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 2012-01-15

00:01:46 Join Galois [0] (
00:02:38 Quit pixelma (Remote host closed the connection)
00:02:49 Join pixelma [0] (quassel@rockbox/staff/pixelma)
00:02:49 Join amiconn [0] (quassel@rockbox/developer/amiconn)
00:04:07 Join Topy44 [0] (
00:04:22 Quit Topy (Ping timeout: 240 seconds)
00:18:33 Join robin0800 [0] (~robin0800@
00:34:03 Join T44 [0] (
00:34:47 Quit T44 (Client Quit)
00:37:02 Quit Topy44 (Ping timeout: 240 seconds)
00:48:37funman[Saint]: you have your nano2g ?
00:48:57Mirfunman: are you a dev?
00:50:01funmanbluebrother: hm that means i broke the build for other manuals?
00:50:19 Join saratoga_ [0] (980329c2@gateway/web/freenode/ip.
00:50:52funmanah i see it was fixed
00:51:04funmanMir: yes (you can /whois people to see that)
00:51:44funmani didnt add a check for the remote because there was no remote image for the zip
00:52:24saratoga_whats the status of the cabbie port to the Zip?
00:52:35funmani did not try it
00:54:39funman[Saint]: i sent a patch for nano2g on the mailing list some days ago, could you test it?
01:00:14 Join captainkwel [0] (
01:03:23 Quit lebellium (Quit: ChatZilla 0.9.88 [Firefox 10.0/20120104111456])
01:03:29***Saving seen data "./dancer.seen"
01:03:42 Quit bertrik (Quit: And That, My Liege, Is How We Know the Earth to Be Banana Shaped)
01:17:38 Join domonoky1 [0] (
01:17:38 Quit domonoky (Read error: Connection reset by peer)
01:25:17 Join bsidhom [0] (
01:26:47bsidhomI have not checked on any other devices, but the battery benchmark plugin appears not to work for the Sansa Clip Zip.
01:27:24bsidhomIs it possible to debug the battery bench plugin in a simulator?
01:28:04funmanit worked for me last time i tried it
01:28:19bsidhomhow recent was your build?
01:28:46bsidhomthe first one I tried was from Jan. 9th
01:29:43saratoga_what output does it generate?
01:30:24bsidhomit flashes "Cannot create file!" and terminates
01:32:35bsidhomI get the same behavior on the most recent git version as well
01:35:05saratoga_oh i know what that is
01:35:09 Quit petur (Quit: Leaving)
01:35:26saratoga_FS #12500
01:35:27fs-bluebot /./ is invalid when not using dircache, so HOME_DIR breaks things (bugs, new)
01:35:32gevaertsAh, the usual
01:35:45gevaertsThat *really* needs to be fixed yesterday
01:38:19jhMikeSI would but I doubt anyone would like how I'd do it, so I'll leave it to another less radical programmer
01:38:32saratoga_whats wrong with dreamlayers patch?
01:38:55funmanah the same "disk full" i had when recording ..
01:39:25*jhMikeS hates that error since it really just an assumption
01:39:35 Quit factor (Read error: Connection reset by peer)
01:40:04funmanjhMikeS: why that first comment?
01:40:16gevaertssaratoga_: I don't know, but I also don't know wht HOME_DIR can't be /
01:40:55bsidhomthe problem still happens for me with dircache enabled; is this expected?
01:42:10gevaertsjhMikeS: to be honest, it's getting to the point where I think radical fixes are fine for this
01:44:42jhMikeSfunman: which first comment?
01:45:18funman19:38 < jhMikeS> I would but I doubt anyone would like how I'd do it, so I'll leave it to another less radical programmer
01:45:34*jhMikeS shrugs
01:46:40jhMikeSit seems like a rabbit hole of sorts
01:46:48funmanalso dreamlayers' patch looks simple
01:47:32 Join perrikwp_ [0] (
01:47:55jhMikeSis it thorough enough?
01:48:08funmanHOME_DIR could be just "/"
01:48:16funmanwhat says "//" is invalid?
01:48:37gevaertsfunman: // is invalid on hypothetical windows-based RaaA builds
01:49:07gevaertsbut I don't see why that would matter here
01:50:09saratoga_what about the windows sim?
01:50:21funmanjhMikeS: looks sane to me
01:50:34gevaertssaratoga_: the windows sim has the simdisk path prepended
01:50:42 Quit perrikwp (Ping timeout: 260 seconds)
01:51:04gevaertsAnd as far as I know, // is fine on windows if it's not at the start
01:51:18jhMikeSwhat tries to access "//"
01:51:22funman says "/" is forbidden, period
01:51:44gevaertsfunman: in a file, not in a path
01:52:17gevaertsjhMikeS: various code does does things like concatenating HOME_DIR and "/something"
01:53:46jhMikeSmaybe it shouldn't?
01:54:03gevaertsFixing that is no fun
01:54:20gevaertsAnd seriously, it's not actually a problem
01:55:23jhMikeStechnically, any number of "/" in a row should be collapsed to "/" except in special circumstances
01:56:46 Join factor [0] (~factor@
01:56:58 Quit domonoky1 (Read error: Connection reset by peer)
02:03:47bsidhomOK, well thank you for the temporary workaround
02:03:58 Quit bsidhom (Quit: leaving)
02:11:29 Quit rasher (Quit: snagra[SF]_)
02:11:44 Join rasher [0] (~rasher@rockbox/developer/rasher)
02:18:02 Quit dfkt (Quit: -= SysReset 2.55=- Sic gorgiamus allos subjectatos nunc.)
02:18:39 Quit KiwiCam (Read error: Connection refused)
02:22:24 Join KiwiCam [0] (
02:29:43CIA-81Commit by amaury.pouly (8cadb58): fuzeplus: fix lcd-target.h (LCD_FRAMEBUF_ADDR must point to lcd_framebuffer and not FRAME)
02:29:44CIA-81Commit by amaury.pouly (35ba39e): imx233: add DCP driver (only memcpy implemented), move channel arbiter to kernel-imx233
02:29:45CIA-81Commit by metaphysiciendouteux (7f26a10): fuzeplus: update plugins keymaps (FS #12405)
02:35:28[Saint]That nick sure is a mouthful.
02:39:21 Quit robin0800 (Ping timeout: 252 seconds)
02:39:39 Quit Thra11 (Ping timeout: 252 seconds)
02:45:46 Join robin0800 [0] (~robin0800@
02:46:13funmani just noticed some commits don't have the gerrit change id
02:46:35funmanshould we just ignore this (not sure what is its purpose anyway) or enforce it on git server if it's really needed?
02:46:49 Quit [Saint] (Quit: I know its a sin to kiss and swallow.)
02:47:09 Join [Saint] [0] (~Saint]@unaffiliated/saint/x-8516940)
02:47:33funmanfor example my manual commit and the subsequent commit by domonoky
02:48:04funman5ef27368f1bcbe31fb27072983d7a29df8de6845 etc .. it's quite inconsistent
02:48:08funmanTorne: ^
02:48:47funmanok cia gets names from the email, it should use real names probably
02:49:26funman[Saint]: ping nano2g
02:56:09 Quit beslayed (Remote host closed the connection)
02:57:18 Join beslayed [0] (~user@
03:00:20 Quit KiwiCam (Ping timeout: 248 seconds)
03:02:31 Quit robin0800 (Ping timeout: 244 seconds)
03:03:32***Saving seen data "./dancer.seen"
03:06:48 Quit beslayed (Remote host closed the connection)
03:07:34 Quit pamaury (Remote host closed the connection)
03:07:40 Join robin0800 [0] (
03:10:02 Quit n1s (Ping timeout: 240 seconds)
03:19:11 Join KiwiCam [0] (
03:27:26 Join beslayed [0] (~user@
03:29:48 Quit robin0800 (Ping timeout: 255 seconds)
03:32:42 Quit jhMikeS (Ping timeout: 240 seconds)
03:34:11 Join jhMikeS [0] (
03:34:11 Quit jhMikeS (Changing host)
03:34:11 Join jhMikeS [0] (~jethead71@rockbox/developer/jhMikeS)
03:40:08 Quit Amqui (Ping timeout: 252 seconds)
03:41:13 Join FrenchVerbs [0] (
03:47:58FrenchVerbsHi everyone, I had a question I thought you might be able to help with. Is there a way I can get a certain font in a larger size? I'm looking for Artwiz-Snap in a 12 or 13 point. How would I get that?
03:49:08 Join robin0800 [0] (~robin0800@
03:49:10[Saint]You'd need to generate it yourself.
03:49:45[Saint]The font .bdf and convbdf are in the sources.
03:49:45FrenchVerbsHow would I go about that?
03:50:01FrenchVerbsWhat is involved in editing them?
03:50:35[Saint]Checking out the source tree in whole or in part, and re-converting the font.
03:51:13FrenchVerbsSorry, I'm sort of a noob. What is that in layman's terms?
03:51:48[Saint]I'm not sure how much more I can dumb it down sorry.
03:52:43FrenchVerbsWell, what do you mean by checking out the source tree, isn't that a little pedantic for just resizing a font?
03:53:16[Saint]You need at least artwiz-snap.bdf and convbdf, and an environment capable of running convbdf.
03:54:02[Saint]The wiki details checking out the source, and setting up a development environment.
03:54:16FrenchVerbsOkay, I'll check it out. Thanks!
03:54:21 Quit FrenchVerbs (Quit: ChatZilla 0.9.88 [Firefox 12.0a1/20120114031054])
04:02:27 Quit KiwiCam (Ping timeout: 248 seconds)
04:05:25 Quit pixelma (Disconnected by services)
04:05:25 Join pixelma_ [0] (quassel@rockbox/staff/pixelma)
04:05:34 Quit amiconn (Disconnected by services)
04:05:35 Join amiconn_ [0] (quassel@rockbox/developer/amiconn)
04:05:45 Nick pixelma_ is now known as pixelma (quassel@rockbox/staff/pixelma)
04:05:57 Nick amiconn_ is now known as amiconn (quassel@rockbox/developer/amiconn)
04:08:32 Quit TheSeven (Disconnected by services)
04:08:49 Join [7] [0] (~TheSeven@rockbox/developer/TheSeven)
04:12:38 Join Strife89 [0] (~Strife89@
04:14:31 Join KiwiCam [0] (
04:17:58 Join JdGord [0] (~AndChat@
04:28:15 Quit robin0800 (Ping timeout: 240 seconds)
04:28:51 Nick Jack87|Away is now known as Jack87 (Jack87@nasadmin/admin/jack87)
04:34:45 Quit bluebrother (Disconnected by services)
04:34:46 Join bluebrother^ [0] (~dom@rockbox/developer/bluebrother)
04:36:55 Quit fs-bluebot (Ping timeout: 240 seconds)
04:38:30 Join fs-bluebot [0] (
04:41:56 Join robin0800 [0] (
04:59:32 Join dys` [0] (
05:01:13 Quit dys (Ping timeout: 252 seconds)
05:03:34***Saving seen data "./dancer.seen"
05:13:30 Join Rob2223 [0] (
05:17:10 Quit Rob2222 (Ping timeout: 260 seconds)
05:23:26 Quit DerPapst (Quit: Leaving.)
05:32:54CIA-81Commit by boris.gjenero (12da352): Fix FS #12391 - PP502x cpucache_invalidate() causes memory corruption
05:40:03 Quit [Saint] (Remote host closed the connection)
05:40:51 Quit anewuser_ (Read error: Connection reset by peer)
05:41:59 Join [Saint] [0] (~Saint]@unaffiliated/saint/x-8516940)
05:45:01 Join dreamlayers [0] (~bgjenero@rockbox/developer/dreamlayers)
05:51:13 Quit Strife89 (Quit: Zoom)
05:54:21jhMikeSwhat's with showing sandbox commits in IRC?
05:54:48 Join Amqui [0] (
05:56:21dreamlayersI guess it can't be due to a misconfiguration on my end?
06:07:35 Quit [Saint] (Remote host closed the connection)
06:11:42dreamlayersI guess sandbox has a cia post-commit hook, and it shouldn't. I don't think there's anything I can do about that; it must be on the git server.
06:25:44 Quit robin0800 (Ping timeout: 252 seconds)
06:34:32 Join jdgord_ [0] (~AndChat@
06:34:35 Quit JdGord (Read error: Connection reset by peer)
06:43:33 Part dreamlayers
06:44:43 Join Keripo [0] (
06:48:13 Join robin0800 [0] (
07:03:35***Saving seen data "./dancer.seen"
07:10:16 Join [Saint] [0] (~Saint]@unaffiliated/saint/x-8516940)
07:29:41 Quit robin0800 (Quit: Leaving)
07:51:35 Join Bruice [0] (
07:51:43 Quit Bruice (Client Quit)
07:57:02 Nick Jack87 is now known as Jack87|Away (Jack87@nasadmin/admin/jack87)
07:58:41 Quit Amqui (Ping timeout: 248 seconds)
08:06:09 Nick Jack87|Away is now known as Jack87 (Jack87@nasadmin/admin/jack87)
08:45:12 Join nosa [0] (
08:48:20 Quit nosa-j (Ping timeout: 260 seconds)
08:48:21 Nick nosa is now known as nosa-j (
09:03:36***Saving seen data "./dancer.seen"
09:11:13 Quit evilnick (Ping timeout: 248 seconds)
09:19:46 Quit jdgord_ (Ping timeout: 252 seconds)
09:34:00bluebrother^funman: yes.
09:34:25bluebrother^or more exactly, the existing manuals would build but have an error shown because the image file is missing (which is wrong)
09:37:06 Join n1s [0] (~n1s@rockbox/developer/n1s)
09:59:48 Join perrikwp [0] (
10:02:45 Quit perrikwp_ (Ping timeout: 248 seconds)
10:09:42 Join chkktri_ [0] (~user@unaffiliated/chkktri)
10:31:43 Quit captainkwel (Ping timeout: 240 seconds)
10:46:05 Join pamaury [0] (
10:46:05 Quit pamaury (Changing host)
10:46:05 Join pamaury [0] (~quassel@rockbox/developer/pamaury)
10:48:02 Join y4n [0] (y4n@unaffiliated/y4ndexx)
10:52:47 Join liar [0] (
10:57:03kugelpamaury: 8cadb58 is strange
10:57:35kugelI don't think that's right either
10:58:37pamaurykugel: why ? The fuze+ uses a double buffering scheme, lcd_framebuffer is the buffer and FRAME is the copy that is used for dma
11:00:00kugellcd-memframe copies to that back buffer
11:00:50kugelit's basically a memcpy, where src is lcd_framebuffer and dest the double-buffer for the lcd
11:03:39***Saving seen data "./dancer.seen"
11:04:27pamaurythat makes no sense, mcd_bit_yuv takes a planar yuv representation as input, so src can't be lcd_framebuffer.
11:04:27kugellcd-memframe.c uses LCD_FRAMEBUF_ADDR() for src
11:04:39kugelerr, dest
11:04:43 Quit liar (Ping timeout: 245 seconds)
11:05:09pamaurythe fuze+ uses a special implementation for lcd_update_rect
11:05:22kugeli saw that
11:06:32kugelanyway, LCD_FRAMEBUF_ADDR() is the dest buffer, i.e. what you do dma from
11:08:53pamaurythen that's stupid, that's destroy the use of having two buffers
11:09:00pamaury*that destroys
11:09:52kugelpamaury: no why?
11:10:45pamaurybecause the fuze+ uses FRAME for the dma but does not wait for completion, it's a background refresh. If you copy to the buffer being dma'ed, your output will be crap
11:10:50kugelwell yes, it's converting yuv directly to the back buffer without an intermediate copy in lcd_framebuffer
11:11:07kugelbut that's the intention, to speed up mpegplayer
11:11:39pamauryI don't see the point, the screen is *slow*
11:11:46kugelpamaury: that's the case for all lcd-memframe targets
11:12:40kugelask jhMikeS for the details, but it's intended that yuv doesnt go to lcd_framebuffer
11:13:02kugelhe did that lcd-memframe work btw (I saw you didn't remember)
11:14:01kugelif you do yuv to lcd_framebuffer, then you need an lcd_update() afterwards no? does mpegplayer even call that?
11:14:21pamauryok then tell me: how the hell would playing a YUV work with LCD_FRAMEBUF_ADDR=FRAME ? If I just do lcd_blit_yuv, it won't suffice, I need to lcd_update_rect. But lcd_update_rect copies from lcd_framebuffer so it will overwrite the content of the screen
11:15:02*pamaury thinks lcd_blit_yuv has a stupid semantics
11:15:19 Join liar [0] (
11:16:10kugelyou don't normally mix them, except in the overlay in mpegplayer (where the updates are controlled) (?)
11:17:00pamaurythe fuze+ needs to send commands to update the screen, the memframe implementation assumes this is not the case
11:17:07 Join mortalis [0] (~mortalis@
11:18:03kugelyea it assumes it updates in the background without commands
11:18:30pamauryseems like I will need to define my own "optimized" version of it x-(
11:18:36amiconnlcd_blit_*() copies directly to the lcd, lcd_update_rect() copies from the framebuffer. You can mix them, but the areas shouldn't overlap, otherwise the last write "wins"
11:18:38kugelif you need commands you can control when updates happen, and then you don't need lcd-memframe, I'd say
11:19:49pamauryhaving a memframe is useful: you have one buffer for the software and one for the hardware, otherwise you need to wait for refresh to complete all the time
11:20:28amiconnThere's an exception when using the greylib - in this case you *must not* call other functions writing directly to the lcd while the overlay is running, but there's a replacement for lcd_update_rect() in the greylib api which can be used instead
11:21:09amiconnFor memframe implementations this means you need two buffers
11:21:18kugelpamaury: I guess you can continue this way. It's just the commit is strange because it uses lcd-memframe in a non-intended way
11:21:41*amiconn considers memframe implementations a "last resort solution" for when we don't know how to control the lcd controller directly
11:22:46pamauryyou are right on one point: the commit doesn't solve the issue, I will need to rewrite lcd_blit_yuv
11:24:43amiconnOn another matter, is there an eta for bringing the build system back up, i.e. finishing the git transition server side?
11:25:25*amiconn doesn't like the number of commits which went in without being build checked
11:27:09pamauryI admit this is a problem but refraining from commiting is not a solution either :)
11:27:37*[Saint] kinda assumed commits would halt until the build system is back online.
11:27:51[Saint]J seem to remember this being discussed.
11:27:57[Saint]*I seem
11:28:17amiconnThat's why I'm asking for an eta, not for holding back commits (although you probably won't see a commit from me for a while, I first have to figure out this git ****)
11:28:54pamauryI don't remember this beeing discussed.
11:29:43 Quit chkktri_ (Ping timeout: 245 seconds)
11:30:08[Saint]I think I do, briefly. But I don't think anyone thought it would take so long
11:32:49amiconnIirc roolku asked for a script, but if it can be done manually now I'd rather do it manually on a few clients and live with the fact that builds will take a bit longer until the script for roolku's "power farm" is ready instead of having no working build system at all
11:33:49JdGordona "bit longer"? :D
11:33:57JdGordonback to the bad old days of 45min build rounds
11:34:05 Quit [Saint] (Read error: Connection reset by peer)
11:34:20 Join [Saint] [0] (~Saint]@unaffiliated/saint/x-8516940)
11:34:32amiconnI'd expect more like 20 minutes. Still better than nothing
11:35:32 Join lovasoa [0] (~olojkine@2a01:e35:8a2e:8080:e2b9:a5ff:fe5b:ca7b)
11:40:04desowingevaerts: what were the conclusions from battery benches on m:robe 500?
11:45:23 Join stoffel [0] (
11:55:04 Join Topy44 [0] (
11:55:39Tornefunman: the change id is to identify uploads to gerrit that are subsequent versions of the same change. if people don't have the hook installed it will behave brokenly if they upload anything for review.
11:56:26Tornefunman: it's not needed for things you push directly, but there's no way for the hook to know whether you are going to push directly or not
11:57:09Tornefunman: there is a checkbox to require it, thoguh, so i dunno
11:58:03kugelcan we check-in the hook somehow?
11:58:45 Join TheLemonMan [0] (
11:59:04Tornegit doesn't have any way of doing that
11:59:35Tornei think if i require it, it will reject it in allpushes if you don't have it
12:02:44Tornei guess that's probably reasonable
12:02:51Torneif you've followed the instructions you will have one :p
12:06:51*[Saint] wants to say something witty about the likelihood of that happening, though it probably need not be said ;)
12:07:19gevaertsdesowin: that my mr500 battery is in a bad shape. I can't get consistent results
12:07:41JdGordonis there a standard git way of referring to another git commit? i.e the equivilant of rXXXX in svn?
12:07:52TorneJdGordon: the hash
12:08:02Tornetraditionally, 7 characters
12:08:15Tornethough if you are unlucky you may need to include more to be unambiguous
12:08:33JdGordonyeah, but the random looking hash isnt so easy for scripts to figure out its a hash and not a word
12:08:56Torneoh. then no.
12:09:22gevaertsit's reasonably likely to have a digit in it
12:09:28Tornealso, [0-9a-f]{7,} is unlikely to match many words
12:10:11Torneyou can reasonably assume a minimum of 7; git won't print a shorter hash by default anywhere even if a shorter one is unique
12:10:36Tornesee git rev-parse −−short 7f26a
12:10:38Torneor similar :)
12:12:15Torneacceded, defaced and effaced are the only matches in my /usr/share/dict/words
12:12:57gevaertsYay, my trigger works :)
12:13:55JdGordonthats abot the same false-positives as the current r[0-9]* regex
12:14:05JdGordonmost commonly hi with yp-r0 and mr500 :)
12:14:52Tornei'd suggest you also include a wordboundary on either side
12:15:00Tornewhich would stop mr500 hitting :p
12:15:31kugeli mostly write ypr0 too :)
12:16:06Torneis a much better regex for svn revision numbers
12:16:13Torneand won't match either of those players :0
12:16:20JdGordontell Zagor :p
12:16:28Torneso yeah, \b[0-9a-f]{7,}\b for git SHAs
12:16:51JdGordonr0 and r1 in asm instructions were also a "funny" false-positive
12:17:06Tornewell mine will still match r1, but hey
12:17:22Torneyou could probably require three digits at this point without losing anything interesting :p
12:17:57kugelTorne: rXXXX regexes became irrelevant :)
12:18:02kugelstop talking about them :P
12:18:19kugel(even the other project where I used them for switched to git recently)
12:19:20 Join evilnick [0] (~evilnick@rockbox/staff/evilnick)
12:19:51pixelmawould have been cool if the build system was also prepared beforehand but whatever
12:23:20*[Saint] wonders, if asked at the time, if anyone at devcon saw the transition taking this long.
12:24:17*bluebrother^ guesses a no
12:24:34bluebrother^wow, wavtrim in voicefile creation in Rockbox Utility is completely ignored :o
12:24:38Tornei expected it to take more or less this long, yes :)
12:25:03Tornebecause i guessed that i would end up writing all the docs and stuff :)
12:25:39[Saint]That's not really been the major holdup though.
12:26:23Tornethere haven't been major holdups, really
12:26:35Tornethere's been a list of things that need doing and they have just happened weeks apart from each other
12:26:44[Saint]A general feeling of disinterest seems to be a
12:26:59[Saint]major player.
12:27:18bluebrother^well, svn was working nicely for us, wasn't it?
12:27:52[Saint]Many seemed to think so, more didnt.
12:28:06[Saint]Or, more vocal people didn't
12:28:20bluebrother^mostly the latter :)
12:28:25 Quit liar (Read error: Connection timed out)
12:28:45bluebrother^and while git definitely has advantages svn wasn't that bad. Especially when compared to cvs :D
12:29:14bluebrother^hmm. Is there a see which commits I've already pushed?
12:29:56Torne`git log origin/master..` will show any commits that exist in your local branch but not in the remote
12:30:45y4ngit <3
12:30:48bluebrother^ah, nice. Exactly what I was looking for
12:31:10 Join chkktri_ [0] (~user@unaffiliated/chkktri)
12:32:36CIA-81Commit by Dominik.Riebeling (9db5c12): Fix wavtrim on voicefile creation.
12:35:15Tornebluebrother^: further to the discussion fractionally earlier about change-id, you don't have the change-id hook installed :)
12:35:26Tornewhich means you didn't read the instructions ;p
12:36:03*pixelma did but has nothing to commit and even if wouldn't have the courage yet
12:36:31pixelmabesides I don't know if I'd be allowed to yet
12:37:02bluebrother^Torne: oh. Will fix.
12:37:17bluebrother^but to be more exact: I don't have the hook on *this machine* installed :)
12:39:13bluebrother^the (slighly) annoying thing when (having to) use multiple machines
12:39:42Torneit would be nice if you could git clone −−with-hooks or something, yes
12:40:01 Join bertrik [0] (
12:40:01 Quit bertrik (Changing host)
12:40:01 Join bertrik [0] (~bertrik@rockbox/developer/bertrik)
12:40:03Tornei mean, it doesn't normally for fairly obvious reasons, but *choosing* to isn't unreasonable :)
12:42:06bluebrother^I wouldn't mind gerrit requiring the hook before pushing. Would make it much more obvious if one forgot to install the hook
12:42:30Torneyeah, i dunno. i didn't require it because frequently it's *not required*
12:43:01Tornebut it might be nicer to
12:43:31kugelsaratoga_: you just change your local git config
12:44:55TorneSynrG: the arguments in the thread are mostly irrelevant :)
12:45:21*SynrG shrugs
12:45:30SynrGjust found it interesting, including some suggested workarounds
12:45:50SynrGe.g. ln -s ../git-hooks
12:46:06Tornethere aren't any workarounds
12:46:10Tornethat doesn't solve anything
12:46:17Torneyou still have to create the symlink in every repo
12:46:21 Quit Mineo (Read error: Connection reset by peer)
12:46:26bluebrother^hmm, maybe it makes sense to put the hook into something like tools/git, and a simple shell script that copies it to the right location?
12:46:28Torneat which point you can just as easily copy hte hook from the server
12:46:38bluebrother^or have a shell script that creates the hook?
12:46:51SynrGor make it a part of the Makefile for the project
12:46:54bluebrother^would at least save from looking up the server location each time :)
12:47:09TorneSynrG: that's not a useful/relevant concept for a commit-msg hook
12:47:14bluebrother^SynrG: doesn't work.
12:47:19Tornethey are talking about hooks that are validating whether the code is okay to submit
12:47:33Tornethe hook we are using here is purely about altering commit messages
12:47:46SynrGso? one project i am in has a 'make commit' for that purpose
12:48:21bluebrother^SynrG: still doesn't work. We have no single "entry-point" Makefile
12:48:23Tornethat's a truly awful idea
12:48:42Tornealso, people may not even have make, potentially :)
12:48:44bluebrother^for example, if you're working on Rockbox Utility (like me) you generate Makefiles using qmake
12:48:53Tornethere are things you can usefully change that do not need you to be able to build at all
12:49:07Torneabusing make like that is terrible, anyway
12:49:09bluebrother^or manual or voices
12:49:19*bluebrother^ agrees with Torne on that
12:49:26Tornean also doesn't handle other cases where commit messages are created
12:49:30Tornee.g. git cherry-pick, git merge
12:51:11Tornebluebrother^: feel free to check in a copy of the hook to the tree, and/or a script to install it
12:51:24Tornethe hook is not expected to change, so it doesn't matter if you fetch it from gerrit
12:51:38Tornebut i don't think that will improve compliance particularly :p
12:51:53Torneactually, one downside of requiring it would be that the hook *doesn't* fire on amend
12:52:21Torneso i fyou commit without it, try and push and get rejected, then install the hook, you need to actually undo your commit and commit again to get it applied
12:52:42Tornewhich involves, e.g. copypasting your commit message :/
12:52:53bluebrother^hmm, that's a bit ... not so nice
12:54:42Tornethere's not really an alterantive that i know of
12:55:05Tornethis all comes down to "git fairly deliberately fails to provide any way to actually track changes"
12:55:29Tornethe slightly hacky trick being used to add that feature is not entirely perfect :)
12:56:42Torne(linux didn't need this feature, because their "patch queues" are mailing lists and thus only humans need to be able to identify successive versions of a change, not machines
12:57:55Torneanyway, yeah. check in the hook if you want
12:58:32Tornebut i suspect that people will not think about this until they've experienced gerrit creating new change reviews when they didn't want it to a couple of times :p
12:58:43Tornethey'll either happen to hit that step in the instructions, or they won't
13:01:59bertrikI don't even know what gerrit change reviews are
13:02:26Tornethen play with the demo :)
13:02:27bertrikSo I'll likely either break things accidentally or get frustrated with it :)
13:03:40***Saving seen data "./dancer.seen"
13:08:27 Join perrikwp_ [0] (
13:11:00 Quit perrikwp (Ping timeout: 248 seconds)
13:13:24*amiconn wonders whether there's a howto somewhere that describes how to use git as similar as possible to svn
13:13:55amiconnNot necessarily using the same commands, but the same workflow, especially also the same number of commands
13:14:51amiconnE.g. how can I commit (push, whatever) from my working copy to the (local and) central repo with one command? Not the multi-stage stuff git uses as a default.
13:15:32JdGordondon't tinhk you can
13:15:40JdGordonbut one extra command isnt such a big deal, surely
13:15:53JdGordonand really, you should look into changing toa git work flow because it is far superior
13:16:18JdGordongit commit && git push == svn ci
13:16:50amiconnWell, that's your point of view. Mine is different. That doesn't belong here
13:17:30amiconnAnd iiuc git commit isn't even the only step locally, as there are two. Or am I misunderstanding things here?
13:17:51JdGordongit commit -a then
13:18:18Torneamiconn: there is no such thing
13:18:37Tornecommit && push doesn't work, because if the push fails due to nonfastforward the commit will still have happened locally
13:18:48Torneso the process to try again is not the same
13:18:58Tornelearn to use git; sorry
13:21:04Torneanyway, yes, you need to also add changes to the index before committing.
13:21:15Tornegit commit -a will automatically add all changes to existing files to the index for you, though
13:21:21Torneso, like, svn, you would only need to run git add for new files
13:21:42amiconnI don't want to commit all changes, just changes from source files I'm specifying on the command line
13:22:01*gevaerts demands to know why openvpn doesn't want to work with udp on his touchpad with CM
13:22:07gevaertsoops, sorry
13:22:20Torneamiconn: Then it already works
13:22:31Tornegit commit foo.c bar.c commits the changes to those two files, ignoring everything else
13:22:52TorneYou still need to do 'git add' for new files, though
13:23:12amiconnYeah, that's the same with svn
13:23:29TorneOK, so if you always commit that way you don't need to think about the index
13:23:42Tornethe only difference for that way of working is that instead of "svn commit" with no arguments, you need "git commit -a"
13:24:25*amiconn rarely used 'svn commit' with no further arguments
13:24:44Torneright; that's a different way of working
13:24:49Tornebut yes, that works fine in git
13:24:59amiconnIn fact, almost never. The only cases were when I checked out a release branch and ported a single fix to it
13:25:09bluebrother^amiconn: something like this?
13:25:23amiconnBut then I still need to push separately
13:26:01Torneyes. that's not avoidable.
13:26:06TorneThere is no atomic commit-to-remote in git
13:26:10TorneCannot be done
13:26:20Torneyou can write an alias that commits and pushes in one go
13:26:28Tornebut if someone else committed since you last updated that will fail halfway
13:26:34TorneWell, actually
13:26:43TorneYou could have it detect the push failure and undo the commit, actually
13:27:22bluebrother^having the possibility to commit without pushing is quite useful imo. Especially combined with amending and squashing
13:27:37Tornegit commit "$@" && git push origin HEAD:master || git reset HEAD^
13:27:37bluebrother^makes it much easier to save in intermediate state
13:27:48Tornethat will undo the commit if the push fails.
13:27:54amiconnbluebrother^: That really depends on the workflow. You assume that one always wants local version control
13:27:55Tornebut.. you probably don't want to do that, seriously.
13:28:09Tornebecause fetching new changes from the remote side is different
13:28:18amiconnI can live with the extra command, although it's sub optimal imo
13:28:30Torneamiconn: It's not just an extra command
13:28:44TorneThe way you need to deal with someone else having committed in the meantime is different
13:29:12Tornewhich is why tutorials that try to map svn commands to git commands are missing the point :/
13:29:40bluebrother^amiconn: well, yes. But once one gets used to it you *do* want local version control. At least that's my experience.
13:30:03bluebrother^Torne: we could run git cvsserver :P
13:30:18Torneamiconn: the difference that's important there is that if you do "svn update" with local changes, it will insert conflict markers/etc and merge the remote changes with what you've done
13:30:22Tornewhich you then have to fix yourself
13:30:37Torneif you "git pull" with local changes, it will abort if any file you have changed has been changed remotely
13:30:45Torneit will not merge them
13:31:00Torneit expects that you will commit your local changes first, and then either actually do a git merge, or do a git rebase.
13:31:15Torneso, you can't really just decide to treat those two steps as a single operation.
13:31:35 Join fyrestorm [0] (
13:32:31Torneyou don't have to use exciting new features of git (heh) but you do need to understand how it actually *works*.
13:33:18Torneincidentally this is why i traditionally used bzr, which has a perfectly functional atomic commit-and-push operation tha tworks exactly how you want
13:33:27Tornebut also supports all the distributed stuff otehr people like :)
13:34:39 Quit fyre^OS (Ping timeout: 248 seconds)
13:41:23 Quit chkktri_ (Ping timeout: 245 seconds)
13:43:38 Join mcuelenaere [0] (~mcuelenae@rockbox/developer/mcuelenaere)
13:43:51mcuelenaerehi, I get "[remote rejected] HEAD -> refs/for/master (prohibited by Gerrit)" when pushing to www.git; is this normal?
13:45:10Tornei haven't enabled code reviews
13:45:20Tornebecause i haven't written up a policy doc/instructions yet
13:45:25Tornegonna do it real soon, promise :)
13:46:08mcuelenaerehmm ok, will post it as a patch then
13:49:56 Quit mcuelenaere (Quit: Ik ga weg)
13:50:47 Join dfkt [0] (dfkt@unaffiliated/dfkt)
13:51:59 Join benedikt93 [0] (~benedikt9@unaffiliated/benedikt93)
14:11:44 Join domonoky [0] (~Domonoky@rockbox/developer/domonoky)
14:12:56soapso how do we quickly have users know what version they are running and how current it is?
14:13:41 Quit factor (Read error: Connection reset by peer)
14:13:53Tornewhat version they are running is hte short hash
14:13:54soapit was easy to tell how far off current and what the chronology of changes (and possible fixes) was with Rxxxxx numbers, but what "language" do we use to discuss such with question askers in the the forum now?
14:13:57Tornehow current it is.. er, the date
14:14:12Tornei'm poking at a script that inspects the branch structure to work out something more useful
14:14:26soapdate isn't always fine-grained enough to know if said user is running a bug-fixed version or not.
14:16:13Torneanyway, yeah. trying to do something better, but haven't had time to do all the entertaining logic yet
14:18:56gevaertssoap: for now those users are running their own builds anyway, but they can still provide the same data (abcdef12-2012-01-12). The only annoyance is that we (i.e. the support people) have to do a bit more lookup work
14:20:03 Join jlbiasini [0] (
14:20:10soapI (think I) totally grok the hash, just don't know why a sequencing number isn't assigned as well.
14:20:29soapI'm sure that's been talked about in the git world before. Google-ho.
14:21:11gevaertsWe probably can't ever assign a proper sequence number for modified builds, but for the builds we provide there are several possibilities. It's just not done yet
14:21:35jlbiasinipamaury: gevaerts once said you should commit this FS #12526
14:21:35fs-bluebot put Fuze+ in unstable (patches, unconfirmed)
14:21:48pamauryah yeah
14:22:33*gevaerts would phrase that slightly differently :)
14:22:34jlbiasinifor the too other bug it would be nice to have bluebrother (manual) and kugel (PLA) opinion first
14:22:47pamaurythat also requires changes the frontpage and perhaps the wiki. I'm busy right now but I'll do that later in the afternoon
14:22:56pamaurygevaerts: how would you phrase that ;)
14:23:14*jlbiasini my english is not evoluate since I live in Austria
14:23:27Tornesoap: it's been talked about, yes, but git people generally hate the idea
14:23:43Tornebecause you cannot do it in a reliable way without breaking git's "perfect" symmetry
14:23:52Torneand so most people give up in the face of extreme discouragement :)
14:24:19pamauryjlbiasini: we still don't have the bootloader file online right ?
14:24:30gevaertspamaury: I think only people closely involved with a port should commit unusable->unstable moves and similar things, so I won't commit that. I didn't mean you have to commit it because *I* think it's time :)
14:24:31jlbiasinipamaury: yes we do
14:25:18pamauryanyway yes I think we can move to unstable, the device is really usable.
14:25:43jlbiasinigevaerts: Ah yeah now I remember that it was what I was also wondering: in which sense you meant that...
14:25:52pamauryjlbiasini: did you try Rockbox Utility with it ? I should test it myself also
14:26:04jlbiasiniit works
14:26:11pamaurynice :)
14:26:28jlbiasiniI had zazog uploading your tagged file a few day ago
14:27:41jlbiasinithe only remaining is the server saying Rockbox Utility that it is in unstable so that the complete installation works also
14:28:47jlbiasiniBut Rockbox Utility has problem that bluebrrother want to solve before making any new official release. So he meant that there is no hurry on this for now
14:30:10jlbiasinifor the moment Rockbox Utility won't recognize the status of the fuze+, declare it unknowed so you have to go on the the second tab of the interface to select bootloader install
14:30:32 Join factor [0] (~factor@
14:31:44jlbiasinipamaury: anyway I have another question: is there anyway one could upgrade the firmware without using the internal memory? And is it possible to have a bootloader loading RB from sd card? I'm askiing because my internal memory just died...
14:32:47jlbiasiniso I wonder I just throw away my device/send it for testing/ or wait for a turn around
14:32:54pamauryyou mean upgrading the bootloader ? It is stored in internal memory but it's rarely written I admit. what happened to your internal memory ?
14:33:16pamauryAnyway yes, you can upgrade the bootloader just using the recovery mode
14:35:01jlbiasinimy internal memory is now read only
14:35:19jlbiasiniwith some cluster being corrupted
14:35:30jlbiasinino way to format it
14:36:08jlbiasiniI tried all the possible fsck.vfat/dosfsck option no result
14:36:31pamaurywhy can't you format it ?
14:36:35jlbiasiniI just can't get anything writed on that
14:37:42pamauryis the whole internal memory read-only ?
14:37:46jlbiasinion windows it say it didn't succeed (wether quick or long format mode) gparted format and then show me the very same partition on releod
14:39:10pamaurywhat is the output of dmesg ?
14:39:39jlbiasiniActually not read only, it never say so, it just copy file and show them correctly copied on it but after unplug replug it, the file are not there
14:39:47jlbiasiniwait I'll test that
14:40:27jlbiasinithe nice thing is that as long has it doen't have to write on the internal memory, RB is still working!!
14:42:57jlbiasinipamaury: on connection
14:47:31jlbiasinino dmesg while rm file of umount
14:51:42pamaurynothing strange.
14:53:13pamauryit's really unexpected, I would except the internal storage to replace bad blocks, it should take a fair amount of writes to end up in this situation
14:55:00 Quit stoffel (Ping timeout: 244 seconds)
14:56:31jlbiasiniyes I really didn't expect that but fsck do report bad cluster (a lot) so I think this is just dead? Or do you have another explaination?
14:56:45jlbiasinianyway now this should be ready: FS #12529
14:56:46fs-bluebot Lamp plugins PLA integration (patches, unconfirmed)
14:57:46jlbiasinikugel: thank's for your comment you were right I just corrected it now it compile
14:58:15jlbiasiniI didn't saw there was a scroll_repeat value
14:59:27jlbiasinibluebrother: sorry for asking again could we have your point about FS #12492
14:59:28fs-bluebot add fuze+ manual (patches, unconfirmed)
15:00:42jlbiasinibluebrother: did you already start something on the zip support for rockbox utility or can i start from the usual git working on that
15:01:30pamaurydid you try the OF ?
15:03:44***Saving seen data "./dancer.seen"
15:06:01 Quit [Saint] (Remote host closed the connection)
15:08:44jlbiasinipamaury: fsck result. Yes OF reboot after saying it needs more place on the device, rockbox go into panic if he try to write on the device
15:09:09jlbiasinifsck result give no change
15:10:28pamauryI don't see any mention of bad clusters, can't it just be a corrupted fat ?
15:10:36jlbiasiniIf I try tesk disk to erase partition table it says: Write error: Can't clear partition table.
15:10:43 Join Horscht [0] (
15:10:43 Quit Horscht (Changing host)
15:10:43 Join Horscht [0] (~Horscht@xbmc/user/horscht)
15:11:02jlbiasinibad cluster are reported with -t option
15:11:13jlbiasinijust a very long list of them
15:11:55pamauryhum, it would be interesting to see if one can get statistics by mmc about the bad blocks and so
15:12:28jlbiasiniwell If I can test anything just let me know :)
15:13:22jlbiasiniI think that remapping the whole device without know about simulatorui was my main mistake...
15:13:24pamaurylet me have a look at the mmc spec
15:13:49bluebrother^jlbiasini: that Quick Start vs. Installation tab thing is something I want to get rid of anyway. People don't seem to understand that installation is done using the Installation tab, not Quick Start (which in fact is only intended for a first and quick start)
15:14:29bluebrother^as for the zip support for reading OF files, I haven't done anything usable yet. It needs quite a bit of thinking since BootloaderInstallBase is pure virtual on pretty much everything except static functions
15:14:40 Join anewuser [0] (~anewuser@
15:14:40 Quit anewuser (Changing host)
15:14:40 Join anewuser [0] (~anewuser@unaffiliated/anewuser)
15:15:07bluebrother^and we also need some way indicate the progress to the user, as well as handling the extracted file in a usable way.
15:15:35bluebrother^feel free to try to figure something, but I really don't want to duplicate stuff for each bootloader class.
15:15:35jlbiasinibluebrother: so you recommand me to want on this before knowing how to do it the reight way?
15:15:58bluebrother^if you find a good way just let me know. Or provide a patch :)
15:16:35bluebrother^but since I'm in the process of cursing the multithread stuff for TTS that went in 1 1/2 years ago and is causing problems since I really don't want to have to rip up things again later and redo it :)
15:16:37jlbiasiniok I just wanted to know If you've already started something to be sure I wouldn't work for nothing
15:17:16bluebrother^why did you change Play/Pause to Play-Pause in the Fuze+ manual patch?
15:18:14jlbiasiniyes this is a big consistency issue: in manual there are some table like volume up/down
15:18:41jlbiasinithen It would lead to some entry like play/pause/cancel
15:18:58jlbiasiniso now it will be play-pause/cancel
15:19:40jlbiasinilike for other two words button (button bottom-left for example)
15:20:16jlbiasiniwhich also means that I have to update the image accordingly
15:20:17bluebrother^that's not a common way to write this though.
15:20:24bluebrother^and it looks awful IMO :)
15:21:00jlbiasiniyes but this is consistent and something like play/pause/cancel is also awful :)
15:21:04bluebrother^I would rather do it the way it's done for e.g. the Ipods: just name the button "Play", even if it has both a play and pause symbol on it
15:21:26bluebrother^Play-Pause/Cancel is pretty much unreadable
15:21:42bluebrother^does that mean Play and Pause/Cancel? Or Play/Pause and Cancel?
15:21:42jlbiasinithat's right too! :D
15:22:31jlbiasinibut then the same problem occurs with bottom-left/bottom-right!
15:22:43bluebrother^and why it is \ButtonPlayPause? It's \ButtonPlay for other targets, so if you make it \ButtonPlayPause for the fuze+ you need to adjust every use of \ButtonPlay in the manual. Which pretty much defeats the purpose of \ButtonPlay
15:23:32bluebrother^true, but just because of those bottom-something we shouldn't make it more complicated for play
15:23:33jlbiasiniYeah that is plainly right and I was very tempted to change that too
15:24:07bluebrother^I would *really* just make it ButtonPlay that simply displays Play. It's absolutely correct and way less changes.
15:24:53jlbiasinibut the button itself is play pause (no slash just both symbol)
15:25:34bluebrother^so what?
15:25:48bluebrother^it's the same for Ipods, and on Ipods we simply call that button "Play"
15:25:59bluebrother^after all we name every button in the player image.
15:26:01jlbiasinibut the consistent way is clearly the one you say... gee! seems I off for a new manual writing week!! :(
15:26:05bluebrother^so it's unambiguous
15:26:20n1sother sansas are that way too
15:27:01bluebrother^and for that bottom-something: we label the buttons. We could even call them "X" and "Y" and would be fine.
15:27:14bluebrother^though I guess there are better ways to name those :)
15:28:17bluebrother^you should be able to replace that ButtonPlayPause with ButtonPlay in the patch and then reapply it :)
15:28:30bluebrother^so at least that shouldn't be a week of work
15:28:45jlbiasiniI thing that bottom-something is good: there are no label on the player for those: they are just virtual on the touchpad like on player, so it avoid the user that read the manual to have all the time to go back to the player picture
15:29:36jlbiasiniyes but I also have to undo some separation where it was possible to add the fuze+ to some play button like other target
15:30:13bluebrother^not sure if I got that
15:30:27jlbiasinithose /opt(fuzeplus) have to go away in such case and be share with other player
15:31:32bluebrother^but if you do the renaming in the patch first it should be easier :)
15:31:39bluebrother^ymmv of course
15:32:49jlbiasinithat's right but whenever I edit patch I'm quite good at ending with "malformed line" stuff... I'll try anyway thanks for the comment
15:33:46jlbiasiniand I will have to rewrite lamp now that it is in PLA
15:34:00jlbiasinipacman PLA is coming soon
15:36:59jlbiasinipamaury: I have to go for a while, if you don't find me here for those internal memory test, you can pm me
15:47:59 Part jlbiasini
15:52:17 Quit perrikwp_ (Read error: Connection reset by peer)
15:53:29 Join perrikwp [0] (
16:02:44 Join stoffel [0] (
16:16:55CIA-81Commit by torne (1114a2b): Convert svn ignores into .gitignore.
16:22:30kugelpamaury: what is dcp?
16:25:20CIA-81Commit by torne (cb4e333): Convert svn ignores into .gitignore
16:25:21CIA-81Commit by torne (39a7a0f): Convert svn ignores into .gitignore
16:25:22DBUGEnqueued KICK CIA-81
16:25:22CIA-81Commit by torne (8e8c1dd): Convert svn ignores into .gitignore
16:27:55kugelthe .gitignores don't look useful for the other repos
16:29:36kugelso is it OK if I push the repo to github as a mirror (not automatically updated yet)?
16:29:51kugeljust so github people have something to clone from
16:30:02Tornekugel: they may not be; someone who cares can delete them
16:30:18Torneand i guess;
16:30:31Torneif we want to mirror it there then prod zagor about setting up gerrit to replicate
16:30:34Tornethe docs should explain it :)
16:30:56Tornebut if you do it manually for now that won't hurt
16:36:29Tornekugel: i just converted what was there so we don't lose it; if someone wants ot make an intelligent decision about which ones are useful/current/interesting they can just change them :)
16:40:23pamaurykugel: data co processor
16:40:35kugelwhat can it do?
16:41:28pamaurymemcpy, blit, encryption, hashing and (apparently undocumented) color space conversion+scaling
16:49:42 Quit SynrG (Read error: Connection reset by peer)
16:49:48 Quit perrikwp (Read error: Connection reset by peer)
16:51:02 Join perrikwp [0] (
17:03:46***Saving seen data "./dancer.seen"
17:03:49 Join Amqui [0] (
17:04:24 Quit Unhelpful (Quit: - Chat comfortably. Anywhere.)
17:04:47 Join Unhelpful [0] (~quassel@rockbox/developer/Unhelpful)
17:07:05 Quit Horscht (Quit: Verlassend)
17:08:54 Join SynrG [0] (
17:18:48 Quit perrikwp (Read error: Connection reset by peer)
17:20:02 Join perrikwp [0] (
17:25:54 Quit TheLemonMan (Quit: WeeChat 0.3.6)
17:42:54 Quit jordan` (Quit: Coyote finally caught me)
17:43:16 Join jordan` [0] (~gromit@2001:660:3302:2826:225:90ff:fe20:d9a8)
17:44:59 Quit jordan` (Client Quit)
17:45:20 Join jordan` [0] (~gromit@2001:660:3302:2826:225:90ff:fe20:d9a8)
17:53:47 Join perrikwp_ [0] (
17:56:20 Quit perrikwp (Ping timeout: 252 seconds)
18:05:54 Quit domonoky (Quit: Leaving.)
18:17:33 Join domonoky [0] (~Domonoky@rockbox/developer/domonoky)
18:19:02 Join Wardje [0] (~ward@unaffiliated/wardje)
18:22:56 Join perrikwp [0] (
18:25:15 Quit perrikwp_ (Ping timeout: 252 seconds)
18:30:46 Join TheLemonMan [0] (
18:38:48 Quit stoffel (Ping timeout: 255 seconds)
18:42:34 Join perrikwp_ [0] (
18:45:15pamauryshouldn't .gitignore be in .gitignore ? that way anyone can modify it for its need without always have it reported as changed
18:45:33 Quit perrikwp (Ping timeout: 255 seconds)
18:46:33Wardjepamaury: you don't want to have to manually copy/find a .gitignore after cloning a repo either though
18:47:20 Join lebellium [0] (
18:47:53pamauryhum, indeed, so there is no solution
18:48:11 Join jlbiasini [0] (
18:48:36 Join perrikwp [0] (
18:48:55pamauryjlbiasini: I'm looking at your problem, I will try to modify rockbox to panic on read/write error and dump the mmc error code, it will be much more precise
18:49:22 Quit perrikwp_ (Read error: Operation timed out)
18:58:05 Quit jlbiasini (Remote host closed the connection)
18:59:53 Join jlbiasini [0] (
19:02:34jlbiasinipamaury: thanks
19:03:48***Saving seen data "./dancer.seen"
19:06:48jlbiasinifor the moment if I try to write something on the internat it does give some info. Trying to create a bookmark give me: "Updating size on empty dir entry 87" same with "dir entry 4" while trying to save a text file with text_editor plugin
19:11:28 Quit jordan` (Quit: Coyote finally caught me)
19:11:42 Join jordan` [0] (~gromit@2001:660:3302:2826:225:90ff:fe20:d9a8)
19:23:52 Part jlbiasini
19:23:56 Join jlbiasini [0] (
19:24:38 Join perrikwp_ [0] (
19:26:09 Quit perrikwp (Read error: Operation timed out)
19:49:42kugelpamaury: personal ignores go into .git/info/excluse
19:50:13pamauryah ok, thanks
19:50:42kugelbut you can as well change .gitignore and commit the change
19:51:58 Quit mortalis (Quit: KVIrc 4.1.1 Equilibrium
19:59:49 Join perrikwp [0] (
20:00:29pamauryjlbiasini: I have a patch that you can try
20:01:19pamaurydo a bootloader build, use mkimxboot with the extra option -t recovery and send it using sbloader
20:02:02 Quit lebellium (Ping timeout: 240 seconds)
20:02:12 Quit perrikwp_ (Ping timeout: 245 seconds)
20:02:44 Join lebellium [0] (
20:08:33jlbiasinipamaury: thx, how do I send it with sbloader? where do I find sbloader?
20:09:42pamauryin utils/imxtools
20:10:27 Quit Zarggg (Quit: Rebooting client...)
20:10:42 Join Zarggg [0] (
20:10:54 Join TheLemonM [0] (
20:11:48 Part lovasoa
20:12:44jlbiasinipamaury: make bootloader end with: make: *** Pas de règle pour fabriquer la cible « /home/jean-louis/Bureau/rockbox-devtree/sbtest/rockbox/buildboot/kernel-imx233.h », nécessaire pour « /home/jean-louis/Bureau/rockbox-devtree/sbtest/rockbox/buildboot/firmware/target/arm/imx233/kernel-imx233.o ». Arrêt.
20:12:52jlbiasiniis it normal?
20:13:00pamauryah yeah, I forgot to commit, wait a sec
20:13:12 Quit TheLemonMan (Ping timeout: 260 seconds)
20:13:30CIA-81Commit by amaury.pouly (66c3086): imx233: oops, forgot file
20:15:47funmani'd prefer to see real names instead of a component of email address, what do you think?
20:17:21pamauryyes I agree
20:17:29n1syes, me too
20:23:26jlbiasinipamaury: what should I give for <xfer size> argument
20:25:15jlbiasinilibusb:error [op_open] libusb requires write access to USB device nodes. :( this sometime happened to me I will reboot and see if I can get working
20:25:19 Part jlbiasini
20:25:29funmani think the cia hook is private (i did not see it in svn)
20:25:32pamaurydo it as root
20:25:37funmanbtw no perl hacker could look at the buildsystem yet?
20:26:37 Join jlbiasini [0] (
20:27:28pamauryrewrite it in something else than perl ;)
20:30:42 Quit jlbiasini (Ping timeout: 240 seconds)
20:31:35funmanpamaury: it's in www repo
20:36:12funmanand i don't feel like rewriting everything from scratch
20:36:28funmani thought build scripts would be fixed before we enable git in fact
20:36:45pamauryme too
20:37:42funmanin Torne's mail it was explicit that it'd be done after though
20:38:12Tornezagor and folks suggested it would be easier to just switch and sort stuff out afterward
20:38:41Tornerather than set up more mirroring and try and run a second copy in parallel to test it, or something
20:40:07funmanthe problem seems to be noone knows perl enough
20:40:49Tornewell, i haven't looked :)
20:40:56Tornepeople seemed confident it would be easy to change it
20:40:59 Quit benedikt93 (Quit: Bye ;))
20:41:49 Join perrikwp_ [0] (
20:43:19 Join jlbiasini [0] (
20:44:00pamauryjlbiasini: I just think about it, build it with -t singleboot instead of recovery, otherwise it will no mount over usb
20:44:47 Quit perrikwp (Ping timeout: 244 seconds)
20:44:56jlbiasinipamaury so just singleboot or singleboot and recovery?
20:45:08pamaurysingleboot only
20:50:08jlbiasinipamaury: same but without "Disable MMC windows. No partition found. " what to do next just unplug?
20:50:28pamauryhum, with singleboot it disable MMC windows ?
20:50:44jlbiasinino more mention of mmc
20:51:01jlbiasinijust like booting in usb mode
20:51:12pamauryok, so now try to write something over usb
20:51:34 Join perrikwp [0] (
20:51:43pamaury(note that the recovery mode is only permanent, if you reboot the device, you will need to send it using sbloader again)
20:54:37 Quit perrikwp_ (Ping timeout: 260 seconds)
20:56:31pamauryjlbiasini: any consequence ? it should panic on any read/write failure
20:57:13jlbiasinino change
20:58:14pamauryit never panic ?
20:58:22jlbiasiniit do not say anything like its working ether copy from nautilus or terminal
20:58:25jlbiasinino panic
20:58:46pamauryhmm, are you sure you are running the sb file ?
20:58:54jlbiasinibut after umount remount nothing changed on the partition
20:59:28jlbiasinifrom the bootloader version showed yes
21:00:22pamauryok, can you do another modification ?
21:00:33jlbiasinitell me
21:00:49pamauryin target/arm/imx233/mmc-imx233.c, in mmc_read_sectors
21:01:23pamauryreplace "return bla;" by 'int ret = bla; if(ret != 0)panicf("die: %d", ret); return ret;'
21:01:33pamauryand same thing in mmc_write_sectords
21:02:14PoodlemastahHas there been any discussion regarding ordering database artist view ignoring a starting "The"?
21:02:38 Join captainkwel [0] (
21:03:49***Saving seen data "./dancer.seen"
21:08:14jlbiasinipamaury: same
21:08:44jlbiasiniPoodlemastah: have a look in flyspray there are to discussion regarding database about that
21:08:58pamaurythen it would mean that all read/write are successful
21:09:24 Quit perrikwp (Read error: Connection reset by peer)
21:09:34jlbiasiniis there a way we could force the syncronisation in terminal mode?
21:09:43pamaurywhat do you mean ?
21:10:11jlbiasiniI suspect everything being copied to cache by the os
21:10:41jlbiasinibut not synchronise or perhaps something hang just before syncronisation
21:11:07 Join perrikwp [0] (
21:11:18jlbiasiniI think there a an option for mount doing that
21:11:37pamaurydid you try with a windows host ?
21:11:45pamauryor another machine
21:14:33jlbiasiniwell the problem does not occur on my usb key
21:14:50jlbiasiniand is the very same both on linux and windows
21:15:21jlbiasiniand even with noatime
21:15:26jlbiasinino luck
21:18:02TheLemonMerr, sync would be the param
21:18:18TheLemonMand the command sync flushes the caches to filesystem
21:18:30TheLemonM(and is implicitly called by umount)
21:19:07jlbiasiniwell the only think that give me a write err is testdisk while asking to wipe the partitions table
21:19:35jlbiasinibut no reaction of the firmware, still showing bl info
21:21:11 Join perrikwp_ [0] (
21:23:27 Quit perrikwp (Ping timeout: 245 seconds)
21:25:00jlbiasiniisn't there a situation where the mmc would reply always true to watever is asked?
21:32:12jlbiasinipamaury: there is only one explication then: hardware is dead!
21:33:30 Quit TheLemonM (Quit: WeeChat 0.3.6)
21:41:26pamauryjlbiasini: perhaps...nevertheless, that's strange
21:42:27jlbiasinisdcard still works with your patch
21:42:54jlbiasiniso that is really specific to mmc
21:43:37jlbiasinisome hadware faillure report writing maybe
21:45:02 Quit y4n (Quit: PANTS OFF!)
21:48:27 Part jlbiasini
21:51:50pamauryjlbiasini (for the logs): then I guess you can disable mmc completely by changing CONFIG_STORAGE (remove mmc). That will default to sd card for the firmware then. Updating the bootloader will be a little more tricky though
21:57:48bluebrother^ acting up?
21:57:58saratogajust came in here to ask that, its down for me
21:58:10bluebrother^ok, so it's not just my network connection
21:59:42 Join jlbiasini [0] ( is ok though
22:02:25funmanit's back now
22:03:04 Nick adnap_ is now known as adnap (
22:04:29pamauryjlbiasini: did you see my comment ?
22:05:42jlbiasiniyes I'm on it
22:05:52jlbiasinibut this will work only one time
22:06:15jlbiasinito have it permanently I have to upgrade the firmware right?
22:08:50pamauryyes, it's possible using recovery mode only
22:09:29jlbiasinii'll try!
22:09:35 Quit curtism (Quit: Live Long and Prosper)
22:10:07 Join LambdaCalculus37 [0] (
22:10:07 Quit LambdaCalculus37 (Changing host)
22:10:07 Join LambdaCalculus37 [0] (~rmenes@rockbox/staff/LambdaCalculus37)
22:11:24pamauryjlbiasini: before, I have another idea
22:11:33jlbiasinitell me
22:11:36LambdaCalculus37Can a forum mod bring the banhammer down on user Abramqsh on the forums? It's a persistent spammer and I'm still throwing stuff out from him/her/it.
22:14:12saratogaapparently this is partially based on rockbox?
22:14:30saratoganot sure if its running as an app on a linux system or just using the codecs though
22:15:53pamauryjlbiasini: in the first patch, in the transfer_sectors function, replace if(ret != 0) by if(ret == 0 && resp & ~0x1e00)
22:16:55jlbiasinipamaury can I let the previous modification or should I download the patch again?
22:17:11pamauryyou can let it
22:17:16pamaurywait, the mask is not good
22:17:37pamaurysorry, not 0x1e00 but 0x1f00
22:18:36jlbiasiniok I better change the code directly then
22:19:27 Quit beslayed (Remote host closed the connection)
22:21:33jlbiasiniwait there a problem! I think the patch did not apply at all!!!
22:21:46jlbiasiniit would explain a lot!!
22:21:50jlbiasinisorry about that
22:22:13jlbiasiniis there a different way to apply patch in git?
22:23:12pamauryno, just use patch
22:23:20pamaurypastebin the diff if you are unsure
22:25:34jlbiasinishould I try the first one or directly the modification you told me?
22:26:34 Join beslayed [0] (
22:26:50jlbiasinipamaury: there is the first patch, then first modif, then the last one
22:28:17pamauryok, can you pastebin the diff anyway, it would be a shame that we do not have the same code in mind :) Then try with the same procedure as before
22:28:35jlbiasini/home/jean-louis/Bureau/rockbox-devtree/sbtest/rockbox/firmware/target/arm/imx233/mmc-imx233.c:234: warning: format ‘%x’ expects type ‘unsigned int’, but argument 2 has type ‘long unsigned int
22:28:42jlbiasinicompilation warning
22:29:02jlbiasiniat least it prove that I did patch correctly this time! :)
22:29:33jlbiasiniwell just warning should I ignore rhis?
22:31:50 Join liar [0] (
22:32:06pamauryyes ignore it
22:32:14pamauryyou should have had this warning before :)
22:32:32jlbiasiniyeah that waht i said ;)
22:34:32 Quit n1s (Quit: Ex-Chat)
22:38:54 Quit evilnick (Ping timeout: 240 seconds)
22:42:23jlbiasinipamaury with patch original no change
22:42:35jlbiasinilet's do first modification
22:51:02 Quit jlbiasini (Remote host closed the connection)
22:53:20 Join jlbiasini [0] (
22:54:57bluebrother^jlbiasini: I was wrong before −− looked into the zip thing a bit and it turned out to be in fact rather easy :)
22:55:00 Join zchs [0] (
22:56:24jlbiasinipamaury: second change no luck
22:56:36pamaurythe last change is the interesting one
22:57:04jlbiasiniyeah this is the one I just tested
22:58:02jlbiasini if(ret == 0 && resp & ~0x1f00)
22:58:31pamauryreally, you just tested this ?
23:02:01funmanLambdaCalculus37: you still have your nano2g ?
23:02:30bluebrother^hmm. On e200 we're using e200pa.bin, but that's only the filename for the american firmware version. The others use e200pe.bin and e200pf.bin. Does the device flash it nevertheless?
23:02:32pamauryso you have tested the ret != 0 check, the ret ==0 & ... :-/
23:02:42LambdaCalculus37funman: Yeah, but haven't used it in ages.
23:03:09funmanbluebrother^: yes the last letter indicates some features to enable or not, the files are identical
23:03:11jlbiasiniI'm about to test if(1) just in case :/
23:03:31funmanLambdaCalculus37: could you try the patch i sent on the ML recently? I can make a build for you if you want
23:03:32bluebrother^hmm, I guess I need to make Rockbox Utility check all those filenames then.
23:03:32LambdaCalculus37funman: I no longer have my iPod Classic though; I gave it to a friend.
23:03:45funmanbluebrother^: just rename it to e200pa.bin ?
23:03:48LambdaCalculus37funman: Sure. Can you build for me?
23:03:50***Saving seen data "./dancer.seen"
23:04:13pamauryjlbiasini: you tried to read and write with each variant ?
23:04:44jlbiasinireading is ok
23:05:02jlbiasiniwriting seems ok but never actually happens
23:05:15pamauryis the battery sufficiently charged ?
23:05:40jlbiasiniI guess so how much should it be?
23:06:07pamauryI don't know but you were close to empty battery, the voltage might be affected, just a random idea
23:06:07bluebrother^funman: I'm trying to read the file directly from the zip you download from sandisk. I can't simply rename that file since I first need to find the correct one, and when I found it I can also simply use it :)
23:06:30bluebrother^but fortunately handling of multiple firmware filenames is already implemented
23:06:54jlbiasinipamaury: on unplug with the last patch I get panic resp: 80900
23:07:57jlbiasiniI still have a third og battery
23:08:18pamaurythat means unspecified error :-/
23:08:40jlbiasininice! :D
23:08:47pamauryso the device indeed reports something but we don't know what
23:09:14pamauryand it might on unplug because of delayed write I guess
23:10:14funmanLambdaCalculus37: funman/"> : can you just tell me if USB still works ?
23:11:20 Quit jxb091000 (Quit: Leaving)
23:11:29LambdaCalculus37funman: Sure, let me get the nano and a cable,
23:14:10jlbiasinipamaury: let me check I did right: I made the modification to the code. run configure in an empty dir selct fuze+ ask for a Bootloader build. run mkimxboot -i OF -b rbbl -i -t singleboot, sbloader 1024, right?
23:18:11jlbiasinipamaury after I rm all file and dir on the volume it told me it is 18,9 Mo left
23:19:24pamaurywhich means ?.
23:19:36jlbiasinibut still no error whil umount
23:19:57jlbiasinithat something is wrong but I think we already know that
23:20:46scorcheforums going down for a quick bit
23:22:02 Join stooo [0] (
23:22:11jlbiasinipamaury: well nothing new any other idea?
23:22:29pamauryno, if the mmc reports error, we can't investigate much further
23:22:33 Part stooo
23:23:34jlbiasiniwell I will turn mmc off then If I can use rockbox on sd card that already a big improvement for me thanks very much for your time
23:25:02 Nick Jack87 is now known as Jack87|Away (Jack87@nasadmin/admin/jack87)
23:25:03pamauryyou still need to install a new bootloader, but maybe we can see this tomorrow
23:25:05jlbiasiniat least now I know how to used sbloader and will be able to implement recovery mode for Rockbox utility!
23:25:24 Quit Staphylo (Ping timeout: 248 seconds)
23:25:33jlbiasinipamaury: oh? I though this was was -t recovery?
23:26:23pamaury-t recovery only build a file that gives you access to the firmware partition
23:27:29jlbiasiniI suposed this is not as easy as copying sb file to it then is it?
23:27:30pamaurythen you need to copy the real sb file (built with singleboot or dualboot) to the firmware partion, with a little subtlety (the first 4 sectors are skipped)
23:28:29jlbiasiniok some dd stuff?
23:28:31pamauryso basically: 1) build with -t recovery 2) sbload to device 3) build with -t dualboot 4) dd bs=512 skip=4 if= of= 5) pray ;)
23:29:09jlbiasiniI could try now
23:29:35jlbiasiniif=bl compiled ?* of=/dev/sdXX
23:29:49pamauryI have never tested building a bootloader with sd support only though so I can only guess that it will work
23:30:08pamauryyou directly send the sb file built with -t dualboot or -t singleboot
23:30:08LambdaCalculus37funman: I think it started to work, but the battery ran out. Had to put the nano on the charger for a while.
23:30:29jlbiasiniI'll try then
23:31:01jlbiasinianyway else I will buy a new one so I guess It worth a try and good to know anyway
23:31:06 Join Scromple [0] (~Simon@
23:31:29funmanLambdaCalculus37: oh ok, ping me when it works
23:31:36pamauryI hope you will be only one to suffer from this problem
23:31:46jlbiasiniwe could even have a special procedure on rbutil later for those who burned the internal
23:31:46pamauryit would be problematic if rockbox destroyed mmc :-s
23:32:28bluebrother^special procedure in Rockbox Utility?
23:32:36jlbiasiniDidn't you saw that someone reported a problem with internal partition on forum?
23:32:47CIA-81Commit by Dominik.Riebeling (92fa7a8): Add alternate firmware filenames for e200v2.
23:32:49CIA-81Commit by Dominik.Riebeling (b45cc0a): Support reading OF files from zip.
23:33:04pamaurywhen ?
23:33:07bluebrother^I'd really like to avoid having *more* special cases in Rockbox Utility. We already have too much
23:33:41jlbiasinibluebrother^: in bootloaderinstallimx actually
23:33:46bluebrother^funny, the CIA notifications show up in the opposite order the commits were made
23:33:47pamauryyes, and doing so is tricky because you need raw disk access and the procedure is not trivial
23:33:53jlbiasinia week ago or something like that
23:34:00pamauryin the Fuze+ thread ?
23:34:18pamauryunfortunately the forum is in maintenance
23:34:21jlbiasiniunless some admin did some cleaning
23:35:08bluebrother^jlbiasini: if that means we need to pass additional information from outside to it (i.e. some information the user or configuration provides) that's a bad thing
23:35:19bluebrother^since that would mean breaking the interface.
23:35:39jlbiasinibut I'm having my fuze+ since over one year and while working on keymaps I was like installing new build of rb 20 time a days... so I guess this could explain stuff...
23:36:45pamaurywell, I have written the rockbox.sansa file an impressive amount of time also
23:36:57 Quit Scromple (Quit: Leaving)
23:37:35bluebrother^wow, it's over 3 years now the bootloader abstraction proved itself to be sufficient :)
23:38:02jlbiasiniAnd I did some big copy on the device to like 50 gb in a few week
23:38:04pamaurybluebrother^: what is the most complicated bootloader installation rbutil has ?
23:38:29pamauryjlbiasini: do you follow the sandisk forum by chance ?
23:39:42jlbiasiniyes why?
23:39:45pamauryerr sansa
23:40:24jlbiasinisince a few month less, I don't know why :D
23:40:33jlbiasinibut I used to
23:41:23pamaurydid you see any report of an internal memory failing ?
23:41:32jlbiasinia few yes
23:42:13jlbiasinibut that right we should care about this problem I hope to be an exception
23:42:44jlbiasini2 or 3 maybe But I was not going there all the time...
23:43:42 Join t0rc [0] (~t0rc@unaffiliated/t0rc/x-5233201)
23:46:23pamaurydid you succeed in installing a new bootloader ?
23:46:33jlbiasinii'm on it
23:47:06 Quit saratoga_ (Quit: Page closed)
23:47:53pamauryok, tell me when you're done, so I can go to sleep :)
23:50:20 Join Thra11 [0] (
23:53:28jlbiasinipamaury: does the recovery bootloader I send throught sbloader has to have mmc enable?

Previous day | Next day