Thundercloud [0] (
19:54:36CIA-80New commit by bluebrother (r23685): Use f_frsize when calculating the free disk space from the block count. Fixes wrong value on implementations with f_bsize != f_frsize.
19:55:34kugelbertrik: how about committing the dbop patch (the one before the latest one)?
19:56:17kugelif needed, we can still work on improving it in SVN. but fixing the sporadic lcd glitches makes it worth having in on its own
19:58:44 Join froggyman [0] (
20:00:21bertrikyes, I think we can commit parts of the DBOP stuff
20:01:10 Part AlacerFox
20:09:10 Quit GeekShadow (Read error: 54 (Connection reset by peer))
20:11:39kugelbertrik: ^
20:22:59 Quit faemir (Read error: 104 (Connection reset by peer))
20:24:16n1syay, mod codec aliasing warnings were easy to fix
20:25:19saratogawhat are aliasing warnings ?
20:27:20saratogaor i guess more directly why would the compiler need to warn about them
20:28:26kugelI think aliasing is actually illegal as of C99
20:28:51n1ssaratoga: because they prevent the compiler from doing some of it's optimizations (and are not valid C actually)
20:28:52Lloreansaratoga: I don't find the Clip keymap too bad for the most part
20:29:11kugel(where aliasing: having different types of pointer pointing to the same address)
20:29:15saratogaLlorean: me neither but i haven't tried the plugins
20:29:48 Join Omlet05 [0] (
20:29:52LloreanAh, might want to specify that you mean the plugins then, so we don't get a flood of "here's how to make it more like the OF" posts.
20:30:56saratogai added a don
20:31:09saratoga't talk about the OF clause to the forum rules so that i can just delete those posts :)
20:34:30kugelmc2739: ping
20:34:45kugelsaratoga: you have on e200v2 right?
20:34:50saratogakugel: yes
20:35:34saratogais the clip radio broken for everyone or only some players?
20:36:17 Quit watto (Read error: 60 (Operation timed out))
20:36:23Unhelpfuln1s: you're working on this then? i hadn't started on codecs, and i don't expect much rockbox time until wednesday anyway.
20:36:32gevaerts18:16 < funman> ah it reminds me FM radio still doesn't work on the Clip
20:36:40gevaertsso I guess all
20:36:51 Quit Omlet (Read error: 145 (Connection timed out))
20:36:55saratogathe wiki page said that they do until i changed it today
20:37:00saratogaso did we break it at some point?
20:37:26kugelno idea, it doesn't work on mine as well
20:37:27n1sUnhelpful: yeah, trying to at least, mod was easy, psc seems fixed but i'm not really sure, looking at mp3_enc now, those were the only ones i got warnings from
20:37:45 Quit Omlet05 (Read error: 60 (Operation timed out))
20:38:45kugelsaratoga: can you test fs10603 a bit? the patch from 10 oct, with and without the patch funman uploaded right after?
20:38:49kugelI guess it's out of date though
20:39:10kugelit's in question whether the increased polling rate is actually useful for the e200v2
20:41:13saratogakugel: was that file removed?
20:41:41kugeluhm no
20:43:05kugelthat never existed in svn
20:44:30saratogasorry wrong oct 10 patch
20:46:28kugelit's probably out of sync
20:46:47n1sUnhelpful: looks like i got the warnings in mp3_enc too so i'll test a bit and probably post a patch tomorrow
20:47:57n1sgot them fixed, that is
20:53:32saratogakugel: it seems to break the wheel unless i synced it badly
20:53:35 Join phanboy4 [0] (
20:54:48kugelyou mean it doesn't work at all?
20:54:54saratogakugel: correct
20:55:00saratogait doesn't even bring the backlight back on
20:55:12kugelI assume you sync'd it badly then :)
20:55:22saratogawell you're welcome to try yourself :)
20:55:36saratogai just deleted the fuze changes and synced debugmenu by hand
20:56:02saratogawas lcd-fuze.c important?
20:56:21kugelthat were the only conflicts?
20:56:37kugelthen it really doesn't work
20:57:00kugelI'll try to post a working patch
20:57:08saratogaok i'll test it when you do
21:03:09 Join PSPdemon [0] (
21:10:35*B4gder added committer #86
21:11:11*petur stares at
21:11:47peturJdGordon can fix it himself
21:13:56*bluebroth3r wonders why there's a point in having a separate statusbar anyway
21:14:24bluebroth3rif it's considered a skin then the skin could contain that. Making the statusbar simply a viewport
21:15:08gevaerts"the skin"?
21:16:37bluebroth3rwell, we now have a skin buffer and thus some kind of skin engine, right?
21:16:42 Quit linuxstb (Read error: 148 (No route to host))
21:16:46 Nick bluebroth3r is now known as bluebrother (n=dom@rockbox/developer/bluebrother)
21:17:31bluebrotherwhich means we're moving towards full skinnability. At least that's my impression.
21:18:08 Join toffe82 [0] (
21:18:44***Saving seen data "./dancer.seen"
21:18:45gevaertsmaybe, but we're a *long* way from that
21:19:33PSPdemoni almost want picture flow included in the browser ( as in a view option instead of a demo rock )
21:20:35bluebrotherthat's possible, but wouldn't considering the non-wps screens a skin (which currently can only get adjusted in parts, which right now is the statusbar) the first start?
21:21:22kugelsaratoga: I hope that one works
21:21:46gevaertsbluebrother: does that differ so much from what's happening?
21:23:24kugelbluebrother: the classic sb is on the way out anyway
21:23:29kugelthere's a task about it
21:24:06bluebrotherwell, it's kinda the other way round: coming from the skin view would mean that we have a skin, which defines two viewports: the statusbar, and a main one (which can't get changed). As far as I understand the current attempt is to add a single viewport that can get changed and is this customizable viewport
21:24:32*bluebrother stresses that he really likes the old statusbar. Just in case JdGordon wonders again if anybody would want to use that at all
21:26:27kugelbluebrother: not a single viewport
21:28:30 Join sh90 [0] (i=cpu98@
21:28:30kugelwhat you mean is implemented (too, kind of)
21:32:12 Join linuxstb [0] (n=linuxstb@rockbox/developer/linuxstb)
21:32:36 Quit bertrik ("De groeten")
21:36:39 Join Reiska666 [0] (
21:37:24 Part Reiska666
21:38:53pixelmapetur: I'm looking into making the more classic anyways, maybe I spot something about the recording frequencies
21:43:30peturI've never liked WPS files and this sbs stuff can go the same way
21:45:39Unhelpfuln1s: excellent. i'll look into using the best -O settings from the benchmark runs. armv5 behaved enough like armv6 that i expect i can just treat them the same, and select slighly differently for armv4
21:46:19Unhelpfulthough i should probably re-run after your patch to see if strict aliasing interacts differently with different levels
21:46:28kugelyou have an armv5?
21:47:08Unhelpfulkugel: beaft is, isn't it? i had krazykit run tests.
21:47:23kugelgigabeaft is v4
21:47:26 Quit Horscht ("Verlassend")
21:47:52Unhelpfulbah! ok. it behaves *quite* differently from e200.
21:48:03Unhelpful(with regard to -O results)
21:48:08kugelI think only (some of) the telechips are v5
21:48:46kugelthat can have various other reasons, like caches and iram
21:52:46n1sUnhelpful: beaft is arm9 though so it has a different pipeline to the arm7 IIRC even if it's still v4
21:53:24Tornev4/v5 is not that important, i wouldn't think
21:53:35Tornethe changed instructions are generally not something a compiler generates (DSP and the like)
21:53:51Tornethe differences ebtween arm7/9/11 will dominate
21:58:28 Join Horscht [0] (n=Horscht2@xbmc/user/horscht)
22:01:03Unhelpfulis the pipeline closer to the beast's arm11 pipeline in some way? the changes for best -O levels for codecs for them are *very* similar.
22:01:21Tornearm9 is more like arm11 than it is like arm7, yes
22:01:27saratogathe gigabeat F chip is the same as the AMS core
22:01:41saratoga5 stage classic risc with seperate I and D caches
22:01:47saratogabut the same multiplier as ARM7
22:02:22 Join Tomis [0] (n=Tomis@
22:02:28saratoganothing will scale like PP because it has 128KB of actual SRAM built into the core itself and not just hung off some slow peripheral bus
22:02:32TorneARM9 has branch prediction, which makes a lot of difference :)
22:02:49TorneARM7 just stalls because it's thick
22:02:58saratogaarm7tdmi has negligible pipelineing which is like good branch prediction only better :)
22:03:20 Quit Thundercloud (Remote closed the connection)
22:03:40Tornesaratoga: you still get a one stage stall, no?
22:03:53Tornewhich affects cache
22:04:01saratogayeah i think its 1 clock worth of stall
22:04:06Torneassuming you have something bizarre like PP which has a cache.
22:04:53 Join Tomis2 [0] (n=Tomis@
22:04:57Torneit won't be a one *clock* stall unless you are executing from cache :)
22:05:02saratogakugel: no difference, still no wheel
22:06:34saratogabasically nothing will ever be as fast as PP unless it can have a very fast memory system and probably a 1 clock MULL instruction
22:06:57saratogaper clock that is
22:07:23saratogaso probably the TCC chips with SRAM and ARMv5E and the beast if your code sticks in cache well
22:07:57Torneif you were particularly brave you could experiment with writing codecs to use TCM :)
22:08:06Torneinstead of relying on cache behaviour
22:08:47saratogaactually some of those TCC chips look fantastic, Coldfire like multiplier performance with PP memory + cache memory and ARM9 clock speeds
22:09:08Tornewe already fiddle a bit like that to do iram at all
22:09:19saratogafor which target?
22:09:24Torneon any target, i mean
22:09:32Torneusing TCM is the same principle
22:09:38Tornethere's just gonna be less
22:09:42saratogai also always wanted to try one of those ARM7E cores where you get the ARM7TDMI pipeline but the ARMv6 DSP unit
22:10:11saratogaTorne: isn't that what we already do with IRAM?
22:10:21Tornesaratoga: ..yes. that's what i said.
22:10:38Torneinstead/aswell of using IRAM you could shove code/data into TCM on ARM11
22:10:54Torneeliminating any dependency on cache behaviour
22:10:57saratogaTCM is the beast's IRAM?
22:11:15Torneno, TCM is when you disable automatic cache management and have the L1 cache available in the memory map
22:11:22Torneto use as regular instruction/data memory
22:11:29saratoga16KB IIRC?
22:11:48Torneyou can use both though
22:11:59Torneinstruction cache becomes ITCM, data cache becomes DTCM
22:12:15saratogai doubt its worthwhile for most codecs
22:12:28Torneprobably not
22:12:58saratogait might be useful in mp3 for putting some of the mdct and synthesis coefficients into TCM since they're so small
22:13:01Torne(ITCM is hilarious; memory you can execute but not read)
22:13:21saratogathats actually really common on DSP systems
22:13:31Torneyah, it just is kinda odd to have on ARM
22:13:33saratogai think theres better ways to make libmad not suck
22:13:45Unhelpfulhow best to distingush between arm7 vs arm9+ in the makefile, then? there are actually codecs for which -O3 works better on beast and beaft, and it's pretty much always a huge loser on arm7...
22:13:57saratogait badly needs a complete code review
22:14:05Torneprobably. it's only really worth looking at if you can demonstrate you are being hurt by cache misses
22:14:20saratogaUnhelpful: I would just check for NOT PP
22:14:22Tornewhich reminds me, do we have the perfctrs on beast?
22:14:43UnhelpfulTorne: it has clock and clock/64 counters, which we don't use
22:14:59Unhelpfulas well as counters for things like branch prediction misses iirc
22:15:00TorneUnhelpful: ARM11 can count memory accesses, cache misses, etc, i mean.
22:15:15TorneThat's very useful for analysis ;)
22:15:34Unhelpfulit is, i suppose, but we've only the one arm11 target :/
22:15:48Torneit's partially there on some ARM9s also
22:15:49Unhelpfulie its optimizations won't be of very broad use
22:15:50Tornejust not documented
22:16:09CIA-80New commit by mc2739 (r23686): FS #10796 - Clip - prevent power switch activation when coming out of hold
22:16:13Tornethe ARMv5 debug/perfctr system is *almost* the same as what's in the ARMv6 ARM
22:16:29TorneARM just weren't ready to commit to it at the time I think, so they left it as a private detail for their debugger
22:17:22Unhelpfuleven cycle counters can be pretty useful, especially as we have such a low-resolution system timer (for speed benchmarks)
22:17:33Tornethat's true.
22:18:24saratogaoptimizing most of our targets isn't that hard once you know the DRAM and IRAM access times as well as the latency on the multiplier
22:18:51saratogait just comes down to load/store/mul as little as possible
22:19:53saratogathat said a plugin for testing mul, DRAM, and IRAM access times in clocks would be very useful
22:20:05n1sUnhelpful: i think TheSeven said the nano4g is arm11 too so we *might* get another one
22:20:37saratogamaybe also for cache associativity and cacheline size
22:20:45 Quit Tomis (Connection timed out)
22:20:45 Nick Tomis2 is now known as Tomis (n=Tomis@
22:21:07Tornecacheline size would be very good to know for sure on some platforms ;)
22:21:27saratogawell its a pretty trivial plugin to write if you're bored :)
22:21:39 Quit fdinel (Read error: 104 (Connection reset by peer))
22:21:44saratogai once hacked one together for PP but never got around to making it really useful and since lost the code
22:21:53Torneoh? ;)
22:22:03Torneis the cacheline size on pp502x definately 16 bytes? :)
22:22:16saratogaIIRC yes, theres a document from MrH explaining it
22:22:19saratogahe did some extensive testing
22:22:26saratogaor else is one of the PP designers
22:22:34Torneyah, i saw that
22:23:06Tornesomeone just threw me a bit by claiming that their ipod stopped being unstable with dma when they switched it to 32
22:23:24Tornebut it turned out not to really be reproducible:)
22:23:34saratogawell theres a lot PP revisions floating around, including two different caches, so its possible they changed it at some point
22:23:43saratogaMrH only claimed that for the PP5024
22:24:43Tornewell, i believed it becuase it's fairly typical for arm at that time
22:24:52Tornethe XScales all have 16-byte cachelines as well
22:25:11mc2739kugel: pong
22:25:17saratogaits a pretty obvious size to use given the bus widths and clock speeds
22:25:41saratogafor AMS i wonder about the memory, funman's measurements suggested some really odd timings, and we have some ability to tweak the internal clocks
22:26:10*Torne should actually unify ata_{read,write}_sectors and see if that makes his ipod stable
22:26:33Tornei've had the DMA patch on for ages and it *normally* works fine but I can get it to die by doing certain things while the player is maybe being shaken a bit
22:26:37saratoganot to mention theres some differences between the AMS targets using internal DRAM and those using external DRAM
22:26:45Tornewhich suggests it's probably write failures not being retried that causes it
22:26:50Tornerather than anything wrong with dma
22:26:52Tornebut it's hard to prove ;)
22:28:08saratogado they make line out cables for the gigabeat F that I can order?
22:29:12saratogaor maybe I can use the cables that came with my S on the F?
22:29:38gevaertsmy homebuilt F lineout cable didn't work with my S
22:29:57gevaertsthat doesn't prove much of course
22:30:22saratogai've got a bunch of F and S boxes lying around, i should dig through and see if any came with cables
22:31:03toffe82saratoga: no line out cable available
22:31:20saratogathe internet shows line out cables coming with the S
22:31:51toffe82through the headphone plug
22:31:58*n1s didn't get one with his S afahr
22:32:08gevaertssaratoga: I believe there's someone who can sell you a compatible connector :)
22:32:27saratogaoh you're right
22:32:35saratogadoes anything connect to the dock?
22:32:51*toffe82 try to finish the break out board for the plug , but can't......
22:33:05Tornehm, my poll is not going so well
22:33:14Torneabsolutely nobody has voted for any player other than the Video
22:33:23n1sTorne: ?
22:33:43Torneanyone here have a PP ipod, other than the video, that they use enough to have noticed whether it has the startup bug? :)
22:33:55saratogadid toshiba even make a dock cable?
22:34:22LloreanTorne: I haven't seen it in Rockbox on my Nano, but I *think* I've seen it once when I let the OF sleep long enough to fully shut down.
22:34:30kugelmc2739: we noticed that the dbop doesn't work on the e200v2 anymore (I know the latest one didn't, but the one before also didn't work already=
22:34:42TorneLlorean: that's unfortunately not a data point i can do a lot with ;)
22:35:16LloreanAnother useless data point - off the top of my head I only explicitly remember bug reports regarding this problem on 5Gs
22:35:40LloreanI *may* remember one instance of it being reported on a Mini, but I'm not certain. The bug may be exclusive to the PP5021/22
22:36:00toffe82saratoga: therei a dock for the F and X but not for the S
22:36:18saratogatoffe82: seen this thing:
22:36:19Tornepeople on the FS have reported that it happens on 4g color, and on mini
22:36:24Torneand that the workaround fixes it
22:36:28saratogalooks like it has access to the line out pins
22:36:32TorneThe workaround is only for 502x anyway
22:36:47n1sTorne: sorry, can't help you there, my only fully working ipod is the nano4g
22:37:04Tornen1s: yah, it's ok. i hoped actual users might respond, and some have, but only for video
22:37:38toffe82saratoga: the U doesn't have a dock connector, just usb and headphone
22:38:21saratogatoffe82: that device has a dock connector and an AUX in port
22:38:27saratogapresumably the aux in is for the U
22:38:58toffe82the T doesn't have dock connectoer too
22:40:10saratogawhats actually on the F dock?
22:40:16saratogajust a USB port?
22:40:31gevaertsUSB (host and device), line out, and power
22:41:41mc2739kugel: dbop_buttons3.patch did not work on e200v2, the last patch I tested before that was dbop_buttons.patch (sync after r22863 by funman)
22:42:17saratogagevaerts: it has line out???
22:42:32toffe82on the dock yes
22:42:36 Quit petur ("Zzzz")
22:42:38saratogabah thats what i basked before
22:42:51toffe82no , you ask for a cable :)
22:42:55saratogaHAVE_TAGCACHE is actually for the database
22:43:00saratogawell wire + dock = cable
22:43:07saratogaand wires are easy to find
22:43:09toffe82there is the line out on the dock connector for the F S and X
22:43:32saratogasorry, HAVE_TAGCACHE actually enables the database, correct?
22:43:50gevaertssaratoga: well yes, but when I needed a cable, it was because a dock wasn't suitable for that situation :)
22:44:26CIA-80New commit by mc2739 (r23687): FS #10781 - Scrolling doesn't stop when entering USB screen from WPS
22:54:23 Part freqmod_qu (" - Chat komfortabelt. Hvor som helst.")
22:55:30 Quit Grahack ("Leaving.")
22:58:53linuxstbsaratoga: Did you tag your Clip bootloader build in SVN?
23:00:00amiconnsaratoga: Regarding differentiation, there is a specific ARM7TDMI macro
23:00:17saratogalinuxstb: its already tagged in SVN, I just checked it out and compiled
23:00:41saratogaamiconn: ok then thats obviously better then checking for PP
23:01:19saratogawe tagged all the AMS 1.0 bootloaders at the same time, even for the ones that weren't released
23:01:21linuxstbsaratoga: Hmm, weird...
23:02:31saratogaon AMS the bootloader does so little its unlikely they'll be a reason to change them soon, at least not until we want to have bootloader USB or else completely rework the clocks
23:03:00 Quit bluebrother^ ("Computer has gone to sleep")
23:03:25saratogaactually putting up c200v2 and m200v4 bootloaders now might make sense
23:03:41saratogasave us trouble later
23:04:17amiconnsaratoga: CPU_ARM7TDMI to be precise. Check with #ifdef
23:04:45saratogaI suspect Unhelpful will need that more then me
23:04:46*amiconn introduced it for the libdemac predictor
23:04:48linuxstbsaratoga: Yes, that's what I meant by weird. If bootloaders are tagged and working, we may as well make them available.
23:04:55saratogaok i'll compile them
23:05:05linuxstb(even if there are other problems preventing them being "unstable")
23:05:45saratogathe c200v2 is probably ready to be released, we're just waiting on someone to try it and make sure nothing else is broken
23:06:00saratogathe m200v4 apparently has some issues
23:07:36kugelmc2739: not the correct fix IMO
23:08:15linuxstbsaratoga: Thanks, but I don't have the power....
23:08:17*linuxstb pings B4gder
23:08:48*B4gder crawls out of his cave and blinks in the light
23:09:17mc2739kugel: why?
23:10:02 Join bekabug [0] (
23:10:59B4gderc200v2, clip and m200v4 bootloaders uploaded to the download server
23:11:20*linuxstb pushes B4gder back into the dark
23:11:38 Quit bekabug (Client Quit)
23:13:00kugelmc2739: well, clear_display does more than just stop scrolling
23:13:08 Part froggyman
23:13:24kugelscreen.stop_scroll() does only stop scrolling on the whole screen
23:14:06Unhelpfulamiconn: is that macro available in makefiles? that's where i'd be setting the CFLAGS...
23:14:11kugelbut nevertheless, I try to "deprecate" stopping scrolling on the whole display. because that also stops scrolling on unrelated screen parts (sbs for example)
23:14:31amiconnUnhelpful: Probably not. It's set in config.h
23:15:25mc2739kugel: so, if you do not wish to stop scrolling, what other options are available?
23:16:07kugelthe prefered way is to call scroll_stop for a certain viewport
23:17:12kugelthat only stops scrolling in that viewport, but it has the complication that the passed viewport has to be the one that was active (via lcd_set_viewport()) when puts_scroll() was called
23:17:22Unhelpfuli was going to start putting my tuning results into the codecs makefiles. i'll leave existing tunings unchanged on non-ARM targets, and maybe also if we're not using the new gcc, for codecs that already set -O for ARM.
23:17:32kugelfurthermore, I think the leaving screen should stop scrolling
23:18:48***Saving seen data "./dancer.seen"
23:19:11Unhelpfullooking at the generated makefiles we have CPU=arm. perhaps we could also define CPU_ARM7TDMI in the makefile when configure is run?
23:20:06Unhelpfulwhat exactly does GCCNUM represent? i can't really see how setting it to 404 for 4.4.2 EABI relates to... anything?
23:20:35B4gdermajor + minor * 100
23:20:41B4gderuh, the other way
23:20:42amiconnIt represents gcc main version and sub-version
23:20:45B4gdermajor *100 + minor
23:21:09Torneata_read_sectors says "wait at least 400ns before writing command and reading status" then has five asm nops. this is slightly worrying. :)
23:21:21Torneis that really okay? :)
23:21:28Torne(also ata_write_sectors *doesn't* have this) :)
23:22:16kugelI assume a udelay(1) would be more appropriate, or is the extra 600ns something to worry about?
23:23:25gevaertsTorne: you mean not all CPUs take the same time to execute a nop?
23:23:26Tornewell, i was more worried about hte hfact that this delay is missing in write
23:23:32Tornealso that
23:23:43kugel5 nops are 625ns, so it doesn't seem to hurt
23:24:04kugel(assuming 1 nop = 1 cycle)
23:24:08amiconnThen it's very likely not necessary. The 400ns -> 5 nops is from SH1 running at 12MHz
23:24:19amiconnThe ata specs should tell you
23:24:41kugelah no, 62.5 actually
23:25:03saratogaBa4gder: can you also update the website?
23:27:02Tornespec says writes also should have a 400ns delay, or possibly an entire PIO transfer cycle, dependingon crazy state stuff i don't really understand :)
23:27:34Tornei was gonna go with the assumption that it wouldn't hurt to have writes also wait, but "5 nops" seems a bit arbitrary ;)
23:28:26saratogaB4gder: hmm I screwed up, can you make that say "Clip v1"
23:28:33gevaertsindeed. Better make it 50 :)
23:28:54saratogaor maybe "Clip v1 (not v2 or Clip+)"
23:29:32B4gderok, me fix
23:29:48saratogai suppose I should clean up that useful logf to disk patch and commit it
23:30:18CIA-80New commit by bagder (r23688): clarify which clip
23:30:30 Join Thundercloud [0] (
23:30:56 Quit tchan ("WeeChat 0.3.1-dev")
23:31:50 Join tchan [0] (n=tchan@lunar-linux/developer/tchan)
23:32:46Unhelpfulah... ok, so i think if there's an existing tuned -O option for arm (and it differs from my and krazykit's benchmark results), i should select based on gccnum.
23:33:05linuxstbB4gder: You seem unsure whether to write "v" or "V"... ;)
23:33:12B4gderhehe, yes I am...
23:33:13Unhelpfuli'm working on being able to add a TDMI and an ARM... level? version? to distinguish between 7/9/11 etc
23:33:13CtcpPing from gevaerts!n=fg@rockbox/developer/gevaerts
23:33:22 Join bekabug [0] (
23:33:37B4gderbut I wasn't the first one I see
23:33:46linuxstbB4gder: Indeed not. There is also "Recorder V2".
23:33:58B4gderand e200 v2 vs Fuze V1
23:34:20Unhelpfulperhaps ARM_CORE would be most descriptive?
23:34:23B4gdera healthy mix! ;-)
23:34:55kugelUnhelpful: how about keeping it simple :)
23:35:09Unhelpfulkugel: and that would be?
23:35:16kugelwhat speed difference are we talking about ?
23:36:25 Quit saratoga ("Page closed")
23:36:31pixelmaJdGordon: what's the %?Rh tag in the (it looks to me that it could be an "if hwcodec recording tag" but couldn't find it on CustomWPS and you committed an extra sbs for the Archoses)
23:36:59Unhelpfulkugel: there's a chart ;)
23:37:53kugelI mean, which codec(s) need another O level on arm7 than on arm9/11?
23:39:28Unhelpfulape, ac3... aache, but that's the same codec as other aac levels, right?
23:39:37PSPdemonwhy does it seem that my gigabeat s is running better than my sansa e200 >_>
23:39:51PSPdemonoh wait wrong place to say that....
23:40:04*PSPdemon runs to #rockbox-communitt
23:42:41 Join saratoga [0] (i=9803c6dd@gateway/web/freenode/x-cdngvbdqeewpovpa)
23:53:49saratogado we have a graphic for the clip on the download page?
23:55:41*PSPdemon could make one
23:55:43Unhelpfulkugel: i think it might be worth adding the mechanism. it's not like every single codec needs to care, the bulk can just be if CPU==ARM and not go any further
23:55:58 Join Omlet [0] (

Previous day | Next day