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

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

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

#rockbox log for 2010-09-02

00:00:13 Part Umrtvovacz
00:04:57 Quit jgarvey (Quit: Leaving)
00:06:08CIA-81New commit by kugel (r27972): Redo previous commit to not break android builds.
00:06:25amiconn[13:17:00] <AlexP> Also, you know with all targets that the button that does select also does context menu <== Not entirely true
00:08:02CIA-81r27972 build result: All green
00:09:19AlexPamiconn: Vast majority then?
00:09:45CIA-81New commit by kugel (r27973): Android: Don't compile sim_tasks.c
00:10:10amiconnAlexP: Yes, certaily.
00:10:37amiconnBtw, one (the only?) exception is the Ondio, which has no dedicated select
00:10:45AlexPheh, OK :)
00:11:36CIA-81r27973 build result: All green
00:11:37amiconnActually it's inconsistent
00:12:25amiconnIn the browser the context menu it's true, but in the wps the context menu is on a different button due to severe lack of buttons
00:12:49amiconns/the context menu//
00:15:49 Quit ender` (Quit: It is difficult to produce a television documentary that is both incisive and probing when every twelve minutes one is interrupted by dancing rabbits singing about toilet paper.)
00:18:30 Quit _s1gma (Quit: irl ragequit)
00:25:46gevaertspamaury: did you see FS #11332 ? It might be helpful
00:27:05 Join krazykit [0] (~kkit@
00:29:41 Join Ramsey[LC] [0] (~RamseyLC]@
00:31:51pamauryit's partly othorgonal
00:32:00pamauryI have a problem: I need lseek to seek the storage file
00:32:10pamaurybut lseek is defined by rockbox for files....
00:32:28gevaertsah, I see...
00:32:36gevaertsA big macro?
00:32:55pamauryI don't understand
00:33:29gevaertsHave the rockbox header that has the prototype for lseek say #define lseek rockbox_lseek
00:33:52pamauryyes, you're right
00:35:45 Join Judas_PhD [0] (
00:36:18pamauryHum, there is a problem with strlcpy in the sim, I have 3 implicit defs each time I compile dir_uncached.c, file.c, filefuncs.c
00:36:25pamaurysomeone wants to have a look ? :D
00:43:06pixelmaselect and context menu on files or menu entries is the on the same button on the Ondio, a screen's context menu (e.g. WPS or radio) is a different story as you don't have a "select" anyway. The Ondio's keymapping is still quite intuitive
00:51:00 Join Dreamxtreme [0] (~Dreamxtre@
00:53:18 Join JdGordon [0] (~jonno@rockbox/developer/JdGordon)
00:54:01 Quit bluebrother (Read error: Connection reset by peer)
00:54:30 Join bluebrother [0] (~dom@rockbox/developer/bluebrother)
00:54:51pamauryhum, either my code is buggy in a highly subtle way, either the fat code chokes on my FAT16 usb key fs
00:55:51pamaurycould there be a bug in the FAT16 code ?
00:55:52***Saving seen data "./dancer.seen"
00:56:53pixelmais there something somewhere I could look up which revision the releases are based of?
00:57:21pixelma(mostly interested in 3.5 and 3.4 currently)
00:57:44gevaertspixelma: you want the branch point?
01:00:42gevaertspixelma: 3.5 was branched at r24344, 3.4 at r22748
01:02:28gevaerts3.5 was tagged at r24501, 3.4 at r22818
01:11:20 Join funm4n [0] (56428fe5@gateway/web/freenode/ip.
01:11:44funm4ngevaerts: still shows 2 different ipod video builds
01:11:53kugel:O powermgmt on android :)
01:12:08gevaertsfunm4n: yes, I know
01:12:20gevaertsThe released rockbox utility expects both to be there
01:12:31gevaertsOr maybe the released config file
01:12:35gevaertsThe builds are identical
01:12:42funm4ni commented HAVE_ADJUSTABLE_CPUFREQ on clipv1 and got 6h30 of runtime (instead of 8 hours last time i had tried)
01:12:45 Quit Judas_PhD (Quit: This is a quitting message)
01:13:07funm4ngevaerts: that means we need to wait the next version of rbutil to change the location?
01:13:16pamaurykugel: did you fiddle with file.h recently ? It seems to me that the these lines are redundant, the test is always true:
01:13:21pamaury#if !defined(CODEC) && !defined(PLUGIN)
01:13:24gevaerts"change the location"?
01:13:28pamaury#define open(x, y, ...) file_open(x,y)
01:14:17funm4ngevaerts: "there" referred to a location, didn't it?
01:14:34kugelpamaury: it's only true for the core isn't it?
01:14:59gevaertsfunm4n: "expects both to exist"
01:15:19funm4nwe can change next version of rbutil
01:16:24 Quit webguest88 (Quit: CGI:IRC (Ping timeout))
01:16:24funm4nah bluebrother already did that
01:16:54gevaertsThere should be a new rbutil release very soon, so waiting for that seems nice
01:17:10funm4nwhat about remove the 64mb build and make a symlink on the server?
01:17:26kugelpamaury: that's so that you can call open via ci-> and rb-> using the standard signature
01:17:29 Join kaiscene [0] (
01:18:00pamauryno I mean, the whole file.h code is already in a #if !defined(PLUGIN) && !defined(CODEC), so doing a nested #if !defined(PLUGIN) && !defined(CODEC) doesn't change anything
01:20:19gevaertsfunm4n: that would work, but is it really worth the effort?
01:21:15*gevaerts doesn't see the point
01:21:53 Join CGL [0] (~CGL@
01:21:53funm4nthere's only one build, why propose 2 ?
01:22:47gevaertsbecause it will get resolved without doing anything in a week or so anyway
01:22:56funm4nalso, why *build* 2 times the same code?
01:23:31gevaertsSame reason
01:23:35funm4nyeah that works too
01:24:29gevaertsWithout a planned rbutil release I'd agree
01:25:32 Quit xchrisx (Quit: CGI:IRC)
01:28:25pamaurygevaerts: I have nearly working code, I'll post it on FS tomorrow if I have time
01:28:27 Quit funm4n (Quit: Page closed)
01:29:15pamauryIt still panics on the fat but perhaps I should run fsck before concluding anything
01:29:53 Quit Utchybann (Ping timeout: 276 seconds)
01:30:38 Part toffe82
01:31:16 Quit pamaury (Remote host closed the connection)
01:32:17 Quit seani (Ping timeout: 258 seconds)
01:33:01pixelmagevaerts: thanks for the revision numbers by the way. Where did you look them up?
01:36:18CIA-81New commit by kugel (r27974): Android: don't compile powermgmt-sim.c ...
01:36:22CIA-81New commit by kugel (r27975): Android: Fix two warnings
01:36:58kugelfinally no randomized battery bar anymore :/
01:38:11kugelunfortunately cabbiev2 doesn't show the battery on sizes!
01:38:39CIA-81r27974 build result: 22 errors, 0 warnings (kugel committed)
01:40:44CIA-81r27975 build result: 22 errors, 0 warnings (kugel committed)
01:41:35kugelah, I get the trick
01:42:39kugelhm, no I don't
01:45:10 Join seani [0] (~seani@
01:49:30CIA-81New commit by kugel (r27976): Fix reds. The battery_capacity setting is used even for fixed-battery targets.
01:51:41CIA-81r27976 build result: All green
01:55:11 Quit JdGordon (Ping timeout: 245 seconds)
01:55:47 Quit user890104 ()
02:00:55 Join WalkGood [0] (~wg@unaffiliated/walkgood)
02:01:03 Join JdGordon| [0] (
02:01:04 Quit JdGordon| (Changing host)
02:01:04 Join JdGordon| [0] (~jonno@rockbox/developer/JdGordon)
02:05:05 Join fdinel [0] (
02:05:43kugelgevaerts: ping
02:06:29WalkGoodi have a strange case on abi member, has anyone ever heard of the "blue ring of death" on a v1 clip?
02:06:57WalkGoodi've only ever seen that on the e200 series
02:07:49WalkGoodalso the lock/hold, center button trick to connect into msc mode doesn't work for him
02:09:18WalkGoodwell with the e200 you could bring it back
02:11:16 Join Judas_PhD [0] (
02:12:33 Join Bice [0] (
02:14:29BiceAnyone available who can help me with replaygain? I understand that, even if LAME is writing replaygain tags, RockBox can't read them.
02:14:31BiceSo I downloaded MP3Gain and ran it on some of my MP3s, only to find that the tags it writes aren't read by RockBox either. How can I get replaygain into into my MP3s in a format that RockBox can read?
02:15:22BiceI'm using Windows BTW, so Linux scripts and the like won't help me.
02:16:02JdGordon|have you enabled replay gain in the rockbox settings?
02:17:02BiceFrom what I've been reading, MP3Gain writes the data to APEv2 tags, and RockBox is looking for it in ID3 that right? Is there any way I can copy the info to ID3 tags?
02:19:39kugelyay, killed libuisimulator!
02:19:46BiceI know RockBox is looking for ReplayGain info, because when I look at the file info of FLAC files that I encoded with FLAC Frontend, it shows the track and album replaygain values. But it doesn't show them for MP3s.
02:24:42CIA-81New commit by kugel (r27977): Android: don't compile stubs.c and kill off libuisimulator entirely. ...
02:24:45CIA-81New commit by kugel (r27978): Move comment
02:26:42CIA-81r27977 build result: All green
02:27:53BiceIn case anyone is thinking of replying the my replaygain question - I'm going to be afk for a while. I'll check back later.
02:27:58 Nick Bice is now known as Bice-afk (
02:28:21kugelwith android.make, I can finally have a go at make apk et all
02:28:38CIA-81r27978 build result: All green
02:34:20 Quit bmbl (Quit: Bye!)
02:46:45 Join Utchybann [0] (
02:55:20 Join JdGordon [0] (~jonno@rockbox/developer/JdGordon)
02:55:53***Saving seen data "./dancer.seen"
03:01:49 Join fyrestorm [0] (~nnscript@
03:04:11 Nick Bice-afk is now known as Bice (
03:04:47BiceSo, anyone have any thoughts on how I can get MP3Gain's replaygain info from the APE tag to the ID3 tag?
03:05:42BiceI just downloaded the most recent verson of WinAmp (the one I was using was ancient and didn't show replaygain info at all) and confirmed that MP3Gain is definitely putting the info in APEv2 tags, but not the ID3v2 Track Gain and Album Gain tags.
03:06:15BiceAnnoyingly, WinAmp will let me change every other ID3v2 tag EXCEPT track gain and album gain. So even if I wanted to do it manually, I can't.
03:06:45BiceSomeone must know some way of getting replay gain into an MP3 file in a format that RockBox can use. Beuller? Beuller? Anyone?
03:10:38 Quit BlakeJohnson86 (Remote host closed the connection)
03:12:44 Join freddyb [0] (
03:24:22 Join saratoga [0] (9803c22e@gateway/web/freenode/ip.
03:24:42saratogaBice: lots of apps can write ID3v2 tags, I recommend foobar2000
03:25:44 Quit kaiscene (Ping timeout: 260 seconds)
03:26:49BiceSaratoga - thanks for the advice...I just finished installing foobar2000 and using it to write the replaygain tags, and it does exactly what I want.
03:27:09BiceI wish I had tried it before I installed MP3Gain - would have saved me a lot of headaches.
03:30:31saratogait should be able to read the mp3gain tags and convert them to id3 as well as scan fresh tags fwiw
03:31:41 Part WalkGood
03:34:24literalBice: I wrote a script which runs mp3gain to calculate the values and then uses eyed3 to add the relevent ID3 replaygain tags
03:34:32literalif you're interested:
03:34:55literalthough since you're using WinAmp and foobar2000 I suspect you're running Windows so it won't be of much use to you...
03:36:06 Quit dfkt (Quit: -= SysReset 2.53=- Ph'nglui mglw'nafh Cthulhu R'lyeh wgah'nagl fhtagn.)
03:38:54kugelhmm, I don't get make sometimes
03:39:22kugelI have $^ in my make command, but it always expands to the first item in the dependencies
03:41:03kugelerr, I mean $<
03:43:24 Quit Zarggg (Quit: Zarggg)
03:47:10BiceYeah, I'm stuck with Windows. Foobar seems to be doing the trick. Fortunately I only did a handful of CDs with MP3Gain.
03:47:44BiceNow I just have to add replaygain to the 100 CDs I've already ripped and remember to do it for the 1600+ CDs that I haven't gotten to yet. Joy.
03:49:22BiceThanks for the help folks. Goodnight.
03:49:23 Quit Bice (Quit: CGI:IRC 0.5.9 (2006/06/06))
03:56:56kugelit passes the wrong file with $< (always the same) but the correct files are generated :\
04:08:36 Join Rob2223 [0] (
04:11:34 Quit Rob2222 (Ping timeout: 240 seconds)
04:14:21 Join steve|m1 [0] (
04:14:21 Quit steve|m (Disconnected by services)
04:14:38 Nick steve|m1 is now known as steve|m (
04:20:16 Quit amiconn (Disconnected by services)
04:20:18 Quit pixelma (Disconnected by services)
04:20:18 Join amiconn_ [0] (quassel@rockbox/developer/amiconn)
04:20:20 Join pixelma_ [0] (quassel@rockbox/staff/pixelma)
04:20:22 Nick pixelma_ is now known as pixelma (quassel@rockbox/staff/pixelma)
04:20:38 Nick amiconn_ is now known as amiconn (quassel@rockbox/developer/amiconn)
04:25:17 Join BlakeJohnson86 [0] (
04:26:29CIA-81New commit by funman (r27979): battery_bench: make functions/struct/variables static ...
04:28:21CIA-81r27979 build result: All green
04:33:05JdGordon|any ideas how to go about profiling the skin display code?
04:33:21JdGordon|can i run the unmodified sim through a program to get numbers?
04:34:01saratogasprinkle LOGFs around the code that print the timer tick?
04:34:59JdGordon|I was hoping for less effort than that :)
04:35:16JdGordon|I really want function call counts and time spent in each func
04:35:40saratogawas going to suggest a test_skin plugin
04:43:07 Quit Utchybann (Ping timeout: 260 seconds)
04:45:19 Join FlynDice [0] (
04:51:07 Join Barahir [0] (
04:54:17 Quit Barahir_ (Ping timeout: 240 seconds)
04:54:49FlynDicefunman: (logs) re: AMSv1 MCI_CLOCK_POWERSAVE (bit 9). I had this working for reads at one point by disabling sd_enable(), writes would still fail with this bit set though. Never got far enough to be able to do a battery bench to see if it used less power.
04:55:51 Quit FlynDice (Remote host closed the connection)
04:55:56***Saving seen data "./dancer.seen"
04:58:23 Join kugel_ [0] (
04:58:45 Quit kugel (Disconnected by services)
04:58:49 Nick kugel_ is now known as kugel (
04:58:53 Quit kugel (Changing host)
04:58:54 Join kugel [0] (~kugel@rockbox/developer/kugel)
04:59:59 Quit freddyb (Ping timeout: 258 seconds)
05:00:06 Join Utchybann [0] (
05:07:16 Join BHSPitMonkey [0] (~stephen@unaffiliated/bhspitmonkey)
05:26:21 Join funman [0] (~fun@rockbox/developer/funman)
05:27:07funmanFlynDice: perhaps we could set this bit only when we're not using the SD (if it proves to be useful at all)
05:27:28funmannow i'm testing set_cpu_frequency() modified to change CGU_PROC and lower FCLK instead of using fastbus
05:31:37 Join CaptainKewl [0] (
05:33:18 Quit ps-auxw (Ping timeout: 240 seconds)
05:35:02 Join S_a_i_n_t [0] (S_a_i_n_t@
05:37:58 Quit panni_ (Quit: ( :: NoNameScript 3.81 :: ))
05:40:05CIA-81New commit by kugel (r27980): Extend android.make so that it can generate a debug signed apk. ...
05:41:42CIA-81r27980 build result: All green
05:42:03 Quit Horscht (Quit: Verlassend)
05:45:03 Join ps-auxw [0] (
05:53:11CIA-81New commit by kugel (r27981): Fix typo
05:54:47CIA-81r27981 build result: All green
05:56:13 Quit anewuser (Quit: FAMICOMPO 7 - August 15 Through August 31 JST -
05:56:57 Quit kugel (Remote host closed the connection)
05:59:23funmankugel: why is there a red delta for some models in r27976 ?
06:06:04 Quit bieber (Remote host closed the connection)
06:12:44 Quit fdinel (Quit: Miranda IM! Smaller, Faster, Easier.
06:29:29 Quit CaptainKewl (Ping timeout: 276 seconds)
06:51:12 Quit saratoga (Quit: Page closed)
06:55:59***Saving seen data "./dancer.seen"
07:09:32 Join AndyI [0] (~pasha_int@
07:20:57 Quit l403 (Ping timeout: 265 seconds)
07:22:39 Join l403 [0] (
07:29:30S_a_i_n_tmc2739: ping?
07:29:59S_a_i_n_tYou converted the Fuze cabbieV2 .wps...yes?
07:31:06S_a_i_n_tI had the idea of using these themes as somewhat of a "teaching tool" and, I kinda wanted them to all have a similarity with oneanother.
07:31:42S_a_i_n_tI was looking in the forums, and I thought I had messed something up, but then I saw "%xd(E, %mm, -1)" and...that definitely wasn't me ;)
07:33:07S_a_i_n_tI *really* don't want to start another flamewar about it, but I personally find "%xd(E<blah>)" a lot more readable...especially if there is only one icon to be displayed.
07:33:37 Join stoffel [0] (
07:35:29S_a_i_n_tand, hold for example being "%xd(A, %mh)" is all well and good, but "%?mh<%xd(Aa|%xd(Ab>" I find again to be more readable...especially to one who is learning.
07:35:53S_a_i_n_twhatever way it goes...I think all the cabbies should be the same..approximately.'
07:36:15 Quit Judas_PhD (Quit: This is a quitting message)
07:42:39JdGordon|wihtout starting a bitter argument again, I agree that the first is wrong but disagree with the second
07:43:03JdGordon|and full names should be used if you dont use %xd(Aa) type
07:43:10S_a_i_n_tyeah...the debateable.
07:43:45S_a_i_n_tan experienced themer will see it straight away...a novice, probably not.
07:44:07S_a_i_n_ta totally agree that using that to display only one icon is quite wrong, though.
07:44:12JdGordon|just fyi, the %xd(label, tag) for is more efficient to render, i.e better battery life
07:44:13 Quit funman (Quit: leaving)
07:44:42S_a_i_n_tefficient to render != efficient to read, though...and thats my main concern.
07:45:07JdGordon|%?mh<%xd(E, 1)> is better than %xd(E, %mh, -1) for sure :)
07:45:44S_a_i_n_t%?mh<%xd(Aa|%xd(Ab> better still, for a novice.
07:46:03S_a_i_n_tnovice coders is what I had in mind...hence the heavy commenting.
07:46:21JdGordon|%xd(E, 1) ======== %xd(Aa)
07:46:31S_a_i_n_tI kinda knew I should have done the whole lot...but I had that whole "Mum dying" thing ;)
07:47:12JdGordon|%xd(Aa) slowly becomes more difficult to work with after about 5 subimages though
07:47:25*JdGordon| not arguing
07:48:29S_a_i_n_tNo, nor am I...I understand your argument, but I hope you understand that I'm trying to make it dead easy to see whats going on. and to me, a condition with X segments, matching a bitmap strip with X subimages is about as easy as it gets.
07:48:49JdGordon|and there is where we have a fundamental difference :)
07:49:06JdGordon|and you are wrong! :D
07:49:11S_a_i_n_tYes, but..lets *not* revisit that, shall we ;)
07:49:19S_a_i_n_tand, you're wrong!
07:49:37Mode"#rockbox +o JdGordon|" by ChanServ (ChanServ@services.)
07:49:40Kick(#rockbox S_a_i_n_t ::yes, maybe, but I can do this) by JdGordon|!~jonno@rockbox/developer/JdGordon
07:49:41 Join S_a_i_n_t [0] (S_a_i_n_t@
07:49:42Mode"#rockbox -o JdGordon|" by ChanServ (ChanServ@services.)
07:50:01S_a_i_n_tneener, neener, neener!
07:52:15S_a_i_n_tI'm confused as hell by
07:52:23S_a_i_n_t(which, made me notice all this)
07:52:44JdGordon|i didnt understand the question when I read it
07:52:50JdGordon|so just skipped the skin text :p
07:52:55S_a_i_n_tThere is a HUGE difference between his file and SVN, but, it has my commenting style so must have been done by mc2739 recently.
07:53:48S_a_i_n_tI know it wasn't me, (i thought it was initially), because it has the subimage style I don't like.
07:54:25JdGordon|this is where a proper skin reformatter/sizer would be very useful
07:54:52JdGordon|even if it each theme needs 3 versions (portraight, landscape, square)
07:55:09S_a_i_n_ticons almost certainly display incorrectly...and in totally different places to SVN...I'm about 99% sure those are changes the user made and is not aware of though.
07:56:26S_a_i_n_tHis .wps cuts off at least 1px from the bottom of almost every icon...I hope he reads my post.
07:57:42S_a_i_n_tvolume has 5px cut from it...very weird. I'm not sure how he sould look at it and say "yeah, that looks fine"
07:57:42JdGordon|I misread you before.. %xd(E, %mm, -1) isnt a true/false... so I disagree that that is worse than the conditional form
07:58:08JdGordon|the -1 does make it a bit tricky, but hardly worse than %?mm<|%xd(Ea)|%xd(Eb)|%xd(Ec)|%xd(Ed)>
07:58:14S_a_i_n_tJdGordon: Yeah, I misremembered the icon also...but kept quiet about it ;)
07:58:23S_a_i_n_tI thought it was true/false initially
07:58:24JdGordon|worse being typo/error/counting pron
07:58:31JdGordon|prone* :)
08:04:35 Quit solexx_ (Ping timeout: 255 seconds)
08:04:46 Quit BHSPitMonkey (Remote host closed the connection)
08:06:25 Join solexx_ [0] (
08:07:24 Quit stoffel (Remote host closed the connection)
08:09:23 Join funman [0] (~fun@rockbox/developer/funman)
08:10:07S_a_i_n_tOMG...super fail.
08:10:19S_a_i_n_tI mixed up 220X176 and 176x220
08:11:42*S_a_i_n_t whistles a merry tune of failure.
08:13:29S_a_i_n_tnow *that* is why his code way soo far from SVN.
08:17:04S_a_i_n_tI can't however, see the difference between his code and SVN..perhaps I'm blind. I'll have to diff the two.
08:18:18JdGordon|is his svn?
08:18:25funmana lot people complain about the clip keymap change: pitchscreen on down button
08:20:20JdGordon|well, you can't please everyone
08:20:40JdGordon|what did it used to be?
08:21:06funmandown was nothing and pitchscreen was on a weird combo: select+right or something
08:21:37S_a_i_n_ta combo seems better to me, then you can't accidentally enter pitchscreen.
08:21:56S_a_i_n_tperhaps down+select as opposed to just down?
08:22:06funmanwe might as well remove any way to go to pitchscreen
08:22:16funmancombo are not usable on clip
08:23:04S_a_i_n_twhy is that exactly? <genuinely curious>
08:23:22funmanthe keypad is too small
08:23:51S_a_i_n_tIts been a while since I sold mine, but, its not much smaller than the Nano IIRC
08:23:51JdGordon|does clip have a hotkey button?
08:24:08funmanJdGordon|: no
08:24:19JdGordon|then long down should be hotkey so it is up to the user what it does
08:24:38S_a_i_n_tJdGordon: Just checking I'm not missing it here...
08:24:41funmanhm long down might work
08:24:54S_a_i_n_tthat's the diff from SVN and the forum guys "changes"
08:25:09JdGordon|the diff is useless :)
08:25:13S_a_i_n_tthat tells me there's narry a single difference, correct?
08:25:17 Join Zagor [0] (~bjst@rockbox/developer/Zagor)
08:25:26JdGordon|no, it says every line is different
08:25:51S_a_i_n_tAha...which isn't so...crappy diff :/
08:26:05JdGordon|probably trialing spaces
08:26:19S_a_i_n_tis there a better way to do it? I just did "diff file file > file.diff
08:26:32JdGordon|-u gives a nicer output slightly
08:26:42 Quit AndyI ()
08:26:50funmandiff -w also (ignore whitespace)
08:27:20S_a_i_n_taha, thanks.
08:27:59S_a_i_n_taha, whitespace was the problem...he either changed nothing, or posted the wrong file.
08:28:25funmanor \r\n vs \n
08:28:37JdGordon|or it is broken in svn
08:29:13funmanpamaury: ping
08:29:22S_a_i_n_tdiff -u -w 1.wps 2.wps > diff.diff gives a completely empty diff file, so either I did it wrong, or, there is no changes.
08:33:05 Quit lestatar (Remote host closed the connection)
08:34:27JdGordon|funman: on the keymap thing, I tihnk (apart from the ondio) hotkey should be required and so it has a much higher "priority" in the keymap fiddling than pitch/playlist/etc
08:34:44JdGordon|the need to #define HAVE_HOTKEY in each target config is just silly imo
08:35:19funmani think blue_dude didn't want to upset people
08:35:31funmanso if targets don't have a hotkey it just means devs with these targets are lazy
08:35:55JdGordon|imo, that was the wrong thing to do.
08:36:08funmanhe should have chosen a key as random?
08:36:33JdGordon|or at least with some discussion...
08:36:39S_a_i_n_tHe got enough hassle for that commit, I think he was glad to be rid of it in the end.
08:37:00funmandiscussion for the million of supported targets?
08:37:03S_a_i_n_tthere was *plenty* of keymap discussion much.
08:37:12JdGordon|it anoys me just as much that the android cabbies link to the playlist viewer instead of hotkey just because that is what kugel thought is more useful
08:37:13funmaniirc he asked that people tell him which key to use and nobody answered
08:37:41S_a_i_n_tfunman: I answered, just, only for the targets I had.
08:37:59S_a_i_n_tI happen to think my solution for iPod works well.
08:42:32 Quit tchan (Ping timeout: 260 seconds)
08:43:03S_a_i_n_tre: pitchscreen..the forumgoer's thread I'm reading says that down was "submenu", whereas funman just said down was nothing...what is "submenu" anyway?
08:43:20JdGordon|wps context menu maybe
08:43:41S_a_i_n_tAha...wait, isn;t that long select on all targets?
08:44:16JdGordon|should be
08:44:28JdGordon|or quickscreen? check the old code
08:44:37S_a_i_n_tHmmm...I wonder what "submenu" is.
08:44:46S_a_i_n_tAh, right...could be quickscreen.
08:44:50 Join LinusN [0] (~linus@rockbox/developer/LinusN)
08:46:30funmanit was nothing in the wps
08:46:57funmanah, long down is playlist
08:47:05funmanlike center o:
08:47:42 Join tchan [0] (~tchan@lunar-linux/developer/tchan)
08:53:03funmanprogress on USB for AMSv2!
08:53:09funmanit still doesn't work, but in a different way ;)
08:54:06 Join bertrik [0] (
08:54:08 Quit bertrik (Changing host)
08:54:08 Join bertrik [0] (~bertrik@rockbox/developer/bertrik)
08:54:17S_a_i_n_tis the SVN Nano2G bootloader known broken?
08:54:33S_a_i_n_tTheSeven doesn't use it, so, I guess not.
08:56:00***Saving seen data "./dancer.seen"
08:58:35 Quit scorche (Ping timeout: 255 seconds)
08:58:55funmanpamaury: <- i get panic: EP 0 out on clipv2, but i don't know what to do next
09:01:55funmanEP0 out, then EP0 in, EP0 out again, EP0 out again... (i stopped there)
09:04:11funmanonly for EP0 though
09:04:16 Join scorche [0] (~scorche@rockbox/administrator/scorche)
09:06:59 Join ender` [0] (
09:10:52 Join bmbl [0] (~Miranda@unaffiliated/bmbl)
09:11:48 Quit bertrik (Ping timeout: 245 seconds)
09:13:06JdGordon|S_a_i_n_t: AH, I'm tihnking he means the submenu button as in the image on the button is of a submenu
09:13:45S_a_i_n_tJdGordon: that makes *some* sense I guess.
09:14:01funmanthe manual refers to it as 'Submenu'
09:14:06S_a_i_n_tHe said he liked the old functionality...he liked it doing nothing?
09:14:16JdGordon|turns on the backlight
09:14:57JdGordon|hotkey is long down (or long down is playlist if no hotkey enabled)
09:15:20funmanlooks cool
09:16:20 Join Judas_PhD [0] (
09:18:53 Quit Judas_PhD (Client Quit)
09:21:50 Quit sinthetek (Ping timeout: 240 seconds)
09:24:08 Join Rob2222 [0] (
09:28:02 Quit Rob2223 (Ping timeout: 276 seconds)
09:29:08 Join sinthetek [0] (~sinthetek@unaffiliated/sinthetek)
09:29:32S_a_i_n_tCan anyone think of a better way I can do "delete "non-SVN" files" than: svn st | awk '/^?/ {print substr($0,8)}' | (while read a ; do rm -r "$a" ; done)
09:30:06S_a_i_n_tI would preferably like a way for it to prompt me about each file it finds, and ask me if I really want to delete it or not.
09:30:23funmanlinux apparently gets a ETIMEDOUT when reading device descriptor
09:30:38funmanyet i get an 'enumeration done' interrupt on the clipv2
09:34:33funmanpamaury: dwc_otg_pcd_handle_enum_done_intr() seems to do a lot more than prepare_setup_ep0()
09:38:20 Join petur [0] (d408b802@rockbox/developer/petur)
09:40:09funmanlong down hotkey doesn't work in the file browser
09:41:11funmanand long home is quickscreen :o
09:41:36JdGordon|why doesnt it work?
09:41:43funmanit just goes down
09:41:49JdGordon|oh obviously :p
09:42:25 Join swilde [0] (
09:42:25funmanwhich context is used in file browser?
09:42:46JdGordon|long down needs to keep being scroll down
09:43:02JdGordon|the hotkey button doesnt *need* to be the same on all screens
09:43:21S_a_i_n_tIt should be, though.
09:43:26funmanbut there's no button available on std context then
09:43:33funman.. except power
09:43:41funmanonly long power, even
09:43:46JdGordon|the browser doesnt *need* hotkey even
09:43:54*JdGordon| likes bolindg need
09:43:55funmanwhich is also "mapped" to "hard power down"
09:44:06funmanJdGordon|: you don't *need* bolding
09:44:18 Join Judas_PhD [0] (
09:46:32funmanwhat about this: move PITCHSCREEN to long UP
09:50:05CIA-81New commit by funman (r27982): Clip WPS keymap: move pitchscreen to long up ...
09:50:16funmanproblem solved!
09:50:32S_a_i_n_tCan someone have a look at committing FS #5855 please? It may need to be edited slightly, as I'm not too much a fan of the paddles changing colour according to music played, but I *really* like the fact that it enables a (pretty good, actually) AI to pong.
09:50:48S_a_i_n_ta solely two player game on a DAP is a bit of a FAIL IMO
09:51:54CIA-81r27982 build result: All green
10:00:03 Quit utanapischti (Quit: WeeChat 0.3.2)
10:00:27 Join sasquatch [0] (
10:03:49funmanso far seems to give a bit more battery life on AMSv1, more results when they power off completely, and tests welcome
10:09:21 Quit funman (Quit: free(random());)
10:11:37S_a_i_n_tJdGordon: I'm thinking about the new icons you want in the cabbie touchscreen popup can differentiate between "touch" and "long touch" no?
10:12:06S_a_i_n_tI'm wondering if the functionality of the existing icons could be changed without adding new icons.
10:12:13 Quit Judas_PhD (Remote host closed the connection)
10:12:31S_a_i_n_tperhaps the icons appearence would need to be changed, though.
10:12:58 Join Judas_PhD [0] (
10:16:10JdGordon|S_a_i_n_t: you want to actually change the icon while the user is pressing it?
10:16:26JdGordon|a new row would be far more useable
10:16:41S_a_i_n_tno, what I mean is that a short press of the icon could do one thing, and a long press another.
10:16:57JdGordon|yes, that is doable, but how would you show it?
10:16:58S_a_i_n_tthe icons appearence would need to be changed I imagine, though.
10:17:15JdGordon|there is plenty of screen space so a second row would be better
10:17:28pixelmaS_a_i_n_t: the "%xd(E, %mm, -1)" was added to the one cabbiev2.wps by JdGordon| during this commit
10:18:39S_a_i_n_twas it intentional?
10:18:56S_a_i_n_tIt seems very odd to have one different, and not all.
10:19:12pixelmaI guess so
10:19:27*S_a_i_n_t apologises to mc2739 for thinking it was he who did so.
10:19:47S_a_i_n_tI thought it was just him and myself working on the cabbie themes.
10:20:05S_a_i_n_ttoo many cooks, spoiling broth, etc. etc.
10:20:54pixelmaas a "show how it works", I too think that it is weird having one different but didn't say because I don't want all others to change this way too
10:21:38JdGordon|why do you make assumptions? "cabbiev2.176x220x16.wps is an example of how to use this"
10:21:39 Join kugel [0] (~kugel@rockbox/developer/kugel)
10:22:06S_a_i_n_tpixelma: agreed. I voiced my opinion...but don't want to press it, as, it went *very* badly the last time I had a strong difference of opinion.
10:22:23gevaertspixelma: for the tags. For branches, I looked at (and v3_5) and looked for the oldest number ("branched for xxx") message
10:22:32S_a_i_n_tJdGordon: That seems kinda odd though, to have it in one, and not all.
10:22:43S_a_i_n_tNot saying I want it in all, though ;)
10:22:44JdGordon|they are funcionally identical
10:23:02S_a_i_n_tbut, harder to read for a novice coder. (IMO)
10:23:14S_a_i_n_tbut I *really* don't want to go there agian...*sigh*
10:23:50S_a_i_n_tperhaps not harder to read...but, less obvious as to what the code actually does.
10:24:15JdGordon|that is entirely subjective, and please don't :)
10:24:54S_a_i_n_tespecially as there is a mix of both ways of doing things in the one theme, if all were done that would be "better"(?)
10:26:34S_a_i_n_tI'm not really sure I like the fact that I need to avoid voicing my opinions because it might start a huge flamewar, though.
10:26:38pixelmaaha, sorry. This line seems to not be in SVN cabbiev2.220x176x16 at all
10:26:42*JdGordon| suddenly realises he didnt update CustomWPS, probably because of the keffufle
10:27:03JdGordon|S_a_i_n_t: non subjective opinions are fine
10:27:23S_a_i_n_tthere's no such thing as a non subjective opinion!
10:27:37 Quit Judas_PhD (Quit: This is a quitting message)
10:27:40pixelmathat is interesting :\
10:27:41S_a_i_n_tall opinions are subjective, or, is that what you're pointing at?
10:27:53JdGordon|saying X is easier than Y is just asking for trouble
10:28:09S_a_i_n_twell, it shouldn't be...but, apparently it is.
10:28:39 Join Judas_PhD [0] (
10:28:45JdGordon|grr.. the whole image section actually needs to be rewritten
10:31:22S_a_i_n_tpixelma: it's 176x200x16 that has the odd mix of different ways of doing sublimages
10:31:39pixelmayeah, found it
10:31:46S_a_i_n_tI'm not sure now if it was mc2739 or JdGordon that added them.
10:32:01S_a_i_n_tI _think_ it was mc2739
10:32:24S_a_i_n_tI initially thought it was me, because whoever did it kept my coding/commenting style intact.
10:32:36JdGordon|if that is e200 then it was me.. probably... maybe...
10:32:57 Quit JdGordon| (Quit: leaving)
10:33:06S_a_i_n_t*someone* did it ;)
10:33:16pixelmait's not in the cabbiev2.220c176x16.wps in SVN
10:33:36S_a_i_n_tI'm looking at it in SVN now :/
10:33:56S_a_i_n_tits 176x220x16...I messed that up earlier
10:35:13pixelmasee above then - and that's not used on the Fuze
10:35:53S_a_i_n_tyes, I messed that up...I said the wrong target as I mixed up 22x176 and 176x220
10:38:32S_a_i_n_tIf I do a quick change to cabbieV2.220x176 is someone able to commit it for me?
10:39:01S_a_i_n_tI'll leave the weird subimage style...but I wasnt to get rid of the %?c's in the album info
10:39:51S_a_i_n_tlines 66~73 in cabbiev2.176x220x16
10:40:05S_a_i_n_tthe conditions there are irrelevant.
10:41:56 Join JdGord [0] (
10:42:16 Join webguest74 [0] (
10:46:34webguest74running diagnostics on a ipod 5.5g. No firewire port on this model, but it wants me to plug in the firewire so it can test the power. Tried all the buttons, and several button combinations, but I can't seem to bypass this test. Can't even power down. Hard reset is the only thing that appears to work. Anyone know any tricks to get out of here without a reset?
10:47:19JdGordWhat diagnostics?
10:47:40webguest74I put it into diagnostic mode
10:48:14webguest74can't really tell you more cause this is the first time I've done it myself
10:48:19S_a_i_n_tI've never heard of it asking to plug firewire if the model doesn't support firewire.
10:48:27JdGordWe don't support the apple os.
10:48:40S_a_i_n_t <−− commit please?
10:49:13webguest74JdGord: no kidding
10:49:30JdGordSo why would you ask us?
10:49:39S_a_i_n_tJdGord: really? I can reach it fine.
10:50:02JdGordStupid phone... pastebin?
10:50:41JdGordActually... ill look when I get home
10:50:51webguest74because I'm having trouble with rockbox and someone here suggested I run some diagnostics
10:51:35gevaertswebguest74: yes, we suggested looking at the smart data for the disk, not running the full set
10:52:21S_a_i_n_tOk, take two: <−−commit please?
10:52:21 Quit balintx (Read error: Connection reset by peer)
10:53:03JdGordBoring change :)
10:53:13JdGordIs that all inside a cond VP?
10:53:18S_a_i_n_tboring, but, not confusing/irrelevant
10:53:49S_a_i_n_tJdGord: yeah...the VP is it doesn't need the %?C condition.
10:54:03JdGordOk. When I get home
10:54:10S_a_i_n_ts/conditional/conditional to %?C"
10:54:35 Join balintx [0] (
10:56:02***Saving seen data "./dancer.seen"
10:59:16webguest74gevaerts: thought I might as well play around since I was there.
10:59:43gevaertswebguest74: well sure, but we don't know anything about the rest :)
10:59:48S_a_i_n_t"playing around" == "asking for trouble"
11:00:10webguest74gevaerts: so I just ran the smart data and there are no errors reported
11:01:29webguest74gevaerts: is the data that was produced useful to you?
11:02:03gevaertsgood question
11:05:27S_a_i_n_twebguest74: can you post the SMART data to pastebin?
11:05:49S_a_i_n_tand, is there an excessive ammount of reallocations?
11:06:41 Quit kugel (Ping timeout: 265 seconds)
11:07:35S_a_i_n_twebguest74: if you're wondering ^
11:08:06webguest74S_a_i_n_t: zero reallocations
11:09:01S_a_i_n_tHmmm...I assume you've tried running a disk check?
11:11:43 Quit JdGord (Quit: Bye)
11:14:12webguest74S_a_i_n_t: what's a disk check?
11:14:53 Quit S_a_i_n_t (Ping timeout: 252 seconds)
11:15:02webguest74I thought that is what I was doing get the smart data
11:16:26 Quit webguest74 (Quit: CGI:IRC)
11:16:33 Join webguest24 [0] (
11:21:22webguest24when converting an ipod from hfs+ to fat32, should you end up with two partitions? the first one is only the first four sectors. The rest of the disk is the fat32 one. I used the wiki instructions and a linux command line to convert, downloaded the partition table and dd'ed it
11:21:55 Join kugel [0] (~kugel@rockbox/developer/kugel)
11:22:34 Join Stephen__ [0] (
11:22:35gevaertsYou're supposed to have two partitions, yes, but the first one shouldn't be four sectors...
11:23:09webguest24gevaerts: i may be wrong about the size, I just remember it was small
11:23:32gevaertsA few tens of megabytes I think
11:25:17Stephen__JdGordon - just going through the logs and there should be two bug reports ith this issue FS #11583 and FS #11562, both seem to be related to remotes triggering the issue. is your crashes on targets with remotes ? Just also wanted to bring these two to your attention. I also think these would hold back a realease should it be pending soon ?
11:25:31 Quit Stephen__ (Client Quit)
11:26:36 Join pamaury [0] (~quassel@rockbox/developer/pamaury)
11:28:57 Join S_a_i_n_t [0] (S_a_i_n_t@
11:29:06 Quit S_a_i_n_t (Client Quit)
11:38:31 Join einhirn [0] (
11:51:09 Quit kugel (Ping timeout: 265 seconds)
12:02:39 Quit Kohlrabi (Read error: Connection reset by peer)
12:02:43 Join Kohlrabi [0] (~kohlrabi@2a01:608:ffff:3644:216:3eff:fe00:7e5a)
12:03:24 Join watto [0] (~watto@
12:08:24 Quit einhirn (Read error: Connection reset by peer)
12:12:30 Quit MethoS- (Remote host closed the connection)
12:14:30 Join MethoS- [0] (~clemens@
12:25:53 Join liar [0] (
12:26:33 Quit Kohlrabi (Ping timeout: 260 seconds)
12:26:59 Join Kohlrabi [0] (
12:36:55 Join evilnick [0] (
12:42:43 Quit webguest24 (Quit: CGI:IRC (Ping timeout))
12:43:06 Join Kitr88 [0] (
12:45:35 Quit Kitar|st (Ping timeout: 240 seconds)
12:47:14 Quit Kitr88 (Ping timeout: 245 seconds)
12:48:32 Join user890104 [0] (
12:50:24 Quit evilnick (Quit: evilnick)
12:50:48 Join webguest82 [0] (
12:51:49 Join Jaykay [0] (
12:53:06 Join Kitar|st [0] (Kitarist@
12:53:26 Quit webguest82 (Client Quit)
12:56:04***Saving seen data "./dancer.seen"
13:05:17 Quit JdGordon (Ping timeout: 260 seconds)
13:05:24 Part LinusN
13:20:16 Quit l403 (Ping timeout: 258 seconds)
13:21:19 Join kugel [0] (~kugel@rockbox/developer/kugel)
13:23:33 Join l403 [0] (
13:24:00 Join JdGordon [0] (~jonno@
13:24:09 Quit JdGordon (Changing host)
13:24:09 Join JdGordon [0] (~jonno@rockbox/developer/JdGordon)
13:30:31 Join funman [0] (~fun@rockbox/developer/funman)
13:32:07JdGordonanyone want to quickly test a bugfix before i commit it?
13:32:21JdGordonbit of a biggish timeout handling change in the skin renderer
13:32:25funmanfuzev1 runtime: 18h (instead of a bit less than 15h previously)
13:32:45funmanclipv1 still running but it should be ~10h instead of 8h so around the same improvement
13:39:44 Quit JdGordon (Ping timeout: 272 seconds)
13:41:23 Join CaptainKewl [0] (
13:42:22 Join giovanni_ [0] (~giovanni@
13:42:41 Nick shai_ is now known as shai (
13:43:27 Quit Jaykay (Ping timeout: 255 seconds)
13:43:35CIA-81New commit by jdgordon (r27983): fix FS #11588 - %t(0) inside conditionals wasnt making that subline skip
13:43:45CIA-81New commit by jdgordon (r27984): Small cabbie update by Hayden Pearce for cabbiev2.176x220x16
13:45:26CIA-81r27983 build result: All green
13:46:13 Join JdGordon [0] (~jonno@rockbox/developer/JdGordon)
13:47:06CIA-81r27984 build result: All green
13:47:11JdGordontrying one last time before writing a ton of code... any ideas how to profile the skin renderer (in the sim is good enough)?
13:49:30funmanJdGordon: using a profiler maybe?
13:49:40JdGordonthanks :) which?
13:49:59funmanthere's plenty, i know gprof and oprofile
13:53:35funman < i asked a free fuze+ from sandisk ;)
13:54:31pixelmait wasn't an e200v2
13:54:40pixelmait was a v1 ;)
13:55:38funmanoops :/
13:59:05JdGordonhaha fat chance
14:00:34 Join dfkt [0] (dfkt@unaffiliated/dfkt)
14:04:12 Quit z35 (Ping timeout: 265 seconds)
14:04:30 Join xnyhps [0] (
14:06:04 Quit JdGordon (Quit: Leaving.)
14:08:59 Join JdGordon [0] (6fdcf38f@gateway/web/freenode/ip.
14:09:53 Quit storm` ()
14:10:40CIA-81New commit by funman (r27985): fix test_codec broken after r27968
14:10:56funmanclipv1 runtime: 9h45
14:12:21CIA-81r27985 build result: All green
14:16:22 Join z35 [0] (
14:21:13funmanpamaury: ping
14:33:22 Join Luca_S [0] (
14:34:19pamauryMy major problem with usbv2 is corruption. My current code breaks because it often happens that the data of a setup packet refers to the one of the previous setup packet. I already clean teh cache so I don't understand the problem. I will try with your patch
14:34:40funmani have seen this structure
14:35:02funmanit might be worth to force it to be aligned up and down, using attribute aligned and union with char[32]
14:35:15funmanis there a way i could check for corruption without the USB hardware you have ?
14:35:46pamauryI already align it on 16, I can try with 32 but I can only align down
14:35:47pamaurydid you extract these settings from the firmware ?
14:37:13funmanyou can align up with the union
14:37:24funmanand the cache line size is 32 bytes
14:37:31funmanyes it comes from OF
14:38:02pamaurywhy would I want to align up ? With a 8 byte buffer, either I align down or up
14:38:04funmanthe bits i changed in CCU_USB are only there for a certain clocking of CGU_USB
14:38:25funmanso cleaning the cache will have no interference with other code/data
14:38:54funmanand it's not aligned in svn
14:39:21pamauryyes, the svn is not up to date, perhaps it's not an alignment problem because it seems quite erratic for me
14:39:40funmancan you try with my changes and tell me if it changes anything?
14:45:48pamauryI can't see a difference on my clip+. That's always the same thing: I get a GetDeviceDescriptor -> Ok, then a reset, then a set address and for an unknown reason, the content of the setup packet for the SetAddress is the one of the GetDeviceDescriptor
14:46:18funmanwhere is the setup packet written?
14:46:24pamaurystatic struct usb_ctrlrequest ep0_setup_pkt __attribute__((aligned(32))
14:46:36funmanplease commit your changes..
14:47:08 Quit giovanni_ (Read error: Connection reset by peer)
14:48:30 Join giovanni_ [0] (~giovanni@
14:49:25funmanmy change to set_cpu_frequency() seems to break long home press on fuzev1 ??
14:49:29 Quit giovanni_ (Read error: Connection reset by peer)
14:49:53 Join Jaykay [0] (
14:51:28 Join giovanni_ [0] (~giovanni@
14:52:08funmanFS #11597 - AMSv1: improve battery life
14:53:27pamauryDo I integrate your settings for the clock ?
14:53:47 Join n1s [0] (~n1s@rockbox/developer/n1s)
14:54:18kugelfunman: interesting, what are the new clock settings with that?
14:54:20funmanpamaury: no
14:54:23funmankugel: same
14:54:38kugelI'm surprised it makes a difference
14:55:01funmanfastbus just bypass fclk and uses pclk as cpu clock
14:55:08funmanwith this patch fclk is really reduced
14:55:32kugeloh I see
14:55:40kugelwhat was fclk again? :>
14:55:41funmanthis home button thing is weird, can you reproduce it?
14:56:08***Saving seen data "./dancer.seen"
14:56:10funmanCPU clock
14:56:26TheSevenfclk = arm core clock (if not in fastbus mode), hclk = ahb clock, pclk = apb clock
14:57:00TheSevenand IIRC it switches to hclk, not pclk, so you should probably change the hclk divider so that hclk = pclk
14:57:20funmanhclk = pclk already
14:57:25funmani think it's not configurable
14:57:38kugelis it possible to get pclk@32MHz but mem@64MHz to work with that?
14:57:47funmanand btw my patch is buggy, in synchronous mode fclk must be higher than hclk/pclk
14:57:47TheSevenon the nano2g there's an optional /2 or /4 divider between hclk and pclk
14:58:19TheSeven(in some undocumented bits of a clock control reg :P )
15:00:19CIA-81New commit by pamaury (r27986): as3525v2: ...
15:02:04CIA-81r27986 build result: All green
15:02:26pamauryfunman: committed
15:03:50kugelpamaury: btw, #undef doesn't cause a warning if the symbol is undefined so putting #ifdefs around is not needed
15:04:14pamauryI didn't know that :) Thanks, I'll change that one day :)
15:04:29pamauryI really don't understand why it doesn't work. If the controller signal a setup packet, the data must be there, so how can I read the data of the *previous* one ?
15:04:52kugelmaybe try with dcache disabled entirely?
15:05:53pamauryI have a single check before this, how can I disable dcache ?
15:07:27funmanmodify enable_mmu
15:07:44kugelmemory-init.S I think
15:07:47funmanor put your structures in uncached memory
15:08:21pamauryok, let's try the second first, how do I do this ?
15:08:28kugelfunman: could we get rid of memory-init.S with using a spare stack before the real stack?
15:08:36funmanstruct x pkt0_; struct x *pkt0 = UNCACHED_ADDR(&pkt0);
15:08:48funmankugel: what's wrong with it?
15:09:08funmanand why does it need a stack?
15:09:12kugelI just dislike asm when you can achieve the same in C
15:09:24pamauryfunman: I continue to give PHYISCAL_ADDr to dma and use UNCACHED_ADDR to read ?
15:09:27kugelthe C version needs a stack, wasn't that the reason the .S was introduced?
15:09:32 Quit CaptainKewl (Read error: Operation timed out)
15:09:40funmanpamaury: yes, use all accesses through UNCACHED_ADDR
15:09:56funmankugel: true, but IMO it's better in assembler so no bugs will slip through
15:11:16kugelI think asm is more error prone, although neither seems to buggy right now (and probably won't change any time soon)
15:11:49funmandid you see my message at 5:59?
15:11:58kugelyes, and I have no idea
15:12:10kugelif you're interested use bloat-o-meter
15:12:55pamaurywhere is define UNCACHED_ADDR ?
15:15:57kugels/CACHE_ALL/CACHE_NONE/ in memory-init.S seems easier :)
15:16:29funmanpamaury: use grep! system-target.h
15:17:15pamaurynope, system-target doesn't provide it to me, I need to copy the definition
15:17:31pamauryok, it doesn't change anything, let's try with the whole cache
15:18:19pamauryI forgot the AS3525 :)
15:18:23funmanwe changed it to hide it from apps/usb*
15:19:05 Quit Gabe_G23 (Ping timeout: 252 seconds)
15:19:12 Join anewuser [0] (anewuser@unaffiliated/anewuser)
15:21:47pamaurywith no caching it doesn't do anything, it even failes to enumerate
15:21:47funmanthis HOME button not working on repeat is weird, if it works on single press the common driver should handle repeats :/
15:21:55funmanpamaury: too slow perhaps?
15:22:11pamaurydon't, it also crashes the usb analyzer
15:22:34 Quit krazykit (Quit: Connection reset by beer)
15:22:51funmandid you _dcache*() ?
15:23:23pamauryyes I still use *_dcache, it is wrong
15:23:34kugelpamaury: rockbox is a good deal slower w/o dcache, maybe a timing problem?
15:23:52funmanyes it's wrong, not sure if there are nasty side effects but remove them to be sure
15:24:03pamaurypossible, looking at the code, I get a usb reset before getting a chance to answer the setup packet, that's probably a timing problem
15:24:27funmandid you disable also icache?
15:25:29funmanUSB_DEVBSS_ATTR isn't intended for alignement
15:25:53funmanit's to put BSS in iram (or not)
15:26:11kugelor to not put it at all, but allocate on the audio buffer
15:26:13funmanhow do you check logf log btw?
15:26:33pamaurythe debug menu
15:26:44funmani don't get control back on disconnect
15:27:00pamaurywhy USB_DEVBSS_ATTR isn't for alignement, how do force alignment then ? The firmware needs to provide correctly align buffer
15:27:22funmanjust use __attribute__((aligned)) (but i see usb-drv-arc.c does that also)
15:27:46funmanand nano2g
15:27:57funmankugel: which target does that? (and how, and why?)
15:28:22kugelall do, except creative zvm IIRC
15:28:38pamauryyes but for example, the code in usb_core needs to provide correctly aligned buffer, and the only used attribute is USB_DEVBSS_ATTR, so you need to work with this
15:28:39funmani think you're mistaken with something else
15:29:03funmanpamaury: hmm
15:29:15pamauryokay, doesn't work too, caching might not be the problem
15:30:34kugelfunman: quite possibly
15:30:46pamauryfunman: look at the code: in usb_core.c
15:30:46pamaurystatic unsigned char response_data[256] USB_DEVBSS_ATTR;
15:31:04pamaurythe response_data is a buffer given to usb code, so it's not aligned correctly it doesn't work
15:31:06kugeldo we implement NOCACHEBSS_ATTR? that could also be useful
15:31:50funmanno, it just requires linker magic
15:32:09funmanpamaury: might be worth adding a new define then
15:32:33pamauryor renaming this one
15:33:00 Quit jae (Ping timeout: 265 seconds)
15:33:16funmanno, it does what its name suggests (unless where it is abused for something else)
15:34:47pamauryAs far as I know, the only uses of this are for usb buffer in usb_core, and in usb drivers. And why can't it include alignment ? If you put it in BSS, there is no reason to skip an alignment attribute
15:35:16funmanbut there is no reason to do it either?
15:35:33funmanjust create an USB_ALIGN or something
15:35:43pamauryyes there is, if every buffer must be aligned to something, this is part of the attribute
15:35:50pamauryno reason to split
15:35:59funmanit's a different requirement -> a different define
15:36:39pamauryI disagree, the define say: put it in bss in a way the usb code can use it, so alignment is part of the requirement
15:36:40*kugel is with pamaury here but it's totally insignificant :)
15:36:59kugelpamaury: I see no reason why resonse_data[] can't be aligned for all targets
15:37:16 Quit CGL (Remote host closed the connection)
15:37:44funmanjust add aligned(32) like most things do in firmware/usb*
15:37:46pamaurybecause it's ugly, each target specifies the necessary alignment
15:38:27kugelit still can with USB_DEVBSS_ATTR
15:38:53kugel32byte align should work for all targets though
15:38:56pamaurybut it should be the only way to do it, there should be no need for an extra alignement attribute is the rest of the code
15:39:11pamaurybut anyway, it's not important as far as it works currently
15:39:50pamauryI'll revert my USB_DEVBSS_ATTR if you wish
15:39:56funmanit works but it's ugly
15:40:13 Quit andash (Read error: Operation timed out)
15:40:14pamauryyes, I agree
15:40:32funmanjust add alignement everywhere in firmware/usb* and remove it from USBDEV_BSS_ATTR
15:41:33funman(for all targets)
15:41:45pamauryit should be made a define, then
15:41:50funmanand btw the use of iram might not be for speed reasons but only because iram is uncached on those models
15:41:58funmanpamaury: yeah it'd be better i think
15:42:10pamauryBtw, why as3525 iusb doesn't use IBSS_ATTR for USB_DEVBSS_ATTR ?
15:42:40funmanwhy should we do that?
15:42:53pamauryI don't know, nearly all target do, there might a reason :)
15:43:03funman(also please remove your define to panicf and change it to USB_DEBUG or something, because i use panicf for debugging)
15:43:18funmanpamaury: the reason might also be that nobody still knows why we do that
15:43:19 Join JdGordon1 [0] (~jonno@
15:43:29pamaurythe panicf is bad because you can't read a log, that's hardly a good debugging
15:43:37funmanit's good enough for me
15:44:29 Quit JdGordon (Ping timeout: 252 seconds)
15:44:45funmanhm i get a suspend interrupt
15:45:09 Join andash [0] (
15:45:10 Quit andash (Changing host)
15:45:10 Join andash [0] (~andash@unaffiliated/andash)
15:46:30 Join krazykit [0] (~kkit@
15:46:36CIA-81New commit by pamaury (r27987): as3525v2: revert the USB_DEVBSS_ATTR and the panicf redefinition, don't protected #undef of USB_NUM_ENDPOINTS
15:47:01pamauryyes there are suspend interrupts but there are of no use basically, we don't implement suspend in usb mode
15:47:46funmanthings go that way: enumeration -> prepare setup packet for ep0 -> send device descriptor?
15:48:20CIA-81r27987 build result: All green
15:49:09pamaurythings go:
15:49:10 Join jae [0] (
15:49:48pamauryreset -> enumeration -> reset -> enum -> setup (ask desc) -> send desc -> reset after 18 bytes -> enum -> setup (set addr) -> ack -> ...
15:50:54pamauryI can't believe my eyes, for the first time in a long time, I got a proper set address. Perhaps that's a sign :)
15:52:43pamauryDon't be to happy, the next run might not work :) And it crashed again the analyzer /o\
15:53:29 Quit jae (Ping timeout: 245 seconds)
15:54:48 Join jae [0] (
15:55:27funman <- hack to align up
15:56:30pamauryWhy would you want to align up ?
15:56:40funmanto prevent cache line interference
15:56:51 Quit n1s (Ping timeout: 240 seconds)
15:57:06funmanit should work without but you never know
15:59:12 Join Gabe_G23 [0] (~gabe@bzflag/player/GabrielG)
15:59:29funmanso when you do something with the cache lines of this structure, it will not affect any other data
16:00:04pamauryhum, it doesn't work with it :) I might have do somethinf wrong
16:00:57funmancan you have 2 DMA transfers active at the same time ? (in and out)
16:02:23pamauryi don't think, it's a usb restriction iic. you can setup two DMA transfers, but it will one first and then the other one.
16:05:01funmani'll let you work and not confuse you, it might the best thing i can do ;)
16:05:55pamauryWith the current code, it's erratic. Sometimes set address works, sometimes not, and in all cases, it chokes on the get device descriptor after the set address
16:06:17funmanbtw the function in linux which handles enumdone interrupt is quite longer than yours
16:08:45pamauryit's "Set USB turnaround time based on device speed and PHY interface." but I checked and the default value is the good one iirc
16:10:11funmanmy fuzev1 home button doesn't work (for long press) on unmodified svn, so my patch is not the problem :)
16:10:28pamauryI'm bored by this driver, I'll do a pause
16:10:40*funman unbores pamaury
16:13:09 Quit funman (Quit: free(random());)
16:14:13 Join milz [0] (~kyle@
16:17:55JdGordon1 comments?
16:18:05JdGordon1hopefully that explains everything clearly enough
16:24:12 Quit Luca_S (Quit: CGI:IRC (Ping timeout))
16:29:29 Quit MethoS- (Read error: Connection reset by peer)
16:29:48 Join Faux_Pas [0] (
16:31:00giovanni_JdGordon1: i don't want to bother you, but have you seen the latest update to my fulltouch skin? I would really appreciate if you'll help me for the internal speaker tag to be added to the wps system.I've written it already in the forums.
16:31:11kugelJdGordon1: why is there a comma before filename in %x(,filename,x,y)?
16:32:33kugeloh, the identifier should be there, I see
16:33:05kugel"<label>" apparently doesn't work well with the wiki
16:35:06 Quit giovanni_ (Quit: Sto andando via)
16:36:58 Quit Faux_Pas (Quit: ChatZilla 0.9.86 [Firefox 3.6.4/20100611143157])
16:39:07 Quit kadoban (Read error: Operation timed out)
16:39:50 Quit Judas_PhD (Quit: This is a quitting message)
16:43:01amiconnkugel: Iirc USB_DEVBSS_ATTR was introduced because the PP usb controller needs an alignment of 2K
16:44:49 Join toffe82 [0] (~chatzilla@
16:56:11***Saving seen data "./dancer.seen"
16:58:39CIA-81New commit by kugel (r27988): Android: Allow "out-of-tree" builds (outside of android/, e.g. build-android/).
16:59:44kugelrasher: there's at least two problems for the android build in the build system
17:00:12CIA-81r27988 build result: All green
17:00:17kugelyou cannot batch build yet because configure has no cli options for the lcd resolution and, well, no build client is able to build it :)
17:00:28kugelI might fix the first one in 12 revisions :p
17:01:34 Join kadoban [0] (
17:09:21 Part Zagor
17:10:56 Join ep0ch [0] (
17:11:41ep0chHello! Can someone tell me if there are any recent Android builds available?
17:12:33 Join Strife89 [0] (
17:14:41 Quit Strife89 (Disconnected by services)
17:16:28 Join _s1gma [0] (~d.d.derp@
17:16:49 Quit _s1gma (Max SendQ exceeded)
17:17:01 Join sam_s [0] (
17:17:24sam_sI was wondering if rockbox supports ipod shuffle 2G?
17:18:24 Join panni_ [0] (
17:19:17gevaertssam_s: it doesn't
17:19:52sam_si hate using itunes
17:25:21 Quit ep0ch (Quit: CGI:IRC (EOF))
17:27:13 Quit kadoban (Ping timeout: 265 seconds)
17:27:35 Join saratoga [0] (9803c22e@gateway/web/freenode/ip.
17:28:03saratoganot sure if this is useful, but theres a GPL'ed android MTP driver out there:
17:28:09 Join kadoban [0] (~mud@
17:32:58 Join TheSphinX^ [0] (
17:34:54 Join Luca_S [0] (
17:38:27 Join pelupes1 [0] (
17:41:36 Join roolku [0] (
17:42:19roolkukugel: what is required to get a build environment for the android port?
17:42:31kugelandroid sdk and ndk
17:43:17roolkukugel: regarding the screen size - since we only want to see if a commit breaks the build, maybe you can have a default resolution to fall back on if none is specified?
17:43:38roolkukugel: is there a wikipage with instructions/links?
17:43:44kugelthere's one, but you need to press enter twice
17:44:28kugelthere's a readme in the android folder, maybe that helps (it also has links to the android docs for sdk and ndk)
17:45:20roolkukugel: that (pressing enter twice) shouldn't be to hard to work around? ;p
17:45:43kugelthe build systems only works with the cli options
17:45:45gevaertsCan the resolution be specified on the tools/configure command line?
17:45:53kugelit's not hard too add them, I'm just lazy right now
17:46:07 Join S_a_i_n_t [0] (S_a_i_n_t@
17:46:51 Join [sko] [0] (~sko]
17:49:37 Quit anewuser (Quit: FAMICOMPO 7 - August 15 Through August 31 JST -
17:50:10 Join [Saint] [0] (S_a_i_n_t@
17:50:31 Quit S_a_i_n_t (Ping timeout: 255 seconds)
17:50:45roolkukugel: do I read this right, you need an x server to set up the SDK? "install the actual platform sdk from the available packages tab"
17:51:48kugeluhm, I think so
17:51:56kugelcan you do ssh -X?
17:52:55 Quit petur (Quit: Page closed)
17:54:15 Nick [Saint] is now known as S_a_i_n_t (S_a_i_n_t@
17:55:28roolkunot easily, the nodes are behind a NAT router - also I don't really want to do it interactively for all the nodes...ah well
17:56:14kugelmaybe you can do it once and copy the sdk folder afterwards
17:56:23kugelI can't really help it
17:57:50roolkuI'll give it a try when I am bored with work :)
17:57:51 Quit TheSphinX^ (Quit: XChat)
18:03:19 Quit [sko] (Read error: Connection reset by peer)
18:03:23saratogado we need android build servers?
18:06:20roolkusaratoga: maybe not many, but it would be good to see if an unrelated commit breaks the andriod build?
18:09:33 Quit fyrestorm (Read error: Connection reset by peer)
18:11:42 Join domonoky [0] (~Domonoky@rockbox/developer/domonoky)
18:17:05 Join Samosa [0] (~samosaphi@unaffiliated/samosa)
18:17:13SamosaIs it true ?
18:17:25SamosaRockbox on android ?
18:17:55 Quit pelupes1 (Quit: Leaving.)
18:18:48 Join mischasworld [0] (
18:18:54 Join pelupes1 [0] (
18:19:32 Quit YPSY (Ping timeout: 240 seconds)
18:19:59 Quit pelupes1 (Client Quit)
18:21:08 Join Ypsy [0] (
18:21:14 Nick Ypsy is now known as YPSY (
18:24:24kugelcan't we just make absolute point mode default?
18:24:57S_a_i_n_tNot when there are still screens that don't support absolute mode, I wouldn;t say.
18:25:32kugelI rather opt for showing a splash "Switched to grid mode" for the time being
18:25:56 Quit Luca_S (Quit: CGI:IRC)
18:26:24S_a_i_n_tAh, so when you enter a screen that doesn't support absolute point, it shows a splash re: switching to grid mode?
18:26:45saratogawhat doesn't support pointing?
18:26:47kugelplus actually switching to grid mode (temporarily)
18:27:11kugelsaratoga: pitchscreen, graphical eq, buffering debug from the top of my head
18:27:19S_a_i_n_twouldn't it be less work to actually make the screens that don't use absolute point mode, support absolute point mode?
18:27:23 Quit sam_s (Quit: CGI:IRC)
18:27:37kugelcertainly not
18:27:43*S_a_i_n_t doesn't actually know the answer to that.
18:28:13kugelI don't use those screens so I'm not strongly motivated
18:28:27 Join stoffel [0] (
18:28:33 Quit user890104 ()
18:29:39S_a_i_n_tthe splash and mode switching seems a bit of a kludge though, and a lot like avoiding the real offence.
18:29:52 Join n1s [0] (~n1s@rockbox/developer/n1s)
18:30:01 Quit stoffel (Remote host closed the connection)
18:30:07kugelyea, of course
18:33:49 Join stoffel [0] (
18:35:04 Join Parsi [0] (~Maysam@
18:36:15 Part Parsi
18:38:26 Join Lonie [0] (
18:41:16LonieQuestion: If I have a USB flash drive (F:) and a iPod (E:) connected to my computer and run RB utility and RB utility detects the iPod as drive (F:) is this a bug?
18:41:32 Quit scorche (Ping timeout: 252 seconds)
18:41:33n1sLonie: sounds like it
18:41:42S_a_i_n_tLonie: Certainly sounds so.
18:41:50LonieOk. I should make a bug report then?
18:42:07LonieOk. thats all I needed, thanks.
18:42:22S_a_i_n_tAnd, a recipie to reproduce, or, as much detail as possibly about OS/etc
18:42:33 Join scorche [0] (~scorche@rockbox/administrator/scorche)
18:42:54 Quit Lonie (Client Quit)
18:43:20 Quit roolku ()
18:56:12***Saving seen data "./dancer.seen"
18:57:30*S_a_i_n_t reads (the rather long winded)
18:57:53S_a_i_n_tIs it really the case that DB auto-update doesn't work unless dircache is enabled?
18:59:43n1sno, that was changed a pretty long time ago
18:59:58n1sit used to be that way but it confused users
19:00:04S_a_i_n_tapparently he learnt if from the, that may need changing.
19:00:05n1snow it's just dog slow
19:00:31n1smay need indeed
19:01:02S_a_i_n_tI shall look forst, and see if it indeed does say that.
19:02:10n1salso as to why it isn't default is because it would be annoying to filebrowser users if that thing kept updating all the time
19:02:46 Quit GodEater (Ping timeout: 260 seconds)
19:03:38S_a_i_n_tHmmm..maybe DB auto-update should be set to yes (and, removed as a setting altogether) if the DB is initialised?
19:04:06S_a_i_n_tthat way, noone suffers
19:05:10pixelmathere are also targets that have the database but not dircache - and as has been said, the auto-update is dog slow with it.
19:05:35S_a_i_n_tAh, yes...good point.
19:06:08n1sS_a_i_n_t: there's no way to disable the database after initializing it short of deleting the files
19:07:02pixelmathere is also this long-standing bug that the database doesn't initialise if there is no audio file on the available disk(s), not sure what happens if e.g. someone has music only on the microSD, database initalised and for some reason ejects the card with auto-update enabled
19:07:15S_a_i_n_tand yes, the manual does indeed say that dircache needs to be anabled for this to work.
19:07:41n1sthen it's lying(sp?)!
19:08:21S_a_i_n_tI'll post that much at least, to that thread..though, with a flash target I don't see why anyone wouldn't have dircache enabled.
19:08:44gevaertsS_a_i_n_t: huh? With a flash target dircache is *less* needed
19:09:02S_a_i_n_tgah...I meant to say auto-update.
19:09:07S_a_i_n_tI confused myself ;)
19:09:17n1si never saw the reason to have dircache on flash targets at all but that is apparently just me...
19:09:40S_a_i_n_tyeah...I ballsed that up.
19:09:58S_a_i_n_tIt provided a speedup, but, it is rather negligible
19:10:01n1swe can just ask the user whether to enable suto update after initing the db
19:10:33S_a_i_n_tn1s: That seems a good idea.
19:10:36gevaertsn1s: "after" won't work, since it might be in background
19:10:44gevaerts"at the time of" would though
19:11:17S_a_i_n_twell, after the "decision" to init the DB has been made, but *before* the actual init.
19:11:28 Join GodEater [0] (
19:11:28 Quit GodEater (Changing host)
19:11:28 Join GodEater [0] (~bibble@rockbox/staff/GodEater)
19:11:53 Quit n1s (Read error: Operation timed out)
19:12:37S_a_i_n_tThat actually seems like rather a nice solution, if you can call it that, as there isn't really a "problem" as such.
19:12:51 Quit mischasworld (Ping timeout: 252 seconds)
19:17:35 Join lestatar [0] (
19:17:50 Join bertrik [0] (~bertrik@rockbox/developer/bertrik)
19:18:36saratogawhich targets don't have dircache?
19:20:12 Join Jerom [0] (~heidi@
19:20:35 Join mischasworld [0] (
19:21:39docgnomeyay! my fuze should be here soon
19:23:22bertrikhm, +20% runtime on AMSv1 with FS #11597 sounds nice
19:23:23 Quit swilde (Quit: ERC Version 5.3 (IRC client for Emacs))
19:23:37 Join Horscht [0] (~Horscht@xbmc/user/horscht)
19:24:36bertrikI should benchmark that on my clipv1
19:24:54 Part Samosa ("Leaving")
19:26:05pixelmasaratoga: low mem
19:26:09saratogaah ok
19:26:15saratogawell why not make it target specific then
19:26:22saratogashould be easy enough to ifdef the defaults
19:26:57S_a_i_n_tsaratoga: Is that in reference to my "enable dircache on DB init" suggestion?
19:27:24saratogai was going to say just make dircache on by default on all the targets where it helps
19:27:48S_a_i_n_tapparently it messes with Filebrowser though?
19:28:05gevaertsdircache is slightly annoying on players with too many files
19:28:07pixelmadircache can take away quite some RAM, and if someone is fine with initialising and re-initialising instead of updating the DB (as e.g. I am)
19:28:12S_a_i_n_tso I thought, why not enable it if the user has chosen to use DB
19:28:16saratogahow much RAM does it take?
19:28:27pixelmadepends on the number of files
19:28:47saratogais there an upper limit on how much it will allocate?
19:28:48gevaerts400K on my player
19:29:29saratogais it actually a cache or is it just remembering the entire directory structure?
19:29:55gevaertsIt remembers the entire structure IIUC
19:30:08pixelmagevaerts: on your "player"?
19:30:18gevaertspixelma: yes, with a small p :)
19:30:25gevaertsGigabeat F to be precise
19:30:46gevaertsThat's current usage by the way. Maximum seems to be around 7MB there
19:31:24gevaertsAnd it's for 5273 files
19:33:23pixelmawow, over 2 MB currently on my M5 (with a ton of data files currently though)... makes a different with just 16 MB RAM (where Rockbox itself already takes some)
19:34:05gevaertsYes, there *are* valid reasons to disable dircache
19:34:27pixelmadifference too
19:37:08pixelmadircache info states "Limit: 6291456 B" and "Entry count: 33475"
19:38:31kugelthe battery savings through dircache propably outdo savings with that bit more available ram regardless of the size
19:38:42 Quit Rob2222 (Ping timeout: 260 seconds)
19:39:11pixelmaI doubt it in this case... most music is flac and it's an SSD
19:39:48gevaertsIf you usually listen to long tracks, but you happen to have the hvsc on your player and you also use it to transport unrelated files, dircache is probably a loss
19:40:02pixelmawould be worse when just listening if it was an HDD
19:40:16 Join jgarvey [0] (
19:50:38kugelwell, I assumed hdd when I made my claim
19:51:33 Quit bertrik (Read error: Connection timed out)
19:52:07 Join bertrik [0] (~bertrik@rockbox/developer/bertrik)
19:52:55pixelma1/8th less audio buffer (even more) is probably quite noticeable if it has to spinup at least once on every typical "popular" audio track as flac
19:53:03 Quit mischasworld (Ping timeout: 265 seconds)
19:53:43pixelmaand not browsing too much
19:57:52*amiconn only enabled dircache on targets with 32MB+ ram and a real hdd
19:58:22amiconnFor me it makes no sense to enable it on a 16MB target, and neither for ssd/ cf
19:59:11amiconnOtoh if someone uses the db, it makes sense to enable dircache even if not using auto-update, because dircache enabled allows the db to be committed "live"
20:00:45gevaertsWasn't there also something about detecting removed files?
20:25:15saratogathe manual says its needed to detect when a file is removed
20:25:21saratogafor the database that is
20:26:53saratogait sounds like enabling direcache by default is a bad idea except possibly on a few players (64MB players ...), but enabling it when the database is on makes more sense
20:27:20 Quit bmbl (Quit: Bye!)
20:27:39S_a_i_n_tI think asking whether you'd like to enable it, when initialising the DB is a good idea.
20:27:51S_a_i_n_t(if it isn't enabled already)
20:28:48gevaertsExcept that asking many questions is often considered bad UI design isn't exactly going to be often.
20:29:36S_a_i_n_tonce per theory.
20:29:44saratogai'd be ok with just forcing it on whenever the database is enabled
20:30:15saratogaat least until the dependency on the dircache is removed from the database
20:30:31S_a_i_n_twell, so would I...but, I can bet there'll be those that'd like to be given the option.
20:30:45saratogalet them improve the database then :)
20:30:54S_a_i_n_tI don't mind it being forced, but, I think being asked would be nicer.
20:30:55saratogawhere is dircache anyway
20:31:08S_a_i_n_tyou mean...the setting?
20:32:10S_a_i_n_tsettigns - general settings - system - disk - directory cache.
20:32:10saratogaah filetree.c
20:32:18saratogaone of those not entirely obvious file names the source, gotcha
20:34:16saratogaalso tree.c
20:34:37 Quit bertrik (Read error: Connection timed out)
20:35:07saratogaso basically all the memory used by dircache is just storing the filenames
20:35:11 Join bertrik [0] (~bertrik@rockbox/developer/bertrik)
20:35:30 Join phixxor [0] (
20:36:31saratogai wonder if dircache could be improved to (optionally) only cache music files
20:36:35saratogaand folders with music
20:36:53AlexPdircache.ignore.unignore :)
20:36:56saratogathat might save a lot of memory in the case that you store other files on a hard disk, not to mention save memory when indexing the .rockbox folder
20:37:07saratogado we have dircache.ignore?
20:37:17S_a_i_n_tprobably...perhaps another "magic" file? like database.ignore
20:37:25gevaertsActually, I'd say indexing .rockbox shouldn't be disabled
20:37:27S_a_i_n_tNo, there isn't.
20:37:37*gevaerts wants to get his codecs *fast*
20:38:06S_a_i_n_tget a flash player then ;)
20:38:09phixxorhey −− I don't want to interrupt, so don't feel like you have to answer right away or at all −− but can I get help with this BASH sentence so I can understand how to use it for my needs?: find . -type f | sed -e 's#^.#//music#g' > allfiles.m3u
20:38:55S_a_i_n_tphixxor: Not really rockbox it?
20:39:16S_a_i_n_tthere is probably better places for help with bash scripts than here
20:39:32phixxorI'm trying to make a .m3u playlist for rockbox that I read about here: so yes.
20:39:34kugelsaratoga: that would defeat its purpose, which is ,for the most part, speed up file browsing
20:40:17saratogakugel: yeah but you don't need .rockbox/codecs/* to browse fast, just folders with actual music in them
20:40:28gevaertssaratoga: uhm you don't?
20:41:01phixxorS_a_i_n_t, i'm just trying to figure out how to make playlists for rockbox using the faq, and I came here b/c I didn't understand it
20:41:13kugelno you need to cache everything, because otherwise you'd always need to look for changed files in the folders, meaning a disk spin up for every folder
20:41:54saratogayou need to cache everything in a folder, but if you don't cache some folders you don't have to spin up until someone browses them
20:42:05AlexPphixxor: #rockboc-community would be a better place for that
20:42:20phixxorAlexP, thanks, I didn't know that channel existed :)
20:42:26AlexPno worries :)
20:42:35AlexPer, #rockbox-community of course
20:42:37S_a_i_n_tphixxor: If you just want to create a playlist of all your music it on the player itself.
20:42:43S_a_i_n_tit'll take like a few seconds.
20:42:59phixxorS_a_i_n_t, that's not what I want to do really −−- are you in -community too?
20:45:30kugelsaratoga: I don't think the few couple dozen files in .rockbox make a difference vs 10k+ music files
20:46:56kugelmanaging both cached and non-cached folders will make the code bigger, slower and probably buggier too
20:47:06saratogawell its more like 300 files, but it would also be useful in cases where you have other files on your player
20:47:25saratogai imagine dumping a few thousand text files on a lower memory target probably screws up dircache
20:53:49 Part watto
20:54:46saratogawhy is updating dircache more efficient then updating the database? it seems to me that if the dircache can figure out if a file changed well enough, shouldn't the database be able to do the same in roughly as much time?
20:55:07 Quit bertrik (Read error: Connection timed out)
20:55:37 Join bertrik [0] (~bertrik@rockbox/developer/bertrik)
20:56:13***Saving seen data "./dancer.seen"
20:57:12 Join Lear [0] (chatzilla@rockbox/developer/lear)
21:04:46kugelsaratoga: dircache reads only the fat, tagcache reads the file's content
21:05:23saratogakugel: yeah but why is detecting deleted files so hard without dircache then? presumably it just has to read the fat?
21:05:34 Join mischasworld [0] (
21:05:44saratogaas i understand it the issue is that the database needs dircache to find deleted files otherwise its unusablely slow
21:05:52kugeldircache is blazing fast (it's already rebuilt after usb disconnect), tagcache is a lot slower
21:06:37 Join fyrestorm [0] (
21:07:11kugelsaratoga: because the file_exists() function is a dircache lookup; if there's no dircache it opens the file (and closes again) and returns open()s success (I don't know if open already exceeds reading only the fat or not)
21:07:44saratogakugel: yeah but why is that fast for dircache, but doing the same operation in tagcache isn't?
21:07:49*gevaerts points out that dircache does *not* look only at the fat
21:07:59 Join Kitr88 [0] (
21:08:07 Quit fyrestorm (Client Quit)
21:08:19gevaertssaratoga: I suspect there must be something in dircache that marks deleted files while it does its on-boot scan
21:09:28gevaertsWhile if you take the naive approach, detecting deleted files means doing a lookup for each file in the database
21:09:55kugelgevaerts: where else?
21:10:13saratogamaybe this wouldn't work, but what if "dircache" kept a 4 byte hash of each file name whenever the database was enabled (but only kept the file name if dircache was enabled), that would make the size of the cache negligible if dircache was "disabled", but still make it pretty fast for the database
21:10:16gevaertsBasically dircache is indexed per directory, which makes this easy, while the dababase is indexed in tag-optimised ways
21:10:20gevaertskugel: the directories
21:10:49gevaertsThe fat is really only this block bitmap like structure
21:10:55saratogathen you always update dircache, but only save the file names if the directory cache option is on
21:11:15kugelI thought directories only exist in the fat
21:11:49kugelsaratoga: dircache doesn't scan for deleted files. it's simply rebuilt
21:11:53saratogathey're files on the hard disk IIRC
21:12:05 Quit Kitar|st (Ping timeout: 276 seconds)
21:12:20saratogakugel: i know
21:12:36gevaertsYes, basically directories are files containing file information. The fat only tells you that a file that starts at cluster X occupies clusters X, Y and Z
21:14:03kugelsaratoga: so, tagcache and dircache aren't doing the same operation
21:14:58saratogakugel: i realize they're not doing the same thing, but if dircache can figure out when a file is deleted, I'm confused why its so hard for the database to perform that same specific task
21:15:14saratogaintuitively i would expect the same task on the system to take the same time
21:15:43 Quit Kitr88 (Ping timeout: 255 seconds)
21:15:50kugelit figures out because the delete operation goes through the dircache (remove it from the cache), then to the filesystem (remove it from the disk)
21:16:15saratogai don't think thats true in UMS mode at least
21:16:32kugelright, that's why dircache is rebuilt on usb disconnect
21:17:02saratogawell why is scanning dircache on USB disconnect apparently fast, but rescanning for deleted files in the database not?
21:17:11Learsaratoga: But as mentioned above, to detect if a file is deleted you have to see if it exists or not. A lookup in the dircache is quick, stepping through FAT folders on disk isn't.
21:17:44saratogaif dircache is out of date you have to step through the FAT folders ...
21:18:11kugeldircache never out of date
21:18:33saratogait is on USB disconnect
21:18:39saratogaat least until it finishes updating
21:18:42kugelall file operations go through dircache, and it's rebuilt when we lose control over the disk (power off, UMS)
21:19:06LearHm, I think I see your point now. The tagcache should do something like this: scan the disk, similar to dircache, and flag any cache entries it finds. When done, remove all unflagged entries.
21:20:14saratogaif dircache and tagcache both need to do this operation, and dircache somehow does it better, maybe it makes sense to merge them and just not allocate memory for disabled features
21:20:36kugelthat's completely different though. dircache doesn't need to flag existing entries, it just trashes and rebuilds.
21:20:56LearCurrently it does it the other way around: scan the tags, and try to open each file (this involves navigating the FAT structures many times per folder).
21:20:57 Join Kitar|st [0] (
21:21:02kugelit doesn't need file_exists(), which is slow without dircache
21:21:45saratogaah I see, its because its doing the operation in database order, rather then FAT order?
21:22:29kugelLear: what about non-music files?
21:22:30LearYes. And it can be done even if the tagcache isn't in RAM, it seems, so you either need to search the FAT or the tagcache for the right entry.
21:22:41Learkugel: What about them?
21:23:01kugeleither you scan/mark all files or you need to look at the file content to determine
21:24:05 Quit bertrik (Read error: Connection timed out)
21:24:46 Join bmbl [0] (~Miranda@unaffiliated/bmbl)
21:28:42 Join Judas_PhD [0] (
21:28:55 Quit Tuplis (Ping timeout: 245 seconds)
21:29:39 Join Kitr88 [0] (
21:32:59 Quit Kitar|st (Read error: Connection reset by peer)
21:33:18 Quit Judas_PhD (Client Quit)
21:37:00 Quit esperegu (Read error: Connection reset by peer)
21:37:15 Quit Kitr88 (Ping timeout: 240 seconds)
21:38:11 Join Kitar|st [0] (
21:39:12 Join Kitr88 [0] (
21:42:21 Quit Kitar|st (Ping timeout: 240 seconds)
21:43:32 Join wombat23_ [0] (
21:46:31 Join balintx_ [0] (
21:46:32 Join bertrik [0] (~bertrik@rockbox/developer/bertrik)
21:47:06 Join ack [0] (
21:47:20 Quit ack` (Remote host closed the connection)
21:47:25 Quit balintx (Write error: Broken pipe)
21:47:33 Quit wombat23 (Remote host closed the connection)
21:48:42 Join user890104 [0] (
21:56:16 Join robin0800 [0] (
22:22:45 Join mischa [0] (
22:24:20 Quit mischasworld (Ping timeout: 265 seconds)
22:24:55 Quit mischa (Client Quit)
22:25:01 Join mischasworld [0] (
22:28:07 Quit kugel (Remote host closed the connection)
22:29:01 Quit mischasworld (Client Quit)
22:30:03 Join mischasworld [0] (
22:31:23 Quit Kitr88 (Ping timeout: 260 seconds)
22:38:01 Join Kitar|st [0] (
22:38:07 Quit Lear (Quit: ChatZilla 0.9.86 [Firefox 4.0b6pre/20100901031143])
22:40:02 Quit robin0800 (Quit: Leaving)
22:41:09 Quit Jaykay (Ping timeout: 240 seconds)
22:50:35 Quit Kitar|st (Ping timeout: 240 seconds)
22:56:17***Saving seen data "./dancer.seen"
22:56:32 Join Kitar|st [0] (
23:03:51 Quit Kitar|st (Ping timeout: 240 seconds)
23:14:01 Join webguest77 [0] (
23:15:50webguest77hi folks
23:16:22webguest77im tryin to compile rbox for philips hdd1620 in ubuntu
23:16:29webguest77like this
23:16:58webguest77but it gives "arm-elf-eabi-gcc not in path" error
23:17:28Bagderand it _is_ in your path?
23:18:28 Quit stoffel (Remote host closed the connection)
23:18:30webguest77i guess
23:18:39webguest77i guess
23:18:43Bagderthen we know what the problem is!
23:18:51AlexPDoes it exist at all? What dev environment are you using?
23:19:22webguest77sorry im kinda new in linux
23:19:34webguest77i simply run terminal in ubuntu 10.04
23:19:50AlexPSo you ran tools/ to build the cross compilers?
23:20:29webguest77i didnt
23:20:40webguest77so is that the problem
23:20:46AlexPThen you don't have any of the compilers
23:20:55AlexPSo yes, you can't compile without the compilers :)
23:21:10AlexPGo into the directory where you checked out the Rockbox source code
23:21:22AlexPAnd run "sudo tools/"
23:21:42webguest77ok i try
23:22:21 Join Kitar|st [0] (
23:22:54AlexPIf you just want to build for the hdd1620, the only option you need from the list of compilers is "e"
23:23:38webguest77it gives target not found
23:23:58S_a_i_n_twebguest77: the link above should give you some pointers
23:24:00AlexPwebguest77: What does?
23:24:26AlexPwebguest77: When you type "sudo tools/", what happens?
23:24:40webguest77that happens
23:24:44webguest77cpu not fund
23:24:50webguest77manifacturer notfound
23:24:55AlexPI can't see your screen, you need to be precise
23:25:10webguest77yzx@yz:~$ sudo "/home/yzx/rockbox/tools/" /home/yzx/rockbox/tools/ 1: Target:: not found /home/yzx/rockbox/tools/ 2: Target: not found /home/yzx/rockbox/tools/ 3: Target: not found /home/yzx/rockbox/tools/ 4: Memory:: not found /home/yzx/rockbox/tools/ 5: CPU:: not found /home/yzx/rockbox/tools/ 6: Manufacturer:: not found /home/yzx/rockbox/tools/rockboxde
23:26:08AlexPDo you have build-essential installed?
23:26:19AlexP"sudo apt-get build-essential"
23:26:49AlexPer, apt-get install
23:27:04AlexPIt's a while since I used a debian based distro :)
23:28:12webguest77okay im installing it now
23:28:26AlexPYou might need to make executable as well "chmod +x tools/"
23:28:50AlexPThere also might be some more dependencies needed - have a read of the page S_a_i_n_t linked, and work through it
23:29:35 Quit Kitar|st (Ping timeout: 265 seconds)
23:29:40webguest77okay thanks
23:30:01webguest77none of these are mentioned in hdd6330 page
23:30:14AlexPNo, as it assumes you have a build environment set up
23:30:20AlexPThis is the same for all targets
23:30:34AlexP(well, different cross-compilers, but same process)
23:33:49 Quit domonoky (Read error: Connection reset by peer)
23:34:41 Quit krazykit (Quit: Connection reset by beer)
23:37:50 Join krabador [0] (
23:40:32CIA-81New commit by alle (r27989): Change the file name extension for the shopper plugin from 'list' to a more descriptive 'shopper'
23:41:21krabadorwhat's are focusing you developers this times?
23:42:23CIA-81r27989 build result: All green
23:42:39AlexPWhatever people want
23:43:05AlexPThere is no roadmap or whatever, devs work on what they fancy
23:47:28 Join Kitar|st [0] (
23:47:34 Join krazykit [0] (~kkit@
23:49:37S_a_i_n_twow....I think ".shopper" needs to be awarded a trophy for "most long-winded file extension in Rockbox so far"
23:50:06S_a_i_n_tshoppinglist.shopper seems a little weird, but, understandable at least.
23:50:33S_a_i_n_tI think I would have gone for ".shop"
23:52:10 Join fml [0] (
23:53:06fmlS_a_i_n_t: I was unsure about "shopping". I didn't like "shop"
23:53:48krabadorAlexP, no 4.0 milestones?
23:54:15 Join wodz [0] (
23:54:16AlexPwell, android probably
23:54:29AlexPBut there really aren't any plans or anything
23:55:38wodzI bought "faulty" MPIO HD300 on ebay so I hope to back it to life and port rb to another target :-)
23:55:49CIA-81New commit by alle (r27990): Add a manual entry for the shopper plugin. More details are in the wiki.
23:56:01gevaertsAh, a plan! :)
23:57:14CIA-81r27990 build result: All green
23:57:22 Quit webguest77 (Quit: CGI:IRC (Ping timeout))
23:58:52 Quit jgarvey (Quit: Leaving)

Previous day | Next day