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 2009-11-20

00:00:30 Part evilnick
00:03:23 Quit Thundercloud (Remote closed the connection)
00:03:47 Quit Lss (Read error: 104 (Connection reset by peer))
00:04:07 Join stripwax [0] (
00:07:47 Quit jdgordon (Read error: 110 (Connection timed out))
00:08:45 Join DerPapst [0] (
00:12:09 Quit petur ("Zzzzz")
00:14:14 Join gtkspert [0] (
00:14:49kugelUnhelpful: I forgot again that binutils needs patching, it's broken :/
00:16:03kugelUnhelpful: that's fixed in upstream, I guess we wait for 2.20.1
00:18:51 Quit YPSY ("Oh noes! My ZNC obviously just shut down :(")
00:19:04 Quit Biont ("ChatZilla 0.9.85 [Firefox 3.5.5/20091102152451]")
00:22:42 Join YPSY [0] (
00:23:39 Join Thundercloud [0] (
00:26:23 Quit gtkspert_ (Read error: 101 (Network is unreachable))
00:29:33 Join Blue_Dude [0] (n=chatzill@rockbox/developer/Blue-Dude)
00:30:33Blue_DudeI'm looking at doing non-destructive realtime mixing in pcmbuf, but I'm going to need to take 64K of RAM for buffer space to do it. Are there any targets where this will be a real problem?
00:39:06kugelBlue_Dude: the clip has 176400 bytes of pcm buffer
00:39:35Blue_DudeOne second of audio? Ouch.
00:39:42kugelbut it's a flash target, we can possibly increase that if needed
00:40:02Blue_DudeI need a bit more than a third of that.
00:40:19kugeland it's a fast target, it doesn't even need to boost to keep it full in most sitations
00:41:03kugelall other targets should have some 500k, I don't think those will be a problem
00:41:28Blue_DudeRight now I'm trying to allocate space. I've still got to figure out how to keep mixing in real time without killing the CPU or other threads.
00:42:32Blue_DudeIt's possible I'll have to steal time from the pcm interrupt callback. I don't know yet how that will affect performance.
00:44:55Blue_DudeI need a process that runs every 5 to 10 ticks to check up on whether a mix is necessary, then perform one if it is. Is there a mechanism to do that? Create a thread and just sleep it for a while if it's not needed?
00:45:34bughunter2spawning processes is cheap on unix, you could simply use a cron job + shell script
00:46:06Blue_DudeI mean a process within Rockbox on the target.
00:46:16bughunter2heh, of course
00:46:56 Join togetic [0] (n=togetic@unaffiliated/ibuffy)
00:50:15Blue_DudeThinking out loud: I could have a mixing thread that stays frozen until an event comes along that requires mixing (voice playing at the same time as playback), then thaws and begins a non-destructive mix. As soon as the chunk is beginning to be built, it becomes the next chunk to be read by the pcm, and this continues as long as there are multiple inputs. When we're down to one input, the...
00:50:17Blue_Dude...pcm is redirected to the actual buffered material and playback (or voice) resumes as normal, and the thread freezes.
00:50:19Blue_DudeHow's that?
00:50:31 Quit DerPapst ("Leaving.")
00:50:40kugela thread seems apprpriate
00:51:44Blue_DudeBuilding a chunk is pretty fast, but it can't be late, or stop in the middle or the dma will starve.
00:52:13kugelyour idea sounds good to me
00:52:19Blue_DudeWe don't have thread preemption do we?
00:56:57kugelwhy do you need that?
00:57:27 Quit salty-horse ("Leaving")
00:57:44Blue_DudeI don't. I want to make sure this one isn't preempted. It needs to run a procedure until it yields.
00:58:01 Join stripwax_ [0] (
00:58:20 Quit stripwax (
00:58:20 Quit rasher (
00:58:20 Quit jasio_ (
00:58:20 Quit Utchybann (
00:58:41Blue_DudeHow often does a thread have a chance to run?
00:59:01 Join Llorean1 [0] (
01:00:12NJoinstripwax [0] (
01:00:12 Join rasher [50] (n=rasher@rockbox/developer/rasher)
01:00:12NJoinjasio_ [0] (
01:00:12NJoinUtchybann [0] (
01:01:02 Quit Llorean (Nick collision from services.)
01:01:04 Nick Llorean1 is now known as Llorean (
01:02:00kugelI'm not sure here, but if it's by another thread (yield, sleep, etc) a thread can run multiple times per tick; but if you are scheduling the next run from the thread itself you're limited to multiple of ticks
01:03:08kugeleither way, often enough :p
01:04:48 Quit stripwax_ ("")
01:06:38Blue_DudeWell, OK then. All it needs to do is monitor a watermark and mix some more when it drops. As long as the buffer doesn't go dry, we're OK.
01:08:03Blue_DudeMaybe it's time to think about building smaller chunks. That increases overhead since there's some housekeeping for every chunk change, but it would increase responsiveness.
01:08:19Blue_DudeIt also means I'd need less RAM.
01:09:31Unhelpfulkugel: binutils needs patching?
01:10:19kugelit doesn't compile. there's a warning about suggesting braces for an if statement; binutils is compiled with -Werror
01:10:47Unhelpfulit compiles fine with gcc-4.4.1...
01:11:06Unhelpfulalthough, i've also still been using the last 2.20 snapshot
01:12:13 Quit bmbl ("Bye!")
01:12:21saratogaBlue_Dude: what are you working on?
01:12:38kugeland I looked at that very piece of code, ignoring the warning seems risky
01:14:14Blue_Dudesaratoga: realtime software mixer in pcmbuf for playback, voice (& maybe an effects channel)
01:14:36 Quit dfkt ("-= SysReset 2.53=- Ph'nglui mglw'nafh Cthulhu R'lyeh wgah'nagl fhtagn.")
01:14:39saratogawhat do we do now?
01:15:29Blue_DudeWe play whatever shows up in the pcmbuf by throwing it directly at the DMA. If there's voice coming in it's destructively mixed into the buffer.
01:15:51saratogawhat do you mean by destructively mixed?
01:16:18Blue_DudeMeaning that the mixed material is forever changed. You can't "unmix" it after it's done.
01:16:23 Quit stripwax (Connection timed out)
01:16:39saratogawhy would you want to do that?
01:16:49Blue_DudeWhat if you pause playback?
01:17:21Blue_DudeOr even stop playback. That also stops voice and effects.
01:17:37saratogaso if rockbox is stopped you lose the voice ui?
01:17:43***Saving seen data "./dancer.seen"
01:18:32kugelI don't think the "destructiveness" is a huge problem
01:18:48Blue_DudeNo. If you're playing a voice clip while playback is stopped, the clip plays. But if you're playing a voice clip *while* playback is playing, then the voice clip also stops.
01:18:54kugelit doesn't really matter if we'd lose the current voice clip due to (un)pausing
01:19:07kugelthe big problem is that there's no voice while being paused
01:19:11Blue_Dudekugel: it is if you try to continue a voice clip and also stop playback.
01:19:32kugelyes, but that's a very improbable situation
01:19:42kugel(I think)
01:19:54 Quit lostlogic ("leaving")
01:20:10LloreanAlso, just moving up or down an entry in the menu again gets the next voice clip playing in that situation, no?
01:20:25LloreanSo while it's a problem, it's a quirk that can be worked around rather than something that drastically hinders usability, right?
01:20:29Blue_DudeSame problem. Pausing currently paused the DMA. I'd like to just stop the input to the mixer and let the voice playback continue.
01:21:11kugelI'd rather have the bigger issue fixed first. the destructiveness can be dealt later with
01:21:56Blue_DudeThe bigger issue can't be solved without the destructiveness issue. If you stop playback, then the voice clip that has already been buffered is lost.
01:23:00Blue_DudeAnd when playback resumes, any voice or effects that were already mixed into the buffer will play.
01:23:45LloreanBlue_Dude: When resuming from *stop*?
01:24:16Blue_DudeNo. Resuming from pause. Stop clears the buffer.
01:24:27LloreanOkay, I was going to say that seemed strange.
01:24:37LloreanI was confused because the line before you were talking about "stop" behaviour
01:24:58Unhelpfulkugel: if eabi builds are possible with svn, we need to figure out the most acceptable way to tackle the enum issues in doom and mpegplayer
01:25:31kugelreplace -1 with an actual enum member seems most sensible to me
01:25:51Blue_DudeMostly it's a pause issue, but that also takes into account "pauses" such as seeks.
01:26:41LloreanWhat's the problem with seeks beyond "no voice during the seeking"?
01:27:18Blue_DudeI wonder how much audio is internally buffered in the hardware? Meaning that after it's read by the hardware, what's the latency until it's played?
01:27:20kugel" If you stop playback, then the voice clip that has already been buffered is lost." doesn't seem like a problem to me. How many clips are buffered at a time? More than 1?
01:27:48kugel"And when playback resumes, any voice or effects that were already mixed into the buffer will play." - can't that be fixed without dealing with the destrictiveness?
01:27:55Blue_Dudekugel: A stop cuts off voice playback mid-syllable.
01:28:23kugelBlue_Dude: I understand that, and I don't think it's a big deal
01:28:26Blue_DudeOnly if you resume by going back to the codec and having it decode again.
01:28:41kugelit could even be wanted actually
01:28:43Blue_DudeIOW, a resume from stop, not a resume from pause.
01:29:56Blue_DudeLlorean: none as far as I know.
01:30:34LloreanI think losing unplayed voice during stopping isn't too bad, but the issues relating to Pause are a real problem.
01:30:44Blue_DudeIn the back of my mind though, I'm thinking about event driven sound effects. Why not play a short clip for button presses, ff/rew, etc?
01:31:15LloreanIsn't there keypress sounds as an option already?
01:31:32Blue_DudeIf you decouple dma operation from codec playback, you can do a lot.
01:32:02Blue_DudeThere's a keyclick. That's also handled by direct mixing or the hardware.
01:32:35Blue_DudeI'd like to make pcmbuf_beep mix non-destructively also.
01:32:58Blue_DudeThat's less painful though, almost an afterthought.
01:33:10kugelI fear getting rid of the destructiveness is going to be a performance hit for very little gain
01:33:15Unhelpfulkugel: a result of doing that is that switches on variables of the enum type will need to handle that member, either explicitly or with a default, to avoid a warning. even if the switch code can never execute with the -1 flag value set. :/
01:33:45Lloreankugel: If it's the "best" way to enable voice while playback is paused, the gain is pretty significant.
01:33:46Blue_Dudekugel: I can't figure out another way around the pause problem though.
01:33:49 Quit fyrestorm ("Ur skills' fireproof like a wooden panel -- U got feds talking leet on your IRC channel!")
01:34:16LloreanFor iPod users especially, who aren't used to the idea that there's a "stop" function, many times they're left somewhat lost by the fact that voice seems to just stop working for them entirely.
01:34:22kugelthe pause problem is that, while being paused already, there's no way to have voice right?
01:34:33LloreanYes, if playback is paused, voice cannot play.
01:34:37kugelI don't see how you need non-destructiveness for that to fix
01:35:00 Join lostlogic [50] (n=lostlogi@rockbox/developer/lostlogic)
01:35:18kugelas I understood, the non-destructiveness is needed to not lose voice in the moment of pausing/stopping
01:36:03Blue_DudeTrue. But if you unpause during a clip, you lose it again.
01:36:16kugelUnhelpful: is that a problem?
01:36:41kugelBlue_Dude: I still think that this is not a huge problem.
01:37:31kugeleven less, since you (naturually) wait for the voice clip to finish before resuming (which could make the voice clip unhearable)
01:37:47Blue_DudeIs it possible to tell the dma to quit reading where it is and go play something else? Or is a chunk committed until its end?
01:38:19 Quit Thundercloud (Remote closed the connection)
01:38:36Blue_DudeCodec playback is ducked during voice playback. You should be able to hear it anyway.
01:38:57kugelthe big problem is getting voice while pased to work, not having it work in the very moment of pressing a button for pause/resume (which is rather unlikely to ever happen)
01:39:05Unhelpfulkugel: testing for cases that can't happen doesn't sound good, to me, for code size *or* readability.
01:39:53kugelanother switch case is surely negligible
01:40:04kugelcode size isn't really an excuse to not do that
01:40:50 Part toffe82
01:41:06Unhelpfuland readability? adding a default: or FLAG_VALUE: case to switches that *actually* already have everything covered?
01:41:10Blue_Dudekugel: that could be fixed by pausing the DMA, reading its current position in the chunk and saving that as a pause resume point. Then if a voice clip needs to play, the dma could actually stop, play the voice clip, and pause again. When unpausing, the dma is pointed to the resume point and voila...
01:41:14Blue_DudeWow, that works.
01:41:24LloreanIn most cases I don't see how someone would pause or resume while voice is playing.
01:41:55LloreanYou can't pause outside of the WPS, and resuming playback also requires returning to the WPS generally, and there's no voicing of the WPS so any voice still playing there is probably unnecessary.
01:42:03kugelUnhelpful: they obviously don't have everything covered
01:42:05Blue_DudeIt's still going to sound like crap if pausing during a voice clip though.
01:42:33kugeldefault: /* should never happen */ decreases readability? I disagree
01:44:32Blue_DudeGot to go in a few minutes, my battery is going.
01:45:06kugelBlue_Dude: having that work seems quite unimportant to me, even bad if it comes with a (performance) drawback
01:45:06Blue_Dudekugel: I'll take a look at a smarter pause feature. Thanks for thinking out loud.
01:45:39Blue_DudeYeah, it's just going to be a little clunky.
01:45:43Unhelpfulkugel: if you like that so much, i'll go with it, then ;)
01:48:13 Join jdgordon [0] (n=JdGordon@rockbox/developer/JdGordon)
01:49:51 Quit Blue_Dude ("ChatZilla 0.9.85 [Firefox 3.5.5/20091102152451]")
01:52:07kugelgcc and binutils don't seem to have roadmaps
01:59:22 Quit TopyMobile__ (Remote closed the connection)
01:59:47 Quit kugel (Remote closed the connection)
02:00:02 Quit Llorean (Read error: 104 (Connection reset by peer))
02:00:49 Join Llorean [0] (
02:03:54 Quit StealthyXIIGer (Read error: 131 (Connection reset by peer))
02:04:09 Join StealthyXIIGer [0] (n=stealthy@
02:05:52 Quit jdgordon (Read error: 145 (Connection timed out))
02:30:06 Quit Strife89DS ("ClIRC - IRC client for Nintendo DS")
02:32:22 Quit mirak_ ("Ex-Chat")
03:01:18 Quit LambdaCalculus37 ("Leaving")
03:08:17Unhelpfuldoom appears to need -Os on arm eabi
03:17:47***Saving seen data "./dancer.seen"
03:24:38 Quit StealthyXIIGer (Read error: 131 (Connection reset by peer))
03:24:41 Join StealthyXIIGer [0] (
03:41:10 Join jdgordon [0] (n=JdGordon@rockbox/developer/JdGordon)
03:43:58 Quit StealthyXIIGer (Read error: 104 (Connection reset by peer))
03:44:01 Join StealthyXIIGer [0] (
03:47:55 Quit StealthyXIIGer (Read error: 104 (Connection reset by peer))
03:48:01 Join StealthyXIIGer [0] (
03:48:27 Quit StealthyXIIGer (Read error: 104 (Connection reset by peer))
03:48:49 Join StealthyXIIGer [0] (
03:51:13CIA-80New commit by unhelpful (r23670): Fix warning on new gcc for checking value or truth of undefined _WIN32 and MIPS.
03:51:19CIA-80New commit by unhelpful (r23671): Short enum fixes for doom plugin.
03:51:25CIA-80New commit by unhelpful (r23672): Disable -mlong-calls for ARM EABI.
03:52:22Unhelpfularm eabi doom does *not* need -Os if you remember not to use -mlong-calls :P
03:56:53 Join T44 [0] (
04:00:15 Join CaptainKewl [0] (
04:02:49 Quit BlakeJohnson86 ("Leaving.")
04:03:09 Join BlakeJohnson86 [0] (
04:05:02Tomisoh man , playing doom on the nano is an exercise in absurdity
04:05:24 Quit jdgordon (Read error: 110 (Connection timed out))
04:05:50kkurbjunUnhelpful: are the eabi builds stable?
04:08:23 Quit Rondom (Nick collision from services.)
04:08:34 Join Rondom [0] (n=Rondom@
04:09:11 Quit StealthyXIIGer (Read error: 110 (Connection timed out))
04:11:16Unhelpfulkkurbjun: i don't think anybody has committed a work-around for the section offset issue we've seen with the newer binutils. this has something like 7/8 chance of breaking any particular iram-using codec or plugin.
04:12:00 Quit TheSeven (Nick collision from services.)
04:12:16kkurbjunso it works with older versions of binutils?
04:12:19 Join The_Seven [0] (n=theseven@rockbox/developer/TheSeven)
04:12:31 Nick The_Seven is now known as TheSeven (n=theseven@rockbox/developer/TheSeven)
04:15:15 Quit Topy44 (Read error: 110 (Connection timed out))
04:15:15Unhelpfulolder binutils don't fully support the relocation types used in eabi objects. i think you need at least 2.19.
04:17:04Unhelpfuli filed a bug report for the section offset bug against binutils 2.19pre but was never able to produce a small test case, preferably one C file and one linker script, to demo the bug, so it's not *too* surprising that they've done nothing at all. :/
04:17:59Unhelpfulalso i have 4 warnings in mpegplayer to fix, and at least one is potentially behavior-changing
04:19:07kkurbjundo you have a link to the bug report, I would be interested to read more about what you are seeing with the offsets
04:19:25Unhelpfulone of the "always false" warnings in doom was for a loop exit condition. that really should be promoted to an error if there's no break in the loop :/
04:22:03kkurbjunwas that a loop that you fixed in the last few commits you did? Doom was kind of hacked to add exits for loops - before it relied on exit(), but that was not present in the rockbox api till recently.
04:23:16Unhelpfulkkurbjun: this was a fixed-length loop. it's just that the comparison used to terminate it is always false with short enums
04:23:28Unhelpfuland yes, it's one i just did
04:23:39kkurbjunahh, I see
04:23:51Unhelpfuli've not found the binutils bug yet, but here's the that i pastebinned a while ago:
04:24:51Unhelpfulcompare the address at which iramcopy is set with the load address of .iram - these are *supposed* to be the same, and are if the end of the previous section happens to align the right way.
04:26:03 Join cowgarden [0] (
04:26:07kkurbjunoh interesting
04:26:13Unhelpfulhere's my poor bug report:
04:26:34cowgardenMy Iaudio7 broke so I had to buy a D2, now I'm in the boat!
04:26:37kkurbjunit looks like it is using the unaligned end of the last section
04:27:33kkurbjundoes .ncdata actually contain anything? what is that section used for?
04:27:43Unhelpfulkkurbjun: yes, but sometimes it gets it right. the bizarre thing is that the .iram load address is assigned directly from iramcopy, not calculated separately.
04:28:23Unhelpfulso it's a matter of val1 = expr (works), val2 = val1 (fails). wtf? ;)
04:32:12Unhelpful.ncdata is for non-cacheable data. it's used in core for the tick counter on multicore targets, and i don't recall what else. no codecs or plugins use it, but they certainly could...
04:32:29cowgardenwhere do I download plugins like rockboy?
04:32:45kkurbjuncowgarden: they are all included in the
04:32:50kkurbjunthe main download
04:33:05cowgardenall? ok I'll check , thx :) must have overseen rockboy
04:33:49kkurbjunUnhelpful: does the bug just happen ot occur around ncdata.. I mean if you mode that to a different part of the linker script does it have a psitive effect?
04:34:24kkurbjuncowgarden: you won't see it in the games list, you just select the .gb file in the file browser you want to play and it will start up rockboy.
04:34:34cowgardenah, I remeber :)
04:34:47Unhelpfulkkurbjun: removing .ncdata works, as do a few other tricks i tried. some things you might *think* would work don't.
04:38:10kkurbjunhmm, it looks like it happened around .ncbss too in that pastebin link you had
04:38:39kkurbjunWould this be an issue with 0 size sections?
04:39:43Unhelpfulkkurbjun: i suspect it has something to do with 0-sized sections, yes. i'm rather at a loss as to how to fix.
04:39:58Unhelpfulother than work-arounds and hacks, anyway :)
04:40:04cowgardenhow to show hidden folders?
04:40:05kkurbjunso the glue sections are 0 size also
04:40:24kkurbjunbut they are not manipulating the memory pointer
04:40:28Unhelpfulif you assign iramcopy, with the same expression, *inside* .ncdata, it works.
04:41:01kkurbjunit seems that both .ncdata and .ncbss are zero and change the . memory address...
04:41:23 Join BHSPitMonkey [0] (n=stephen@unaffiliated/bhspitmonkey)
04:43:42kkurbjunUnhelpful: do we need the .nc sections?
04:44:26Unhelpfulno plugin or codec uses them, but amiconn was strongly against removing them, and would rather see the bug fixed upstream or a different work-around used.
04:45:37 Quit liar (Client Quit)
04:54:37kkurbjunUnhelpful: does the behavior change if you change the memory pointer before the .ncdata section? I'm not sure if you can do this, but I was thinking . = . + 0x1000000 right before the .ncdata section
04:59:18kkurbjunUnhelpful: we could add an elf loader :-D, then we wouldn't need to have quite such crazy linker scripts
05:00:50kkurbjunthat would allow us to have objects with non-linear memory layouts
05:05:20cowgardenoh thank you all so much for this wonderful peace of software, goodnight!
05:05:30 Quit cowgarden (Client Quit)
05:17:49***Saving seen data "./dancer.seen"
05:20:18 Quit AaronM ("Emo Time In My Corner... //_-")
05:26:00 Quit at0m|c ("Lost terminal")
05:28:50 Join phanboy4 [0] (
05:44:46 Join jdgordon [0] (n=JdGordon@rockbox/developer/JdGordon)
05:48:22 Quit Horscht ("Verlassend")
05:54:41 Quit MethoS- (Remote closed the connection)
06:01:12 Quit bubsy ()
06:04:12Unhelpfulkkurbjun: couldn't we then also have relocations to handle calls into core, with the callsites in the plugin code given the correct offsets on load?
06:08:32 Quit jdgordon (Read error: 110 (Connection timed out))
06:10:16 Join jdgordon [0] (n=JdGordon@rockbox/developer/JdGordon)
06:10:54 Quit jdgordon (Client Quit)
06:11:36 Quit fdinel ("Miranda IM! Smaller, Faster, Easier.")
06:11:52 Join JdGordon [0] (n=jonno@rockbox/developer/JdGordon)
06:22:12 Quit Strife89 ("Need sleep...")
06:23:21 Join Tomis2 [0] (n=Tomis@
06:24:28JdGordonWTF?! my fuze has the label "t|%xdKu|%xd" !
06:24:49Unhelpfulweird. amiconn's suggested fix of moving the second . = ALIGN(CACHEALIGN_SIZE) out of .ncdata produces *map* files that don't show the offset bug... but the codecs still break in what appears to be the same way
06:24:58JdGordonthis is the OF usb... no idea where that came from.. but it looks like part of a skin
06:25:07kkurbjunUnhelpful: I'm not sure offhand, but I believe so, that could make standalone plugins and codec's more feasible too
06:25:38kkurbjunI eman, a plugin that is distributed outside of the main rockbox build
06:25:43kkurbjunmean that is
06:26:21Unhelpfulhrm. whenever i've attempted to distribute a plugin, despite having it built from the current svn rev, it has failed to work properly on the recipients' devices :/
06:26:25 Quit Tomis (Read error: 60 (Operation timed out))
06:26:25 Nick Tomis2 is now known as Tomis (n=Tomis@
06:27:17Unhelpfuli had gevaerts try to run a custom benchmark plugin for some changes in the bitmap scaler. despite both of us using current svn, it wouldn't work until i sent him a complete build to install.
06:27:41kkurbjunthat's a bit strange, I would expect that if you were both on the same revision it would work
06:28:19kkurbjunI can't think of what would cause that unless the api structure was somehow setup different in memory
06:28:46kkurbjunJdGordon: did you have a patch for me to try
06:29:16kkurbjunthere are some bugs when exiting plugins with the ui viewport set in the sbs by the way that kugel says are related to a change you made
06:29:57kkurbjunif you use the latest Cleangreen theme on the MR500 and enter/exit a plugin you will see the rendering issues
06:30:06 Join Tomis2 [0] (n=Tomis@
06:30:20JdGordonbut apparently its screwey on target and fine in the sim
06:31:17JdGordonand yeah, sbs being broken after exiting plugins is known
06:33:22kkurbjunI'll give that patch a try sometime
06:33:24*JdGordon sees why its broken on target
06:33:40kkurbjunwhy is that, I was just looking for that?
06:35:07JdGordonthe skin buffer isnt init yet
06:35:14JdGordonalthough not entirely sure why it works on sim
06:35:22JdGordonneed to fix the init order
06:35:35kkurbjun:), I wouldn't have seen that
06:38:44JdGordonyay spagetti code
06:43:05JdGordonsweet... simple and incorrect (in the long run) fix
06:46:51 Quit Tomis (Read error: 110 (Connection timed out))
06:46:52 Nick Tomis2 is now known as Tomis (n=Tomis@
06:49:16 Join bluebrother [0] (
06:51:31 Quit Tomis (Read error: 131 (Connection reset by peer))
06:51:47 Join bubsy [0] (
06:52:29 Join Tomis [0] (n=Tomis@
06:56:16Unhelpfulthis seems to be the only hack, of many i've tried, that actually fixes the iram offset issue on newer binutils. acceptable for commit?
06:57:56 Quit CaptainKewl ("( :: NoNameScript 4.22 :: )")
07:01:18JdGordonkkurbjun: if you can be bothered... could really use some testing :) otherwise ill just coimmit it so I can move on to the next bits of it
07:03:43 Quit dmb (Read error: 104 (Connection reset by peer))
07:06:14 Join n1s [0] (n=n1s@rockbox/developer/n1s)
07:06:16 Quit bluebroth3r (Read error: 110 (Connection timed out))
07:12:13JdGordonah crap
07:17:52***Saving seen data "./dancer.seen"
07:19:33JdGordonright, this patch is "ready" for commit.... it does cause some minor glitches (which were sort of always there anyway) and they should be simple-ish to fix... but those fixes are heavily dependant on this patch, and I dont want to do both big changes in one commit
07:19:51JdGordonand I dont want to work on the 2nd patch without the first being in svn....
07:20:15JdGordonso do I just annoy people and commit early? or make it harder later and do it all in one big commit?
07:22:40pixelmado you keep something built-in for early USB ( I seem to remember that was necessary) and has that been tested?
07:23:26JdGordonits there... I need to charge my rec to make sure it works
07:23:32JdGordonit did when I tried a few days ago
07:27:30JdGordonthere goes that dilema.... totally not working :p
07:33:29JdGordonhow nice... stupid sbs doesnt work at all on the ajbr
07:34:22JdGordonit does get loaded though! 4kb of images
07:38:39 Join einhirn [0] (
07:38:42JdGordonpixelma: the images you fixed definelty load on the archos?
07:42:37JdGordonah crap
07:42:52JdGordonstupid .sbs is only for 16bit displays
07:43:33UnhelpfulJdGordon: use git, get both ready, and then commit x2? ;)
07:43:41JdGordonyeah... no :p
07:44:44Unhelpfulit's nice to be able to do batch commits :)
07:46:03JdGordoni cbf trying to learn git
07:46:07 Join AEnima1577 [0] (
07:47:58 Join Blue_Dude [0] (n=chatzill@rockbox/developer/Blue-Dude)
07:48:14Blue_DudeHey, throw in a fix for highlighted lines persisting on the USB connect screen...
07:48:49Blue_DudeI'd do the commit incrementally. Sure, you'll be busy with bug fixes for a few days, but so what?
07:50:00Blue_DudeAnyway, my 2 cents.
07:50:03 Quit Blue_Dude (Client Quit)
07:58:34UnhelpfulJdGordon: how *much* work do you expect between these two patches? i'd say if the "fix" part is fairly trivial then go ahead and commit the first part, it can be reverted if the fix turns out to be ugly.
08:04:19amiconnUnhelpful: No codec or plugin uses them *atm*, that doesn't mean it will stay this way
08:05:13JdGordonUnhelpful: more than trivial
08:05:16Unhelpfulamiconn: right, and the pastie i linked to has a work-around. oddly, a number of *other* work-arounds result in a "correct" mapfile but a codec that is still broken
08:05:25CIA-80New commit by jdgordon (r23673): fix mono and hwcodec classic_statusbar so they actually load
08:06:06UnhelpfulJdGordon: duplicate svn tree? svk perhaps? (i don't know if you can "create" an svk local repo from an upstream svn server, but it seems like that would be perhaps the most desirable feature)
08:06:13 Quit AEnima1577 ("Leaving.")
08:06:27JdGordonmore effort than its worth
08:06:56JdGordonI tihnk its too late to consider doing it tonight now.. which means it wont be commited till the weekend anyway
08:08:34pixelmaabout BlueDude's statemen - it's not highlighted lines in general, only scrolling ones (and lines only scroll when they are highlighted and longer than would fit on the display). Besides that I don't understand this statement completely... is it how JdGordon should do? Because the issue has nothing to do with the statusbar but with errors when viewportifiying the USB screen
08:09:17JdGordonyeah, I'm not too happy about the usb screen code.. it has some issues which need fixing
08:09:48amiconnUnhelpful: That fix looks weird. Defining iramcopy twice? And !NOACHE_BASE?
08:09:52 Quit phanboy4 ("Leaving")
08:09:57 Join stoffel [0] (
08:11:44Unhelpfulamiconn: it's ensured earlier on that NOCACHE_BASE is always defined. essentially the fix defines iramcopy exactly as before if NOCACHE_BASE is 0, but defines it inside .ncdata if NOCACHE_BASE is non-zero
08:12:12Unhelpfuli suppose the NOCACHE_BASE-is-zero case could remove any reference to NOCACHE_BASE aside from the #if, and just read as iramcopy = .
08:12:56Unhelpfulor maybe you were just suggesting that NOCACHE_BASE == 0 would be more readable?
08:13:13pixelmaJdGordon: does the classic_statusbar work on greyscale then? Greyscale uses different viewport parameters too - 4 shades of grey instead of RGB colour...
08:13:26JdGordonit should... they are all -'s
08:14:00 Join linuxstb [0] (n=linuxstb@rockbox/developer/linuxstb)
08:14:32pixelmaok, on greyscale fore- and background shades aren't settable yet though
08:14:39amiconnUnhelpful: The latter, or just use an #else block
08:15:13amiconnA comment explaining the trickery is probably a good idea as well
08:15:39Unhelpfulindeed! #elif defined(IRAMSIZE) would do the trick, and yes, a comment is an excellent idea :)
08:16:37amiconnpixelma, JdGordon: Does that mean the 'classic' status bar isn't really looking like the real thing on anything but mono?
08:16:40Unhelpfulalso it looks like IRAMSIZE can be defined as 0... it's probably harmless but should there even be a .iram or iramcopy in that case?
08:16:57JdGordonamiconn: ?
08:17:26amiconnThe classic bar used grey in tthe charging animation to distinguish the moving part from the current battery level (shown in black)
08:17:33 Join AEnima1577 [0] (
08:18:38JdGordonthere is no charging animation in that sbs
08:18:42JdGordonfeel free to add it
08:18:44Unhelpfulhrm, actually .iram doesn't exist in that case, and iramcopy is harmless unless something actually references it...
08:19:46JdGordonfeel free to add it
08:20:59amiconnDid I remove it?
08:21:12*amiconn dislikes this removal of working features
08:23:32JdGordonnothing is removed
08:23:37JdGordonits just not reimplemented
08:24:34JdGordonalthough, in all honesty.. I dont know if an animation while chargin is possible with skins... might be something to look into
08:25:13JdGordonbut you're going to have to come up with a much better reason to stop this commit happening than that...
08:26:20 Quit TheSeven ("ChatZilla 0.9.85 [Firefox 3.5.5/20091102152451]")
08:26:42 Join dmb [0] (n=Dmb@unaffiliated/dmb)
08:26:44JdGordonit would probably have to be done with some conditional viewport magic and sublines
08:26:50Unhelpfulamiconn: this looks better, i think:
08:27:45Unhelpfuland that will be everything needed for "anybody" to play with eabi, except that i still need to fix mpegplayer, and somebody needs to decide on a means for building a toolchain.
08:29:52amiconnThat does indeed look more readable
08:31:31 Join Zagor [242] (n=bjorn@rockbox/developer/Zagor)
08:31:34amiconnJdGordon: I deem the charging animation a rather helpful feature. Removing features without having an adequate replacement is a bad idea, imo
08:32:08JdGordonthe charging animation is a nice addition... but an icon showing its charging does the same
08:32:25amiconnIf the skin stuff doesn't allow to have a charging animation, it looks to me like a step back instead of forward
08:32:40amiconnAn extra icon needs extra space
08:33:20amiconn...and is less intuitive. All mobile devices I've seen so far indicate charging by means of a charging animation
08:34:19amiconn(although they're often showing less useful information that the rockbox charging animation)
08:37:56JdGordonthe icon doesnt use any extra space that something else was using
08:38:41amiconnHmm? How would that work?
08:39:00CIA-80New commit by unhelpful (r23674): Fix codec and plugin issues on ARM EABI caused by .iram sometimes being assigned an incorrect load address.
08:42:00 Quit grndslm (Client Quit)
08:42:06 Quit dmb (Read error: 60 (Operation timed out))
08:42:38 Quit SUSaiyan (Read error: 113 (No route to host))
08:43:19 Join flydutch [0] (
08:43:29 Join dmb [0] (n=Dmb@unaffiliated/dmb)
08:44:46 Join bertrik [0] (
08:45:25pixelmaJdGordon: the way it worked in the built-in can not be reproduced with current tags (because you can't get the current level (I think) only charging or not. I'm not sure at the moment - didn't the classic also have a grey animation on mono screens?
08:46:11JdGordonsure you can get current level and charing yes/no
08:46:32JdGordonyou cant get it precise, but maybe we add a new tag for that
08:46:46JdGordonif its really that important (pretty low on my priority list though...)
08:47:50pixelmathe also doesn't look like the original thing yet (but I'd like to work on it. One thing I already noticed is that it probably won't be possible to have different disk activity icons depending on whether the target is flash or hard disk based
08:48:28JdGordonsure you can... just use a different image
08:49:00pixelmayes, and how do I make the skin parser aware if the target is flash or hard disk based?
08:49:00JdGordonclassic_statusbar is only meant to be the lowest common denominator to keep it simple... everyone is welcome to go nuts and redo it for each target
08:49:23JdGordonyou woudn't do it in the parser
08:49:40JdGordonor add a tag (which imo is a waste)
08:49:50 Join grndslm [0] (
08:50:36pixelmaamiconn: different battery icons if it is charging or not (using the same spot)
08:51:24JdGordonthere is a 8pixel (or so) gap between battery and volume indicators
08:51:27JdGordonthe icon goes there
08:52:28pixelmathe gap is probably because your version is wrong (and not as fine grained as the classic) and why would it need a different icon?
08:53:12Unhelpfulhrm, i assume that if adding an arm-eabi toolchain target to requires a new gcc patch (and it does) that that can also be checked in via svn?
08:54:21Unhelpfulalso would calling it "e" for now (in terms of option needed to build it) be acceptable? it probably goes away in the future if we switch to EABI for general use on ARM.
08:54:22JdGordonwhy would what need a different icon?
08:54:41 Join maruk [0] (
08:54:46JdGordonand how can it be wrong? I also never said it was a 1:1 pixel perfect replacement
08:55:14amiconnThe rockbox charging animation conveys both the fact that it's charging and the current battery level, with no extra space
08:55:31 Join Rob2223 [0] (
08:56:02pixelmaI'll try that out, could be possible with the current available "skin" tags
08:57:03amiconnDoes the skinning engine allow to use arbitrary fractions of a bitmap?
08:57:12JdGordonnot yet
08:57:51amiconnI can't find a gap where an additional icon would fit
08:59:06JdGordonhave you actually tried yet?
08:59:29amiconnBetween battery and volume indicators is where the charger/ usb power icon is shown
08:59:32pixelmain the current sbs there is one... but as I said it's not that classic
08:59:55JdGordonso add almost_ onto the name
09:00:09amiconnThose icons are also important
09:03:04pixelmacharger icon is already possible, USB power not yet but should be added (JdGordon said earlier that it is probably quite easy, my question is if it should use a different tag or reuse the charging on (depends on if there are targets that have both not either or)
09:03:42pixelmaand charging is already indicated
09:04:22amiconnJdGordon: No, I didn't try that patch yet. I fail to see why I should try a patch that isn't feature complete, adds nothing of value to me, but in turn removes features I'm relying on
09:04:45JdGordonbecause its being forced on you is a pretty good reason I tihnk
09:04:57JdGordonand coimplainig before trying it out is poor form
09:05:34 Join petur [0] (
09:05:38JdGordonanyway... I'm going to bed... but before I do... would you rather I spend time fixing the skin (which anyone with simple skin understanding can do)... or fix actual bugs?
09:06:08amiconnNo, that would be a good reason to either stop the force, or working around it
09:07:57pixelmaas I said, I'll work on getting the sbs as close to the original as possible currently (and there won't be missing much)
09:08:11pixelmamaybe nothing
09:08:24 Join liar [0] (n=liar@
09:08:25JdGordonthis is the first time I've heard complaints from you... and I've been very open about my plans here....
09:13:24 Quit Rob2222 (Read error: 110 (Connection timed out))
09:14:12amiconnI'm not against the skin engine as such, as it seems there are people interested in it. But what I absolutely don't like is that you're obviously thinking that it is okay to take away features which *you* think are not important
09:15:10JdGordonmeh, if it was important at all to the themer community then there would have been a request ages ago
09:16:15JdGordonthe loss of the animation is tiny because the exact same info can be got with the charge % and "charign?" tags
09:16:38JdGordonbut anyway.. there are so many more ACTUAL SBS BUGS that the fact you want to leave because of this one boggles my mind
09:17:06JdGordonand 90% of those bugs will disappear once this is in and the follow up patch is in
09:17:45*amiconn would throw out most of this skinning stuff in case of a fork, but would actually prefer to avoid going that route
09:17:54***Saving seen data "./dancer.seen"
09:19:12pixelmaJdGordon: charge % ? I only know of current level % (and the conditional)
09:19:37JdGordonisnt charge level the percent?
09:19:43pixelmaand charging yes/no - which probably can be combined in a way to make the animation work
09:20:09pixelmawith sublines
09:20:13 Join Grahack [0] (
09:20:52JdGordon(11:26:53 PM) JdGordon: it would probably have to be done with some conditional viewport magic and sublines
09:21:52n1sJdGordon: i don't care about themes that much but i understand amiconns point,
09:21:59pixelmawhy would I need conditional viewports there?
09:22:38JdGordonjust saying I said it was possible an hour ago.. that was my initial answer how to do it
09:23:13n1s(sigh) ... i usually don't check patches i don't have any big interest in but then if they change aomething i do use i usually dont' find out untill they are in because the people interested in that patch didn't care about this other feature
09:23:19 Join Thundercloud [0] (
09:24:57JdGordon(sigh and yawn) the feature is a silly little animation, which can be duplicated, which isnt a core feature, and most people will have disabled anyway... its not like I've gone and removed voice or something
09:26:28AlexPJdGordon: I agree that for me it doesn't matter, but I don't think anyone should presume to know what is important for others
09:27:04n1sJdGordon: i don't mean this as criticism agianst you but it seems to be a common problem of larger patches that do a lot of changes, the people interested in the core of the patch see the other things the pach does (for whatever reason) as unimportant and the people who do care abotu these unimportant changes don't know about them untill patches go in
09:27:36pixelmaI wonder where you take the knowledge from that most people will have disabled the classic statusbar, on small screen targets it's still quite nice and at the start it should be the same. As I said earlier (and even in the commit) I'll work on it
09:28:41pixelmaduring the weekend, got to work now actually...
09:29:23JdGordonmy seemingly briliant reasoning is.. how many people are using the inbuilt wps?... and before anyone can refute that.. im going to bed.. for real this time
09:29:43AlexPThat matters, but isn't the be all and end all
09:29:50pixelmaWPS and statusbar are two different things
09:30:14JdGordonnot any more
09:30:19JdGordonthats the whole point of this
09:33:59pixelmaI thought the point of this was to get rid of the built-in thing
09:34:07JdGordonthe line between the wps and the sbs and the theme and the skin is alll getting very blurry
09:34:35pixelmaI won't be able to chose statusbar and WPS independently from each other?
09:34:55JdGordoncourse ytou will
09:36:05pixelmaso? Then there should be the used as default as long as there is no cabbiev2 replacement for it yet (or other default theme if there ever will be one)
09:36:26pixelmaand then it is still two different things
09:36:34JdGordonand if everyone was happy with cabbie there would be no need for
09:36:54JdGordonif the problem is there is no skin tag to do this then file the bug...
09:37:30JdGordonif the problem is the quick reimplementation I did isnt good enough.. then you (being anyone...) know what to do
09:38:29martian67cpmplain untill it magically improves >_>?
09:39:16*martian67 hides
09:42:24 Join kugel [0] (
09:43:10 Join Bagder [0] (
09:43:15kugelit is possible to do different charging and discharging animations with skins already
09:43:56kugelone of my themes does that (azure ultimate I think)
09:44:00 Quit bertrik (Read error: 113 (No route to host))
09:44:51 Quit freqmod_qu (Remote closed the connection)
09:45:20 Quit kugel (Remote closed the connection)
09:45:22 Quit Thundercloud (Remote closed the connection)
09:59:06 Join pamaury [0] (n=pamaury@
09:59:20 Quit stoffel (Remote closed the connection)
09:59:53Unhelpfulhrm, success with a minimalist change to svn to add an "e" option to to build arm-elf-eabi w/ gcc-4.4.2 and binutils-2.20. binutils requires −−disable-werror to compile, and building gcc needs some magic... i'll make sure it doesn't destroy anything else :)
10:07:32 Quit einhirn (Read error: 110 (Connection timed out))
10:13:55 Join LinusN [0] (n=linus@rockbox/developer/LinusN)
10:16:26 Join freqmod_qu [0] (
10:17:52 Quit Grahack ("Leaving.")
10:19:34 Quit mt (Read error: 104 (Connection reset by peer))
10:20:05 Join mt [0] (n=mtee@rockbox/developer/mt)
10:20:06 Join pondlife [50] (n=Steve@rockbox/developer/pondlife)
10:20:25 Join DerPapst [0] (
10:20:38pondlifeAnyone read that book - "101 uses for a bricked H300" ??
10:20:58*pondlife might write it
10:21:45pondlifeJust wanted to warn any others not to try an SVN H300 bootloader until I've post-mortemed.
10:22:15pondlifeWhite screen, no button recognition
10:22:18pondlifeTime to eBay
10:24:02Unhelpfulare they *that* hard to recover? :/
10:24:42pondlifeI posted on the ML what I did - maybe I missed some step out... don't think so though
10:25:07pondlifeUnhelpful: They need hardware that only lives in Sweden
10:25:19pondlife(Or something like that.)
10:25:41*Unhelpful will remain content with his nigh-unbrickable beast
10:26:10Unhelpfulat least until somebody is ready to attempt rockbox-in-flash on it
10:26:28pondlifeI am thinking a beast might be a good idea - will it take an MK8007GAH?
10:27:01pondlifeBut only if there's nothing else I can do that doesn't involve pestering others
10:27:41Unhelpfuli don't know which drives will fit, sorry
10:34:45*petur wonders if LinusN really has time to unbrick his h120 *and* pondlife's h300
10:35:17pondlifeRelated cause in the code?
10:35:32pondlifeI have to go to work now, just wanted to warn
10:35:36peturjust my stupidity
10:36:58 Join av500 [0] (
10:39:51 Quit mikroflops (Remote closed the connection)
10:40:10 Join mikroflops [0] (n=yogurt@
10:51:22amiconnpondlife: No. Beast has ZIF, MK8007GAH is 50-pin (as the H1x0, H300, iAudios and iPods up to G4 need)
10:52:29 Quit pamaury ("exit(*(int *)0 / 0);")
10:52:36 Part pondlife
10:56:25 Quit BHSPitMonkey (Remote closed the connection)
10:56:29 Quit DerPapst ("Leaving.")
11:04:02CIA-80New commit by unhelpful (r23675): Patch supporting gcc-4.4.2, disabling exception support in libgcc and enabling the same multilib options as other versions.
11:07:19Unhelpfulhrm, why can't i download the patch i just added to www/gcc?
11:07:21 Join fyrestorm [0] (
11:09:00Unhelpfulthe server doesn't have a thing about filename lengths or something, does it? or need me to set some property first?
11:09:23Bagderwhere do you try to download it from then?
11:09:34Bagderweb site? tries to fetch:
11:10:18ZagorUnhelpful: the web page is not automatically updated from svn. you need to shout to me or bagder to "svn up" it
11:10:31*Bagder svn up'ed the gcc dir
11:10:43Unhelpfulah! i was not aware. thanks. :D
11:11:42Unhelpfulworks now. i've built everything else, once i verify i've not broken mips i'll push the changes... should make it *much* easier for anybody to try this eabi stuff. :)
11:17:56***Saving seen data "./dancer.seen"
11:24:32 Quit ej0rge (Read error: 145 (Connection timed out))
11:27:55CIA-80New commit by unhelpful (r23676): Add arm-elf-eabi target to, selected with "e".
11:28:59 Quit fyrestorm ("lamers envy me like they envy bill g -- main boot xp, just the way it should be!")
11:35:54 Part LinusN
11:35:57 Join LinusN [0] (n=linus@rockbox/developer/LinusN)
11:41:36 Quit liar (Read error: 113 (No route to host))
11:41:49 Join MethoS- [0] (n=clemens@
11:44:18 Quit MethoS- (Remote closed the connection)
12:02:20 Quit Kopfgeldjaeger (Read error: 113 (No route to host))
12:12:36 Quit bubsy ()
12:48:03 Join seani_ [0] (n=seani@
12:48:10 Quit seani (Read error: 110 (Connection timed out))
12:58:13CIA-80New commit by unhelpful (r23677): Merge mpeg_malloc_reason_t and mpeg2_alloc_t, add STATE_INTERNAL_NORETURN to mpeg2_state_t, to fix warnings with short enums.
12:59:26 Join Horscht [0] (n=Horscht2@xbmc/user/horscht)
13:13:57 Join hebz0rl [0] (
13:17:57***Saving seen data "./dancer.seen"
13:20:42 Join watto [0] (n=watto@
13:27:34 Join salty-horse [0] (
13:28:31 Join DerPapst [0] (
13:32:34 Join teru [0] (
13:42:45 Join cowgarden [0] (n=qgar10@
13:43:40cowgardenhi, as the D2s touchscreen seems to be working really good, how hard is it to write plugins like the exploding/expanding pixel thing? I'd really like ot have a piano
13:47:56gevaertscowgarden: have you looked at the source?
13:48:21cowgardengevaerts, no, I never programmed C or anything :) I just know a bit java and actionscript
13:49:18cowgardenbut I guess it should be quite easy to extract and modify what I need. Is multiple sound output supported?
13:49:54cowgardenI'd just have to change the click to button areas and assign a soundevent somehow I guess
13:50:36cowgardenand since there is allready a collision system I could simply take that (that way drag might be supported to for those keyboard shwooshes :)
13:50:41gevaertscowgarden: well, most (or all?) current plugins are written in C, but you can write them in lua these days
13:51:08cowgardenwhen all are C I'd stay with C to be able to C&P
13:51:37gevaertsyou also need some in P then :)
13:52:59cowgardenare multiple sounds at the same time possible?
13:53:04gevaertsThere's no sound mixing though. Rockbox is still primarily an audio player. Someone (Blue_Dude) is working on this, but I don't know if that work will be general enough for this sort of thing. The primary purpose is to be able to mix in voice clips and key clicks with the normal audio stream
13:53:48cowgardenah, ok, well I was pleased with some 6bit single tone sound :) I just need something responsive to check out some melodies
13:54:06*gevaerts adds a note to the effect that he isn't a sound specialist
13:54:10cowgardenand the original firmware allwas lags when you play sound in flash lite
13:54:18gevaertsYou could do your own mixing I uess
13:54:49cowgardenyeah, but that was way to much work, sadly I have a job and a portfolio for the university to finish :)
13:55:41cowgardenare the emulators all ported to rockbox or written specially for it?
14:00:43Bagderthey are ports
14:00:51cowgardenah, k thx
14:08:59 Quit antil33t1 (Read error: 54 (Connection reset by peer))
14:14:10 Join antil33t [0] (
14:22:08 Quit T44 (Read error: 104 (Connection reset by peer))
14:25:37n1scowgarden: you can do your own miging pretty easily by just adding and shifting samples from the different sounds, maybe not the prettiest solution
14:25:55 Join Topy44 [0] (
14:33:47 Quit seani_ ("Leaving")
14:34:04 Join seani [0] (n=seani@
14:34:08 Join seani_ [0] (n=seani@
14:34:32 Part seani_ ("Leaving")
14:36:03Unhelpfuln1s: you can build an arm-elf-eabi toolchain by telling to build for "e", and it will install alongside your normal toolchain, since they all get target-named subdirectories. you'll need to run configure with −−eabi to get set up to build with it.
14:36:15 Join liar [0] (n=liar@
14:36:19Unhelpfuland there are probably still some targets for which work needs done
14:36:34n1sok, i'll try that
14:38:03cowgardennls, thx
14:39:28Unhelpfulgigabeat f needed a change due to adr being invalid across section boundaries... i'm not sure what else might be an issue. with the recent svn commits things *should* work on at least e200, the only thing i've had handy to test on.
14:41:52n1smaybe it's time to upgrade coldfire gcc too soon :)
14:44:11markunI wonder how clang will do on the ARM targets.
14:44:30markunnot sure if they support the inline asm yet
14:45:27markun(clang of llvm)
14:45:48n1sUnhelpful: i'll let you know how it works out on my beast once the netbook finishes building the toolchain :)
14:46:38Unhelpfuln1s: yes... i need to locate a mini-usb cable, my beast is not good for much lately, unless playing music counts.
14:47:35gevaertsUnhelpful: you use a DAP to play music?
14:55:12 Quit Rob2223 ()
14:56:30Unhelpfulthere's also a warning in lua, but i don't believe it's a big deal...
15:17:41 Quit teru ("Quit")
15:18:01***Saving seen data "./dancer.seen"
15:23:21 Join dfkt [0] (i=dfkt@unaffiliated/dfkt)
15:30:13 Quit salty-horse ("Leaving")
15:32:15 Join T44 [0] (
15:37:50 Quit Topy44 (Read error: 60 (Operation timed out))
15:39:44 Quit AEnima1577 ("Leaving.")
15:51:42 Quit antil33t (Read error: 104 (Connection reset by peer))
15:51:55 Join antil33t [0] (n=Mudkips@
15:52:03 Join ender` [0] (i=krneki@
15:56:47 Join Ubuntuxer [0] (
15:59:50 Join evilnick [0] (i=0c140464@rockbox/staff/evilnick)
16:13:12 Quit antil33t (Read error: 131 (Connection reset by peer))
16:15:23 Join antil33t [0] (n=Mudkips@
16:17:22 Quit ender` (Read error: 131 (Connection reset by peer))
16:18:39 Part Bagder
16:18:57 Quit n1s (Read error: 60 (Operation timed out))
16:31:12 Quit antil33t (Read error: 131 (Connection reset by peer))
16:34:19 Join n1s [0] (
16:36:30 Join antil33t [0] (n=Mudkips@
16:36:59n1sUnhelpful: built fine for the beast, got the lua warning about non void function not returning a statement or whatever too
16:38:50 Quit Ubuntuxer ("Leaving.")
16:39:36 Join solexx [0] (
16:39:39n1s...and *PANIC* Stkov headphone
16:40:17 Join toffe82 [0] (n=chatzill@
16:41:19n1sbtw, seems to be no way to reboot from a panic without the battery switch
16:45:42 Part LinusN
16:46:45 Join ender` [0] (i=krneki@
16:47:29 Quit ender` (Client Quit)
16:49:20n1sincreasing the stack size of that thread helped
16:50:43 Quit shai ("Leaving")
16:51:43n1shmm, isn't the "L" powerup in brickmania supposed to make the paddle larger?
16:55:31 Join einhirn [0] (
16:55:40evilnickn1s: No
16:55:45evilnickL = gains a life
16:56:07 Quit solexx_ (Read error: 110 (Connection timed out))
16:56:42evilnickAlthough E isn't currently used, in case you want to add that to it! E = Extend.
16:58:17 Quit petur ("connection reset by beer (Leffe)")
17:01:58n1sUnhelpful: everything seems fine, running test codec now
17:09:56 Quit mt (Read error: 104 (Connection reset by peer))
17:10:32 Join polobricolo [0] (
17:18:04***Saving seen data "./dancer.seen"
17:20:20 Quit Tomis (Remote closed the connection)
17:24:20n1sI hate when someone replies to one bug and starts out with "i think i have this bug too" and then describes entirely different symptoms
17:36:53 Join phanboy4 [0] (
17:38:15 Join AEnima1577 [0] (
17:41:44 Join Tomis [0] (n=Tomis@
17:42:10CIA-80New commit by nls (r23678): Increase stack space for the Gigabeat S specific headphone thread to fix stack overflow when compiled with gcc 4.4
17:43:48 Join bmbl [0] (n=Miranda@unaffiliated/bmbl)
17:44:10 Join seani_ [0] (n=seani@
17:44:56 Quit seani (Read error: 110 (Connection timed out))
17:45:13saratogaare we planing to update gcc versions soon?
17:45:36 Quit martian67 (Read error: 110 (Connection timed out))
17:48:33n1ssaratoga: i think so, IIUC it turned out we need to to be able to use the eabi stuff
17:48:42n1sfor arm that is
17:50:11 Join mt [0] (n=mtee@rockbox/developer/mt)
17:50:25 Nick YPSY is now known as Ypsy (
17:52:49Tornesaratoga: the ARM EABI linker can fix up long/short calls automatically
17:53:11Tornethis is a potential large win for targets where iram can't be mapped close to sdram
17:56:00n1son the beast mp3, wma and wv are slightly faster and the rest is slightly to a bit slower (using the test set from
18:00:11saratogayeah i saw the test results, it seemed like newer gcc was a small win for codecs
18:00:29 Join AEnima15771 [0] (
18:01:51 Join Rob2222 [0] (
18:02:54n1si'd say it's a small loss on average on the beast
18:03:08 Quit Rob2222 (Client Quit)
18:03:25n1sthe faster codecs are just very slightly faster while some are a bit slower
18:03:52 Quit einhirn ("Miranda IM! Smaller, Faster, Easier.")
18:04:10 Quit AEnima1577 (Read error: 60 (Operation timed out))
18:06:33 Join Rob2222 [0] (
18:09:21n1sflac for example is a whole MHz slower
18:10:01 Join pamaury [0] (
18:10:31 Quit saratoga (Ping timeout: 180 seconds)
18:10:39n1sthe core binary shrunk by about 10kB
18:11:28n1s...which is probably more due to newer gcc than eabi since the beast was using only shortcalls already
18:12:30Torneyah, the beast isn't one that will benefit from eabi linkage
18:13:03n1shmm, that reminds me of one thing to test
18:16:14 Join Omlet [0] (i=omlet05@
18:17:36 Quit jon-kha (Remote closed the connection)
18:19:43pamaurygevaerts: ping
18:20:44gevaertspamaury: pong
18:22:06pamaurygevaerts: I have a question about unhandled requets in usb: what the spec says ? Should the device stall endpoints, or do something or just ignore the requets. Indeed, by not responding to some OS descriptors requets (extended properties), the detection hangs out during a long time (until timeout)
18:23:18gevaertspamaury: stall sounds safest
18:23:32 Join at0m [0] (
18:23:47gevaertsor maybe just implement extended properties :)
18:24:33pamauryYes, I will perhaps do it but I don't know which properties to implement ! The doc is not very precise. Anyway usb_core simply ignores unhandled requets, no ?
18:25:52 Join jspeter [0] (
18:25:54gevaertsit stalls them, at least in most cases
18:26:24gevaertssee e.g. the !handled bit in control_request_handler_drivers()
18:29:32 Join funman [0] (n=fun@rockbox/developer/funman)
18:33:51 Quit jspeter ("No Ping reply in 180 seconds.")
18:34:11 Join jspeter [0] (n=quassel@
18:35:42 Quit maruk ("Leaving.")
18:36:21 Join Tomis2 [0] (n=Tomis@
18:42:07 Join bertrik [0] (
18:43:15 Quit Tomis (Connection timed out)
18:43:15 Nick Tomis2 is now known as Tomis (n=Tomis@
18:43:43pamaurygevaerts: I just found a libmtp thread about the control messages and I have the feeling that they just pick gPhoto2 code and didn't really understand it. It's not even sure that the correct device would respond something to extended properties requets. We should try it on several devices to see
18:46:42gevaertspamaury: I'll have a go at trying it on my gigabeat S. That runs Windows CE as the OF, so I'd guess it should behave like the MTP specifiers intended
18:50:07pamauryNote that in the extended properties spec, it is written that Windows first send a request with enough size for the header and then by reading the answer sends a second request with the required size. We never know, perhaps the device will need this magic.
18:50:14 Quit mt (Read error: 104 (Connection reset by peer))
18:50:45 Join mt [0] (n=mtee@rockbox/developer/mt)
18:51:32gevaertsthat's actually standard behaviour for descriptors in USB, and a device should be prepared to always handle sending partial descriptors
18:52:17 Quit jspeter (Remote closed the connection)
18:54:07n1saha! reenabling strict aliasing improves speed A BIT
18:54:46pamauryOk. Then it would be nice if you could try it and tell me the result of your experiences. I have to go
18:56:24n1sor i'm blind
18:56:44 Join AEnima1577 [0] (
18:57:20n1salac is about 4MHz faster in fact regardless of aliasing
18:58:52 Join kugel [0] (n=kugel@rockbox/developer/kugel)
18:59:14*gevaerts probably has to charge the thing first
18:59:22kugeln1s: IIRC Unhelpful found that using -O2 for flac makes it equally fast
19:00:01 Quit AEnima15771 (Read error: 60 (Operation timed out))
19:00:35kugelflac is a good deal slower on all tested devices so far
19:00:37n1skugel: aha, it's using the core setting
19:01:03n1s(which is -O for arm targets)
19:01:07gevaertss/charge/find/ :\
19:01:51n1sreenabling the stric aliasing makes aac about a MHz faster
19:02:35 Join kugel__ [0] (
19:05:20 Nick hd is now known as HellDragon (n=jd@Wikipedia/HellDragon)
19:05:42n1sok so with fstrict-aliasing and if O2 gets the flac speed back it seems like it's an overall win for the beast too, only aache seems noticably slower
19:05:43 Quit kugel (Nick collision from services.)
19:05:56 Nick kugel__ is now known as kugel (
19:06:27 Quit flydutch ("/* empty */")
19:13:12 Quit pamaury (Read error: 113 (No route to host))
19:13:49 Quit amiconn (Nick collision from services.)
19:13:51 Join amiconn_ [0] (i=quassel@rockbox/developer/amiconn)
19:14:11 Nick amiconn_ is now known as amiconn (i=quassel@rockbox/developer/amiconn)
19:15:35 Join pixelma_ [0] (
19:15:39 Quit pixelma (Nick collision from services.)
19:15:48 Nick pixelma_ is now known as pixelma (i=quassel@rockbox/staff/pixelma)
19:18:06***Saving seen data "./dancer.seen"
19:28:06n1skugel: yeah, -O2 made flac quite a bit faster ~12MHz on 4.0 (-O) to ~10MHz on 4.4 (-O2)
19:28:58 Join Tomis2 [0] (n=Tomis@
19:33:19 Join Thundercloud [0] (
19:34:29 Join Tomis2_ [0] (n=Tomis@
19:36:38 Quit JdGordon| (Read error: 110 (Connection timed out))
19:36:52 Quit Tomis (Read error: 110 (Connection timed out))
19:36:52 Nick Tomis2_ is now known as Tomis (n=Tomis@
19:38:04 Quit Tomis (Read error: 104 (Connection reset by peer))
19:38:24 Quit DerPapst ("Leaving.")
19:38:59 Join Tomis [0] (n=Tomis@
19:41:34 Join einhirn [0] (
19:41:37n1slibcook is also built with the core setting
19:41:57n1shave no test file for that though
19:42:59 Join stoffel [0] (
19:45:07 Join Tomis2_ [0] (n=Tomis@
19:48:51 Quit Tomis2 (Read error: 110 (Connection timed out))
19:53:09 Quit parafin (Read error: 60 (Operation timed out))
19:53:32 Nick Ypsy is now known as YPSY (
19:57:12 Join parafin [0] (
19:58:07 Quit Tomis2_ (Read error: 104 (Connection reset by peer))
19:58:54 Quit at0m ("it is now safe to turn of your machine.")
20:01:24 Quit Tomis (Read error: 110 (Connection timed out))
20:02:23 Join petur [0] (
20:05:41 Join Tomis [0] (n=Tomis@
20:06:57 Join JdGordon| [0] (n=Miranda@nat/microsoft/x-bbhbsrbaelcgessu)
20:07:15 Quit polobricolo (Read error: 110 (Connection timed out))
20:07:32 Quit AEnima1577 ("Leaving.")
20:12:02 Join Tomis2 [0] (n=Tomis@
20:13:18 Join GeekShadow [0] (n=Antoine@reactos/tester/GeekShadow)
20:14:05 Join pondlife [50] (n=Steve@rockbox/developer/pondlife)
20:15:30CtcpPing from gevaerts!n=fg@rockbox/developer/gevaerts
20:23:41 Quit cowgarden (Connection reset by peer)
20:24:16 Join AEnima1577 [0] (
20:29:08 Quit Tomis (Read error: 110 (Connection timed out))
20:29:09 Nick Tomis2 is now known as Tomis (n=Tomis@
20:35:30 Quit antil33t (Read error: 131 (Connection reset by peer))
20:37:13 Quit bertrik (Remote closed the connection)
20:37:19 Join ruckus1 [0] (n=Earworm@
20:37:35 Join antil33t [0] (
20:39:02ruckus1hey peoples. i modded my 30gb ipod 5.5 with a 120gb disk. i am trying to use the utility and i don't know what "audio player" to select in the rockbox utility. iPod video 30gb, or iPod video 60/80gb ?
20:39:03 Join p3tur [50] (n=petur@rockbox/developer/petur)
20:40:04ruckus1that's what i was thinking, though i though the size difference may require the one with the larger capacities
20:40:08gevaertsThe 30gb and 60/80gb builds are different because the latter have more RAM
20:40:19gevaertschanging the disk doesn't change that
20:40:33ruckus1ahhh i see. that's all i wanted to know! THANK YOU!
20:40:38 Join TheSeven [0] (n=theseven@rockbox/developer/TheSeven)
20:40:54 Part watto
20:42:38ruckus1gevaerts: error reading the partition table
20:42:58ruckus1(possibly not an ipod)
20:43:04ruckus1that's what the utility says
20:44:01TorneHave you installed the Apple firmware on the disk yet?
20:44:10TorneRockbox can't install to a blank hard disk
20:44:21Torneit needs to be restored in iTunes, or manually restored using the instructions on the wiki
20:44:50ruckus1it did have the firmware installed Torne
20:44:50 Quit antil33t (Read error: 104 (Connection reset by peer))
20:45:10Tornewhat OS?
20:45:37 Join davidfg4 [0] (n=david@CMU-350194.WV.CC.CMU.EDU)
20:45:51gevaerts"error reading the partition table" could be permissions I guess
20:46:01 Join bertrik [0] (
20:46:08ruckus1hold on
20:46:56 Quit Thundercloud (Remote closed the connection)
20:47:46ruckus1that was actually mistaken drive letter. :) i didn't notice the difference "F" and "E" are similar
20:48:10 Join jon-kha [0] (
20:51:49Tornenevermind :)
20:52:59 Join antil33t [0] (
20:55:56 Nick p3tur is now known as himself (n=petur@rockbox/developer/petur)
20:56:13 Nick funman is now known as gevearts (n=fun@rockbox/developer/funman)
20:56:16 Nick himself is now known as p3tur (n=petur@rockbox/developer/petur)
20:56:25 Nick gevearts is now known as funman (n=fun@rockbox/developer/funman)
20:56:29 Quit parafin (Read error: 60 (Operation timed out))
20:56:32davidfg4When developing a plugin, is it possible to only recompile the plugin, to save time?
20:56:47funmanmake $PWD/apps/plugins/plugin.rock
20:56:49davidfg4is it easy?
20:57:30 Part pondlife
20:57:34 Join panni_ [0] (
20:58:00ruckus1by the way, is it okay to delete the leftover folders from the iPod's non-rockboxed state?
20:58:49 Join parafin [0] (
20:59:17Tornethe OF will recreate them, mostly. it's easier to just set them hidden, which will sto pthem appearing when "view files" is set to "supported"
20:59:27davidfg4Thanks funman
21:01:04ruckus1holy crap, the util is eating nearly 50,000 K here
21:01:20 Quit davidfg4 ("Ex-Chat")
21:01:47Tomiswhat util, an 50MB of what
21:02:15ruckus1the rockbox utility
21:05:11ruckus1really weird, now it's lurking in the background even though i closed it
21:05:14 Join Kopfgeldjaeger [0] (
21:07:04 Quit bertrik (Read error: 54 (Connection reset by peer))
21:09:41 Quit ThomasAH ("changing host")
21:10:18ruckus1anyway, thanks to Torne and gevaerts again! have a great day!
21:11:48 Quit ruckus1 ("Leaving.")
21:11:53 Join bertrik [0] (
21:15:25 Join ThomasAH [0] (
21:18:07***Saving seen data "./dancer.seen"
21:20:30 Quit AEnima1577 ("Leaving.")
21:24:05 Join AEnima1577 [0] (
21:25:38 Quit bmbl ("Bye!")
21:33:57 Join Grahack [0] (
21:34:06p3turhrm... when I do make (very)clean, why are there some files staying in my build dir? They cause the build to fail
21:34:35gevaertsdifferent configuration?
21:34:52p3turh300 sim
21:35:08p3turI have a different dir for each target+config
21:35:26p3turit is quite some time since I made builds in there
21:36:02p3turbut lang.c/h/o and max_language_size stay in there
21:36:21p3turanyway, wiping the dir clean by hand fixes this
21:36:42p3turso I suspect something was changed and can't handle the old files being there
21:36:51 Quit AEnima1577 ("Leaving.")
21:37:05gevaertsdidn't those move to apps/ a while ago?
21:43:23 Quit parafin (Read error: 60 (Operation timed out))
21:47:20 Join parafin [0] (
21:51:32 Join AaronM [0] (
21:57:48n1syeah, make clean only deletes known files, files move, old files are not known anymore
22:03:22 Quit stoffel (Remote closed the connection)
22:04:26CIA-80New commit by bluebrother (r23679): Add docs/COPYING to rbutil source folder and make checkout slightly more robust.
22:07:30 Quit Sajber^ (Read error: 104 (Connection reset by peer))
22:07:43 Join GeekShado_ [0] (
22:14:27amiconnp3tur: *Very* old build dir then...
22:15:52 Quit gevaerts (Nick collision from services.)
22:16:01 Join gevaerts [0] (n=fg@rockbox/developer/gevaerts)
22:22:44 Join tornvig [0] (
22:23:46tornvighey.. I just installed a new rockbox on my ipod.. now it stalls when initially building the database.. any clues?
22:24:06 Quit GeekShadow (Read error: 110 (Connection timed out))
22:24:18funmanreboot it and try again
22:25:34tornvigseems to be the same
22:26:46funmanthen boot the OF, plug usb and delete .rockbox/database_tcd* files
22:27:25n1sthat can be a file that breaks one of out metadata parsers, or disk corruption
22:27:34funmanOriginal Firmware (Apple)
22:27:49tornvigah ok.. I'll try it
22:27:50funmann1s: or a bug in database
22:28:48n1sa file breaking the metadata parsers is a bug too, they should handle broken files gracefully
22:28:55 Quit Grahack (Read error: 110 (Connection timed out))
22:30:06n1stornvig: if retrying to build the database fails and a disk scan comes back clean i'd suggest doing a binary search on you music to find which file(s) it breaks on and then file a bug report
22:30:56tornvigholy moly, the dir looks completely f**cked
22:31:24tornvigthis "?p????'?.??p" kind of stuff everywhere
22:31:45 Join Grahack [0] (
22:32:19funmandisk corruption then, you should restore the ipod with itunes and reinstall rockbox
22:32:41funmanperhaps try checking the disk with fdisk/chkdsk.exe first
22:32:41tornviggah! I just did that..
22:34:56 Quit mikroflops (Read error: 145 (Connection timed out))
22:36:56 Quit einhirn (Read error: 104 (Connection reset by peer))
22:39:34 Quit tornvig (Remote closed the connection)
22:47:09Unhelpfuln1s: i think the lua function in question should be void noreturn anyway, shouldn't it? it calls the plugin exit macro...
22:49:23p3turamiconn: yeah, I've gone back looking at that histogram code after a year or so
22:51:39n1sUnhelpful: didn't look at all
22:53:04n1sUnhelpful: did you see my results from the beast? basically all codecs were unchanged or faster except aac he (only tested the ones we have test files for though)
22:54:31Unhelpfuln1s: yes... i have a branch with switchable -O levels for all codecs, and a spreadsheet with times. i'll need to get the beast run and entered now that i have a cable for it again :)
22:55:27n1sUnhelpful: reenabling fstrict-aliasing gave some small speedups too but some warnings in the zxbox plugin
22:55:57Unhelpfuln1s: did anything slow down with it? and are the warnings possibly worth fixing?
22:56:05Unhelpfulwe could also enable it only for codecs :)
22:56:28n1snone of the codecs i tested slowed down
22:57:22n1si think fixing the warnings is the way to go if anyone can be bothered, it's disabled since 4.0 gives a lot more warnings iirc
22:59:34amiconnWould be interesting to do the same tests on coldfire
22:59:45 Join mikroflops [0] (
22:59:50*amiconn would expect some slowdowns there from -O2 and higher
23:00:41Unhelpfuln1s: type-punning not done via a union would be the usual thing that triggers those warnings, right?
23:02:19JdGordon|petur: hey, have you checked out the which is in the latest zips? Its probably broken in the recording screen...
23:02:26 Quit Tomis (Read error: 104 (Connection reset by peer))
23:02:38 Join Tomis [0] (n=Tomis@
23:04:53JdGordon|no patches needed :) just load it from the menu
23:06:48 Join tobs [0] (
23:07:17 Join LambdaCalculus37 [0] (n=LambdaCa@rockbox/staff/LambdaCalculus37)
23:10:00 Quit Grahack ("Leaving.")
23:10:22 Join Thundercloud [0] (
23:13:26 Quit tobs ("Leaving")
23:14:22p3turJdGordon|: from a first look the samplerate shown is wrong and some stuff is missing like the mono/stereo indication
23:15:01JdGordon|I thought I added mono/stereo? :(
23:15:15JdGordon|is samplerate off by one? or way out?
23:17:08p3turit said 96k when configured for 44
23:18:09***Saving seen data "./dancer.seen"
23:19:13 Join at0m [0] (
23:24:39 Quit JdGordon| ("Miranda IM! Smaller, Faster, Easier.")
23:27:46 Join fyrestorm [0] (
23:28:36 Quit n1s (Read error: 110 (Connection timed out))
23:28:57 Join n1s [0] (n=n1s@rockbox/developer/n1s)
23:31:21 Quit dfkt ("-= SysReset 2.53=- Ph'nglui mglw'nafh Cthulhu R'lyeh wgah'nagl fhtagn.")
23:32:04 Quit phanboy4 (Read error: 110 (Connection timed out))
23:38:47 Quit evilnick ("Page closed")
23:39:13 Join Sajber^ [0] (
23:43:12n1sUnhelpful: yeah, that or casts between types that do not alias i think
23:43:21 Join GeekShad__ [0] (
23:43:32 Join ercani [0] (n=ercan-i@
23:44:26ercanirockbox works on jz4725 ?
23:44:48 Join GeekShadow [0] (n=Antoine@reactos/tester/GeekShadow)
23:44:58B4gderrockbox runs on music players
23:45:03B4gdernot single chips
23:45:18ercanineed linux ?
23:45:24AlexPWhat for?
23:45:34ercanimp4 player with jz4725
23:45:46AlexPWhat has linux to do with it?
23:46:13B4gderah no, it has a jz4732
23:46:14 Join saratoga [0] (i=9803c6dd@gateway/web/freenode/x-bbbocttvvetcdwzb)
23:46:16ercanios is linux
23:46:23B4gderno, Rockbox is Rockbox
23:46:35AlexPRockbox has nothing to do with linux
23:46:39saratogaanyone mind if i commit the clip playback fix and make it a supported target tomorrow?
23:46:54B4gder is the better page
23:47:41n1ssaratoga: is that fs#10605?
23:48:09saratogan1s: yeah
23:48:27ercaniit is a firmware
23:48:31ercaniright ?
23:48:35LambdaCalculus37saratoga: Go for it. :)
23:48:36ercanicpu cpecific ?
23:48:47B4gderercani: it depensd on what you define a "firmware" to be
23:49:18ercaniwhat we can do with rockbox ?
23:49:19B4gderRockbox is an OS and applications and more for portable music players
23:49:30 Quit GeekShado_ (Read error: 104 (Connection reset by peer))
23:49:34 Quit GeekShad__ (Read error: 131 (Connection reset by peer))
23:49:34B4gderand yes it has to be customized for every player it runs on
23:49:45 Quit GeekShadow (Read error: 104 (Connection reset by peer))
23:49:47 Join GeekShado_ [0] (
23:49:53B4gderercani: try, it explains some of what Rockbox is
23:50:24saratogathe jz chip isn't really a CPU, its a complete computer on a chip
23:50:36saratogarockbox supports some devices using it, but not others
23:50:41saratogadepending on what they do
23:50:52*bluebrother wonders how people find the IRC channel but not the website
23:53:23ercaniit has TTS ?
23:53:40AlexPbut it'll speak the inerface etc from pre made voice files
23:53:54saratogalots of ffmpeg improvements this week on their AAC-HE decoder
23:54:26*LambdaCalculus37 had a thought today...
23:54:41LambdaCalculus37I was looking at the firmware updates for the Gigabeat MEG50JS and the MEG201.
23:54:56LambdaCalculus37They're both Linux-based firmwares, like the Gigabeat F.
23:55:14LambdaCalculus37So I was thinking... we might be able to get custom code loaded onto each of those.
23:55:30toffe82you can try :)
23:55:40LambdaCalculus37I certainly would like to. :)
23:56:59 Join thegeek_ [0] (
23:57:16 Quit thegeek (Read error: 54 (Connection reset by peer))
23:57:16 Quit HellDragon (Read error: 54 (Connection reset by peer))
23:57:17 Join hd [0] (
23:57:35toffe82LambdaCalculus37: you want the players to test
23:57:50LambdaCalculus37toffe82: Sure, and I promise to return them. :)
23:57:58*LambdaCalculus37 wants something to do during the holiday weekend
23:58:18ercaniI got it
23:58:27*toffe82 was thinking finishing the gogear perhaps....
23:58:56 Quit Sajber^ ("Leaving.")

Previous day | Next day