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).

Notice: Only Gecko based browsers prior to FF4 support the multipart/mixed "server push" method used by this log reader to auto-update. Since you do not appear to use such a browser, this page will simply show the current log, and not automatically update.

#rockbox log for 2008-10-02

00:03:11funmanlinuxstb: did you investigate the decompressing of SansaV2 OF on an emulator ?
00:03:36linuxstbfunman: No, but I think I found the problem...
00:04:20linuxstbIn "uclpack", the so-called "in-place" decompression function requires an "overhead" of (block_size / 8) + 256 - so about 16KB for our image...
00:04:54linuxstbSo I think we need to "memcpy" it somewhere (e.g. the end of DRAM) before uncompressing.
00:05:22funmanwhat's the D in DRAM ?
00:05:22 Quit toffe82 ("ChatZilla 0.9.83 [Firefox 3.0.3/2008092417]")
00:05:41linuxstbErr.... Not sure ;)
00:07:00funmanlet me try this trick
00:07:29Bagder... it's also typically SDRAM these days
00:07:43funmanlike in Static AND Dynamic ?
00:07:54Bagderno, synchronous
00:10:01pixelmawindows Ipod - plays wma too ;)
00:11:32Bagderoh bounce! ;-)
00:11:50 Join toffe82 [0] (
00:12:41 Quit lacrstech (SendQ exceeded)
00:13:23 Join midgey_ [0] (n=tjross@
00:17:15 Quit bertrik ("Leaving")
00:17:46 Quit ender` (" If at first you don't succeed, skydiving is not for you.")
00:19:29 Join MarcGuay_ [0] (
00:20:43funmanlinuxstb: I think you are right ..
00:21:52linuxstbfunman: It's working now?
00:22:20soapanyone else get the following flyspray error when attempting to log in?
00:22:20soapFatal error: Allowed memory size of 8388608 bytes exhausted (tried to allocate 35 bytes) in /usr/share/flyspray/htdocs/adodb/ on line 2859
00:22:23funmanif I use an infinite loop instead of branching to #0, it 'works'
00:22:30funmanlet me dump the memory to check
00:23:44 Join freqmod_qu [0] (
00:24:09 Quit perrikwp (" ajax IRC Client")
00:24:11 Join Zarggg [0] (
00:24:18funmanyes it works
00:24:29*funman looks frag<tab> damn !
00:25:12 Join perrikwp [0] (i=d1a8d351@gateway/web/ajax/
00:25:38linuxstbfunman: OK, now flash it to your Clip ;)
00:25:56*linuxstb looks around for fragilematter
00:25:56funmanhm I'm still a bit reluctant
00:26:24linuxstbSo what happens when you remove the infinite loop?
00:27:01funmanSKYEYE:Error in mem_read_word, no bank found, NumInstrs 12930564, mem_read_word addr = c80f0014 no bank
00:27:26funmanpc was 0x100, maybe we can check what this is in the firmware
00:27:54linuxstbWhich OF version are you using?
00:28:21funmanif it's the same than v29 : vector software interrupt ?
00:29:21funman(an infinite loop)
00:30:10funmanisn't there a way to address a register so it is incremented by the offset we give after the instruction ?
00:30:27soapOk, little bug: killed by me in two different ways. Take your pick! (Requesting commit)
00:30:41 Quit midgey (Read error: 110 (Connection timed out))
00:31:11 Join midgey [0] (n=tjross@
00:31:20 Quit domonoky (Read error: 104 (Connection reset by peer))
00:32:09 Quit mcuelenaere ()
00:32:26 Quit MarcGuay (Read error: 110 (Connection timed out))
00:32:27 Join webguest04 [0] (n=4cf40cba@gateway/web/cgi-irc/
00:32:47 Nick HBK- is now known as HBK (
00:33:51linuxstbfunman: Hmm, yes. So I would be quite confident - it looks like the OF is running fine, and then hitting an exception when an invalid memory access is attempted
00:34:27funmanIt is running fine - in a simulator
00:34:34 Quit webguest04 (Client Quit)
00:34:43funmanchoose your pill
00:35:17kugelWe could just test it on another target
00:36:12funmanhm now it says something else: pc 0x1390 (but I have modified my memcpy code)
00:38:21funmanwell the decompressed firmware is OK
00:39:40MarcGuay_Weird, one of the c100s I got is showing up as a TCC7801...
00:39:53MarcGuay_Err, revert that.
00:40:03funmanwell it loops waiting for some hardware register to change
00:41:07funmanI'm not sure of the ram size ..
00:41:10 Quit FRiZzO (Read error: 110 (Connection timed out))
00:42:25MarcGuay_Does sendfirm just drop whatever you give it in the drive root?
00:42:37RandalSchwartzthis floss weekly should post on friday, watch for details
00:42:41 Quit tvelocity (Read error: 104 (Connection reset by peer))
00:42:52BagderRandalSchwartz: good show!
00:42:53 Quit midgey ()
00:42:54kugelfunman: you copy the whole firmware to the end of the ram? linuxstb only talked about an overhead of 16K
00:43:01RandalSchwartzthat was fun
00:43:02RandalSchwartzI like demo
00:43:05BagderRandalSchwartz: did you spot the conversion link I posted?
00:43:11linuxstbkugel: Read my later comments
00:43:13RandalSchwartzyes thanks
00:43:17funmankugel: why not using the maximum ? :)
00:43:54LloreanRandalSchwartz: Thanks again for having me.
00:44:15 Quit midgey_ (No route to host)
00:44:17RandalSchwartzAhh. thanks for offering yourself!
00:44:24kugel"So I think we need to "memcpy" it somewhere (e.g. the end of DRAM) before uncompressing." did you refer to memcpy to firmware or the overhead?
00:44:30*Bagder offered Llorean, a very handy approach ;-)
00:44:40linuxstbkugel: funman did what I suggested...
00:44:45Bagderanyone else wants him? B)
00:44:46LloreanYes, the world works better when you can volunteer someone else for the job. :)
00:45:01scorcheBagder: im sure some forum users might want him for a bit....
00:45:07funmanyes 0x50000 seems to be 2.5 Mbit . :)
00:45:40linuxstbMaybe 0x40000 would be safer - I seem to recall the OF using memory after that
00:46:01gevaertsscorche: really? Now why would you ever think that? ;)
00:46:48funman[Reg], #1 => will Reg be incremented after being addressed that way ?
00:46:53 Quit bluebrother ("bye")
00:47:23BagderI'm glad we did 3.0 before this interview anyway ;-)
00:47:42linuxstbfunman: Yes, I think so
00:47:55 Quit faemir (Remote closed the connection)
00:47:55LloreanBagder: Yeah, having 3.0 out in time for it is kinda nice.
00:48:05LloreanNext time, though, I'll skip the hurricane and just ask if we can do it a little later.
00:48:30RandalSchwartzyou arranged the hurricane?
00:48:39gevaertsOf course
00:49:03gevaertsYou just need to catch the right butterfly
00:49:26funmanlinuxstb: why do you think 0x40000 would be safer ?
00:51:02 Quit herrwaldo ("Konversation terminated!")
00:51:18amiconnfunman: Why not copy it just far enough that it decompresses without overlap?
00:51:37 Quit DerDome (Read error: 104 (Connection reset by peer))
00:51:50funmanwhy bother ?
00:51:53 Join DerDome [0] (
00:51:57amiconnIiuc the goal is to keep changes in memory to a minimum, because you don't know yet what is where?
00:52:17linuxstbfunman: Those was my thoughts...
00:52:28 Join midgey [0] (n=tjross@
00:52:43funmanwell I assume the AMS bootloader doesn't initialise memory
00:53:03funmanso when the OF runs it is presented with memory in an 'unknown' state, and initialises it when needed
00:53:28linuxstbfunman: I did think that maybe there was more than just the main firmware block loaded into RAM, but I think that's unlikely now.
00:54:13funmanthat may be possible
00:54:46*amiconn wouldn't assume such things about bootloaders
00:55:16amiconnThe ipod OF relies on several things initialised properly by its bootloader (e.g. the LCD)
00:55:37funmanthe clip firmware clears ~ 0x7000 bytes after the end of the firmware
00:55:47 Join tvelocity [0] (
00:56:00amiconnbss, probably
00:56:06funmanamiconn: but those are hardware settings, not memory ?
00:56:19 Join einhirn [0] (
00:56:21linuxstbamiconn: Although the AMS SoC seems to have a more generic bootloader in the internal ROM, rather than a device-specific bootloader like our other targets.
00:56:29 Quit matsl (Remote closed the connection)
00:56:52linuxstbSo I think it will just do the minimal hardware initialisation, and load the firmware from the NAND flash.
00:57:03funmanlinuxstb: do you understand how the compressed and decompressed datas can 'overlap' ?
00:57:12 Quit perrikwp (" ajax IRC Client")
00:57:28linuxstbYes. We basically need to memmove the OF image 16KB forwards in RAM.
00:57:44amiconnJust 16KB?
00:57:47funman0x7000 > 2*16KB
00:57:58Bagderit's funny, but we're still on twice the pre-release daily visitor amount on the main web site
00:58:07funmanmaybe a 4th word in the test.S which says how much we have to shift ?
00:58:09linuxstbYes, if we compress with a block size of 12KB
00:58:13linuxstbI mean 128KB
00:58:13amiconnI thought you need to move the compressed data upwards so that it end lines up with where the decompressed data will end
00:58:22funmanoh, it's fixed
00:58:22amiconn*its end
00:58:28funmanexactly 16kB
00:58:58 Join perrikwp [0] (i=982137d4@gateway/web/ajax/
00:58:58linuxstbfunman: It's technically (block_size / 8) + 256 - so a little more than 16KB if we compress with a 128KB block size
00:59:01funmanamiconn: this appears to not be enough
00:59:33linuxstbamiconn: I was wrong - I missed the "overhead" added in the official uclpack when performing so-called "in-place" decompression
00:59:53linuxstbSee the "do_decompress()" function in uclpack.c
01:00:05LloreanBagder: Well, I mean I would expect it to settle at a point at least somewhat higher than the original value.
01:00:07amiconnIs this an additional overhead?
01:00:35linuxstbYes - the buffer to decompress a block needs to be "block_size + overhead", with the compressed data located at the end of that buffer
01:00:36BagderLlorean: I can't say I expected it, but I'm enjoying it! ;-)
01:01:06 Join JamesR87 [0] (n=5205fcf3@gateway/web/cgi-irc/
01:01:07LloreanBagder: I had the numbers for hits and unique visitors sitting in front of me in case the topic of expanded notice since 3.0 came up
01:01:08amiconnWhy not just compress into a single block, like we do for archos, btw?
01:01:17linuxstbYes, we do.
01:01:27funmanlinuxstb: is 'MAINT' a define of rockbox, or something from the original UCL package ?
01:01:41amiconnlinuxstb: So the OF is <128KB?
01:01:42JamesR87Quick question: Is my 80Gb iPod classic still not supported by Rockbox?
01:02:08linuxstbfunman: Everything in that .S file is from the original UCL package - I committed the unmodified version to SVN first, so you can see what I changed.
01:02:12linuxstbamiconn: Yes, about 110KB
01:02:19 Join n17ikh|Lappy [0] (
01:02:23funmanlinuxstb: no I meant in uclpack.c (I was checking the block size used)
01:02:33linuxstbamiconn: Or rather, there's a main firmware block which is that size, plus lots of library blocks, taking the entire image to about 5MB
01:02:51LloreanJamesR87: The supported players are listed on the front page of the site.
01:02:56LloreanWe'll update that list if things change.
01:03:05linuxstbWe're just manipulating the main firmware block of the OF image.
01:03:54JamesR87Le sigh, just seeing if there was any updates :> Cheers anyway! *Goes to find a different alternate firmware*
01:04:12 Quit bmbl ("Woah!")
01:04:26***Saving seen data "./dancer.seen"
01:04:37Nevtusi don't think there is any alternative firmware for classics yet
01:04:52LloreanJamesR87: If you do find one, let us know. We haven't gotten to the point where we can even load our own code, so any information they have would be useful to us.
01:07:18linuxstbfunman: It works ;)
01:08:17linuxstbfunman: I changed your 0x50000 to 0x40000, flashed it to my Clip, and rebooted fine...
01:08:44 Quit MethoS- (Remote closed the connection)
01:08:55 Quit JamesR87 ("CGI:IRC (Ping timeout)")
01:09:33 Quit n1s ()
01:11:42funmanplease commit and retest, so I'm sure I didn't make a typo since
01:12:24*linuxstb pours a large whisky
01:12:39*funman 'll go for rhum
01:14:18 Quit Thundercloud (Remote closed the connection)
01:14:26funmancheers linuxstb !
01:15:24linuxstbCheers... Second test worked as well, so I'll commit.
01:18:18 Join kugel_ [0] (
01:18:33 Quit kugel (Nick collision from services.)
01:18:46 Nick kugel_ is now known as kugel (
01:18:55gevaertsExactly how unacceptable is increasing RAM usage on PP and iMX31 by about 500 bytes to allow for easier merging of the TCC USB stack? I don't like it at all, but the alternative seems to be lots of static configuration which I'm afraid will bite us later on
01:19:20 Quit einhirn (Read error: 104 (Connection reset by peer))
01:19:54linuxstbgevaerts: If it makes the code simpler and easy to maintain, then I think it's worthwhile.
01:20:43linuxstbfunman: Committed.
01:21:21kugelfunman, linuxstb: great work
01:22:05kugeldoes this work on any firmware and for every v2?
01:23:11funmanlinuxstb: patched
01:23:37funmankugel: no
01:23:50gevaertslinuxstb: it's not exactly simpler than it is now (although it's not really more complicated either), but the current code has some assumptions about how a usb controller works that aren't true for tcc
01:24:10funmancurrent code assumes that the decompressor function will fit in the last block of your firmware
01:24:11linuxstbkugel: It should be easy to adapt to any firmware though. However, currently it requires about 170 bytes in the padding.
01:24:38linuxstbBut we can just memcpy that function as well, so store it in the main firmware block area
01:25:35funmanor just calculate compressed OF offset so the function just fits
01:25:46kugelIIRC there's only 88bytes in the fuze firmware
01:25:59kugelin the firmware block
01:26:09linuxstbfunman: No, the function will then get overwritten when it's busy decompressing the OF
01:26:14 Part toffe82
01:26:21funmanoops right
01:26:50linuxstbBut we could just put it before the compressed OF image, and memcpy them both together
01:26:58funmanbtw can you close FS #9396 ? it's useless now
01:28:17linuxstbfunman: Done.
01:28:17 Quit ZincAlloy ("CGI:IRC (EOF)")
01:28:18 Quit DerDome ("Leaving.")
01:28:48 Join krazykit [0] (
01:28:50 Join Bawitdaba [0] (
01:29:57linuxstbfunman: I think it's now time to create a Clip target in the build system and work on the drivers... Is atomicpunk also hacking the Clip, or another target?
01:30:43funmanlinuxstb: yes it's a good idea. atomicpunk is only hacking the Clip (his m200 is dead)
01:31:09linuxstbgevaerts: I think you're worrying too much - unless others disagree... If you're making the code more general (with good reason), then I don't see any reason to object.
01:31:42*scorche wonders if it is about time to take his clip out of his sock drawer
01:31:44gevaertslinuxstb: ok. I'll make it work tomorrow (it doesn't currently), and then commit.
01:32:22linuxstbscorche: I had forgotten how tiny it is...
01:32:36scorchelinuxstb: indeed...nice screen though
01:32:48funmanlinuxstb: I can start doing that tomorrow
01:33:37 Join saratoga [0] (n=9803c264@gateway/web/cgi-irc/
01:33:38kugelfunman: So any chance for this to work on the fuze firmware?
01:33:59funmankugel: yes, it will require just a bit more hacking
01:34:17saratogacongrats guys
01:35:17funmanthanks ;)
01:36:07kugelsaratoga: seems like we have to get it to work on our fuzes first
01:36:28 Quit moos ("Rockbox rules the DAP world")
01:37:58linuxstbkugel: Did the old mkamsboot work on the fuze?
01:38:30 Quit Schmogel ("Miranda IM! Smaller, Faster, Easier.")
01:38:30funmanonly the c200 was untested I think
01:41:26kugellinuxstb: yes, but the the code inserted wasn't 170KB
01:41:58saratogaso the fuze firmware might not have enough space available?
01:42:16linuxstbsaratoga: With SVN mkamsboot, yes. But we can fix it now.
01:42:24kugelit turned out the padding in the firmware block can give 88bytes
01:42:39linuxstbi.e. we won't use the padding at all.
01:43:02funmanhow about waiting to use proper crt0 and the linker ?
01:43:08kugelwhere to store the decompressor else?
01:43:30funmananywhere in our 40kB available
01:43:47funmanand copy it in RAM, after the OF size, before executing it
01:47:30kugelforgot that the compression gives us space for the decompression :S
01:48:25 Join setkeh [0] (
01:48:42funmanlinuxstb: I think as a first step we can use the git repository for Clip code
01:49:28funmanand when we make significant progress you can check-in into the rockbox SVN repository
01:49:38funmanfdinel: hola (:
01:50:22linuxstbfunman: I would prefer to see things committed to Rockbox SVN - that way, lots of eyes will see the code as it develops.
01:50:26kugelfunman: so why exactly does this not work on the fuze out-of-the-box again?
01:50:29setkehhow can i get .wps files to work ???
01:50:50linuxstbkugel: Because the ucl decompression function is stored in the padding.
01:51:22funmanlinuxstb: ok ..
01:51:28 Quit PaulJam (".")
01:51:29kugelnow I'm confused
01:53:05linuxstbkugel: About what?
01:53:20pixelmasetkeh: what's happening, what did you try, what player do you have, how did you install them etc... please give more information
01:54:16kugelok: is this right: as of the decompression code is in the padding, but in the future, we're going to put it right before the compressed OF where then ~40KB is available
01:54:40setkehpixelma, 5g ipod video in copyed the files for the theme into there respective folders the theme file seems to be working fine but the .wps looks like its not working
01:55:08 Quit krazykit ("Connection reset by beer")
01:55:13funmankugel: yes
01:55:23kugelok cool
01:55:30fdinellinuxstb: yeah I'm also hacking on the clip ;)
01:55:33funmanand we are gonna copy this code somewhere else in case we want to decompress the OF over it
01:55:38*fdinel is atomik punk
01:56:01fdinelis it big?
01:56:04kugelhave you noticed the great success?
01:56:17saratogathe clip port will be very interesting
01:56:28linuxstbfdinel: The uncompress code is about 166 bytes
01:57:08 Join chibiluke [0] (n=516338ab@gateway/web/cgi-irc/
01:57:08linuxstbfunman: I'm working on that now, so I'll be asking you to test on your emulator setup shortly...
01:57:10pixelmasetkeh: I guess you got a broken theme/wps then. There was a change in wps syntax a short while ago that broke a lot of them. You could either try other ones or try to fix them yourself. Where did you get the themes from by the way - Rockbox wiki or something else?
01:57:31fdinellinuxstb: ok nice, then hmm why not put it (in RAM) past the OF "section"?
01:57:54funmansaratoga: hopefully linuxstb and fdinel won't do all the work and leave a bit to the others !
01:57:55 Quit chibiluke (Client Quit)
01:58:13setkehpixelma, thats the place i got them from i think ill just go to the wiki to get them :P
01:58:20saratogafunman: they'll need Nico_P's buffering from flash patch at least
01:58:24funmanlinuxstb: ok - I'll get to bed shortly
01:58:36fdinelfunman: no worry, you're coding too fast for me :P
01:58:43kugelWell, better someone does more than others instead of many people doing nearly nothing
01:58:45saratogafunman: how difficult is it to setup the skyeye anyway?
01:58:58setkehpixelma, does the wiki have an updated tutorial like this for makeing .wps files ??
01:59:01funmansaratoga: what's nico_p's flash patch ?
01:59:26saratogafunman: a different version of the buffering code for ultra-low mem targets
01:59:26funmansaratoga: the most difficult part is finding the rare, badly translated from chinese documentation
01:59:43pixelmasetkeh: yes. the CustomWPS page is up to date and there is also a thread in the forums that explains the changes
01:59:49saratogadoes it run an OS inside the emulator, or do you just point it at an elf binary?
02:00:09setkehpixelma, thank you heaps mate :D
02:00:22funmanit uses an elf binary, but I load the firmware at offset 0 and my elf binary is mov r0, #0; bx r0
02:01:24saratogais it expandable if we wanted to try faking hardware and running binaries?
02:01:47funmanProbably, but it may involve hacking skyeye
02:02:24funmanThere is a number of boards supported, an lcd screen (which I didn't use)
02:03:03saratogawe have documentation for most of the V2 hardware right?
02:03:10saratogamight be fun to try emulating a sansa
02:03:21funmanuart, touchscreen, network, sound ..
02:03:26amiconnDo we have an arm v5(te) target? D2, maybe?
02:03:27funmanflash ..
02:03:35funmanlooks quite complete
02:03:44saratogabah a WMA bug report
02:03:48kugelsaratoga: do we have infos about the different LCDs?
02:03:50saratogai need to get caught up on those
02:04:07saratogakugel: i assume thats handled by the AMS controller?
02:04:24funmanuart, touchscreen, network, sound, flash <= that's what skyeye can emulate, not which docs we have
02:04:33kugelI haven't read the docs (I didn't ask for them yet), so I don't really know
02:05:12saratogatheres no other chips on some of the V2 parts (clip) so I think its all AMS stuff
02:05:22linuxstbfunman: I've done it very quickly, so it's likely to be wrong, but here's a first attempt -
02:05:31funmanSanDisk added their own chips
02:05:57fdinelseems like the Soc got "upgraded"
02:06:14 Join BHSPitMonkey [0] (n=stephen@unaffiliated/bhspitmonkey)
02:06:24fdinelwe suspect they inserted an SD-to-NAND interface somewhere
02:06:27linuxstbamiconn: All the telechips devices are armv5, so yes, the D2, plus the m200, DAX etc
02:06:47amiconnPlain v5 or v5te?
02:07:51funmanlinuxstb: oops
02:07:59funmanIn at91rm92_io_read_word, io error, addr=0xfffffff0
02:08:18funmanskyeye exits so I don't have any control on the debugger
02:08:31saratogaamiconn: I believe its the 946 which is v5te
02:08:31 Join mmadia [0] (
02:08:34linuxstbamiconn: The D2 has an arm946e-s (v5te) and ARM926EJ-S (v5tej)
02:09:19saratogayes should give the Gigabeat F a run for it money, once for each core
02:09:19amiconnOf those, only the D2 already plays music
02:09:45funmannow is time to sleep, and tomorrow time to code :=)
02:09:46*amiconn was asking because of the SMLA<x><y> instruction
02:09:59funmansee you guys
02:10:02funmankeep up !
02:10:12 Quit funman ("leaving")
02:10:20amiconnsaratoga: Yeah, the F and X are arm9, but only v4t
02:10:21linuxstbAnd the tcc77x devices (DAX etc) are arm946es as well
02:10:36 Join Strife89 [0] (n=michael@
02:10:37amiconnIsn't the iaudio7 using a similar cpu?
02:10:44linuxstbYes, tcc77x
02:10:56amiconnAnd it already plays music...
02:11:06linuxstbThe "x" just defines what peripherals there are - codecs, NOR flash, built-in RAM
02:11:14Strife89This is odd. I tried to apply the "Pong Upgrades" patch and this happens.
02:11:49Strife89Removing the patch (with −−remove) doesn't fix it, so...... I'm stuck.
02:12:21linuxstbamiconn: I'm not sure if it plays music - maybe just test sounds... The flash driver isn't working very well.
02:12:54saratogaseems like USB is well along though
02:13:05saratogacan't be long before vitja turns his attention to flash
02:13:43linuxstbEveryone seems to be avoiding the challenge...
02:14:07Strife89Pong upgrades patch is here if someone wants to look:
02:15:05 Quit tvelocity (Remote closed the connection)
02:15:07linuxstbamiconn: Presumably the beast has those instructions as well?
02:15:32amiconnlinuxstb: v6 has even better instructions for speeding up the demac filters
02:16:10linuxstbBut I mean if you wanted to implement it, you could test on the beast...
02:16:18amiconnBut on v5te the mentioned instruction would at least allow a bit of speedup versus C code or the armv4 code
02:17:17linuxstbHave you done any APE benchmarks yet on the S?
02:17:24amiconnBut comparing the speeds on a v6 wouldn't give a realistic comparison of how it would behave on a v5te
02:17:33 Quit maddler ("connection reset by beer!")
02:17:38linuxstbYes, I'm aware of that, but at least you could get it working.
02:17:56amiconnNo, I didn't do speed measurements yet.
02:18:00Strife89For those who didn't catch the pastebin, I applied a patch and, when I make, make skips almost everything. This wouldn't be a problem if −−reverse worked, but.... it didn't. Now what?
02:18:08*amiconn needs to do quite a bit of other stuff before
02:18:13saratogado a make clean
02:18:54kugelStrife89: it only recompiles files which have changed since the previous compilation
02:18:55Strife89Thanks, saratoga. Didn't think about that.....
02:19:06kugelso, everything was fine
02:19:13*amiconn wonders why linuxstb didn't use post-increment in his el cheapo memcpy...
02:19:32linuxstbamiconn: Because it was funman's code, and I just committed it because it worked...
02:19:39Strife89I'm still very new to patching and making (obviously).
02:19:41amiconn(for r0 at least)
02:19:46linuxstb(and my ARM-fu is rusty...)
02:20:14amiconnchange ldrb r3, [r0] to ldrb r3, [r0], #1 and remove adds r0, r0, #1
02:20:36 Join jfb9301 [0] (
02:20:43amiconnEh, you can do that for r2 too, as there's a 'cmp' anyway
02:21:03amiconnHmm, or is that thumb code?
02:21:09linuxstbNo, that ARM code
02:21:37Strife89How did it go for Llorean today?
02:21:50amiconnlinuxstb: Okay, so reclaim those 8 bytes...
02:22:32amiconnHeh, who was squeezing the decompressor?
02:23:23 Quit saratoga ("CGI:IRC (EOF)")
02:23:42kugelStrife89: he performed exceptionally well
02:23:46*amiconn wonders why funman even used 'adds' instead of 'add'
02:23:54Strife89kugel: Good to hear. :)
02:24:56linuxstbamiconn: Yes, I noticed that, but forgot to remove it. But I expect that code will get significantly changed in the near future anyway... Once it gets merged with Rockbox, we can just call memcpy
02:26:08 Quit jfb9301 ()
02:26:39 Join jfb9301 [0] (
02:26:47 Quit culture (Read error: 110 (Connection timed out))
02:26:56*linuxstb sleeps
02:27:45 Quit webguest54 ("CGI:IRC 0.5.9 (2006/06/06)")
02:27:57 Join mc2739 [0] (
02:29:50 Quit Strife89 ("Leaving")
02:31:12 Nick mc2739 is now known as mc2739_ (
02:31:34 Nick mc2739_ is now known as mc2739 (
02:33:39 Quit jfb9301 ()
02:35:08 Join goffa_ [0] (n=goffa@
02:35:50 Part mc2739
02:38:26 Join mc2739 [0] (
02:41:19 Join toffe82 [0] (n=chatzill@
02:44:19 Quit n17ikh|Lappy ()
02:45:30 Quit Wictor (Read error: 104 (Connection reset by peer))
02:45:40 Join Wictor [0] (
02:46:23 Quit goffa (Read error: 110 (Connection timed out))
02:46:31 Join kronflux [0] (
02:49:27 Part pixelma
02:49:59 Join pixelma2 [0] (n=marianne@rockbox/staff/pixelma)
02:50:38 Quit midgey ()
02:57:32 Join Waldo000000 [0] (
03:02:19 Quit kugel ("ChatZilla 0.9.83 [Firefox 3.0.3/2008092417]")
03:04:28***Saving seen data "./dancer.seen"
03:04:31 Join massiveH [0] (
03:06:17 Quit m0f0x (Read error: 60 (Operation timed out))
03:10:06 Join webguest94 [0] (n=4aaa3bc7@gateway/web/cgi-irc/
03:16:26webguest94Hi all, for some reason the word acksaw: was already in the field for the web client. hmm. but that's not why I joined. I jsut updated Rockbox to the latest current build as of 15 minutes ago and now can't get it to say anything. I've tried turning the battery switch off and back on (holding down the power button for over 10 seconds didn't do anything) and it still isn't working. I can browse my player throug
03:17:48 Nick fxb is now known as fxb__ (
03:18:13MarcGuay_webguest94: Did you install the voice file as well?
03:19:18webguest94Also the build I updated from is only maybe 3 weeks old.
03:19:37MarcGuay_And it spoke fine before?
03:20:21MarcGuay_How did you re-install?
03:20:48webguest94I even tried selecting several times to get to something playable and holding down the volume up button and didn't hear anything
03:21:15webguest94Using the latest utility, installed a current build
03:21:23MarcGuay_So there's no sound at all, even music?
03:22:02webguest94Nope. Should I try putting the old .rockbox back? (I backed it up before updating.)
03:22:13MarcGuay_It might help to track down the problem.
03:23:55webguest94If it will recognize the player that is... grrr.
03:24:29MarcGuay_What player do you have?
03:24:50 Quit JdGordon (Read error: 104 (Connection reset by peer))
03:25:11webguest94Ah, there we go, ahd to turn teh battery switch off and back on. A gigabeat F
03:25:16Waldo000000hi, stupid question: from the file browser, i want to play all files in the selected directory - how do i do that?
03:25:26Waldo000000(couldn't see an obvious answer in the manual)
03:26:22webguest94Press and hold select, then select playlsit, insert.
03:26:38MarcGuay_Waldo000000: In the context menu you can choose Playlist->Insert. Too slow.
03:27:08Waldo000000thanks. kinda non-intuitive. but thanks :) just installed rockbox for the first time :D
03:27:24MarcGuay_Waldo000000: You'll get used to it.
03:28:47webguest94and now of course windows Vista won't let me rename the old backup, saying I must type a filename.
03:29:12Waldo000000MarcGuay_: yeah, i'm thinking i definitely will, very impressed so far. especially when i'm using an m:robe which was discontinued years ago by olympus, now i have something that is still being updated! love it
03:29:55mc2739webguest94: you can rename it from a command prompt
03:30:17Waldo000000is anyone here using an olympus m:robe, and if so do you mainly use the file browser or database? is it good to avoid using the database if you don't need to, to save battery?
03:31:09webguest94mc2739: ah. What command would I use? I sometimes mess around with the prompt, but haven't renamed files/folders before
03:31:20 Join MrHacks [0] (
03:31:45MrHacksHey guys. I have a status update on my H120
03:32:24MrHacksMy time is limited. The laptop I am temporarly using is bad. If I were on my main computer it would be different.
03:32:26webguest94mc2729: ah never mind, got it
03:32:29mc2739webguest94: move .rockbox rockbox-bu
03:32:56MrHacksI decided to do the last thing you should ever do with the device: Disassembly
03:33:06 Join montag_ [0] (n=18bd042c@gateway/web/cgi-irc/
03:33:46MrHacksAfter removing the harddrive, I found out that the problems with the firmware
03:34:03MrHacksBattery is OK. Harddrive is OK.
03:34:10webguest94hmm. the old build works perfectly.
03:34:12 Join JdGordon [0] (n=jonno@rockbox/developer/JdGordon)
03:35:17MarcGuay_webguest94: That's odd. I don't have an F but I'll give it a try and see if I have the same problem.
03:35:20 Join kugel [0] (n=chatzill@unaffiliated/kugel)
03:35:27webguest94should I try reinstalling the new build?
03:37:25kugelJdGordon: ping
03:37:38 Join Hillshum [0] (
03:38:09montag_I have been trying to add album art to my music files but they do not show up in my ipod video with Rockbox
03:38:21montag_how do i make it work
03:38:28MarcGuay_AlbumArt wiki page.
03:39:06montag_can't find that page, where is it?
03:40:11kugelJdGordon: (for the logs) It seems you've been quite interested in FS #6800 a while ago. I'm wondering if you still are. I've done some work on it, including enabling it for c200 and gigabeat S. You might want to take a look at it.
03:40:21 Quit kugel ("ChatZilla 0.9.83 [Firefox 3.0.3/2008092510]")
03:41:11MarcGuay_Search the Docs Index.
03:41:36 Join m0f0x [0] (
03:41:56MarcGuay_webguest94: Okay I updated to the most recent build, what am I looking for? I presume you had you config.cfg file set to voice?
03:42:09montag_thank you soooooo much marcguay
03:43:05MarcGuay_webguest94: I just turned voices on and they seem fine. Sansa e200 + r18679
03:43:46MarcGuay_Hmm, just remembered it's not a voice problem, you're getting no sound at all.
03:45:34webguest94Sorry, was in a noterh window. It shouldn't be a problem, even if RB was completely removed and reinstalled, voice would be enabled by default
03:45:45 Quit montag_ ("CGI:IRC (EOF)")
03:46:16*ameyer is not amused by the new sansa usb trick
03:46:36*ameyer rebuilds to see if he just had a bad build
03:46:53MrHacksquick! The commands for mounting a usb device!
03:47:38MarcGuay_webguest94: If you simply install over your old one, the config.cfg which contains your settings should remain. For instance, I have voicing off, and when I update, it remains off.
03:48:26MarcGuay_webguest94: Oh wait, I just got your point. Ignore me.
03:52:29RandalSchwartzok - I think I have my 80G ipod FAT-32 and put the ipod software back on
03:52:42RandalSchwartznext step - rock install!
03:52:45 Join Darksair [0] (n=user@
03:53:42RandalSchwartzoh, I have to restore my music lib as well
03:54:43RandalSchwartzshould I be worried that disk util is reporting it as mac os extended again?
03:54:46RandalSchwartzthat doesn't sound right
03:55:08RandalSchwartzdid itunes reformat it back as mac?
03:55:14RandalSchwartz(on the restore)
03:56:45MarcGuay_RandalSchwartz: I presume you've been following the wiki page...?
03:57:07RandalSchwartzeven "mount" shows it sas hfs too
03:57:10RandalSchwartzthis can't be good
03:57:53MarcGuay_Doesn't sound like it took.
03:57:53RandalSchwartzafter newfs_msdos -F 32 -S 2048 -v iPod /dev/rdisk2s2
03:58:03RandalSchwartzitunes offered to restore it
03:58:13RandalSchwartzso I said yes, beacuse I want the itunes software on there
03:58:23RandalSchwartzapparently itunes forced it back to hfs. :(
03:58:23soapRandalSchwartz, please do not use the "Enter" key in place of a comma. There is no need for you to post 8 lines in a row. Please consolidate your thoughts.
03:58:59LloreanRandalSchwartz: iTunes restore (on mac) reformats to HFS+, yes.
03:59:13RandalSchwartzso I can't use itunes with rockbox?
03:59:19LloreanRandalSchwartz: If you follow all the directions on the wiki page, you *should* end up with iTunes software on there.
03:59:20RandalSchwartzor is there another step?
04:00:04LloreanBut I think I've heard reports similar to yours, where it boots the Apple firmware fine, but iTunes on the Mac just doesn't want to work with it anyway.
04:01:21 Quit Wictor (Read error: 104 (Connection reset by peer))
04:01:32 Join Wictor [0] (
04:03:04RandalSchwartzwell - I'll give it another try tomorrow after I read the rest of everything better
04:03:11MarcGuay_RandalSchwartz: Maybe try the "older, alternative" method...?
04:03:26RandalSchwartzit definitely ended up as HFS though. I don't need that. I need FAT. :)
04:05:56webguest94so should I just use the older build for now?
04:10:10Lloreanwebguest94: Are there any messages on the screen when you try to play music and it fails?
04:10:21webguest94Not sure, can't see it
04:10:37setkehhow do i veiw photos from the ipods photo database
04:10:41Lloreanwebguest94: It would help if you could get someone sighted to check, we need more information before we can give you any better advice.
04:10:54Lloreansetkeh: Rockbox just views jpeg files, accessed through the file browser.
04:11:31setkehLlorean, ok thanks mate ill have to remove the photo database then lol cuz i have like 8gb of pics :P
04:11:32 Quit XavierGr ()
04:13:02 Join miepchen^schlaf_ [0] (
04:14:52 Quit kronflux ("Leaving")
04:20:09 Quit miepchen^schlaf (Connection timed out)
04:20:23 Join webguest64 [0] (n=48afa203@gateway/web/cgi-irc/
04:20:47 Quit webguest64 (Client Quit)
04:21:51setkehdo i need the original ipod firmware to make rockbox work ??? what i mean is if i formatt the ipod using the windows partitioner and install rock box will it still work ??
04:22:47LloreanThe Rockbox install tools expect a "working" iPod
04:23:22LloreanWhile it's possible to install Rockbox on a "clean" iPod, it's considerably more work since you still need to get the partition layout set up in a way that the Apple in-ROM code likes it.
04:24:07 Quit mmadia (Read error: 104 (Connection reset by peer))
04:25:33webguest94k, aparently after booting the new fbuild, it said rockbox and then went completely blank, nothing I pressed or touched did anything
04:25:54webguest94the HD is still spinning though, but doesn't sound like it's trying to access anything
04:27:15Lloreanwebguest94: What do you have the backlight time out set to?
04:28:17webguest94not sure, from looking at it (I have light perception) maybe 5 seconds or so.
04:28:42setkehLlorean, do you know of any partition editors that will format the ipod so the in-ROM code will work ??
04:30:37Lloreansetkeh: After that you need to put the Rockbox bootloader image in the right place in the hidden partition. It's not a simple process, and it doesn't make much sense to try to do it.
04:30:50Lloreanwebguest94: If you press buttons, does the backlight come back on?
04:31:15webguest94hmm. the hd is still spinning. It's been like 5 minutes since I touched the player, nd I have it set to spen down after 3 seconds.
04:31:39setkehLlorean, ty
04:32:53 Quit mc2739 ()
04:33:44webguest94llorean: not that I can tell. It's completely froze with the hd still spinning
04:34:38MarcGuay_webguest94: Do you know the revision numbers of the current build and the previous build?
04:37:54Waldo000000general question: i think i've found a bug with the m:robe - where's my first point of call/where should i first report the bug? specifically, the button backlighting is quite erratic
04:37:59webguest94just a sec, have to turn off and back on the battery and restore the previous build. My pc wwon't even recognize the player so have to "hard boot" the player.
04:38:43MarcGuay_Waldo000000: Check to see if it's a known problem first by looking at the mrobe port page. Then read the FlySprayHowto wiki page.
04:39:15Waldo000000MarcGuay_: will do, thanks
04:39:33 Quit perrikwp (" ajax IRC Client")
04:43:32webguest94k, the old, working build is r18546
04:44:19Waldo000000is r18679 the official 3.0 release? i used the rockbox utility to install, but apparently you can't use that to install the official release, only the latest build (although i have no idea whether the latest build was the official release when i installed)
04:44:33webguest94where can I get the number for the current build?
04:45:04MarcGuay_webguest94: It's 18679.
04:45:15webguest94Waldo000000: the rb Utility has been updated.
04:45:25LloreanWaldo000000: You can use the utility to install the official release, just go to the "Install" tab, choose to install rockbox, and pick "Stable"
04:46:48 Quit Hillshum ("ChatZilla 0.9.83 [Firefox 3.0.3/2008092417]")
04:46:58MarcGuay_webguest94: I suggest you try to update again tomorrow and if the problem persists, file a bug report, or continue searching for someone who can reproduce it.
04:47:15 Join blkhawk- [0] (
04:47:31 Join perrikwp [0] (i=d1a8d351@gateway/web/ajax/
04:47:54 Join potato [0] (
04:48:31webguest94MarcGuay: k, will use the old build for now. Thanks.
04:49:17 Quit webguest94 ("CGI:IRC (EOF)")
04:49:24Waldo000000Llorean: do you mean i have to go "Installation" then "Install Rockbox", rather than "QuickStart"->"Compete Installation" to install a stable release instead of a current build??
04:49:27 Part potato
04:51:47ameyerWaldo000000: if you're running rbutil 1.0.6 or before, you need to upgrade to 1.0.7 to be able to install 3.0
04:52:34ameyerfor what it's worth, I suggest that you run the current CVS if you're on a PP-based target.
04:53:02Lloreanameyer: We use SVN.
04:53:34Waldo000000ameyer: i downloaded and ran 1.0.7 on ubuntu this morning (first time user of rockbox), and the "Complete Installation" prompted me that it would install the "current build", not necessarily a supported release
04:54:42ameyeryou may need to do "Installation"->"Install Rockbox"
04:54:52ameyerLlorean: yeah, sorry
04:55:09*ameyer isn't sure where that CVS came from
04:55:55ameyerstill, there's a legitimate reason to use the svn build if you're on a dual-core target and you run mp3
04:56:01ameyererm, use mp3
04:56:17LloreanWaldo000000: As i said, you need to use it from the Installation tab.
04:56:21Waldo000000ameyer: so you'd recommend i do that before filing a bug? i just ran system->version and i've got r18679-081002 - is that the supported release, or should i do "installation"->"install rockbox"?
04:56:41Waldo000000Llorean: ok, i'll do that now. wierd. but thanks :)
04:56:45ameyerr18679 == current svn
04:57:20Waldo000000ameyer: as opposed to 3.0?
04:57:25*ameyer wonders what string 3.0 gives
04:57:38 Join midgey [0] (
04:57:40ameyerWaldo000000: 3.0 isn't current svn
04:58:19Lloreanameyer: It gives revision-3.0-date, I believe.
04:58:21*Llorean isn't certain.
04:58:56Waldo000000ameyer: k cheers, i'll try it out. hopefully that will fix this problem - basically the button lights on m:robe are completely whack, generally not coming on at all
05:00:53Waldo000000um... should my computer still detect my m:robe? i just booted into rockbox, then plugged in the usb - the m:robe showed a cute little picture of a USB cable, then that disappeared and went to the old "battery charge" screen - no USB device detected by ubuntu - help?
05:01:31Waldo000000ah, n/m, key is to plug in before turning on
05:01:49ameyerit *should* boot into the OF when you plug in the USB
05:02:01ameyershould != does in all cases
05:02:11 Nick jfc^3 is now known as jfc (
05:02:58 Quit blkhawk (Read error: 110 (Connection timed out))
05:03:13 Nick blkhawk- is now known as blkhawk (
05:03:30 Quit ajonat (Read error: 104 (Connection reset by peer))
05:03:37ameyerfor what it's worth, 3.0 seems to be r18607-3.0-080923
05:04:02 Join ajonat [0] (n=ajonat@
05:04:15Waldo000000ameyer: k :). OF = original firmware?
05:04:31***Saving seen data "./dancer.seen"
05:05:51Lloreanameyer: Only about half of our MP3 players boot into the OF when USB inserted. M:Robe-100 happens to be one of them, but many don't need to.
05:06:36ameyerI think the "boot to OF" is a PP thing, and I'm pretty sure the M:Robe is PP
05:06:42Waldo000000i guess i've already made it obvious, but if any of you are involved in the documentation, it might be a good idea to have the rockbox utility "Complete Installation" install the supported release by default - because 99.9% of newbies will be doing this, and then later whinging about bugs
05:06:57 Join Skail [0] (
05:07:06Waldo000000PP = ?
05:07:15Waldo000000ah, k
05:07:46 Join sarixe [0] (
05:10:32Lloreanameyer: It is a PortalPlayer thing, but I just wanted to make sure you knew it wasn't an all-players thing
05:11:39ameyerright, it's not. you could probably even argue it's not all PortalPlayer targets, depending on how you classify the iPod's disk mode.
05:12:17Skailhey guys, I'm trying to write a plugin, and it says "Incompatible version" when I run it. Do I need to install my compiled .mi4 just to run my compiled .rock ?
05:12:25LloreaniPod's disk mode still qualifies as the OF.
05:12:51LloreanHopefully soon, a significantly smaller portion of PP devices will need rebooting.
05:12:51soapOF part deux
05:13:14Waldo000000well, the button light is still completely whack. guess i'll file a bug, unless you guys have any debugging advice?...
05:13:17LloreanIf anything the disk mode is more "firm" than the normal firmware, which is really just software
05:13:27LloreanSkail: Often, yes.
05:13:35*ameyer wonders out loud whether ipodlinux needs the of's disk mode
05:13:41Lloreanameyer: Yes.
05:14:08LloreanWaldo000000: "whack" isn't a particularly useful phrase. I'd recommend being more descriptive in the bug report.
05:14:43*ameyer vaguely remembers seeing a bug about the mrobe's lights being broken
05:15:19 Join potato [0] (
05:17:00ameyer...or not.
05:17:05potatowhy has the (s)nes flyspray "2008-02-04: A task closure has been requested." ? i thouhgt it was a good project =[
05:17:29Waldo000000Llorean: basically, they turn on momentarily on bootup, for a little while they turn on, turn off as expected depending on input, but then at some point they flash off/on quickly, then don't turn back on again, ever, except for some strange little flashes here and there. sometimes they momentarily come on when you flick the hold button on or off
05:18:00Waldo000000in other words "whack"
05:21:05LloreanWaldo000000: No, "whack" really doesn't meant that, and we ask you try to use real English in here.
05:21:09 Join Hillshum [0] (n=4ba5ee28@gateway/web/cgi-irc/
05:21:47Lloreanpotato: That just means some person _requested_ closure. We can't really read everyone's minds. If a developer wanted it closed, there'd be no "requested" about it, it'd just be closed.
05:21:48 Quit Hillshum (Client Quit)
05:22:00 Join Hillshum [0] (
05:22:45potatooh, ic.
05:22:47SkailLlorean: ty. (yay, I've now surmounted "Hello, World")
05:22:47 Quit Hillshum (Client Quit)
05:23:20LloreanRemember guys, Real English. That includes "ic" and "ty"
05:23:38Waldo000000Llorean: lol that was a joke dude, i mean, that was a joke, fellow :P so was that. i'm just filing a bug - category for button lights would be.... LCD?
05:23:59LloreanWaldo000000: The LCD category is for the LCD screen.
05:24:25LloreanProbably drivers.
05:26:50Waldo000000Llorean: Drivers, roger that
05:32:50Waldo000000just posted a bug report at Thanks for your help guys, let me know if there's anything i can do or further information i can provide to help get this resolved
05:34:03 Join Glert [0] (
05:35:43 Quit MarcGuay_ ("ChatZilla 0.9.83 [Firefox 3.0.3/2008092417]")
05:36:00 Quit massiveH ("Leaving")
05:37:13potatook, i have posted antibomb.c and fingersofrock.c to flyspray as patches. I hope i have made the patches crorectly...
05:40:41 Part potato
05:44:28 Join Transience39 [0] (
05:45:38Transience39i've got an iPod video with rockbox which is getting stuck during boot
05:45:51Transience39my xp machine won't recognize it either
05:45:57Transience39can anybody help me?
05:46:20LloreanTransience39: What happens exactly?
05:46:47 Part Waldo000000
05:48:05Transience39i get a black screen with some text in the top corner
05:48:16Transience39one sec, i'll get what it says
05:49:45Transience39ok, so the bootloader starts, it reads off the hardware, and then says "error! can't load rockbox iPod, file not found."
05:49:51Transience39and then tells me how to put it into disc mode
05:49:52LloreanHow did you install?
05:50:03Transience39i used the automated installer
05:50:19LloreanWhich install choice?
05:50:48 Quit Mooloo (Read error: 110 (Connection timed out))
05:51:25Transience39iPod video 30gb
05:52:08 Quit MrHacks ("Leaving")
05:52:48LloreanTransience39: No, I mean, "Quick Install" or did you go to the Install tab and choose one of the other options?
05:54:02Transience39quick install
05:54:51LloreanWhere is the file "rockbox.ipod" located on your device?'s not coming up when i search
05:55:55LloreanHow are you searching?
05:55:59Transience39should i just re-copy the .rockbox directory?
05:56:05Transience39windows explorer
05:56:12Transience39i got it to connect finally
05:56:16 Quit Zarggg ()
05:56:27LloreanYou can just re-run the utility, go to the "Install" tab, and just choose "Install Rockbox" since the bootloader is fine.
05:57:10 Part Transience39
05:59:04 Quit Skail ("Concoction recent by Pier.")
06:13:52setkehdoes anyone know where i can get the source code for the ipod patcher???
06:15:56*ameyer points at
06:16:07ameyermore specifically, the svn command given there
06:21:52Lloreansetkeh: Basically, all the official project tools, and the manual, and the Rockbox and bootloader source code, are all in Rockbox SVN
06:22:52 Join n17ikh|Lappy [0] (
06:23:04ameyerit should also be in the compressed source code archive rockbox releases
06:23:41LloreanI don't know how much that includes.
06:25:01setkehLlorean, yeah im trying to work it out :P i only need it to know how it scans for the ipod because im going to attemp to build a rock boxinstaller that you click on and ita all done lol
06:26:50Lloreansetkeh: How's that going to handle whether or not to download all the extra stuff (fonts, Doom, etc)?
06:27:00LloreanWhy not contribute by improving the user experience of the existing installer?
06:27:47 Join jfb9301 [0] (
06:31:01setkehLlorean, i can add it to the pacage so when the exe file extracts it will extract thoes files too
06:32:26Lloreansetkeh: So the point is to distribute outdated files, just for iPods?
06:33:02setkehLlorean, what do you mean outdated ???
06:34:11 Quit midgey ()
06:34:15Lloreansetkeh: Well, if you package them with the installer, it's only good once. If those files change, your installer will still be using old ones.
06:34:49LloreanAs well, auto-detect could easily mess up another device if it mistakes it for an iPod, so it doesn't seem very safe not prompting the user at all.
06:34:58setkehLlorean, well once i build the pacage ill just repack it when an update comes out and suply that file also
06:34:59LloreanWhat, exactly, is your goal, and why?
06:35:15LloreanAre you going to repack it every time Rockbox updates, or is it going to download the Rockbox build?
06:35:55setkehLlorean, ill have to repack it i have no idea how to make it download the latest relese
06:36:07LloreanRockbox is updated several times a day.
06:36:58LloreanAnd there are 9 different versions for iPods alone.
06:37:53setkehLlorean, do you seriously update you rockbox several times a day tho ??
06:38:18LloreanNo, but that means that the vast majority of the time, you'll be giving people an outdated version.
06:38:38LloreanEven if you update your installer once a day, something like 20 hours a day, it'll be giving old and possibly buggier versions.
06:38:46LloreanYou still haven't told me why you don't want to just contribute to the official installer
06:41:15setkehbecause im not an immensly experianced coder i know enough c++ to package a folder and make it extract to a folder but i cant put F:\ because it varys from pc to pc what driveletter the ipod is
06:42:44LloreanSo what are you going to do with this installer? It's either only going to support one iPod, or it's going to be immensely bigger than it needs to be, and it's basically just going to extract the zip?
06:43:28LloreanAnd being bigger, people will have to redownload it every time they want to install, even if they don't need 90% of the stuff taking up its size.
06:46:05setkehLlorean, diferent relese for diferent ipod so its not stupidly big
06:46:18LloreanAnd then a version without extras for every iPod as well?
06:47:32LloreanHonestly, it sounds like you're setting yourself up for an awful lot of work that serves no purpose.
06:48:18LloreanOnce you download RBUtil, you need to set it up once. Then you just click "Install" and it downloads the latest version and installs it for you, you don't have to keep going to a webpage, and then hoping the version he packaged has the feature you wanted listed on the front page.
06:49:34 Join saratoga [0] (n=41becb3b@gateway/web/cgi-irc/
06:59:30 Quit saratoga ("CGI:IRC (Ping timeout)")
07:04:34***Saving seen data "./dancer.seen"
07:05:13 Join XavierGr [0] (n=xavier@rockbox/staff/XavierGr)
07:07:38 Quit Darksair ("ERC Version 5.3 (IRC client for Emacs)")
07:08:38 Join tvelocity [0] (n=tony@
07:17:47 Join n1s [0] (n=nils@rockbox/developer/n1s)
07:20:02 Quit jhulst (Read error: 113 (No route to host))
07:21:09 Quit jfb9301 ()
07:24:04 Join Adduc^Desktop [0] (
07:25:52 Quit setkeh (Read error: 110 (Connection timed out))
07:26:11 Join J-23 [0] (
07:26:46 Quit Adduc^Desktop (Client Quit)
07:30:16 Quit fdinel (Read error: 110 (Connection timed out))
07:33:03 Join Bagderr [241] (n=daniel@rockbox/developer/bagder)
07:33:30 Nick Bagderr is now known as B4gder (n=daniel@rockbox/developer/bagder)
07:41:31 Part J-23
07:46:07 Quit AndyIL (Read error: 60 (Operation timed out))
07:46:13 Join AndyI [0] (i=AndyI@
07:50:22 Part toffe82
08:12:44 Quit miepchen^schlaf_ ()
08:19:10 Join LinusN [0] (n=linus@rockbox/developer/LinusN)
08:19:45 Join reacocard [0] (n=reacocar@WL-431.CINE.HMC.Edu)
08:30:00 Quit Glert ("Leaving")
08:39:43 Join Lambdumb [0] (
08:39:58amiconnLlorean: Hmm, maybe the problem with the wrong disksize reported in the bootloader and the "rockbox.ipod" not found problem when it's actually there is due to the blocksize guessing
08:41:01amiconnSomehow disk.c finds a sector that resembles a proper fat bootsector close enough to try to mount it, even if it's not the correct one
08:42:02Lloreanamiconn: It seems it's always the same wrong size when it happens, could this have something to do with, maybe, the installed OF version?
08:42:09amiconnIf that happens again, a dump of such a disk would be helpful (sector 0 .... a bit after the start of the fat32 partition should be enough)
08:42:42 Join Rob2223 [0] (
08:42:48amiconnI don't remember the exact size of the G5.5 firmware partition, so dump maybe 100MB to be on the safe side
08:42:49LloreanI mean, my first guess, originally, was that it had to do with the way we were having people format to fat32, but it seems to happen on "properly" restored iPods as well
08:43:09GodEaterand it's happened on the Gigabeat S too don't forget
08:43:12LloreanG5.5 FW partitions are as much as 120 MB aren't they (on the 64MB units)
08:43:33LloreanGodEater: That one might relate to the fact that it's not a pure FAT32 partition, though?
08:43:43LloreanIsn't it some FAT variant?
08:43:59linuxstbYes, TFAT
08:45:43GodEatervery possibly yes
08:45:56GodEaterbut if we're ever to support the beast we need to handle it right ?
08:46:15 Join Waldo000000 [0] (
08:46:20LloreanGodEater: I just meant to say, there could be other explanations on the Beast as well
08:46:44 Quit BigBambi (Read error: 113 (No route to host))
08:47:03GodEaterLlorean: oh certainly
08:47:25amiconnIt would be best if we wouldn't need to guess at all
08:47:35linuxstbBut isn't the blocksize guessing that amiconn is referring to only enabled for the 5g?
08:47:50amiconnlinuxstb: correct
08:48:05amiconnPerhaps using the ipod hardware revision...
08:48:25n1sthere's a guy posting in FS #9369 with the ipod.rockbox not found error on a 1g iiuc
08:48:35amiconnCould a G5.5 (both 30GB and 80GB) consistently be distinguished from a G5 using that?
08:49:15GodEateramiconn: you mean the checking that ipodpatcher does ?
08:49:41linuxstbNo, ipodpatcher doesn't know the difference between the different Video models
08:50:07GodEaterit does check the sector sizes though still right ?
08:50:09Lloreanlinuxstb: I think though he's asking *could* we in the future using the hardware revision number?
08:50:31linuxstbGodEater: Yes, using the scsi "get disk geometry" ioctl or similar.
08:50:36LloreanIf we could check the sector size from the PC, couldn't we just write the value somewhere during bootloader install for the bootloader to use later?
08:51:07GodEaterlinuxstb: yes I remember writing that patch orignally ;)
08:52:08linuxstbLlorean: It may make sense to build two different bootloaders in that case - it would be easy for ipodpatcher to install the right one.
08:52:45GodEaterthat makes more sense imo too
08:52:48amiconnThe problem is that the MBR just contains the start sector and number of sectors of each partition, but these are logical sector numbers/counts, and there's no hint to what logical sectors size these values are based on
08:53:08amiconnlinuxstb: Main rockbox also needs to know the difference
08:53:37LloreanAnd the 7700 or so value is 1/4 the size the 30G iPod actually should be.
08:53:52linuxstbLlorean: Unless you have a CF mod or....
08:54:00*GodEater foresees complications if we start offering yet another set of rockbox builds to differentiate the models
08:54:11Lloreanlinuxstb: Yeah, but I'm just talking, "they're getting reports of 7700 or close to that from the bootloader"
08:54:40amiconnSo disk.c just probes when large logical sector support is enabled: First it tries to mount using 512 byte sectors. If it cannot mount a partition using that, it retries with 1024 byte sectors, and then 2048 byte
08:54:47linuxstbGodEater: Yes, I wouldn't want to split the builds, just the bootloaders (which the user never needs to choose). But a better solution would be to fix the auto-detection,...
08:55:27GodEateramiconn: but doesn't ata.c already know by that stage what size the sectors are ?
08:55:28Lloreanamiconn: Would it be possible then, if rockbox.ipod isn't found with 512 to then try mounting with the next size up?
08:55:29amiconnProblem is that with the (effectively) halved or quartered (?) offsets, it tries sectors within the firmware partition. They can contain anything
08:55:58GodEateramiconn: shouldn't we reverse the order we try mounting in then ?
08:56:03amiconnGodEater: That's the physical sector size, which could give a *hint* on an 80GB G5.5, but not on the 30GB
08:56:04GodEateri.e. start larger and work down
08:56:15GodEateramiconn: ah yes - I see
08:56:17linuxstbMaybe it should detect the Apple (C) at the start of the firmware block, rather than the start of a FAT32 partition
08:56:27 Join ender` [0] (
08:56:33 Quit Lambduh (Read error: 110 (Connection timed out))
08:57:15amiconnMaybe there actually *is* a hint in the mbr. We'd need a bunch of dumps from G5s and G5.5's to compare
08:57:49amiconnIf there is, it's not an official standard though
08:58:30amiconnGodEater: Trying the reverse order could also cause false positives (on plain G5s)
09:00:18 Quit Rob2222 (Read error: 110 (Connection timed out))
09:00:20linuxstbA simple hack could be to make ipodpatcher/rbutil write the sector size to sector 1...
09:00:53linuxstbBut I think that's a hack...
09:01:03amiconnWe could compare the partition size and the total disk size, but that's not really reliable either (partition could be smaller on purpose, e.g. if ipl is installed as well, or a user could have an extra partition for data, formatted ntfs or somesuch)
09:01:42 Quit nuonguy ("Leaving")
09:01:52linuxstbAs I said, I think locating the firmware partition would be more reliable than locating a FAT partition.
09:02:02GodEateryes, I think you're right there
09:02:10linuxstbThat has lots of unique magic - e.g. the Apple (C) header
09:02:26LloreanThat would make Rockbox absolutely dependent on the presence of Apple though
09:02:46linuxstbLlorean: But it is - the flash-based bootloader requires that firmware partition.
09:03:22LloreanI guess, it just seems something that'll have to be rethought if we ever start flashing them
09:03:24amiconnWell, our bootloader resides in that partition
09:03:36*amiconn wonders how apple does it
09:03:42linuxstbLlorean: Yes, but that will be the easiest thing to solve...
09:04:36***Saving seen data "./dancer.seen"
09:05:48*linuxstb notices Apple have different firmware downloads for the 5g and 5.5g
09:06:16GodEaterthat's cheating!
09:06:39linuxstbBut could make it easier for us to detect things...
09:06:39 Join BigBambi [0] (i=86ceaf40@rockbox/staff/BigBambi)
09:06:43 Join kushal_12_27_200 [0] (n=kushal@
09:08:04 Join petur [50] (n=petur@rockbox/developer/petur)
09:08:24*linuxstb compares the flash dumps from a 5g and 5.5g and sees the 5g has the text MA146, and the 5.5g has MA446
09:09:20GodEaterthat would do then
09:09:42linuxstbThe hwrev also appears to differ - 0x000b0005 (5g) and 0x000b0011 (5.5g)
09:10:11linuxstbBut we should do more checks - I think the debug screen shows the hw revision if there are any 5g/5.5g owners here....
09:11:30amiconnDidn't you say the hw rev can't be used?
09:12:00linuxstbIf I did, then I've forgotten why. Or maybe I was thinking of the RAM detection...
09:12:28*GodEater doesn't remember linuxstb saying we couldn't use the hardware rev
09:13:49amiconnMaybe I misread that
09:14:38amiconnIirc LinusN's G5.5 has 0x000b0010
09:14:54*GodEater will check on his 5.5G
09:15:11amiconnCould have been 0x000b0011 though - Nico_P could check that...
09:15:31GodEaterI have a very early 5.5G
09:15:37GodEaterI bought it pretty much as they came out
09:16:03GodEaterb0011 here
09:23:09 Join AndyIL [0] (i=AndyI@
09:23:28 Quit AndyI (Read error: 104 (Connection reset by peer))
09:25:31amiconnLlorean: Btw, if the displayed partition size is exactly 1/4 of what it should be, it cannot be a coincidence
09:26:17amiconnThis could imo only happen if some (buggy) conversion program writes the fat bootsector into the wrong position (512-byte based instead of logical sector size based)
09:26:24Lloreanamiconn: I don't know if it's exactly one fourth, but I'm sure it's close so it seems likely it was.
09:27:02amiconnThis would also mean it's being written into the firmware partition, and hence do some damage to the OF
09:27:50 Quit dan_a ()
09:28:22amiconnDid this happen only on manually converted ipods, or also ones which aleady were winpods (or converted by restoring via windows itunes)?
09:29:00LloreanI am absolutely positive it happens on converted ones. I know the "rockbox.ipod not found" problem happens on restored via itunes ones, but I don't recall what disk size those ones have reported.
09:29:50 Quit sarixe ("Ex-Chat")
09:29:55linuxstbIf the user incorrectly converted with the wrong MBR, then formatted as FAT32, then converted correctly with itunes, that could leave the FAT signature in the wrong place...
09:30:40 Join Abzy [0] (
09:30:47GodEaterI'm pretty sure we've seen this error with people who've always had winpods, and upgraded the apple firmware on them with iTunes too
09:30:58 Join Zagor [0] (n=bjorn@rockbox/developer/Zagor)
09:31:37LloreanI know we've seen the rockbox.ipod error (we just "saw" it today), I just don't know if that includes the wrong size disk message
09:32:07Lloreanamiconn: It's reporting as 7130MB apparently
09:33:02Abzyi need help
09:33:13GodEaterAbzy: just ask, if someone here can help - they will
09:33:14Waldo000000hey, does morse input mode actually work in text editor? "Display + Power" doesn't seem to do anything for me
09:33:16amiconnLlorean: On a 30GB G5.5?
09:33:48Lloreanamiconn: yes.
09:33:55Waldo000000this is on an m:robe, by the way
09:34:06Abzythe wps doesn't work for my 30gb video ipod except the original one
09:34:30GodEaterAbzy: that just means the others you've tried are currently "broken" - keep trying others
09:34:36Lloreanamiconn: It looks like this is happening on ones people haven't had to convert using our manual conversion process.
09:34:55 Quit tvelocity (Read error: 60 (Operation timed out))
09:35:08Abzyis there any way to fix it?
09:35:24LloreanAbzy: They're just outdated WPSes. Some in the gallery will work, some won't.
09:35:32LloreanLook for ones updated recently, in the last couple months.
09:36:11Abzythank you
09:37:11Waldo000000anyone here used morse input mode in the Text Editor plugin? (last time i'll ask, thanks)
09:37:16GodEaterand yes, there are ways to fix it
09:37:31GodEaterAbzy: if you check in the WPS section of the forums, there are posts dealing with how
09:38:04Abzyis it that perl script thing?
09:38:19Abzyi have no idea how to use it
09:38:22GodEaterAbzy: there may be such a script, but you don't need it - they can be fixed manually
09:41:05 Quit pixelma2 ("-")
09:41:19 Join pixelma [50] (i=pixelma@rockbox/staff/pixelma)
09:51:23 Quit AndyIL (Read error: 104 (Connection reset by peer))
09:53:21 Quit Twisty (Read error: 110 (Connection timed out))
09:53:43 Join goffa [0] (n=goffa@
09:53:44kushal_12_27_200hello, I have a sandisk sansa c250. I need to put an mi4 file in the 16 MB recovery partition. I am using Intel Macbook 10.4.11. I cannot use the GUI to do it. Could someone help me copy and paste the mi4 from the terminal, please?
09:54:06 Join AndyI [0] (i=AndyI@
09:54:10 Quit ruskie (Read error: 104 (Connection reset by peer))
09:54:14 Part Waldo000000
09:58:46 Quit Abzy ("Ex-Chat")
10:04:17 Quit goffa_ (Read error: 110 (Connection timed out))
10:04:21 Quit n17ikh|Lappy ()
10:04:34 Quit BHSPitMonkey (Remote closed the connection)
10:07:45 Join Nibbler [0] (
10:19:50 Join PaulJam [0] (
10:23:33 Join pondlife [50] (n=Steve@rockbox/developer/pondlife)
10:23:54 Quit jeffdameth1 (Read error: 110 (Connection timed out))
10:25:17*B4gder hugs the Rockbox logo
10:25:41*pondlife kisses the Rockbox logo
10:25:44 Join jeffdameth [0] (
10:26:07*scorche covers the children's eyes for the next steps
10:26:08JdGordondamn wierdos!
10:26:23pondlifein a gentle and affectionate manner, nothing sexual. How old is the logo anyway?
10:26:40*pondlife feels wrong
10:27:47B4gderthat's 6+ years
10:28:20LloreanSo we could just say "We're keeping it for historic reasons, but may consider a Rockbox Icon or Splash replacement"?
10:28:23pondlifeI like the left-hand-side menu
10:28:47*scorche hugs the Rockbox site
10:29:23LloreanI actually didn't mind his site designs too much, but thought his "Rockbox" header and logo really kinda lacked life.
10:29:40 Quit ajonat ()
10:29:55LloreanIt was like "Some icon, and Rockbox written in a boring font" (subjectivity all mine)
10:29:59n1swow, there was a roadmap back then!
10:30:16pondlifehehe, we learnt...
10:30:23scorchen1s: that was before they leanred their lesson ;)
10:30:30*Llorean wants to re-introduce a very basic roadmap-like concept.
10:30:37*scorche glartes at his keyboard
10:30:49linuxstbn1s: And commits from Bagder and LinusN ;)
10:31:15 Join lasser [0] (
10:33:42Lloreanpondlife: I think if we have a non-mandatory roadmap, like just *saying* "We'd like 3.1 to be focused on getting the recording working better" or whatever in a way just keeps that topic in the air and maybe encourages people who are interested in those areas to pick now as a time to focus a little bit more of their time on Rockbox.
10:34:00pondlifeAbsolutely, I was just joking before..
10:34:46LloreanAnd I think it might be nice to do something like that with the next 3 or 4 releases in mind, then people will see "in 3.3 they want theming improvements, maybe I should try to get my Patch X back into shape, and they might finally commit it for me"
10:35:08LloreanIt encourages the community because they know we'll be looking for specific types of patch right after the prior release.
10:35:28ZagorI think it quickly gets very fuzzy when we say we want things done but don't know who wants to do it
10:36:00JdGordonalso,,, after recording what else is there? its pretty much "fix bugs"...
10:36:19LloreanRecording, FM, Playback, Theming, Stability, Features.
10:36:31Zagorwouldn't a list of things people are working on or "feel strongly about" be more concrete and just as useful?
10:36:40GodEaterCoffee making, and beer brewing need to be on the list
10:36:50B4gderbut what kind of beer?
10:36:52LloreanFor example we could take one release and say "We'd like to introduce several new features for this one, so if you see patches of interest, bring them to our attention." We don't *have* to add new features, but it does help to have "a direction" even if it's not enforced.
10:37:25JdGordonLlorean: yeah RIGHT! there are too many anti-feature-adding people in the group...
10:37:39LloreanZagor: There's something helpful, as a whole, just to say "we'd like to focus on X". It encourages the people who would normally work on it anyway, because you're saying "Now we actually NEED you", even indirectly.
10:37:41JdGordonoh noes.. heaven forbid we have a red delta
10:37:47pixelmaremoving PLA (and improve some keymaps in general)
10:38:37pixelmaPLA in their current form of course
10:39:11LloreanZagor: Since this is all volunteer time, people drift here and back. If we spread the word that we're working on UI improvements in one release, we might get a lot of guys who've lost faith that there patches will ever get accepted becoming more active in trying to clean them up and get them in, for example.
10:39:26JdGordonPLA isnt entirely bad.. it just needs either a rethink, or removal in some plugins
10:39:35LloreanYou have to admit, there are a lot of people who've lost faith that we'll ever accept their patches. If we say "this time we'll, hopefully, look at patches in this category" it's encouraging.
10:39:42B4gderI think a simple "roadmap" is a fair idea to try for the next say two future releases
10:39:50ZagorLlorean: I'm sceptical, but I giess it's worth to try
10:40:06Zagorguess even
10:40:08LloreanZagor: That's the best thing. If it doesn't work, I don't think we can be any worse off, since we're not forcing anyone to work on anything.
10:40:20LloreanIt's very similar to the list of "3.0" bugs we had in the tracker.
10:40:22JdGordonone thing though.. if its in the wiki it will quickly turn into a feature request
10:40:24LloreanIt helped focus attention.
10:41:05LloreanJdGordon: How we present it is very, very important since we don't want to people to misunderstand it as well.
10:41:39Zagorthere's a definite risk of confusing "roadmap" with "release promise"
10:42:31Zagorotoh we should never let PR issues dictate development
10:42:31LloreanI think something along the lines of "For 3.1 we're simply focusing on our usual stability and feature improvements. For 3.2 we're going to be looking at UI improvements, so get your UI patches in shape, and talk to us about what could be improved, over the next three months. For 3.2 we'd like to look at recording, so those of you who've been working on REP and other features, and despair ever seeing it in Rockbox, now is the time to star
10:42:34JdGordondont alot of projects have a "how can you helpout?" type page usually linked prominantly? we could add it to the front page under "why should you run.." and have it a static page
10:43:04JdGordon.. the other thing is that we need to add timeframes onto releases for this to have any hope of working
10:43:04LloreanThen, *we* develop as usual, but encourage more community involvement.
10:43:15LloreanJdGordon: We're aiming for every 3 months still.
10:43:28JdGordonyeah, but people dont nescacarily know that
10:43:37LloreanWe can clarify it in a few places, soon.
10:43:55JdGordonalso, having people think they need to wait 6+ months for their patches to be even considered isnt goign to draw people in
10:44:09LloreanIt is if you pitch it right.
10:44:29JdGordonwanna put up a draft?
10:45:41LloreanJdGordon: Once I figure out exactly how to pitch it, maybe. :0
10:45:44Llorean:) even
10:47:55LloreanSomething like "We try to pick out interesting patches for inclusion all the time. But with thousands of entries in the tracker it can get overwhelming very quickly. That's why with some releases, in addition to our usual oversight of the tracker we'll be looking for well created patches in specific categories that we'll pick for that release. That way we'll be looking at a smaller sampling, and you can know to clean up patches that may
10:49:25LloreanI think if you tell someone "We'd like your patch to be one of our release features, but it still needs X done" in some cases at least, it'll encourage them more than "We can't commit your patch until X is fixed"
10:51:11*pondlife just wants the tdspeed patch sorting out, so it doesn't use massive static buffers...
10:54:18 Join culture [0] (
10:54:50*Llorean would kinda like that patch for his audiobooks.
10:54:59LloreanSounds like some people are experiencing bugs with it too, though.
10:55:15pondlifeYes, but I haven't had a problem with it
10:55:28pondlifeNot recently, anyway
10:55:39LloreanOn a different architecture though.
10:58:54 Join JdGordon_ [0] (
10:59:53 Join fragilematter [0] (n=fragilem@
11:00:28JdGordon_Llorean: yeah that sounds ok..
11:01:29 Quit JdGordon_ (Client Quit)
11:03:37 Join JdGordon1924 [0] (
11:04:11 Join goffa_ [0] (n=goffa@
11:04:38***Saving seen data "./dancer.seen"
11:09:36 Quit JdGordon (Read error: 110 (Connection timed out))
11:14:59 Quit goffa (Read error: 110 (Connection timed out))
11:18:42 Join tvelocity [0] (
11:22:46gevaertsamiconn: the 7700MB display thing is doesn't mean anything. The bootloader code that prints this out just assumes 512-byte sectors (and isn't involved in the actual mounting - this is just a display issue)
11:26:47 Quit culture (Connection timed out)
11:33:13*linuxstb thinks there's only one solution to the problem of rotting patches - experienced devs need to spend time reviewing them and encouraging the patch author to do things needed to get it committed. But none of us do...
11:33:39linuxstb(or we take a decision that the feature isn't wanted, and close the task...)
11:35:12 Quit homielowe (Read error: 104 (Connection reset by peer))
11:35:17 Join homielowe [0] (
11:38:11JdGordon1924linuxstb: well that still leaves the problem of devs with no time and little inclination for reviewing taks which they have no interest in
11:38:16 Nick JdGordon1924 is now known as JdGordon (
11:39:23 Join bmbl [0] (n=Miranda@unaffiliated/bmbl)
11:39:32 Join culture [0] (
11:40:46 Quit JdGordon (Read error: 104 (Connection reset by peer))
11:41:37 Join mcuelenaere [0] (n=mcuelena@rockbox/developer/mcuelenaere)
11:42:58Lloreangevaerts: 9443 is pretty much a known artifact of the fact that it just filters by single tags, so there's only one album with a given name (the other solution would break compilation albums). It may be best to just say "This is how it has to work, give your albums unique names"
11:43:01 Join JdGordon [0] (
11:46:24linuxstbIs pictureflow based on the database, or does it just scan files?
11:46:45peturdatabase iirc
11:47:23pixelmayes, doesn't even work without the database enabled
11:47:37linuxstbSo the core problem is that the database doesn't identify albums correctly?
11:48:07Lloreanlinuxstb: It's impossible to.
11:48:38 Quit fragilematter ("Leaving.")
11:48:46LloreanYou can have a single album with multiple artists, or multiple albums from different artists with the same name, and there's no way to distinguish them unless you have a separate unique "album ID" tag of some sort.
11:49:09linuxstbOf course, but I can't recall people complaining about "album view" in the database - or have they?
11:49:33LloreanI suspect few people have two albums with the exact same name.
11:49:42JdGordonit might be the way tagnavi.config is set up that it actually avoids the issue
11:50:06LloreanJdGordon: True, if you filter artists first, you'll never see the issue.
11:50:34PaulJamyou could check for the albumartist tag and group songs with the same albumartist tag into the same album.
11:50:38linuxstbBut of course "album name" shouldn't be taken as a unique identifier for an album - you want to use a combination all the "album-level" tags.
11:51:04Lloreanlinuxstb: "Album level"?
11:51:22*amiconn has no "album level" tags set apart from album name
11:51:25pixelmabut then you won't see various artist compilations (maybe if you have a "vorious artists" album artist tag
11:51:56*pixelma was too slow
11:52:04amiconnChecking the year would at least reduce the likeliness to hit that issue though
11:52:29pixelmaand can't type
11:52:43Lloreanamiconn: Greatest Hits albums are very, very likely to have different years (and I suspect the most likely to have the same name)
11:52:47*linuxstb also includes a CDDB ID tag in his rips
11:52:49*amiconn actually has the year set for almost all albums, hence his previous statement wasn't entirely correct
11:53:20*linuxstb accepts that breaks with two-CD albums...
11:54:00*linuxstb wonders if the database attempts to assign a unique "album id" to tracks
11:54:21amiconnNot with my way of tagging 2-CD albums...
11:54:25Lloreanlinuxstb: What could you base it on?
11:55:05LloreanThe only valid thing is "Album + Album Artist" and that assumes the existence of a tag not often in use.
11:56:25PaulJamcurrently in rockbox if the albumartist tag is empty it gets the value of the artst tag.
11:56:32soapThere are lots of albums with the same name, Llorean. And even if you meant "few people have two albums /by one artist/ with exactly the same name" there are still collisions.
11:57:08LloreanPaulJam: Which would conflict with trying to use it with indetification.
11:57:45 Join moos [0] (
11:57:57Lloreansoap: I don't think I said few people have two albums with the same name.
11:58:36PaulJamLlorean: why? the only problem would be if you have a compilation and no albumartist tag set, but then you could just tag the files correctly, so albumartist is something like "various artists".
11:59:00LloreanPaulJam: So you're demanding people add extra tags, that they may not even know exist, just so things "Sort normally"?
11:59:17pixelmaLlorean: why? The pure artist tag won't help but at least you would have a chance, now it is the responsibility of the user to tag them
11:59:27markunLlorean: isn't that what we exect of them with the "The .." problem as well?
11:59:27LloreanIt seems like it'd make more sense to have people with duplicate albums just name them "Bob's Greatest Hits" or "Joe's Purple Album"
11:59:31amiconnPaulJam: If you do that, 2 compilations with the same name would collide again
12:00:06Lloreanmarkun: Except that there is no useful solution without that extra tag, and *not* having that tag won't break things that previously sorted one way before.
12:00:32pixelmawell, you can't take care of everything I guess
12:00:39Lloreanpixelma: PaulJam's suggestion would break *ALL* compilation albums until such time as someone retagged every compilation in their collection with the Album Artist "Various Artists"
12:00:49*GodEater sees that Apple has lifted the NDA around the iPhone SDK
12:00:51Lloreanpixelma: I can't see as how drastically breaking current sorting is a _good_ thing.
12:00:56linuxstbLlorean: It's obvious that album ids can't always be guessed by a database builder. But I think that one needs to be there - browsing by album is a fundamental feature of a music database, so we should try to find solutions, not say "it can't be done".
12:01:07pixelmaLlorean: huh?
12:01:47Lloreanpixelma: PaulJam's solution is to use "Album Artist" + "Artist" as a unique identifier, combined with "Use Artist as Album Artist if the latter isn't present". This results in single albums with multiple artists and no album artists being treated as multiple albums.
12:01:50pixelmathey are not sorted correctly now too
12:02:17LloreanThe only thing not sorted correctly right now is if you don't filter by Artist first, all albums with the same name are assumed to be one album with multiple artists.
12:02:49LloreanWe'd be getting rid of that, and replacing it with "If you don't have Album Artist tags, all your compilation albums will be sorted incorrectly"
12:03:01LloreanWhich, I think, would have a negative effect on a significantly higher portion of users.
12:03:02linuxstbGodEater: That's a positive step...
12:03:29pixelmaLlorean: as I understand it, this would have no impact on the "standalone" album sorting - just on artist > album
12:03:35markunlinuxstb, GodEater: but not really rockbox related
12:03:38*amiconn doesn't like the guessing in the db anyway
12:03:58*linuxstb just uses the filesystem, to avoid all these issues...
12:04:07amiconnI have some tracks with no track number where the db guesses the track number from the artist name....
12:04:18pixelmaLlorean: and the "artist > album" sorting is now broken anyway with compilations
12:04:26Lloreanpixelma: That would mean the "artist" filter works differently depending on what precedes it, I think that's even worse...
12:04:34LloreanEr, Album filter
12:05:00Lloreanpixelma: Right now you can use "Album Artist" instead of "Artist" and if the AA tag isn't present, Artist will be used instead.
12:05:01*pixelma has the feeling we are talking about different things
12:05:20LloreanSo while "Artist -> Album" seems broken, "Album Artist -> Album" will get the effect you want.
12:05:49LloreanIf you've set the "various artists" tag.
12:06:49pixelmayes, isn't that what he (and me) said?
12:06:58LloreanWell, I don't know about you
12:07:05LloreanBut the original discussion was for uniquely identifying albums
12:07:10pixelmasorry, can't follow then
12:07:13LloreanAs in, "being able to tell a single album, without filtering by artist first"
12:08:07LloreanAnd the suggestion was to use "Album Artist" as part of that (which I'm okay with) or assuming "Artist" if "Album Artist" is not present, but that latter half will break compilation albums into a dozen separate albums
12:08:22LloreanI think if "Album Artist" is not present, it should be assumed that "Album" alone is a unique identifier.
12:08:34LloreanBasically, "Tag your duplicates" rather than "Tag everything"
12:08:51amiconnIt could use the year if album artist isn't present
12:09:12Lloreanamiconn: Not useful on Greatest Hits, or other collections over time by a single artist.
12:09:49pixelmawell, I think PaulJam and me just talked about "(Album) Artist -> Album" sorting, not using this as identification for "Album" only
12:10:08Lloreanpixelma: Well, the original topic was "Album" only, because we were discussion a specific flyspray task.
12:10:16 Join tree [0] (
12:10:31LloreanSo I didn't realize you weren't discussing what we were, since they're very, very similar.
12:11:01amiconnLlorean: why not?
12:11:04Llorean"Album Artist -> Album" sorting already uses "Album Artist" and falls back to "Artist" if you set your Tagnavi with AA.
12:11:11pixelmayes, but then I think you read too much into PaulJam's first post *I think*
12:11:30Lloreanamiconn: Year is often the year of the song's recording, not the year of album publishing.
12:11:47*amiconn always uses the album year there
12:11:58LloreanYeah, but it's not safe to assume everyone does.
12:12:24Lloreanpixelma: Well, I assumed since he was saying what Rockbox already did, that he was suggesting that it could be done in the area we were discussing too.
12:13:26ZagorLlorean: otoh all cd rippers I've seen only have a field for album year, not for each track
12:13:27LloreanI think the problem with trying to guess is that you can get false positives, and in my mind at least it's better to err on the side of "false positives are nearly impossible" and ask people to use unique tags if it's giving them problems, than to try to get too clever, and start having people file bugs because we're merging (or separating) their albums.
12:14:01amiconnSame as with the track number guessing...
12:14:22LloreanZagor: This is a valid point. I just think it's not solid enough to depend on though, is all.
12:14:43Lloreanlinuxstb: Is there a common tag for CDDB ID?
12:14:43PaulJampixelma: i was suggesting to use the album artist tag to decide if songs with the same album name should be grouped into the same album or different albums.
12:15:22linuxstbLlorean: I'm not sure
12:15:45LloreanIt seems like a truly unique identifier like that might be even better, though it'll split multi-disk albums.
12:15:54pixelmaPaulJam: ok, then maybe I misunderstood
12:18:09 Nick fxb__ is now known as fxb (
12:18:59soapthe foobar2000 forums used to have a long (and long standing) thread on this sort of album sorting mess. There are notable holes in any scheme (outside an "album sort" tag), despite full standard tagging)
12:19:05 Quit tree ("leaving")
12:19:14 Join Nico_P [50] (n=nicolas@rockbox/developer/NicoP)
12:19:49LloreanAnd we'd get an "Album Sort" tag with the sort-tags patch anyway.
12:19:56LloreanWhich could hold any sort of unique ID you wanted.
12:19:57gevaertsfull standard tagging assumes certain music styles, so holes are unavoidable
12:21:13soapyou can do classical (the one I assume you are talking about) with stock ID3 tags. You end up putting lots of info in the track name, but...;)
12:22:08gevaertssoap: you can make it fit somehow, but different people are bound to use different fitting mechanisms
12:22:11soapBut yea, I'm not sure more than "full standard tags" (what you see in most rippers and the basic screen of most taggers) can ever be expected.
12:22:53 Join trisiak [0] (
12:23:10soapI think Itunes' solution is to secretly tag VA albums behind your back, itsn't it?
12:23:24kushal_12_27_200hello, I have a sandisk sansa c250. I need to put an mi4 file in the 16 MB recovery partition. I am using Intel Macbook 10.4.11. I cannot use the GUI to do it. Could someone help me copy and paste the mi4 from the terminal, please?
12:31:27linuxstbsoap: I seem to reacall a "compilation" tickbox to indicate a track is part of a compilation. I've no idea what that does though....
12:32:35 Join DerDome [0] (n=DerDome@
12:32:36soapI'm off to work.
12:32:54soapFS #9428 needs love.
12:33:22linuxstbkushal_12_27_200: Why doesn't the GUI allow you? I wouldn't expect it to work in the terminal if it doesn't work in Finder...
12:34:01kushal_12_27_200because the 16 MB partition is not mounted
12:34:42linuxstbAnd your c250 is definitely in recovery mode?
12:35:23 Join vitja [0] (n=vitja@
12:36:46vitjagevaerts: hi
12:37:18 Quit m0f0x ()
12:37:30kushal_12_27_200yes, afaik
12:40:18gevaertsvitja: hello. I didn't commit anything yet, but my current work is at FS #9436
12:40:59vitjagevaerts: ok i'll take a look
12:41:19kushal_12_27_200It is in Mini-B->Device Mode
12:42:08 Join funman [0] (
12:42:27 Join J-23 [0] (
12:44:08linuxstbfunman: Hi, I found some bugs in my patch from last night, and have just put a new one here:
12:44:40funmanlet me read it
12:45:58linuxstbfunman: I think it would also be nice to round things to a multiple of 4 bytes, and use ldr/str instead of ldrb and strb, but that can be after we get this change working...
12:46:45funmanthe decompression is fast enough, imo it is better to keep the code the simplest
12:47:29 Join JdGordon_ [0] (
12:47:54*linuxstb should test it without the deliberate delay loop, to see if there's a noticable slowdown when booting the oF
12:48:42funmanalso you can extend the firmware block to the maximum size possible
12:48:57funmanthere is no little profit !
12:49:14linuxstbYes, there didn't seem any point, so I just keep it exactly the same size now.
12:49:49funmanAlso, maybe it's worth putting mkamsboot in tools/ , and link it with libucl
12:50:07linuxstbYes, I mentioned the other day that we should do that.
12:50:13funmanI am still thinking of a way to transmit the e200 model from mkamsboot.c to test.S for the button check
12:50:38funmanMaybe mkamsboot.c could write a 'model.h' with "#define CLIP" as a content for example
12:50:58linuxstbIt's *test*.S - so we shouldn't worry about that - users can just set a #define in the Makefile or something.
12:51:04funmanand test.S (or another name) would have .ifdef CLIP <clip button for dual boot> .endif .ifdef e200 ..
12:51:33funmanwell we can now make a 100% safe bootloader
12:51:45linuxstbWe shouldn't try and duplicate the existing Rockbox build system - I think we've reached the stage to start hacking on the main RB source.
12:52:34funmanbut we need dual-boot
12:52:57linuxstbI've been thinking that maybe we should also UCL compress the bootloader - as we're developing this code now, it would seem easier to do it now than later when we run out of space.
12:53:06 Quit JdGordon (Read error: 110 (Connection timed out))
12:53:07funmanthat also
12:53:45 Nick JdGordon_ is now known as JdGordon (n=jonno@rockbox/developer/JdGordon)
12:53:46funmanI mean the "test.S" will become our "dual-boot.S"
12:54:01amiconnlinuxstb: That sounds suspiciosly similar to what the archos flash loader does...
12:54:06funmanand it better should not stay a test, because it will be our recovery mode on Clip and similar
12:54:20linuxstbamiconn: Does it always compress the RB bootloader, or is it optional?
12:54:35linuxstb(or main Rockbox binary, whatever it stores...)
12:55:16amiconnIt handles booting one of 2 images, depending on button presses (plus a serial monitor for low-level recovery work)
12:55:34amiconnEither image can be ucl compressed or uncompressed.
12:56:33funmanlinuxstb: your 2nd patch looks ok
12:56:45amiconnTypically the first (alternate) image is bootbox, ucl compressed, and the second (main) image is full rockbox, either compressed, or (on targets that still have enough room) uncompressed rombox
12:56:49linuxstbfunman: You've tried it in the emulator?
12:56:56funmanyes in the emulator only
12:57:15linuxstbOk, let's give it to kugel to test on his fuze ;)
12:58:18linuxstbfunman: Is the button detection on the various V2s documented anywhere?
12:58:20funmando you want my scripts to test firmware integrity ? (I guess they will be needed a couple more times)
12:58:32funman< here it is very well documented
12:58:59 Quit Xerion (Read error: 113 (No route to host))
12:59:07linuxstbWhat about the wheel light on the Clip, or the LCD backlight?
12:59:10funmanthere is various code snippets on the forum / our git repository
12:59:20 Join Xerion [0] (
12:59:27funmanalso in the patch I had posted on flyspray
12:59:43 Join fragilematter [0] (n=fragilem@
12:59:50funmanyou have to enable the gpio clock (set bit 16 of cgu_peri register)
13:00:09linuxstbDo you disable that clock again before starting the OF?
13:00:16funmanconfigure your pin as input to be on the safe side
13:00:26funmanI never restored the registers state
13:00:33funmanand had no problem
13:01:09funmanwe should unconditionally check A3 (which is either usb or hold on all tested models)
13:01:24funmanand a specific button for each model for dual booting without a computer handy
13:01:35linuxstbWhy "hold" ?
13:01:49funmanwell on the Clip A3 is the state of hold button
13:02:14 Join eight [0] (n=eight_@unaffiliated/eight/x-7825324)
13:02:15funmanso safe way to boot OF: shut it down, turn hold on, plug it on usb
13:02:34 Quit UncleRemus ("leaving")
13:02:40funmanA3 is usb on Fuze and E200
13:02:45mcuelenaerefunman: do you have the English translation of Skyeye - Internals?
13:02:48funmanand might be hold on m200
13:03:00linuxstbfunman: Yes, but why do you want to boot the OF if hold is on?
13:03:25funmanmcuelenaere: nope :/ but I found that skyeye.conf is documented in README of the tarball
13:03:43mcuelenaerehmm bugger
13:03:52funmanlinuxstb: in case we want to flash another rockbox image on it
13:04:24linuxstbfunman: Huh? Isn't that covered by the "insert USB" or "press a specific button" ?
13:04:40funmanyou can't power the Clip if hold is set anyway (except if you powered it by usb)
13:04:41***Saving seen data "./dancer.seen"
13:05:10funmanwell it's only an extra safety
13:05:53funmanbutton led is D7 on E200/Clip
13:05:57funmanI don't remember on Fuze
13:07:06linuxstbCan you detect the "home" button on the Clip? That would seem an easy dual-boot button.
13:07:20funmanyes it is B2
13:07:32funmanbut you want to check that on your side
13:07:52funmancheck the paste: it is the list of known buttons
13:08:28funmanlinuxstb: what can I do to not duplicate work ?
13:09:02linuxstbI'm not doing anything at the moment...
13:09:38funmanI'll move mkamsboot in it's own directory in tools/
13:09:47linuxstbMy "quick2.diff" patch needs testing on target... ;)
13:09:52linuxstb(and then I can commit it)
13:10:29funmanIMO it's safer to test on e200 first
13:10:57funmanno need to hurry :)
13:12:31kushal_12_27_200hello, does anyone have the e200tool compiled compatible with Intel mac 10.4.11
13:14:53linuxstbfunman: How about powering off the Clip? Have you investigated that?
13:15:05JdGordonkushal_12_27_200: I can compile it for you...
13:15:23funmanlinuxstb: what do you mean exactly ?
13:15:35kushal_12_27_200Thank you very much JDGordon. How long does it take to compile it?
13:15:45linuxstbfunman: I mean how to make our code shut down.
13:15:56funmanno I don't know
13:16:23funmanYou can force it to power off by holding the button if it's in a 'stuck' state
13:16:37JdGordonkushal_12_27_200: 30s... if i remembered where it is :p
13:16:37JdGordonoh, hmm... 10.4.11... ive got 10.5.something
13:16:58kushal_12_27_200thanks a lot, JDGordon
13:16:59linuxstbfunman: OK, so that's good enough for now - we just end with an infinite loop and the user can shut down manually.
13:17:18funmanhint: infinite blinking of the led is nice (maybe not for the battery)
13:17:55JdGordonkushal_12_27_200: umm... e200tool? you need to recover it?
13:18:10linuxstbYes, maybe a "final" version of test.S before we start work on Rockbox itself could be to implement dual-boot button detection, with the "rockbox" option going into an infinite loop, flashing the LED.
13:18:10kushal_12_27_200JDGordon, yes.
13:19:40JdGordonarg, nup, no libusb on the mac... sorry
13:20:32 Join dan_a [0] (n=d917a37d@gateway/web/cgi-irc/
13:20:45*linuxstb spots a stranger with a Clip
13:20:51*dan_a waves
13:20:53linuxstb(if he remembers right...)
13:21:02funmanhey dan_a
13:21:11kushal_12_27_200are you on a mac, JDGordon?
13:21:14dan_aHi funman
13:21:29funmanI wanted to ask you something for some time
13:21:38JdGordonkushal_12_27_200: no, i have a mac.. im on linux atm
13:21:42funmanyou wrote the NAND driver for sansaV1 ?
13:21:51 Join mc2739 [0] (
13:21:53dan_aDo we know anything about the I2C on the V2s yet? The version of the datasheet which I have doesn't talk about it
13:22:15dan_afunman: I did the initial version, based on MrH's work
13:22:31funmanis the exact same code used both for NAND & SD cards ?
13:22:42kushal_12_27_200do you delete files from your website often? do you think you still have a file from February this year on your website?
13:22:46amiconnThere is no "nand" on the v1
13:23:05dan_aIt all goes through the SD driver
13:23:08amiconnThe internal flash basically is an SD card in a different package
13:23:16funmanit's the same on the V2
13:23:42funmanwell it's a real NAND chip, but (apparently) connected to a SD-NAND interface chip made by SanDisk
13:24:28kushal_12_27_200JDGordon, do you delete files from your website often? do you think you still have a file from February this year on your website?
13:24:44amiconnAt least it's a 1000% better solution than the bare-metal nand interface on the various tcc targets
13:25:19funmanwell when this interface is documented :o)
13:25:21JdGordonkushal_12_27_200: which was it? i can have a look
13:26:08dan_afunman: Is it similar to the v1's SD chip, or don't we know?
13:26:14JdGordonnup, looks like that was before i reformatted it...
13:26:26funmandan_a: no it appears to be a ARM PL180
13:26:28kushal_12_27_200ok, thanks for looking it up for me
13:26:51funman(guessed from RE)
13:29:16 Quit bmbl ("Woah!")
13:29:40dan_aIt has documentation on the ARM site. That is useful
13:30:09funmanbut I couldn't issue a read cid command to it :'(
13:30:31 Join LambdaCalculus37 [0] (
13:32:40dan_aDid I read that someone was close to having an LCD driver ready?
13:32:54funmanyes: fdinel on irc
13:33:21 Quit Nibbler (Read error: 113 (No route to host))
13:33:22dan_aThat will make things a lot easier
13:33:33linuxstbWhat does "close" mean? Has he displayed anything on the LCD?
13:34:13 Join JdGordon_ [0] (
13:35:10funmanI must say I'm a bit confused by your 'quick' patch linuxstb
13:35:19funmanNow I don't know what is where in the firmware
13:36:21linuxstbfunman: Ah yes, I need to update the comments...
13:36:47JdGordon_gevaerts: did you end up doing any investigation into that data abort?
13:36:50funmancan you do that and highlight me with the patch please ?
13:36:59linuxstbtest.bin is at 0x0, and (ucl_unpack + ucl image) are together (in that order) at the end of the firmware block.
13:37:04funmanI'm going, but should be there in the afternoon
13:37:18funmanok, it confirms my suspicions
13:37:35funmansee you
13:37:36linuxstbi.e. I just moved the ucl_unpack function to immediately before the compressed image, and extended the memcpy to copy both.
13:37:36 Join UncleRemus [0] (
13:38:15linuxstbfragilematter: Around to test something? ;)
13:39:13kushal_12_27_200JDGordon, how could I get a header.txt? Do you know what that is supposed be?
13:39:42JdGordon_no idea what that is...
13:43:02JdGordon_gevaerts: im gonna need more info on that data abort.. it didnt happen on my sansa (although im not 100% sure i had a playlist to resume on boot...)
13:43:49kushal_12_27_200gevaerts, I am flashing my sansa c250 right now with the c250reapir-mac tool. I have firmware.mi4, pribootLoader.rom, and c250repair-mac on my desktop but I am missing header.txt . Please advise.
13:49:07 Quit JdGordon (Read error: 110 (Connection timed out))
13:49:11 Nick JdGordon_ is now known as JdGordon (n=jonno@rockbox/developer/JdGordon)
13:54:25 Quit LambdaCalculus37 ("Ka-chunka")
13:54:54gevaertsJdGordon: I havent' looked yet. I ran out of time last night
13:54:56fragilematterlinuxstb: sorry for the delay, I was afk but now I'm back
13:55:12gevaertskushal_12_27_200: sorry, no time right now. I'm at work
13:55:14 Quit dan_a ("CGI:IRC (Ping timeout)")
13:55:23JdGordongevaerts: ok, got a link to the problem reports? i dont see it in my flyspray emails?
13:55:27linuxstbfragilematter: Fancy testing something?
13:55:31fragilematterhit me
13:56:25linuxstb - apply to current SVN
13:57:03linuxstbfragilematter: I don't know if you read the commit message (or the IRC logs), but the SVN version is now working for both me and funman on the Clip.
13:57:25kushal_12_27_200thanks for letting me know, gevaerts
13:57:26linuxstbSo you could try that as well, but this patch improves it (removes the limitation on available firmware padding size)
13:57:32fragilematterthe one I tested last night?
13:57:41fragilemattererr, it was night for me ;))
13:57:52linuxstbThis was about 14 hours ago.
13:57:56gevaertsJdGordon: forums and IRC
13:58:09fragilematternope, I tested something more than 16-17 hours ago
14:00:30 Join mf0102 [0] (
14:02:48fragilematterlinuxstb: damn, I bricked it with the second test yesterday so I'll have to unbrick
14:03:27fragilematterlinuxstb: mind if we do the test in about 30 minutes, I have to go to lunch
14:03:38linuxstbfragilematter: Sure, there is no rush.
14:03:55fragilematterokay, tty in about 30 minutes or so
14:05:49 Join dan_a [0] (n=d917a37d@gateway/web/cgi-irc/
14:11:02kushal_12_27_200I am flashing my sansa c250 right now with the c250reapir-mac tool. I have firmware.mi4, pribootLoader.rom, and c250repair-mac on my desktop but I am missing header.txt . Please advise.
14:11:30JdGordonwhere is header.tx ever mentioned?
14:11:50 Quit mc2739 ()
14:14:04 Join nplus [0] (
14:14:11 Quit mcuelenaere ()
14:15:46 Quit at0m|c (Read error: 104 (Connection reset by peer))
14:18:23fragilematterlinuxstb: tested and working with a 5 second delay on boot
14:18:50fragilematterin fact about 8 seconds, but it's working
14:18:57linuxstbfragilematter: SVN or the patch?
14:19:06kushal_12_27_200 I quote "23.39.07 # <gevaerts> You'll need the other files now. Make sure you have header.txt, firmware.mi4 and pribootLoader.rom on your desktop"
14:19:20linuxstbfragilematter: OK, thanks. I'll test on my Clip now, and then commit (assuming it works...)
14:19:34fragilematterlinuxstb: fingers crossed
14:20:21kushal_12_27_200JDGordon, do you think header.txt is not necessary?
14:23:00linuxstbkushal_12_27_200: Are you the same person who did this in February? There is a conversation with gevaerts in the logs here -
14:23:22linuxstb(with a link to a header.txt)
14:24:33 Quit moos (Read error: 104 (Connection reset by peer))
14:24:47 Join moos [0] (
14:25:09linuxstbfragilematter: Yes, works on my clip as well.
14:25:24fragilematterlinuxstb: great!
14:25:27kushal_12_27_200thanks a lot, linuxstb, I am the same person. I tried looking up the old irc logs but somehow missed that day's log.
14:28:28 Quit tvelocity (Read error: 104 (Connection reset by peer))
14:29:15 Join Mathiasdm [0] (
14:29:44 Quit Mathiasdm (Read error: 104 (Connection reset by peer))
14:33:30 Join wwall [0] (
14:34:00 Quit Seed ("cu, Andre")
14:35:44funmanlinuxstb: I think as a dual boot we should rely on A3 since we don't know buttons for all the models (and also because A3 is the only 'button' accessible for the Fuze), plus model-specific buttons; and remove the check for A3 when we have a reliable method for all other models
14:36:51 Quit JdGordon (Read error: 113 (No route to host))
14:38:08linuxstbfunman: OK. I agree we can't be too safe at this stage.
14:38:37linuxstbI've just committed quick2.diff, with a simplification to the memcpy added. It worked on my Clip, and before I changed the memcpy, on fragilematter's e200
14:40:55funmancool, I'll be able to use last clip firmware
14:41:49 Quit DerDome ("Leaving.")
14:42:10 Join tvelocity [0] (
14:42:26 Join Darksair [0] (n=user@
14:47:12 Part B4gder
14:47:17 Join JdGordon [0] (
14:47:59 Join J-23_ [0] (
14:50:49 Quit culture (Read error: 104 (Connection reset by peer))
14:52:29 Quit JdGordon ("Konversation terminated!")
14:53:13 Join at0m|c [0] (
14:55:01 Join culture_ [0] (
14:55:19 Join JdGordon [0] (
14:56:48wwallthanks for the cool software
14:57:14wwallI just put the manual on my ipod, because I wanted to read/have the manual on my player
14:57:41 Quit J-23 (Read error: 113 (No route to host))
14:57:46wwallI converted the html to text using html2text, it works but is a little messy.
14:58:06wwallso my question is is there a better way to do this?
14:58:14 Nick culture_ is now known as culture (
14:59:27kushal_12_27_200hmm, I am not sure what I did but I did not get it right.
15:01:04 Join Twst [0] (
15:03:08 Part sandsmark
15:04:45***Saving seen data "./dancer.seen"
15:07:45 Join bmbl [0] (n=Miranda@unaffiliated/bmbl)
15:10:14 Join DerDome [0] (
15:10:40preglowwwall: well, for reading on the ipod it neads be to be plain text format one way or the other
15:11:14preglowi expect there are many ways of doing that
15:11:59preglowthere might even be a way of converting the raw latex source code to text
15:12:45 Quit J-23_ (Remote closed the connection)
15:13:21wwallpreglow: that's what I'm thinking of doing now
15:13:46 Join LambdaCalculus37 [0] (i=44a04303@gateway/web/ajax/
15:14:13wwallIdeally I'd want to convert it to reStructured or markdown
15:14:28wwallto get some structure but still have it readable
15:17:14 Join J-23_ [0] (
15:17:24 Join tucoz [0] (i=528612c1@gateway/web/ajax/
15:17:29funmanfragilematter: don't run away, I'll ask you to test my dual-boot in some time :)
15:17:30 Nick J-23_ is now known as J-23 (
15:17:50fragilematterfunman: I'll be around ;)
15:18:05linuxstbwwall: Rockbox doesn't have any "structured text" viewer at the moment, although I think there are some unofficial patches around, for example a wikipedia viewer.
15:18:22tucozwwall: you could try: make manual-txt
15:18:44linuxstbfunman: You're adding button detection etc?
15:20:04moosdoesnt the wikipedia patch one good candidat for the future comming (hopefully) patch reviewing/commenting/commiting sessions?
15:21:11funmanI really want to remove word-padding of ucl files, the current memcpy+decompression executes in less than 1/2s
15:22:05linuxstbI pad the ucl file to ensure the ucl unpack function is aligned. What's the harm?
15:22:31funmanoh I didn't think about the function
15:22:45funmanWell it adds complexity on the code
15:23:10linuxstbYes, a little, but the code should be quite stable now - i.e. no reason to change things.
15:23:33funmanI add support for a compressed rockbox image
15:24:25linuxstb40KB isn't enough for you? ;)
15:24:53funmanlinuxstb: that way we can create a rockbox image without caring about dual boot and decompression
15:26:53linuxstbYes, but it's easier to do it all in the bootloader - this way, we'll need to add an extra binary into the build system (the "dual_boot.bin"). But I think it makes sense to compress the bootloader, so we'll have to add that.
15:28:24 Join midgey [0] (
15:30:16 Join kugel [0] (
15:31:44linuxstbfunman: There's a "bin2c" tool in tools/, so if you add mkamsboot there, you can also add the "dual_boot.bin" and ucl unpack function, and convert them to .c arrays, and link directly into mkamsboot. There is also a "libucl.a" generated in tools/ucl/src, which can also be linked to mkamsboot - meaning we can do a simple "mkamsboot of.bin bootloader.bin patched.bin"
15:32:16linuxstbOops, seems bin2c isn't there, but in sansapatcher/ipodpatcher...
15:32:29funmanah I didn't think about the bin2c, but I wanted to link with libucl.a in a 2nd approach
15:32:56funmani.e. in 2 steps, for easier potential debugging
15:34:39 Quit Rob2223 ()
15:34:54 Quit JdGordon ("Konversation terminated!")
15:36:42 Part tucoz
15:37:29 Quit dan_a ("CGI:IRC (Ping timeout)")
15:37:57 Quit fragilematter ("Leaving.")
15:43:15 Quit midgey ()
15:50:42 Join JdGordon [0] (
15:52:30 Quit bmbl (Read error: 113 (No route to host))
15:53:45pondlifeHours late, I know, but I'd certainly not like to assume that a year tag is related to albums at all. That assumption can only be made for "Album..." tags, IMHO - so "Artist" is no use either. I don't know if there was a conclusion, but I'd say that "Album artist"-"Album" is a reasonable key, but there should be no substitution or magic for a missing/blank "Album artist" tag; in that case just use "Album".
15:57:13JdGordonthere is always the option of adding some unique id to all files in a folder with the same album name
16:01:30 Join domonoky [0] (n=Domonoky@rockbox/developer/domonoky)
16:02:21JdGordonhey domonoky
16:03:03JdGordonim fudging with my routers so my mac is on and off the net atm... wanna get the rbutil compile done nowish?
16:04:16domonokyi dont need the mac access now, as we already built the new rbutil 1.0.7 mac version... But i will need it again, when its time for the next release...
16:04:40JdGordonoh ok, cool
16:04:46JdGordoni thought it wasnt done yet
16:04:59wwalllinuxstb, tucoz: I can't get make manual-txt to work right now ("Unknown option -no-numbering"), so the converted html will have to do for now
16:06:41pondlifedomonoky: Are the old binaries of RBUtil archived anywhere?
16:06:57pondlife1.0.7 for Windows is crashing for me
16:06:58domonokypondlife: yes, all are on the download server
16:07:01linuxstbThey should still be on the download server - /rbutil/
16:09:09pondlifeThanks, will try 1.0.6
16:09:19pondlifeAh, what's 1.0.7b ?
16:10:37wwallanyway, I wanted to propose to have a text version of the manual included in the default installation, where is the best place to submit such a proposal?
16:10:59domonoky1.0.7b is version 1.0.7 with the sapi bug fixed. :-)
16:11:07pondlifeThat might be my problem ;)
16:11:15pondlifeCrash on SAPI config?
16:11:37domonokypondlife: yes, crash/hang on everything that uses the sapi script...
16:11:54pondlifeOK, problem solved
16:11:59linuxstbwwall: I don't think there's a need to propose it - it's been discussed before, but it's just that no-one has yet found a good way to generate it automatically from the latex source code.
16:12:02domonokythe sapi script changed to expect utf16-le input/output, and we missed that in rbutil...
16:12:20pondlifeFastest fix yet, then, from my point of view..
16:13:14domonokypondlife: time to fix was -5days :-)
16:13:22 Quit coatman (Remote closed the connection)
16:14:19pondlifeI must have picked up from the old wiki (maybe early on the 29th)
16:14:40 Quit vitja (Read error: 104 (Connection reset by peer))
16:15:44pondlifeDoes hold the news that's dropped off the front page? I was expecting to see 3.0 mentioned for some reason.
16:16:29 Join coatman [0] (
16:20:12Zagoris there no wps screenshot in the manual?
16:20:24JdGordonis there any point?
16:21:07ZagorI don't know. I'm just updating our freshmeat entry and was looking for a few screenshots
16:21:40JdGordonah, grab some off the wiki maybe?
16:21:41LambdaCalculus37Zagor: You can always take the WPS screenshot from the Wikipedia entry on Rockbox.
16:22:47ZagorLambdaCalculus37: that's a good one. thanks.
16:26:00wwalllinuxstb: ok, thanks.
16:26:13LambdaCalculus37Zagor: You're welcome. :)
16:27:57pondlifeAny Archos owners tried bookmarks recently? might be worth a look
16:28:59LambdaCalculus37pondlife: Haven't used my Archos for a while now. Thanks for giving me an excuse to fire it up again. :)
16:29:15pondlifeLambdaCalculus37: Shame on you ;)
16:29:45ZagorI use it every time I start my car. don't think I've ever bookmarked anything on it though :)
16:30:07 Join fragilematter [0] (n=fragilem@
16:34:24 Part LinusN
16:36:01 Join funman_ [0] (n=fun@
16:37:46 Quit funman (Read error: 60 (Operation timed out))
16:37:52 Nick funman_ is now known as funman (n=fun@
16:44:36 Quit amiconn (Nick collision from services.)
16:44:42 Join amiconn [50] (n=jens@rockbox/developer/amiconn)
16:47:20 Nick JdGordon is now known as JdGordon|zzz (n=jonno@rockbox/developer/JdGordon)
16:47:41 Part wwall
16:48:29funmanfragilematter: ping
16:48:41fragilematterfunman: pong
16:48:57fragilemattersorry for leaving earlyer, I had an urgent matter to attent do
16:49:11funmanno problem
16:49:22funmancan you test a patch ? (I just have to add button check)
16:49:29fragilematterof course
16:50:53funmanI tested it in the emulator first, it seems fine
16:51:20 Part J-23
16:51:20fragilemattereven if it doesn't that won't be a problem now :D
16:51:22funmanjust give me a button
16:51:34fragilematterstick with A3 :D
16:52:24 Join J-23 [0] (
16:52:34LambdaCalculus37pondlife: Bookmark creation didn't crash my Archos JBRv1.
16:53:01 Quit Zagor ("Client exiting")
16:53:05 Join Rob2222 [0] (
16:55:16funmanfragilematter: you can apply it with git-apply
16:55:35fragilematterI'll have to sync the git forst
16:55:53funmandefault is "rockbox" (infinite blink of button led)
16:57:57 Join mf0102_ [0] (
16:59:30fragilematterI got a "fatal: git-apply: bad git-diff - expected /dev/null on line 73" when trying to apply the patch
16:59:56fragilematterand a bunch or trailing whitespaces, but those are no problem...
17:00:04funmanif you have the rockbox git tree, you should be able to use git-am
17:01:03fragilematterI applied it to the git tree from :))
17:01:24fragilematterI thought that was what you meant
17:01:25funmanhehe no this tree is a bit old
17:01:48fragilematterso, where's the rockbox git tree, I only have svn :D
17:02:11funmanlet me generate a proper patch for svn tree
17:02:29 Join toffe82 [0] (
17:03:06 Join Schmogel [0] (
17:03:49 Quit perrikwp (" ajax IRC Client")
17:04:47***Saving seen data "./dancer.seen"
17:05:26 Join Hillshum [0] (
17:05:54 Quit Hillshum (Client Quit)
17:06:06funmanfragilematter: @@
17:06:14funman * (apply with patch -p4)
17:06:27fragilematteron the root of the svn
17:07:31 Quit mf0102 (Read error: 104 (Connection reset by peer))
17:07:33 Join Hillshum [0] (
17:07:54linuxstbfunman: Any idea if all main firmware blocks (i.e. for the various devices) are less than 128KB?
17:08:03funmanso far yes
17:08:14funmanbut the code wouldn't be hurt by sanity checks
17:08:21fragilematterI think I only need -p 2
17:08:36linuxstbfragilematter: That depends where you are when you apply it...
17:09:02fragilematterthe root of the svn, and it should be working with relative paths
17:09:39funmanp1 then
17:09:51fragilematterp1 worked
17:10:01funmangit uses fictives a/ and b/ directories for original and modified files
17:10:35fragilematteryou used uclpack 2 times?
17:10:43funmanI pack the 'rockbox' image also
17:11:24funmanI made a mistake
17:11:32fragilematterI didn't flash yet :D
17:11:45funmanfragilematter: in boot.S GPIOA value is incorrect, should be 0xC80B0000
17:12:30fragilemattercleaned and made it again
17:12:51fragilematterit boots the test code on A3 = #0 or A3 != #0?
17:12:58funman= #0
17:13:03 Part Hillshum
17:15:04linuxstbkugel: You could test SVN code on your fuze - it's working on the Clip and E200...
17:15:26fragilematternormal boot: blinks like crazy
17:16:08*linuxstb thinks we're getting good at this ;)
17:16:29fragilematterthe blinking stopped quite fast on power button
17:16:38fragilematterit wasn't like it had the time to reset
17:16:51fragilematterbut it boots normal on usb
17:16:54funmancool :)
17:17:09funmanI'll add some code for model-specific button tests and ask you again
17:17:20fragilematterokay :)
17:18:57funmanfragilematter: choose a button for dual boot (one of the 4 directions)
17:19:36fragilematteryou don't have to squeeze your finger between the others to press it :))
17:19:38markunlinuxstb: which player are you guys workig on? The sansa v2's?
17:19:51 Quit Schmogel (Read error: 104 (Connection reset by peer))
17:20:10linuxstbfunman: To be consistent with Rockbox, you could add "TARGET=-DTARGET_NAME" in the Makefile, and then add $(TARGET) to the test.S compile command. e.g. -DSANSA_CLIP, -DSANSA_E200V2
17:20:15linuxstbmarkun: Yes.
17:21:08kugellinuxstb: the procedure is still the same? I.e. edit the make file to point to my fuze firmware file, then just make
17:21:50fragilematterkugel: the test.S from svn has no safety check
17:22:00linuxstbYes, and then copy patched.bin to your device, renaming appropriately.
17:22:07linuxstbfragilematter: What kind of safety check?
17:22:09fragilematterso it is pretty much equal to bricking if you have no recovery mode
17:22:20fragilematterbutton test for dual boot
17:22:35 Join Schmogel [0] (
17:22:39funmanfragilematter: the svn has no dual boot either
17:22:49kugelfragilematter: svn is tested on both clip and e200v2, so I guess there's little chance to brickage
17:23:14fragilematterI forgot that it branches back to the of after delay
17:23:31fragilematterI guess I got used to gpio checks in test.S :D
17:23:44funmanlinuxstb: ok weel, but I need to add the target first in the build management
17:23:59pondlifeAnyone think I should disable the keyclick on PP targets in 3.0 ?
17:24:07linuxstbfunman: What build management? The "user" (i.e. developers...) can just modify the Makefile
17:25:07linuxstbpondlife: Aren't you a bit late to ask that? ;)
17:25:14pondlifeYes ;)
17:25:16*linuxstb will be impressed if pondlife can do it
17:25:25pondlifeI meant in SVN, sorry
17:25:35pondlifeJetlag and DVT - not good
17:25:45*kugel can't find his fuze
17:25:49funmanas has no support for defines
17:25:59funmangiven on the command line I mean
17:26:21funmangcc -nostdlib -c would be fine
17:26:51linuxstbfunman: Ah yes... I had the same problem with the ucl function - took me a while to realise that arm-elf-as wasn't running cpp
17:27:11linuxstbI don't think you need -nostdlib with -c
17:29:09*kugel needs a pointer to his fuze
17:29:45funmanhum cpp replaces the defines: ".ifdef SANSA_CLIP" => ".ifdef "
17:30:19funmanoh I must use the cpp #define, not as .define ..
17:30:37*fragilematter 's pointers are all out of range :(
17:31:08markungevaerts: It might be difficult to find a M6T1/TP: meizu offered to exchange them for a newer model.
17:31:10amiconnlinuxstb, funman: Just let gcc handle the .S files - it will run cpp and then as
17:31:20funmanyes that's what I do
17:31:23amiconnAfaik this is how it's done in various parts of rockbox...
17:31:31kugellinuxstb: 5s delay, then maybe some delay due to the decompression of the OF, then successful boot into the OF <−− is that supposed to happen?
17:31:45funmankugel: nope, your fuze should have been bricked !
17:31:50funmanour evil plan failed :(
17:32:05kugelfragilematter: out of range was good, I found the fuze, but not where it was supposed to be
17:32:06linuxstbamiconn: Yes, that's what I did.
17:32:22kugelfunman: I haven't done anything yet
17:33:06linuxstbkugel: Yes, that's what we're hoping for...
17:33:13funmankugel: he ok, yes this is what should happen (the decompression is nearly not noticeable)
17:33:27kugel"uclpack not found"
17:33:40fragilematteryou need to make uclpack in the tools dir
17:35:33fragilematterokay, I'll clean up first
17:36:10funmanI corrected a size check in mkamsboot.c, and added button checks in Makefile & boot.S
17:36:25funmanyou can run diff on the 2 diffs to get the incremental change ;)
17:36:44fragilematterI cleaned up already
17:36:48funmanor was it only a comment that I corrected .. :o
17:37:50 Join rasher_ [0] (
17:37:59 Quit rasher_ (Client Quit)
17:38:00*kugel crosses fingers
17:38:06fragilematterfunman: do you want to check or should I go ahead with the test?
17:38:17funmanfragilematter: test ! test !
17:38:42kugelgreat guys
17:38:57funmankugel: cool :)
17:39:05fragilematteruncomented the e200 define and changed the infile name... gonna make it
17:39:08lasseryes, cool. really!
17:40:17fragilematterupdating the sucker...
17:40:56fragilematternormal boot by pressing power alone
17:41:15fragilematterblinking like crazy on down
17:41:20funmanhum I missed something then .. it should default to rockbox
17:41:37fragilematteryou set it the other way around :))
17:41:59funmanmaybe the meaning of C6 is reversed (0 = button pressed)
17:42:21funmani'll update my clip
17:42:37funmanfragilematter: what if you change the bne into beq in boot.S ?
17:43:03fragilematterbeq boot_of
17:43:47 Join mcuelenaere [0] (n=mcuelena@rockbox/developer/mcuelenaere)
17:43:50funmanhum my check for home button is broken
17:43:54funman(on the clip)
17:44:23fragilematterit just boots the of all the time?
17:44:25funmanI can't read it - but my check for hold button is correct so I can recover
17:44:34fragilemattergreat then :D
17:44:35funmanno it boots the 'rockbox' code
17:44:54funmanI'll just use down, to be consistent with E200
17:45:17fragilematterweird as hell
17:45:24fragilematterit behaves like before
17:45:34*fragilematter is suspecting a bad flash
17:45:42funmanah no stupid me I forgot to uncomment the define in Makefile
17:46:11*fragilematter is stupid too, edited a different #define section
17:46:17 Quit wpyh (Remote closed the connection)
17:46:20 Join mf0102__ [0] (
17:46:32 Quit mf0102_ (Read error: 110 (Connection timed out))
17:47:02funmanfragilematter: so is this patch correct, or does the check for e200 button need to be changed ?
17:47:20kugelfunman: Let's keep booting into the OF by default.
17:47:21fragilematterthe down check is correct of you want it to boot the OF by default
17:47:37fragilematteryep, I vote for OF by default too
17:47:52funmanwell I think for all the other ports it's rockbox by default
17:47:57 Quit petur ("work->home")
17:48:12fragilematterwhen we'll have rockbox booting we can change it :D
17:48:20kugelfunman: but in early stages it's to the OF by default. We can change later when rockbox is in a usable state
17:48:37funmanfragilematter: and if I want rockbox by default, does changing bne into beq at line 80 makes it behave the correct way ?
17:48:51fragilematterI'll check it now, earlier I edited the clip code
17:48:51kugelfunman: See, a broken butten detection or forgotten
17:48:52funmankugel: how can you run rockbox on your fuze if you don't have any way to dual boot then ?
17:49:13kugelforgotten #define can render the device into a brick, while we can reach the same without having this risk
17:49:25fragilematterkugel: we still have usb/hold check as a failsafe
17:49:42funmankugel: no, in case anything goes wrong you can put on the hold button and plug your device over usb to run the OF (and update the firmware)
17:49:54fragilematterusb for fuze & e200
17:50:03funmanfor e200 and fuze you don't need the hold button
17:50:26funmanbut that makes a procedure common to all sansaV2 models, maybe less confusing to newcomers
17:50:53fragilematterfunman: your @s are broken again
17:51:05kugelBut what if the USB detection is broken? I rather insert the USB to see the blinking LCD, instead of being dependent on the detection in order to boot into the of
17:51:23funmanfragilematter: sorry ... you can change ' ' into ' '
17:51:34funmanI type the first 'space' with altgr+space
17:51:37fragilematterkugel: it can't be broken, it is independent of the #defines
17:51:56 Join wpyh [0] (n=william@
17:52:09kugelfragilematter: Didn't you agree with me some minutes ago?
17:52:41fragilematterI wanted default OF so I won't have to do key combinations when I want to use the e200 :))
17:52:55funmankugel: the USB detection is not broken, it's the very first check made
17:53:12funmanand it is made unconditionally of the device (I have a brickable model too)
17:53:34 Join rasher_ [0] (
17:53:44funmanfragilematter: well just flash it with the OF when you're finished hacking
17:54:02kugelin any case, I'd feel alot more comfortable if we boot into OF by default
17:54:27kugelfragilematter's argument is very valid too
17:54:28fragilematterfunman: flashing the e200 takes a bit more than the clip and I'd rather not
17:54:44fragilematterbeq = reversed behaviour
17:55:02funmanwith beq it boots 'rockbox' by default ?
17:55:30funmanI'll comply with your requests guys :)
17:55:49 Nick rasher_ is now known as rasher (n=rasher@rockbox/developer/rasher)
17:56:09*fragilematter feared funman might have used a full clip on the disobeying crowd :))
17:56:15kugelfunman: let me test the led blinking code. I'm keen on that for some time now :)
17:56:26funmankugel: just wait for the next patch then
17:56:33kugelwill do
17:57:32funmanhm imagine that:
17:57:42funmanmy device has no battery
17:58:06kushal_12_27_200gevaerts, would you be able to help me now?
17:58:07funmanI plug it over usb to recharge it, but then our bootloader will execute 'rockbox' which doesn't support recharging over USB
17:58:12kugelfunman: "mov r2, #0x40000" that's the address in the ram we copy the compressed image too, sin't it?
17:58:24funmankugel: yes, it's the upper limit of our copy
17:58:47fragilematterfunman: the instruction manual says "press hold to recharge" :P
17:58:50funmanso still, I'm reluctant to make OF by default
17:59:05kugelwe have a lot more ram available on the non-clips, how about moving it further behind for this targets?
17:59:06funmanwell for my Clip yes
17:59:10funmannot for kugel's Fuze
17:59:17funmankugel: there is no point
17:59:48funmanwe can overwrite the compressed image since we don't use it anymore after booting
18:00:20kugelokok, I was just asking :) Never change a running system is probably the better way here
18:00:20fragilematterso, we're now down to sd & fat drivers
18:00:35funmanlcd, audio, usb ..
18:00:46fragilematterfor the bootloader :P
18:01:02kugelI assume the sd driver will be nearly the same as on e200v1
18:01:14funmanwe do have the bootloader now !
18:01:28funmanlinuxstb: how about committing this ?
18:01:38fragilematterthe bootloader that will boot rockbox from disk :))
18:01:53gevaertskushal_12_27_200: in half an hour or so. I'm on my way home now
18:02:01funmanfragilematter: we can put rockbox in the firmware file directly
18:02:08kushal_12_27_200thank you
18:02:33fragilematterflashing every new version would be less than ideal
18:02:48linuxstbfunman: How/where?
18:02:53kugelfunman: that's not the goal. we want to be able to update as easily as on existing targets
18:02:54funmanvery true
18:03:02 Part eight
18:03:07funmanlinuxstb: in svn ?
18:03:27fragilematterkugel: patch -p 1 on the svn root
18:03:27linuxstbfunman: I was responding to "we can put rockbox in the firmware file directly"
18:03:38funmankugel: true, but for now we don't have a way to read from the nand
18:03:43kugelfragilematter: did you test this version?
18:03:50fragilematterfunman too
18:03:51funmanlinuxstb: replace test.ucl by a compressed rockbox image
18:04:06linuxstbfunman: It won't fit - a full Rockbox binary will be a few hundred KB
18:04:11funmanah ok
18:04:15kugelfragilematter: I assume funman changed something since the last pastebin
18:04:36fragilematterlet me see...
18:04:51funmanI only reversed the meaning of C6 button for e200 and removed the weird spaces
18:05:02linuxstbfunman: Although obviously it can't be for the Clip.... That's a problem for later though.
18:05:28funmanI meant we can use it to test other code while waiting for a NAND/SD driver
18:05:46fragilematterwell, a nand/sd driver would probably be the test code :D
18:06:09linuxstbfunman: Yes, that's what the Rockbox bootloader will be used for - the bootloaders are just stripped-down versions of Rockbox itself, using the same drivers.
18:06:23kugelfunman: How about just trying the e200v1 SD driver and see if it works
18:06:25funmanSince the AMS bootloader loads the SanDisk firmware from NAND, the nand could be already initialised when we start executing
18:06:34funmankugel: it's just not the same
18:06:43funmanlinuxstb: Ok
18:06:50linuxstbfunman: So the plan will be to start work on writing a Rockbox bootloader, and use that to test/develop the various drivers. Once those drivers are written (which are needed for Rockbox itself anyway), we can switch to loading a Rockbox binary from disk.
18:06:54fragilematterkugel: and it might be dangerous
18:06:57kugelfunman: why exactly are you so sure?
18:07:18linuxstbfunman: And then we're done ;)
18:07:25funmankugel: because I read both code (in rockbox for V1 and in OF for V2)
18:07:58kugelfunman: Better compare OF for v1 with OF for v2
18:08:01fragilematterguys, I'll be out for about 30 minutes... any last words? :))
18:08:19fragilematterkugel: different processors = different OF
18:08:26kugelfragilematter: Have you tested the latest pastebin patch?
18:08:38fragilematteryep, tested and working
18:08:47kugelfragilematter: it's both arm, not too different
18:08:59funmankugel: you can do that if you want but I think it's useless, the C driver of rockbox is much more readable
18:09:34fragilematterboth arm, but they have different hardware architectures
18:09:38 Quit BigBambi (" ajax IRC Client")
18:09:51kugelI strongly assume Sandisk reused the SD interface from the v1 (since they've added it to the AMS package)
18:10:00fragilematteranyways, I'll be back in a while
18:10:06kugelfragilematter: see you
18:10:08*fragilematter waves
18:10:14funmanlinuxstb: now it's time to sync the flyspray patch adding the e200v2 target
18:10:18 Quit z35 ("Leaving")
18:10:40linuxstbfunman: Oh, there's a patch already?...
18:10:43funmanand I think remove the assumptions that it shared drivers with as3514
18:11:22funman#8843 *
18:11:24linuxstbHmm, April 2008...
18:11:28kugelWhy not prove that the assumptions are wrong first?
18:12:00funmankugel: well feel free to test ;)
18:12:17funmancheck which hardware is in the Fuze, etc
18:12:35 Join bmbl [0] (n=Miranda@unaffiliated/bmbl)
18:12:42 Quit goffa_ (Read error: 60 (Operation timed out))
18:13:46kugelI highly doubt that hth posted a nand driver without being at least 90% the driver is correct
18:13:47 Join bughunter2 [0] (n=Jelle@
18:13:50linuxstbfunman: I'm not sure about that patch - I think it's better to just commit things in much smaller pieces.
18:14:17kugel90% sure
18:14:47funmanwell it's tested
18:14:53 Join MethoS- [0] (
18:15:06funmanand now I'm not sure how I can split it
18:15:28linuxstbfunman: We're confused again - I'm talking about #8843
18:15:46funmanyes sure I agree
18:16:01funmanbut at least the defhardwa
18:16:08funmandefine for hardware registers*
18:16:17linuxstbNo, the hardware registers aren't defined in the Rockbox style.
18:16:25linuxstbSee the other files in firmware/export/
18:16:40 Join goffa [0] (n=goffa@
18:16:54 Quit Schmogel ("Miranda IM! Smaller, Faster, Easier.")
18:17:00linuxstbIt looks very useful to use as a reference, but I don't think it's committable.
18:17:11funmanI'll check the NewPort wiki page also
18:17:15funmansee you later
18:17:17 Join Schmogel [0] (
18:17:41 Quit funman ("non pas myleeeeeeeene")
18:18:01 Quit Schmogel (Client Quit)
18:19:55*linuxstb spots a problem in tools/configure for the Clip - the "memory" variable assumes multiples of 1MB...
18:21:00 Quit tvelocity (Read error: 60 (Operation timed out))
18:21:05kugellinuxstb: what do you think? Couldn't it be possible that the sd interface is entirely the same with the v1?
18:21:30linuxstbkugel: I have no idea, I haven't looked at either.
18:25:38amiconnlinuxstb: Why is that a problem? Does the clip come with various different memory sizes?
18:25:51linuxstbamiconn: It has 320KB of RAM
18:26:04amiconnYes, but it *always* has that
18:26:22amiconnSo set 'memory' in configure to 0 ...
18:27:15linuxstbBut that means hacking the .lds files to ignore it...
18:27:46linuxstbBut I guess the Clip will need some special lds anyway...
18:27:47amiconnIsn't the .lds target specific anyway these days?
18:28:44linuxstbPotentially, yes. But it's generally architecture specific...
18:29:47amiconnWell, the other option would be to change configure to use KB. That means touching all targets....
18:30:38J-23linuxstb: heh, "DOS addresses only 1 Megabyte of RAM because we cannot imagine any applications needing more."
18:30:51linuxstbYes, I was thinking of KB, but don't like doing those kind of changes...
18:31:03*amiconn is really curious how rockbox will be made to fit on the Clip...
18:31:16gevaertskushal_12_27_200: I'm here now
18:31:49 Quit fragilematter ("Leaving.")
18:32:49linuxstbamiconn: Moving lots of things to plugins (that's basically what the OF works). Although I've no idea how that can be done cleanly...
18:33:03gevaertskushal_12_27_200: could you quickly summarize the state of your c200?
18:33:44amiconnlinuxstb: Does the clip's CPU have an MMU?
18:33:44linuxstbamiconn: But this work is applicable to all the SansaV2 targets, I just bought a Clip because it's the cheapest such device, and wanted to help get the port started.
18:34:01linuxstbI think so.
18:34:01amiconnDo the others have more RAM?
18:34:15amiconnWell, if it has, you could use virtual memory...
18:35:02linuxstbI think the others all have a "normal" amount of RAM. I don't know for sure though.
18:36:42amiconnHmm. The clip seems to have the same OLED display as the Express
18:37:00amiconnThis is actually monochorme (just with 2 differently coloured areas) right?
18:37:10linuxstbYes, it looks that way.
18:37:14kugelfunman reported that too
18:37:46amiconnWell, monochrome will already save quite some meory (smaller bitmaps, no album art...)
18:38:15amiconnGoing thumb would probably help too
18:38:57amiconnJust compare with the archoses - they're monochrome, and thumb code should be comparable in size to SH1 code
18:39:08amiconn(SH1 istructions happen to be 16 bit too)
18:39:35linuxstbHow big is a typical archos binary?
18:39:36kugelI have no doubt rockox can run on the clip
18:39:44amiconnWell, archos is hwcodec, but they still need a playback engine. Codecs are separate anyway (but need their buffer of course)
18:39:48linuxstbkugel: Not by magic...
18:40:15kugellinuxstb: sure. archos bin size is ~200KB btw
18:40:17amiconnDoes the Clip have recording? Radio?
18:40:43kugelfm recorder 260KB
18:40:46linuxstbI've no idea... Let me turn it on...
18:40:56amiconnSmallest archos binary is the Player (193KB) but no radio and no recording, and charcell
18:41:17linuxstbYes, it has FM
18:41:30amiconnRecorder is 248KB uncompressed
18:41:31linuxstbAnd a "Voice" menu, which I assume is mic recording.
18:42:01amiconnThis is just the binary though - it needs further ram for BSS
18:42:31linuxstbAnd that's with dircache?
18:42:43linuxstbAh yes, lowmem...
18:43:04gevaertskushal_12_27_200: after my repair tool you need to follow the instructions in
18:43:43*linuxstb sees the Archos losing the "lowmem" title in the future
18:44:01linuxstbUnless the clip is just called "nomem"
18:44:47amiconnWell, I still wonder how the Openneo people managed to do it
18:44:54gevaertsCall it "I'm sure there was some memory somewhere!"
18:45:10amiconnThe neo has only 256KB RAM, and openneo is a fork of (very) early rockbox
18:45:10linuxstbOr "fullmem" and "highmem"
18:45:36LambdaCalculus37And here I thought that the iFP was already our "king of the lowmem targets", despite it not being a supported target.
18:45:50amiconnThe neo doesn't need to be conservative with power though, as it's fed by a car battery...
18:46:18amiconnBut they even added voice iirc...
18:46:46 Join create [0] (n=55443911@gateway/web/cgi-irc/
18:48:34amiconnMaybe the Express is just a Clip in a different case?
18:48:51kushal_12_27_200i am holding the record button. progress so far:
18:49:38kugelkushal_12_27_200: you can enter recovery mode or not?
18:49:48linuxstbamiconn: I know nothing about the Express - is that a third type of CPU?
18:49:53gevaertsOK. Now it should be in recovery mode. Does it show anything on the screen?
18:49:56kushal_12_27_200the screen is still blank
18:49:58 Join PaulJam_ [0] (
18:50:19amiconnlinuxstb: I mean the Sansa Express
18:50:24gevaertsThat's weird
18:50:43kugelkushal_12_27_200: did you run "e200tool init" before?
18:50:48linuxstbamiconn: I know - I meant to say "does that use a third type of CPU, or is it either PP or AMS"?
18:50:57kugeltry that
18:51:18amiconnlinuxstb: I don't know. But the technical data looks suspiciously similar to the Clip
18:51:18kushal_12_27_200so can i let go of record button?
18:51:37gevaertskushal_12_27_200: once it's plugged in you don't need to hold any buttons
18:51:49linuxstbamiconn: Seems to be sigmatel of some kind
18:52:09 Quit PaulJam (Read error: 110 (Connection timed out))
18:53:00kushal_12_27_200gavaerts, it seems like e200tool-mac is still alive.
18:53:15 Join jhulst [0] (n=jhulst@
18:53:28kushal_12_27_200the process did not finish, afaik
18:53:42gevaertsPress ctrl-C
18:53:42 Part pondlife
18:53:55kushal_12_27_200ok done
18:54:40kushal_12_27_200e200tool init gives command not found
18:55:01gevaerts./e200tool-mac init
18:55:45kushal_12_27_200its doing a backward count 9 8 7 ...
18:55:50linuxstbamiconn: Specifically, the Sigmatel 3630 - (300MHz ARM926EJ-S core!)
18:56:07kushal_12_27_200not found, do I need to disconnect it and connect again?
18:56:29amiconnOh, hmm
18:57:54 Join funman [0] (n=fun@
18:58:10kushal_12_27_200OK I did that. latest pastebin:
18:58:15 Join BigBambi [0] (n=Alex@rockbox/staff/BigBambi)
18:58:56kugelwell, now you should be able to use the recover command, but it seems to be stuck again
18:59:43 Join ea_suter [0] (
19:00:07*gevaerts doesn't really understand why it's stuck
19:00:44ea_suterhi there
19:01:10kugelis there a sudo/root equivalent on mac? I know on linux you need to run e200tool with root access
19:01:23gevaertsThere is sudo
19:01:24kushal_12_27_200hmm , os x has sudo
19:01:29kugeltry this
19:01:43ea_suteryesterday i had asked what I would need to help with the e200v2 porting, and one of the answers was to be able to "code for example". what does coding for example mean exactly?...
19:01:43kushal_12_27_200do I need to control c first? right?
19:03:01funmanea_suter: write device drivers for the hardware
19:03:12linuxstbea_suter: Just that coding (programming) is one way you can help (i.e. an example of how you can help)
19:04:01kugellinuxstb: Is there a way for me to get the datasheets for the as3525?
19:04:23funmankugel: you can ask them to Bagder
19:04:32funmanwell it's only 1 document
19:04:35kushal_12_27_200ok, done!
19:04:38 Join fragilematter [0] (n=fragilem@
19:04:48***Saving seen data "./dancer.seen"
19:05:01kugelkushal_12_27_200: good, looks way better, now repeat the recover step
19:05:10linuxstbkugel: What funman said
19:05:19kugelBagder: ping
19:05:41kushal_12_27_200is it this one ?
19:06:02kugelkushal_12_27_200: the "./e200tool-mac recover pribootLoader.rom" one, just with sudo this time
19:06:31kushal_12_27_200do I need to disconnect and reconnect the cable for that?
19:08:00kushal_12_27_200It is searching for the device but no luck. :( latest: device not found.
19:08:48gevaertskushal_12_27_200: unplug and try again. I don't think the init step is really needed
19:09:05kugelafaik it is
19:09:41 Join Thundercloud [0] (
19:09:42kugelbut apparently it's not always finding the device. From the pastebin, the first init said not found, the second one directly after found it
19:09:53gevaertskugel: I've never needed it
19:10:59linuxstbI recall someone (maybe barrywardell) saying that e200tool was generally unreliable on a Mac - it needs lots of attempts to get working, and possibly a reboot of the Mac.
19:11:00kushal_12_27_200do I need to unloce before the recover step?
19:11:11 Join miepchen^schlaf [0] (
19:11:26kushal_12_27_200unlock the hold
19:11:35pixelmadoes someone know by chance what font is used in the pegbox - title and game screen?
19:12:07ea_suterpl thanks guys
19:12:07 Join Wictor_ [0] (
19:12:10kushal_12_27_200should the hold be orange or white when I do the recover? I am confused a bit
19:12:14kugelkushal_12_27_200: No, keep it locked, and press REC immediately after the recovery command succeeded
19:12:20ea_suterwell, I have a little mips assembly knowledge
19:12:42ea_suterkind of new to reverse engeneering stuff though
19:13:50kugelgevaerts: btw, I've made the enum for FS #6800, is it what you expected?
19:14:01amiconnmcuelenaere: Is it normal that an Onda vx747 normal build doesn't build, and a bootloader builds but throws a lot of warnings?
19:14:25 Quit jfc (Read error: 104 (Connection reset by peer))
19:14:29mcuelenaereamiconn: yes
19:14:33gevaertskugel: a lot better, yes:)
19:14:37kushal_12_27_200thanks. latest pastebin here :
19:14:43mcuelenaereI haven't tried the normal build yet and bootloader build isn't ready for the build table yet
19:14:45 Join jfc [0] (
19:14:53amiconnOk, so my crosscompilers on Interix are all working as they should
19:14:54mcuelenaereit's, let's say, highly experimental ;)
19:15:02kugelkushal_12_27_200: you are now in the recovery mode, aren't you?
19:15:04gevaertskushal_12_27_200: ok. Anything on the screen?
19:15:25amiconnNow I just need to write gpl'd replacements for stdint.h and inttypes.h, and then write down the instructions...
19:15:37kushal_12_27_200Screen says: Mini-B->Device Mode (next line starts here)USB2.0 MSD (next line starts here)LUN0 locked
19:15:42kugelgevaerts: You could just try the patch with your c200
19:15:48 Quit Wictor (Read error: 60 (Operation timed out))
19:16:12kugelkushal_12_27_200: ok, I expect it keeps adding LUN0 (un)locked lines?
19:16:12kushal_12_27_200gevaerts, now I can see the 16 MB partition in disk utility
19:16:19gevaertskugel: I'll do that later, while I'm testing things on the c200 anyway
19:16:21kushal_12_27_200just one line
19:17:00kushal_12_27_200but I cannot mount the 16 MB partition
19:17:02kugelkushal_12_27_200: Ok, I've experienced that it adds such lines for the time being. Anyway, you should be in recovery mode and see the 16MB FORMAT drive
19:17:10LambdaCalculus37kugel: I'd like to try your newest revision of FS #6800 on my c200. Can you roll a build for me? My laptop is indisposed at the moment.
19:17:17funmanea_suter: maybe a first step is writing a lcd driver, there is some informations on
19:17:24kugelLambdaCalculus37: of course! :)
19:17:25gevaertskushal_12_27_200: you need the mount lines from last time. Let me check
19:18:07kushal_12_27_200yes, I see it but not in finder. only in the disk utility
19:18:35*funman thinks it's time for subscribing to rockbox mailing lists
19:18:49gevaertskushal_12_27_200: run "mkdir /tmp/sansa"
19:19:00LambdaCalculus37funman: Join the club! ;)
19:19:21gevaertskushal_12_27_200: after that, "sudo mount -t msdos /dev/disk1 /tmp/sansa"
19:19:32kushal_12_27_200file exists. Ok moving on to next step
19:19:57funman : the "tools/configure" section is a bit messed up, some problem with wiki update ?
19:20:06kugeldoesn't make feature automount?
19:20:23gevaertskugel: usually it does, but apparently not always.
19:20:28kushal_12_27_200gevaerts, done, no output
19:21:06kugeldrop the firmware file, and empty sansa.fmt file and the bootloader file into that folder
19:21:19kugels/make/mac/ btw :S
19:21:23gevaertskushal_12_27_200: ok. Do you still have firmware.mi4 and pribootLoader.rom?
19:21:23LambdaCalculus37funman: Looks like the layout is all wonked up.
19:21:38kushal_12_27_200yes, gevaerts in desktop
19:21:42gevaertskugel: not that easy. /tmp/ won't appear in the finder...
19:21:44 Join herrwaldo [0] (
19:21:50LambdaCalculus37funman: I'm going to fix the page now.
19:22:16amiconnLambdaCalculus37: Maybe some twiki plugin is missing?
19:22:21kugelgevaerts: well, why did you tell him to create the mount point in tmp then? ;)
19:22:22gevaertskushal_12_27_200: ok. "sudo cp firmware.mi4 pribootLoader.rom /tmp/sansa"
19:22:23amiconn(or needs enabling for that page)
19:22:37createhello all
19:22:40gevaertskugel: I don't have a mac here, so I don't know where else...
19:22:51 Join nuonguy [0] (
19:22:53gevaertskugel: also, I'm pasting from logs :)
19:23:11kushal_12_27_200gevaerts, do I need header.txt?
19:23:32gevaertskushal_12_27_200: do you have it? If so, "sudo cp firmware.mi4 pribootLoader.rom header.txt /tmp/sansa"
19:23:35LambdaCalculus37amiconn: I found %CODE{"bash"}% in the wiki page.
19:23:39gevaertsIf not, I don't think it matters
19:24:12createhas anyone had a look at my #9325 patch ?
19:24:24funmanI think I saw other pages with the same problem
19:24:26gevaertskushal_12_27_200: after that, "ls -l /tmp/sansa", to verify that everything is there
19:24:28kushal_12_27_200done, no output
19:24:58kugelgevaerts, LambdaCalculus37: <−− backlight fading build
19:25:03 Join perrikwp [0] (i=9821373a@gateway/web/ajax/
19:25:32gevaertskushal_12_27_200: can you pastebin the output?
19:25:36kushal_12_27_200latest pastebin :
19:25:50kushal_12_27_200oh, just the output? ok
19:25:56gevaertsthat's fine
19:26:06gevaertskushal_12_27_200: now "sudo umount /dev/disk1"
19:26:15LambdaCalculus37funman: It's fixed now.
19:26:16kugelyou certainly don't need the
19:26:24funmanthanks ;)
19:26:31kugelkushal_12_27_200: add an empty sansa.fmt please
19:26:33kushal_12_27_200no output
19:26:42gevaertskushal_12_27_200: now "sudo hdiutil detach /dev/disk1"
19:26:47gevaertskugel: no point
19:26:52LambdaCalculus37amiconn: I replaced the %CODE{"bash"}% with the <verbatim> tags. The page appears correct now.
19:26:54kushal_12_27_200kugel, I am quite illiterate in this matter
19:26:58gevaertskugel: sansa.fmt doesn't do anything on c200
19:27:06kugelok then
19:27:53kushal_12_27_200disk unmounted, disk ejected says terminal. disk unlocked, disk ejected says sansa
19:28:02gevaertskushal_12_27_200: ok. Now unplug
19:28:50kushal_12_27_200reading main image, writing main image... it rebooted and I guess now I need to unlock the hold switch?
19:29:20gevaertsOnce the OF boots, go to the Settings menu and ask it to format, to make sure
19:29:21kushal_12_27_200it turned on to the sansa
19:29:43kushal_12_27_200I am in language screen. I am choosing English and going to format
19:30:06*pixelma summons midgey
19:30:10 Nick gromit`` is now known as gromit_na_gromit (
19:30:12kushal_12_27_200format complete.
19:30:22gevaertsok. Now it should be as good as new again
19:30:31kushal_12_27_200thank you gevaerts!
19:30:45*kugel doesn't want credits
19:30:58kushal_12_27_200thank you kugel
19:31:04kushal_12_27_200I never forgot you either
19:31:06kugelLambdaCalculus37: How's it?
19:31:09kushal_12_27_200you guys are awesome!
19:31:17gevaertskugel: actually, the fact that sansa.fmt doesn't work is the entire reason you need special tools for the c200
19:31:44kushal_12_27_200i just copied .rockbox into the root folder in sansa.
19:31:58LambdaCalculus37kugel: The backlight fades out but doesn't fade back in.
19:32:07kushal_12_27_200ok ttyl. bye for now!
19:32:15*gevaerts waves
19:32:25*kushal_12_27_200 waves back
19:32:32 Quit kushal_12_27_200 ("Leaving")
19:32:34kugelLambdaCalculus37: really?
19:32:54LambdaCalculus37kugel: As it's said, "ya rly". :)
19:34:05kugelWell, actually my build contains a change which is not in the patch on the tracker
19:34:08kugelin button.c
19:34:15kugeljhMikeS: ping
19:34:48kugeland obviously, backlight_is_on doesn't work so well (i.e. doesn't return false with off backlight)
19:35:23LambdaCalculus37kugel: Can you roll another build without the other change?
19:35:27 Quit moos ("Rockbox rules the DAP world")
19:36:11kugelsure, wait a moment
19:37:16kugelLambdaCalculus37: I tell you why I didn't notice it on my e200
19:37:21 Quit fragilematter ("Relaxing with a bit of TeeWorlds gameplay")
19:37:30kugelit turns on when I use the wheel
19:38:05kugeljhMikeS: Why is backlight_on() in button.c on every button press even if the backlight is on?
19:38:42 Join bertrik [0] (
19:39:03 Quit Wictor_ (Read error: 104 (Connection reset by peer))
19:39:15 Join Wictor [0] (
19:39:49kugelLambdaCalculus37: there you go, same link
19:40:24 Join PaulJam__ [0] (
19:41:40 Quit daurnimator ("Cyas later...")
19:43:01funmanwhat's MODEL_NUMBER define in firmware/export/config-*.h ? c100 and ondavx747 have the same for example
19:44:32 Part ea_suter
19:45:27linuxstbfunman: It _should_ be a unique number... But it's not that important, I think it's just used in the header of main Rockbox binary to identify the model that binary is for.
19:46:09gevaertsThere are lots of duplicates already
19:46:19kugelI assume it's the same id you add in the configure script in the model id field
19:46:37kugelor should be*
19:47:11funmanthere is also target= in tools/configure
19:47:45 Quit Horschti (Read error: 110 (Connection timed out))
19:47:58kugelI was refering to that one
19:48:21 Join Horschti [0] (
19:48:28linuxstbfunman: That one is for the menus in configure...
19:48:47linuxstbAh, no, it's not...
19:49:11 Quit PaulJam_ (Read error: 60 (Operation timed out))
19:50:36pixelmajust realised that we shouldn't use the current pegbox images anyways... :\
19:52:58LambdaCalculus37kugel: This one works.
19:53:02 Join kronflux [0] (
19:53:16 Quit create ("CGI:IRC (EOF)")
19:53:17kronfluxheyy everyone
19:53:27*bertrik waves
19:53:37LambdaCalculus37Hi kronflux!
19:53:41kugelLambdaCalculus37: nice
19:53:49*linuxstb fixes configure to make target_id unique
19:53:58kugelLambdaCalculus37: how's the fading compared to the beast?
19:54:20 Join meven [0] (
19:54:26LambdaCalculus37kugel: Very smooth.
19:54:42 Quit MethoS- (Read error: 104 (Connection reset by peer))
19:56:22 Join MethoS- [0] (
19:58:01 Quit Wictor (Read error: 104 (Connection reset by peer))
19:58:12 Join Wictor [0] (
19:58:15 Quit Darksair ("ERC Version 5.3 (IRC client for Emacs)")
19:58:53 Join setkeh [0] (
19:59:02funmanCONFIG_CPU in firmware/export/config.h has strange numbering, any specific meaning ?
20:00:41setkehany one have anyidea where to start if im going to build a plugin for tv support for the 5g ipod video (i like the feature wanna add it to the box :P)
20:00:58kugelLambdaCalculus37: Well. Currently it's configured to in/decrement the brighntess each HZ/25
20:01:45kugelI was thinking of making it target dependent (or even global_settings dependent)
20:02:11 Join {phoenix} [0] (
20:02:34 Nick PaulJam__ is now known as PaulJam (
20:02:39kronfluxthats interesting. the chip the nano2g uses supports touchscreen.
20:03:46linuxstbfunman: I'm changing MODEL_NUMBER now in all the config files to make them unique. The next free one will be 40.
20:04:08funmanI had chosen 69 to be sure ;)
20:04:12LambdaCalculus37kronflux: Heh... I haven't seen it being used in a touchscreen target (although I'm thinking that it's the Meizu M8).
20:05:07kugelLambdaCalculus37: How exactly do you mean better?
20:05:07setkehany one have anyidea where to start if im going to build a plugin for tv support for the 5g ipod video (i like the feature wanna add it to the box :P)
20:05:14kronfluxLabmdaCalculus37: so what are we trying to do with the info we gather for the lcd? you said collect anything related to the lcd.. so if I knew what we're trying to do, maybe I'd understand what I'm looking for more :p
20:05:25LambdaCalculus37kugel: The effect is very smooth on my c240.
20:05:48kugelYou told me some days ago that it's smooth on the beast as well
20:05:53LambdaCalculus37kronflux: I'm ordering a nano2G LCD screen. I need to see if there's any identifying numbers on it.
20:06:17 Quit {phoenix} (Remote closed the connection)
20:06:28kronfluxLabmdaCalculus37: identifying numbers?
20:06:28funmancan you order spare ipod parts ?
20:07:07 Join tvelocity [0] (n=tony@
20:07:10LambdaCalculus37kronflux: Yes. Most manufacturers will usually label chips, ribbon cables, and LCDs with part numbers for identification.
20:07:22LambdaCalculus37funman: From my favorite iPod repair shop. ;)
20:07:34kronfluxLabmdaCalculus37: so what will the part numbers help with?
20:07:41kugelLambdaCalculus37: answer please :)
20:08:52LambdaCalculus37kugel: The effect is smooth on the beast as well.
20:08:58 Join {phoenix} [0] (
20:09:06kugelBut you said on c200 it's better
20:09:24kugelbetter because it's smoother or better because it's faster (the latter one can be fixed easily)
20:10:11LambdaCalculus37kugel: Better because it's smoother. The beast is smooth on lower brightness settings, but a wee bit clunky on higher settings.
20:10:18linuxstbsetkeh: We're assuming the TV-out is attached (and controlled by) the Broadcom video processor in the 5g. This chip is undocumented...
20:10:51LambdaCalculus37kronflux: Punching part numbers into Google will usually get you information about said part.
20:10:52*setkeh scratches head
20:10:59kugelLambdaCalculus37: come on. I asked you yesterday (or the day before) it's less smooth on higher settings, and you said "no"
20:11:02setkehlinuxstb, elaborate please mate
20:11:26kugelI asked since I expected it to be less smooth
20:11:33markunsetkeh: what exactly do you want to know?
20:11:37kronfluxLabmdaCalculus37: good point :)
20:11:53LambdaCalculus37kugel: I'm comparing again, and I have to revise because now I'm noticing that the effect *is* a teeny bit clunky on higher settings.
20:11:58linuxstbsetkeh: There's not much more to say... You'll need to reverse-engineer the Apple firmware to work out how to control this chip. A far from trivial task - which is why no-one has even started it in the two years those ipods have existed...
20:12:09setkehmarkun, i wanna build a plugin to make tv out work for rockbox
20:12:52LambdaCalculus37kugel: I'm also comparing again because I want to see how the effect is on the Sansa versus the beast, side-by-side.
20:13:22kugelLambdaCalculus37: Ok, accepted :) Now I'd like to now if the shorter fade time on the c200 is preferably over the longer fade time on the beast (or vice-versa respectively)
20:15:19LambdaCalculus37kugel: Having a shorter fade time would be better.
20:16:06kugelOk, I can roll a build with a shorter time for the beast, if you want to test
20:16:21*LambdaCalculus37 has his beast ready to test
20:18:47 Join petur [50] (n=petur@rockbox/developer/petur)
20:20:40 Join Strife89 [0] (n=michael@
20:21:00funmanexport/system.h:197:27: error: system-target.h: No such file or directory
20:21:25funmanwhere can I configure the location where this header is looked for ?
20:21:43funmanthere is one in target/arm
20:22:01 Quit Wictor (Read error: 104 (Connection reset by peer))
20:22:07linuxstbfunman: It's in the target-tree - what you set the manufacturer, cpu etc variables to.
20:22:11 Join Wictor [0] (
20:22:55funmanI copied the e200 config-e200.h but there doesn't seem to be a system-target.h specific for sansav1
20:23:51linuxstbfunman: This may give you a clue - it's the first commit of a new port (in this case the Logik DAX)
20:24:38linuxstbfunman: You should create a directory "firmware/target/arm/as3525" and put the as3525 specific files there.
20:24:59funmanshouldn't it be in sandisk/ ?
20:25:01linuxstbYou can then have target-specific subdirs under there - e.g. "clip", "e200v2"
20:25:51linuxstbfunman: No - we've discovered that devices with the same SoC are closer than devices from the same manufacturer.
20:26:00kugelfunman: the existing sandik should be renamed rather
20:26:31 Quit mf0102__ (Read error: 110 (Connection timed out))
20:26:42funmansvn commit 15339 is 0cb7a302d6dbe167ff981f2677edc5daae2c8312 in the git tree
20:27:07*linuxstb doesn't use the git tree
20:27:19 Join mf0102__ [0] (
20:27:22 Join fophillips [0] (n=fophilli@unaffiliated/fophillips)
20:27:47fophillipsAny idea why my database only gets halfway through D when updating on my 80GB 5th gen. iPod?
20:28:51fophillipsUsing r18682-081002
20:29:03gevaertskugel: I've quickly tried your build, and it seems to work. Just fast enough to not notice, and slow enough to make it a smooth transition
20:29:55kugelgevaerts: You know, it depends on your brightness setting
20:30:04gevaertsdefault settings :)
20:30:17kugelso 6? at 6 it should look just fine
20:30:59gevaerts6, yes. It looks a lot better than I expected
20:31:15linuxstbfophillips: Most common problem seems to be a corrupt filesystem - have you checked it for errors?
20:31:25kugelyea, I'm also surprised again and again how good it looks with only 6 steps
20:31:34LambdaCalculus37gevaerts: Which of kugel's builds? c200 or beast?
20:31:44fophillipslinuxstb: I’ll check now
20:32:09kugelLambdaCalculus37: have you tried the latest beast build? fading time should be ~halfed
20:32:29LambdaCalculus37kugel: I'm extracting it right now.
20:32:37fophillipslinuxstb: Am I right in thinking that vfat is the only supported FS?
20:32:43kugelgevaerts: So, you think it's fading too fast?
20:32:52*Strife89 just popped in and wonders where he can find kugel's builds.
20:33:04gevaertskugel: no. It's just fine
20:33:12kugelStrife89: depends on what build you're searching for
20:33:20gevaertsMake it faster and it will start getting in the way I think
20:33:28LambdaCalculus37Strife89: We're testing a backlight fading patch. Want to help?
20:33:36linuxstbfophillips: Yes
20:33:38Strife89LambdaCalculus37: Sure.
20:33:42fophillipslinuxstb: No errors
20:33:45Strife89LambdaCalculus37: Link it, please.
20:33:47 Quit Wictor (Read error: 104 (Connection reset by peer))
20:33:58kugelwhich target?
20:33:59LambdaCalculus37Strife89: <−− c200 with FS #6800 applied.
20:33:59 Join Wictor [0] (
20:34:08Strife89kugel: c200
20:34:09LambdaCalculus37kugel: Strife89 is a c200 man himself.
20:34:13linuxstbfophillips: Then option 2 is harder to pin down - a specific track that causes Rockbox to choke....
20:34:22kugeloh, nevermind then :P
20:34:34LambdaCalculus37kugel: beast backlight fading looks smoother and a little faster now.
20:34:44*funman is looking for c200v2 people
20:34:49fophillipslinuxstb: I will try nuking the DB and starting again. Since I have already unplugged it :)
20:34:58*LambdaCalculus37 is a c200v1 person, sorry
20:35:01linuxstbfophillips: I don't know much about the database though, so I'm not sure what ways exist to track it down, other than removing some files from your player and using trial and error...
20:35:12kugelLambdaCalculus37: it should be as fast as on the c200 with the default setting
20:35:50Strife89LambdaCalculus37: Should I use _this_ download, or can I just grab and apply the patch?
20:36:01kugelStrife89: download and install
20:36:26linuxstbfophillips: If you do find one (or more) tracks that cause Rockbox to choke, then it would be very helpful if you could share it, so we can try and work out why.
20:36:29funmanlinuxstb: I'm looking at the diff of revision 15339 but I'm not sure I understand what to do
20:36:36*Strife89 backs up his current installation.
20:36:42funmanfirmware/SOURCES contains paths to C files only
20:36:47funmanand firmware/FILES has disappeared
20:36:49LambdaCalculus37kugel: It's now as fast as the c200.
20:37:07linuxstbfunman: Yes, firmware/FILES isn't needed any more.
20:37:26funmanI think I must look in Makefiles for additional include paths
20:37:37linuxstbfunman: But there are lots of .S files referenced in firmware/SOURCES (in current SVN)
20:37:38 Join fragilematter [0] (n=fragilem@
20:37:42 Part J-23
20:38:20*Strife89 has put so many extras into his ./rockbox folder that it's 106 MB in size.
20:38:56linuxstbfunman: What have you set "t_cpu", "t_manufacturer" and "t_model" to?
20:39:00linuxstb(in configure)
20:39:04 Join jfb9301 [0] (
20:39:18funmanah right, I need to change the manufacturer from "sandisk" to "as3525"
20:39:42Strife89Build installed. Let me wait for this @%^#%* "Refresh Database".
20:40:02funmanand t_cpu was set to "am" :(
20:40:35linuxstbfunman: The e200 is probably a bad example to base your changes on - because it is quite old and may not do things in the currently preferred way. Something like the m200 may be better.
20:40:45bertrikStrife89, there's ways to get rid of the refresh database thing
20:41:15Strife89bertrik: Yeah, I saw the Wiki not too long ago.
20:41:37Strife89bertrik: But how do you set "read-only" in Ubuntu?
20:41:51Strife89Booting my c200 now....
20:43:05*Strife89 likes the fade effect.
20:43:27kugeleveryone likes it :D
20:43:48Strife89It's a little too quick at lower brightness settings IMO.
20:43:55Strife89But oh well.
20:44:04Strife89I usually keep it at 3.
20:44:20linuxstbStrife89: Using the "mattrib" utility (part of mtools)
20:44:52 Quit MethoS- (Remote closed the connection)
20:45:28Strife89I take it the patch was improved recently?
20:45:29 Join MethoS- [0] (
20:45:32kugelStrife89: at 1, there won't be fading
20:45:58fophillipslinuxstb: Well the DB initilisation has gotten further than last time
20:46:00kugelthat's because it's looping setting the de/incrementing the backlight by 1
20:46:10Strife89kugel: Ah, okay.
20:46:38kugeland thus, it's less noticable and faster on lower settings
20:47:24Strife89Fine by me.
20:47:42bertrikthis is done purely by setting the backlight brightness in the as3514 and not using any fancy PWM'ing, right?
20:48:39kugelbertrik: well, it's using whatever the target uses to alter backlight
20:49:37Strife89Quick question: What's the best thing to do here?
20:50:33kugelbertrik: on the beast, it's PWN fading actually
20:50:50kugelsince the brightness levels are implemented that way
20:51:32 Join faemir [0] (
20:51:37kugelStrife89: not entirely sure, that shouldn't happen
20:51:56kugelhave you already added the patch previously?
20:51:56Strife89kugel: I _have_ applied an older version of this patch.
20:52:07kugelwell, then remove that first :)
20:52:24Strife89kugel: I will - if I can find it.
20:52:43kugeljust do svn revert if that's the only patch
20:52:53Strife89it's not
20:53:02*kugel wonders why Strife89 has this patch applied previously if he's having a c200
20:54:31Strife89kugel: Because I mistakenly thought the patch worked for the c200.
20:54:45Strife89kugel: Obviously it didn't, and I forgot about it.
20:54:52kugelStrife89: it was actually working for c200 all the time. just the proper #define was missing
20:55:09kugelsince c200 and e200 share their backlight code
20:56:00Strife89Looking at my terminal again. Should I assume -R, say no, or abort with [CTRL]+[C]?
20:56:41Strife89Now, to find that patch.....
20:56:49kugeland skip the patch to config-c200 of you apply it again
20:56:55*funman notes how much he prefers git over svn
20:58:45bertrikwhen switching backlight timeout from off to on with my c240, the backlight comes on but the display is blank
20:59:16Strife89kugel: I might just use SVN revert and dig up patches again.
20:59:43kugelStrife89: I told you already to skip the patch to c200 if you reapply it
21:00:01kugelso type n and then y
21:00:16 Quit perrikwp (" ajax IRC Client")
21:00:57Strife89kugel: Okay, it looks successful.
21:01:27kugelyou should've typed n again :S
21:01:39kugelnow you have the HAVE_BACKLIGHT_THREAD_FADING twice
21:01:57kugeledit config-c200.h and delete one of them
21:03:14funmanwhich condition defines the building of sysfont.h ?
21:03:22 Join avacore [0] (
21:03:24kugelI thought you're "the c200 man"
21:03:25 Quit reacocard (".")
21:03:52Strife89kugel: Those were LambdaCalculus37's words, not mine. :)
21:04:04kugelthe terminal showed you where it is btw
21:04:14Strife89I already closed it.
21:04:52***Saving seen data "./dancer.seen"
21:05:53Strife89kugel: Edited. Preparing to compile.
21:07:28Strife89kugel: Making. I appreciate your help. :)
21:07:40kugelYou're welcome :)
21:09:34Strife89BTW, the other patches I have include the Mikmod patch, and the Sansa charging patch.
21:09:41 Join reacocard [0] (n=reacocar@WL-431.CINE.HMC.Edu)
21:10:31kugelthat's the wrong one .9
21:10:35kugel:) rather
21:10:47Strife89Oops, 8806
21:11:04Strife89I had typed in a different address on that bar.
21:11:11Strife89change 6800 to 8806
21:11:47Strife89Such is the peril that comes from having umpteen tabs open. :)
21:12:02kugeltabbed browsing is evil :)
21:12:13 Join dave__ [0] (
21:12:19Strife89kugel: A necessary one. ;)
21:12:32kugelas well as tabbed irc'ing, which caused me to talk about rockbox in -community
21:12:46kugelseveral times
21:13:17Strife89I'm generally good at keeping tabs (pun intended) on which channel I'm in.
21:13:32*gevaerts thinks that tabs are off-topic here ;)
21:13:58 Quit jfb9301 ("Java user signed off")
21:14:01Strife89%*&$%$( 800MHz CPU.......
21:14:32dave__hey, I submitted a theme to about a week ago
21:14:38Strife89Make is working on Doom at the moment... so slow.....
21:14:51dave__how long does it normally take for the theme to actually go up on the site?
21:15:35dave__well, uploading has been enabled, so there's obviously been some kind of progress
21:15:42kugeloh yea
21:15:45kugelI just noticed
21:16:05kugelbut there's definitely old themes there
21:16:27kugelthat's not the new theme site, is it?
21:16:42dave__is it older than the WPS gallery?
21:17:15 Join ]MrFWorK[ [0] (
21:17:30dave__I thought was like the 2.0 of the wiki's WPSgallery page
21:17:51dave__I like the layout better, at least
21:18:17]MrFWorK[question, how to i quit doom on the ipod 5.5g i cant get the menu up
21:18:47linuxstb]MrFWorK[: Turn on the hold switch, then turn it off again
21:19:04]MrFWorK[thank you
21:19:08 Quit MethoS- (Read error: 104 (Connection reset by peer))
21:19:52Strife89It's finally done. :)
21:20:32pixelma]MrFWorK[: doesn't the manual mention it? ;)
21:20:57dave__well if the maintainer for needs any help putting themes up on the site, I might be able to help
21:21:10 Join MethoS- [0] (
21:21:18linuxstbdave__: That site was never official, and as far as I know is abandoned.
21:21:59linuxstbdave__: Work has been started on a new, official themes site, but no-one is working on it at the moment... If you know php, you could help...
21:22:00dave__so the WPSgallery is the place to go for the latest and greatest then?
21:22:21Strife89Pretty much every theme on is broken now.
21:22:35dave__ha, funny you should mention PHP, I just started reading a tutorial last night
21:23:19 Quit nplus (Remote closed the connection)
21:23:53dave__linuxstb: is there a link to a work-in-progress type of thing, or is it not even up on the web yet
21:24:16 Join massiveH [0] (
21:25:21linuxstbdave__: I think there's a copy somewhere, but I forget the address. The actual PHP code for the new site is in the Rockbox SVN repository - you can browse that here -
21:31:31 Join Hillshum [0] (
21:31:37amiconnfunman: Did you find the SD driver in the Sansa _v1_ OF?
21:31:46 Quit MethoS- (Remote closed the connection)
21:32:00*LambdaCalculus37 casts a "Spell of Summon Swedes"
21:32:38*linuxstb thinks that works better when you mention their nicks...
21:32:38funmanamiconn: I found it in rockbox svn repository
21:32:51LambdaCalculus37Bagder: Ping
21:33:07 Quit ]MrFWorK[ ()
21:33:16amiconnfunman: That's not what I mean...
21:33:24 Join MethoS- [0] (
21:33:31dave__I certainly need to keep reading that tutorial on PHP, but the language looks pretty familiar (I've done stuff with C/C++ and Matlab)
21:33:50amiconnThe driver in rockbox has a bug (not related to the SD protocol but probably related to some timing setup)
21:34:27amiconnI think it would be helpful to know how the OF initialises the PP SD controller
21:34:32kugelwhich driver?
21:34:40funmanwhat do you expect from me?
21:34:46dave__I guess if I'm going to help, I need to learn how to use SVN as well
21:35:27amiconnfunman: I assumed you disassembled the v1 OF as well judging from an earlier comment...
21:35:42funmanno sorry
21:36:03Strife89dave__: Also, see this; it has relevant information:
21:37:24dave__Strife89: thanks. I've actually been compiling my own build using SVN for a while now (since I used to throw patches into my builds)
21:37:45dave__but I've never committed any code or anything like that
21:38:08Strife89dave__: Ah, alright. My bad on that, then. :)
21:38:49dave__yeah, I think there is some information in the wiki about committing stuff into SVN though
21:39:01dave__I'll look into it
21:44:03 Quit bmbl ("Woah!")
21:44:05 Join m0f0x [0] (
21:44:26funmanI must say this build system is quite confusing
21:45:48 Quit jhulst (Remote closed the connection)
21:46:03kugelonly if you're not used to it :)
21:46:12funmansysfont.h is built in firmware/Makefile
21:46:24funmanusing convbdf, which itself is built in tools/Makefile
21:46:44funmanI added convbdf in my 'toolset' but it's not called, and I can't see any condition in firmware/Makefile
21:47:12kugelconvbdf is built by the tools target
21:47:19kugelyou shouldn't worry about it
21:47:52kugelon what are you working right now anyway?
21:48:13funmanI try to create a target for the clip
21:48:50kugeljust have build tools before building the "binary"
21:49:11funmanthey are built
21:49:20funmanbut sysfont.h isn't
21:50:58kugelsysfont.h is created while building isn't it?
21:51:02funmanit is
21:51:11 Nick fyre^OS is now known as fyrestorm (
21:51:18funmansee firmware/Makefile:52
21:51:21kugelmaybe you're just looking in the wrong folder :)
21:51:56 Join nutrientman [0] (n=45b57a7f@gateway/web/cgi-irc/
21:52:59 Quit nutrientman (Client Quit)
21:53:37kugelsysfont.h should appear in your build directory
21:54:35 Quit dave__ ("Ex-Chat")
21:54:42funman% rm -rf *;echo clip\\nB|../rockbox/tools/configure >/dev/null && make |& tail -3|head -2
21:54:49funmanexport/font.h:33:21: error: sysfont.h: No such file or directory
21:54:49funmanmake[1]: *** [/media/bordel/bootloader/firmware/powermgmt.o] Error 1
21:54:59funmanit doesn't, else I wouldn't be looking for it :)
21:55:30 Join maddler [0] (
21:55:39funmanI built the e200 target, and indeed sysfont.h is built
21:56:22*Llorean does not really like the Patch of the Week idea.
21:56:46 Join nutrientman [0] (n=45b57a7f@gateway/web/cgi-irc/
21:57:05 Join Wictor_ [0] (
21:58:50 Quit nutrientman (Client Quit)
21:59:00 Quit LambdaCalculus37 (" ajax IRC Client")
22:00:31*gevaerts has mixed feelings about it. Focusing attention on patches is good, but so far I haven't seen any good idea on how to select which patches
22:00:34funmanand since some files are built in the builddir, some in the sourcedir, I don't know how to force building of this file
22:00:56Lloreangevaerts: I think it's going to be really, really discouraging for some people. It kinda says "We'll only be looking at one patch at a time"
22:01:00kugelAre you sure you are building firmware?
22:01:18 Join shotofadds [0] (n=rob@rockbox/developer/shotofadds)
22:01:27funmankugel: I'm building a bootloader
22:01:38kugelstill, you need firmware/ code for it
22:01:48Lloreangevaerts: And it encourages people *not* to come to us to discuss our patches the normal way. People are just going to nominate them for PoTW as if it's "the usual process to get them committed" rather than trying to discuss them at other times
22:02:25shotofaddsfunman: a missing sysfont.h usually means you've included a file in SOURCES that doesn't exist
22:02:27 Join pondlife [50] (n=Steve@rockbox/developer/pondlife)
22:02:30gevaertsThere is that too, yes. I hadn't thought of those things yet
22:02:32 Part pondlife
22:02:37shotofaddsit's not a very helpful error message ;-)
22:03:06funmanhm right, I didn't create the .c file in bootloader/ yet
22:03:30amiconnfunman: You don't need to force anything, but check the dependencies, and files in the various SOURCES list
22:03:31funmanbut there's no difference
22:03:51Lloreangevaerts: I think the -dev mailing list is a good place for people to say "I think my patch is ready, could someone look at it", and I think people could be doing that all the time (since it's non-realtime discussion, devs can respond/look whenever they have time, and easily flag messages to look at later)
22:04:12funmanI only changed bootloader/SOURCES
22:04:28funmanI might need to modify firmware/SOURCES then
22:05:01gevaertsLlorean: indeed. Except when people start spamming the -dev list about obviously incomplete and buggy patches
22:05:29funmanif I look at SANSA_M200 in firmware/SOURCES for example I only see a list of drivers, nothing about fonts
22:05:40Lloreangevaerts: Well, if _they_ (being, hopefully, the author or person willing to work on it) don't know it's incomplete or buggy, it gives us a chance to tell them.
22:05:44shotofaddsfunman: have a look at fs#7583 - it links to a Makefile change which should tell you which file is failing
22:06:06Lloreangevaerts: We can have a simple guideline, "don't post a patch to the list unless you're volunteering to address the issues that are brought up about it"
22:06:34gevaertsLlorean: good idea. That should keep the crippled-build-people out at least
22:08:08 Quit thegeek (Read error: 104 (Connection reset by peer))
22:08:16 Join perrikwp [0] (i=d1a8d351@gateway/web/ajax/
22:08:35Lloreangevaerts: Someone always has to volunteer to do the work anyway. We could do "Patch of the Week" in here, but unless someone's willing to address our list of issues afterwards, it doesn't really matter what we discuss.
22:09:12funmanin firmware/SOURCES, if CONFIG_CPU doesn't match a specified list, the firmware will be linked against target/arm/crt0.S which is non existent
22:09:30*linuxstb wonders if the bmp resize patch should be closed, as no-one is taking it in the right direction, so it will never (as it is now) be committed -
22:09:52LloreanOne option would be "Patches can only be nominated by someone agreeing in advance to fix what we list" but if we do that, I suspect we'll very quickly run out of patches. If we don't do that, I suspect we'll just slowly work our way through less and less interesting/valuable patches until we all agree it's just not that worth our time to get together weekly to discuss patches fixing typoes in comments.
22:10:25LloreanSo I think, instead of doing it weekly, just interested coders saying 'I think this patch is ready, and I'll work on it if it's not' to the mailing list is the best idea.
22:10:32LloreanAn ongoing process, rather than us cherrypicking.
22:11:14kugellinuxstb: well, it's a bit stuck, since your demands aren't easy to implement
22:11:30linuxstbkugel: They're not "my demands"...
22:11:55kugelnot your as in your personal
22:12:30 Quit reacocard (Read error: 104 (Connection reset by peer))
22:12:44kugelpretty much the same goes for multifont
22:12:46*gevaerts would reply "+1" to Llorean's mail if he were the sort of person who replies "+1" to mails
22:13:21Lloreanlinuxstb: Maybe we need a "good feature idea, implementation not appropriate for Rockbox" reason for closing.
22:14:30kugelthat sounds more of a reason for letting it rot, rather than closing
22:14:37 Quit Wictor (Read error: 110 (Connection timed out))
22:14:47Lloreankugel: Why should we let it rot if it's not going to be fixed?
22:14:49Bagderno leaving it to rot is bad
22:15:09gevaertsLlorean: that sort of reply needs to be well-motivated I think, so I'm not sure if a canned reason is appropriate
22:15:10Strife89kugel: Figured I'd let you know: the compile was successful, and the build works fine. (I let my Sansa charge, that's why it took so long.)
22:15:12LloreanIf we close it, saying "we like the idea, but not the code" it makes it clearer that maybe it should start over from first assumptions.
22:15:39kugelHow can you know it's not going to be fixed? If the idea of the tracker entry is generally accepted, just the current implementation not, I see no reason to close it
22:15:58 Quit fragilematter ("query if you need me")
22:16:02*shotofadds wonders if the blank Flyspray pages will be fixed one day
22:16:13Strife89Thanks again. :) I need to go.
22:16:30kugelStrife89: see you
22:16:31 Quit Strife89 ("Leaving")
22:16:33linuxstbkugel: If someone fixes, they can open a new task with the fixed version of the patch.
22:17:15gevaertskugel: that basically would be equivalent to reopening the feature request tracker
22:17:16shotofaddslinuxstb: but what's the point of closing? isn't it better to have the whole history and reasoning in one task?
22:17:32Bagderthere's a good idea to keep number of open patches low
22:17:35linuxstbBagder: LambdaCalculus37 was asking for Developer rights on the tracker earlier...
22:17:36Bagderto keep focus
22:18:05linuxstbshotofadds: Have you seen the task in question? About 240 comments, with no real discussion of the issues, and a patch that is implementing the feature wrongly.
22:19:30 Quit m0f0x ()
22:19:45shotofaddshmm.. you have a point ;-) (assuming you means the bmp resize patch)
22:21:22*gevaerts doesn't see how this POTW patch selection would work
22:21:55funmanrandom(3) ?
22:22:02Bagderthe point for me is that the selection process wouldn't have to be complicated
22:22:10Bagdergevaerts: you could pick them for all I care
22:22:25Bagderit would help me get interesting patches to look at
22:22:32Bagderinstead of wading through 400
22:22:37LloreanWe all know about the interesting patches.
22:22:42*shotofadds mumbles "user vote" and runs...
22:22:43*linuxstb still wonders when the RSB is going to meet in order to arrange their first meeting
22:22:55LloreanThe users use them in their unsupported builds.
22:23:11BagderLlorean: then those should be fine first contenders
22:23:28LloreanBagder: We've already discussed most of them, and listed things that get in the way on their entries, for a lot of them.
22:23:37BagderI'm seeing how this could help me, that's all
22:23:55 Quit jeffdameth (Read error: 113 (No route to host))
22:24:01LloreanThe real problem, for nearly all the "interesting" patches, is that there's still work to be done and it's not always happening.
22:24:41gevaertsPart of the problem may be that patch authors feel that after they've fixed the current complaints we will come up with new ones
22:25:17LloreanWhich, we probably will.
22:25:21 Join jeffdameth [0] (
22:25:21*gevaerts points out that Software USB is not a tracker patch ;)
22:25:26kugellinuxstb: If I understood correctly, bmpresize isn't doing it entirely wrong. As far as I noticed there're 3 demands, a) fast b) smooth and c) on-load resizing. the latest implementation meets two of them. And on-load resizing sounds rather complicating for me, especially when smooth is also a demand
22:25:33Lloreangevaerts: There's an entry for it on the tracker, no?
22:25:34funmanyou could propose help on future problems to motivate the authors
22:25:39Bagderwell, authors may believe that no matter what process we use
22:26:01LloreanWe can't help but have future demands
22:26:11LloreanUntil they write new code for the previous set of demands, we can't know what it's going to do now.
22:26:43 Join thegeek [0] (
22:26:50LloreanBut I really think a process by which an interested author comes forward and says "I'd like to get this patch in, let's initiate a discussion about it" will help, and the mailing list is good for that.
22:26:57amiconnlinuxstb: There already are APE performance figures at which I can confirm
22:27:12BagderLlorean: yes, but that could happen already but never does
22:27:19LloreanBagder: We don't ask people to do that.
22:27:20 Join jhulst [0] (n=jhulst@unaffiliated/jhulst)
22:27:28amiconnSo the beast copes fairly easily up to -c3000, and would handle -c4000 at max. clock with svn code
22:27:37LloreanWe don't have, written somewhere, "once you think it's committable, and you're willing to fix any issues, post it to the mailing list" as part of our formal process.
22:27:38linuxstbamiconn: Seems I did those ;)
22:27:39gevaertsThere are indeed always new demands, but at some point it's good enough to go in. Defining that point clearly in a discussion, and then committing if that point is reached, may well motivate people to actually work to reach that point
22:27:39kugelIt's especially confusing to me (and possibly other interested people as well), that the patch uses an algorithm which is in SVN, but didn't get itself into it, because it's using this algortihm
22:27:43Bagderpeople these days are too scared of mailing lists...
22:27:44LloreanI think just doing that could improve things vastly.
22:28:26*amiconn much prefers the ML over forums
22:28:28Lloreankugel: Isn't it used only for plugins?
22:28:38kugelYes, but still
22:28:42Llorean"But still" what?
22:28:47Bagderkugel: that's the explanation
22:28:47kugelit's in svn
22:28:59Bagderplugins are our wild west
22:29:02Lloreankugel: There's code in SVN that's not even used in Rockbox (see RBUtil)
22:29:04Bagderyou can do anything there ;-)
22:29:10LloreanPlugins should be considered "not Rockbox, but a separate application"
22:29:12kugelclosing a task for using a feature that is in svn, that's irritating for me
22:29:13linuxstbkugel: Also, not everything in SVN is perfect, sometimes things sneak under the radar...
22:29:56linuxstbkugel: So you're saying the patch should be committed as it is?
22:30:08gevaertskugel: there is an mp3 decoder in svn. That doesn't make it a good idea to use mp3 decoders for caching fonts
22:30:09kugelI say it shouldn't be closed
22:30:16markunwhich patch are we talking about?
22:30:16*gevaerts likes extreme examples
22:30:25kugelit's just one step to "committable"
22:30:31amiconnPlugins do a lot of things which aren't allowed in the core
22:30:34 Nick fxb is now known as fxb__ (
22:30:50Lloreankugel: If it's "one step to" then obviously it isn't by your own words, committable.
22:30:56linuxstbmarkun: bmp resize -
22:31:34Lloreankugel: If there's nobody interested in working on it, we can close it with a _clear_ explanation until someone comes along and says "I want to fix it" at which point they can open a new task or we can reopen it.
22:31:41Lloreankugel: In such a situation, what harm does closing it do?
22:32:15linuxstbkugel: The _feature_ isn't being rejected, simply that specific implementation. Other implementations can live in new tasks.
22:32:56linuxstbkugel: Closing the task may even motivate someone to do it right....
22:33:02Bagderclosing this situation is _good_
22:33:04Bagderin this
22:33:14*linuxstb wonders who wants to be the bad guy
22:33:36*Llorean is always willing to be the bad guy if nobody else wants to.
22:33:52kugelwell, you close the entire task, not just the patches. That looks like "we're rejecting the whole task, so we reject the whole feature". The task doesn't only consist of the code produced, but also for the general idea.
22:33:53LloreanPeople seem to enjoy sending me angry emails for closing tasks.
22:34:05gevaertskugel: we closed the feature request tracker...
22:34:10Lloreankugel: That's why there's a "reason for closing" in which you explain this.
22:34:15*amiconn wonders whether the armv6 simd instructions are sufficiently magic...
22:34:15markunwhat about closing the task and opening a feature request?
22:34:26gevaertsAnd a task without an acceptable implementation is a feature request
22:34:35gevaertsmarkun: where?
22:34:40LloreanReason for Closing: Good feature, but this implementation isn't progressing in the direction necessary for including. Can be reopened or a new task started if work is to resume on an implementation fit for use in the core.
22:34:51markungevaerts: ah yes, we don't do that in the tracker anymore..
22:35:20kugelthen lets just hope someone in the future will actually search closed tasks, browse through loads of pages and happens to find the old bmp resize task, and then also realize that the feature is still wanted even though it was closed
22:35:46Lloreankugel: If they don't search in the first place, then they'll open a new task anyway.
22:36:00gevaertskugel: if it's really wanted, somebody will notice within three days
22:36:03Lloreankugel: If they DO search, they'll probably wonder why we rejected it. If they fall in the small group that doesn't, there's not much we can do.
22:36:08 Quit kronflux ("Leaving")
22:36:46*gevaerts is going to close the USB task. It's becoming more useless to keep open day by day
22:37:10*linuxstb will close the bmp resize task
22:37:34kugelthen close multifont too, the situation is similar
22:37:36 Join ompaul [0] (n=ompaul@gnewsense/friend/ompaul)
22:37:41kugeleven worse imo
22:38:52pixelmaI wonder how long this open a new task will take to happen, maybe even with the old code
22:39:08kugelI wonder how smooth resizing should work with on-load resizing. Don't you need to analyze the source to make it smooth?
22:39:29markunkugel: no, why would you?
22:39:32gevaertsDo we still need FS #9376 after the release?
22:40:39Bagderwe might need one for 3.1 as well, but that'll be a new one by then
22:41:08*gevaerts votes to commit FS #8943
22:41:19markunkugel: I think you only need to know the input and output width and height and read enought pixels to calculate the output pixels.
22:42:05markunkugel: I'll think about it a bit, maybe I can come up with a nice algorithm
22:42:08Bagdergevaerts: seems fine
22:42:11kugelwell, you maybe don't need to analyze the full picture, but at least several rows
22:42:21Lloreangevaerts: I don't see any successful tests beyond "hosing the filesystem"?
22:42:27bertrikI think two output rows should be enough in theory
22:43:02gevaertsLlorean: actually the "hosing the filesystem" issue was my fault, and due to a buggy workaround for FS #8663
22:43:18Lloreangevaerts: Ah, just as long as you're confident then.
22:43:19markunbertrik: I think it should be possible without that even
22:43:29Lloreanmarkun: Just adjacent pixels?
22:44:20markunis memory usage much more important than speed? I think so, since it's only done on load, right?
22:44:23 Quit bughunter2 ("bye")
22:45:12Lloreanmarkun: I'd have to agree, yes.
22:45:29Lloreanmarkun: It's actually more going to happen once per buffering.
22:45:31bertrikmarkun, I think speed is mostly irrelevant
22:45:41LloreanAnd buffering takes long enough that even a few _seconds_ extra isn't going to be noticed by the user in the background.
22:45:52ameyerbertrik: within reason...
22:45:57Lloreanameyer: In this case.
22:45:59LloreanNot in general.
22:46:02markunkugel: I think I'll give it a go
22:46:25kugelmarkun: Nice. And I'll glady help you
22:46:35ameyerwhat exactly is "this case"
22:46:41Lloreanameyer: Bitmap Resize patch.
22:47:09kugelmarkun: fast, post on the tracker, that you're going to do something, before it's closed :)
22:47:20markunnah :)
22:48:10Lloreankugel: He also knows he can start a new task, or re-open the old one, if he accomplishes something.
22:48:32gevaertskugel: clearly he wants a new task all for himself :)
22:48:51kugelmy idea is, to read XxY tiles of the original (into a small buffer), resize this tile, and put it into the output, then take the next tile. I don't think the result would look good though
22:50:21kugelIt would be rather artifactial I assume
22:50:23markundoes the existing code use linear interpolation?
22:50:41kugelI think so, yes
22:50:44*gevaerts checks if he needs to add the author of the FS #8943 patch to CREDITS
22:50:52kugelbut don't count on what I think
22:51:48LloreanBagder: Maybe we could add a "Contribute" link above the patches link, and it could lead to a page that not only describes the contribution process for code, but has links to SimpleGuideToCompiling and other "useful" pages, and talks a little bit about the usual requirements we have for patches to meet.
22:52:23Bagderwe could start with creating such a page
22:53:17LloreanBagder: I'm going to post a mail to -dev to ask for feedback on what people think such a page should contain/say, then maybe write a text draft?
22:53:36Bagdera good plan!
22:54:17markunkugel: you could also try to get jhMikeS interested, but he's probably occupied with other things these days
22:54:35kugelWhy don't you try?
22:54:41 Join reacocard [0] (n=reacocar@WL-431.CINE.HMC.Edu)
22:54:41funmanis there any point using tools/scramble for the SansaV2 ?
22:54:49 Join rsfgt [0] (
22:56:32 Quit Hillshum ("ChatZilla 0.9.83 [Firefox 3.0.3/2008092417]")
22:56:33kugellinuxstb: so, you'll also close multifont now?
22:56:45kugelwould be rather inconsistent if not
22:57:21 Join kronflux [0] (
22:57:36 Quit kronflux (Client Quit)
22:57:45 Quit Wictor_ (Read error: 110 (Connection timed out))
22:58:31markunkugel: who cares if we're inconsistent, we do all of this just to piss you off :)
22:58:33 Quit ompaul (Client Quit)
22:58:47bertrikan idea I've been thinking about is to simply step left-right top-bottom through the input pixels, and accumulate them in a pixel buffer the size of one output row. Possibly a bresenham-like way of stepping would easily tell us when a horizontal pixel boundary is crossed. On a pixel boundary crossing the input pixel is proportionally divided between the two output pixels. Vertically you could do a very similar thing and you wo
22:58:47bertrikuld need to buffer only two output lines, the current line and the next line.
22:58:57kugelmarkun: But I'm not pissed off enough
22:59:05 Quit massiveH ("Leaving")
22:59:07linuxstbkugel: I'm not the only person who can close tasks... bmp resize was just one that I've been following and posting comments to recently. But yes, multi-font probably could get closed with the same reason.
22:59:28Bagderkugel: I doubt the authors of that multifont patch care if we are consistent about that
22:59:51 Join webguest31 [0] (n=57bd56da@gateway/web/cgi-irc/
23:00:06Bagderbut I am in the camp in favour of closing more!
23:00:46funmanit's a new game category: "close'em all"
23:00:59 Quit Thundercloud (Remote closed the connection)
23:01:30kugelpoketask, gotta close'em all
23:02:29*linuxstb thinks Bagder should just run some more of his shell/perl scripts on the flyspray server
23:02:52Bagder"0 open patches"
23:02:52 Join Thundercloud [0] (
23:03:03gevaertslinuxstb: we want to close the tasks, not delete them :)
23:04:16kugelfunman: What will the bootloader do you're working on right now?
23:04:39*domonoky begins to write a flyspray bot which automatically closes all tasks.. for ever :-)
23:04:42funmanwell it will build
23:04:49 Quit webguest31 ("CGI:IRC (Ping timeout)")
23:04:54***Saving seen data "./dancer.seen"
23:05:08markunfunman: I guess you'll use it to run some test code?
23:05:45kugelWhy do you need a lds file?
23:06:09funmanto link it, specifying where exactly are the stack, code, and stuff
23:06:38 Quit bertrik ("Leaving")
23:07:25kugeland the built bootloader would then be passed mksmsboot to patch the firmware?
23:11:01 Join midgey [0] (
23:11:54kugeltime to rename mkamsboot into sansaV2pather, isn't it :)
23:12:01kugelpatcher, rather
23:12:20funmanyes, next step :)
23:13:18 Quit petur ("*plop*")
23:16:55 Join ZincAlloy [0] (n=d9eed7e9@gateway/web/cgi-irc/
23:17:16 Join {-phoenix-} [0] (
23:21:56kugelfunman: I wonder if we would be able to not use an external firmware file, but the one on the flash, for patching the bootloader
23:22:06 Quit midgey ()
23:22:10funmanhm ?
23:22:12 Join webguest38 [0] (n=55b218db@gateway/web/cgi-irc/
23:24:35kugelnow we patch a firmware file, put it on the player and rely on the OF's updating mechanism
23:25:04kugelI think it's desirable to modify the firmware which is on the flash
23:25:34funmanhow could we do that ? and what's wrong with the OF firmware upgrade mechanism ?
23:26:13 Join Munkie [0] (
23:26:15 Quit tvelocity (Remote closed the connection)
23:27:09 Join midgey [0] (
23:28:12 Quit rsfgt ()
23:28:45funmanbootloader built !
23:29:32kugelfunman: great! Do we need a e200v2 dude to test or do trust your code?
23:29:58kugelwell, should actually matter if the bootloader is bugged
23:30:22 Join denes [0] (
23:30:24kugeldoes it include the LED blinking?
23:30:26funmankugel: nothing to test
23:30:31funmanit doesn't do anything at all
23:30:48kugelthen let it do something :)
23:30:54funmanI didn't modify the mkamsboot, so I assume it still works :)
23:31:15denesmarkun gevaerts : so I will try to send in the m3 lcd code into flyspray during the weekend
23:31:17 Quit {phoenix} (Connection timed out)
23:31:48denesgevaerts: just fyi, I checked the qt1106 code on the m3, and it works on default clock
23:31:54 Quit webguest38 ("CGI:IRC (EOF)")
23:31:56markundenes: ah, great
23:32:09denesgevaerts: but on 200 MHz it is kind of buggy for some reason
23:32:14funmankugel: there is some (a lot of) work to do before that
23:32:28markundenes: even with the delay loops adjusted?
23:32:33denesmarkun: yes
23:32:43denesmarkun: the delay function was adjusted
23:33:02 Join AaronShort [0] (n=c6ec2b02@gateway/web/cgi-irc/
23:33:04funmanwho is Paul Louden ?
23:33:15denesmarkun: I looked, but the reason for the bug didn't jump at me
23:33:25funman/whois let me think he was called "purple"
23:33:36*scorche|sh references the IrcNicks wiki page to funman
23:33:37markundenes: well, gevaerts will soon have his M3 to play around, maybe he finds something
23:33:48denesmarkun: great
23:34:18markundenes: there was another M3 owner here who wanted to help out with the port, but I forgot his nick
23:34:37kugelfunman: what work exactly? Don't you have such a thing as bootloader.c, in which you can insert your led blink code?
23:34:47AaronShortHowdy. I'm new to rockbox and so far loving it. I'm a software developer. though i havn't done much c since college. I'd like to correct a few things on the wiki I found can someone please give me write access?
23:35:06markundenes: could you put your LCD code in the tracker now and then just update it later? Then maybe other people can start playing with it
23:35:10denesmarkun: meven , iirc
23:35:21kugelAaronShort: What's your wiki account?
23:35:24markunyes, it was him
23:35:24 Join jhulst_ [0] (n=jhulst@unaffiliated/jhulst)
23:35:40 Quit jhulst (Read error: 113 (No route to host))
23:35:46denesmarkun: it's on my other computer
23:35:52markunah, ok
23:36:19kugelAaronShort: Done, don't spam!
23:36:31markun... like kugel always does
23:36:35 Join webguest23 [0] (n=57bd56da@gateway/web/cgi-irc/
23:36:39funmankugel: yes but I need to add the hardware registers into defines, and fix/implement some (a lot of) functions I just commented out to let the bootloader build
23:38:19AaronShortI wont. I may even think of a minor patch or two. I'm at work and prob shouldn't stay on here so thanks and see you around.
23:38:25Munkieis there any tutorial or can any1 help me with programming plugins for rockbox?
23:38:33 Join saratoga [0] (n=9803c6dd@gateway/web/cgi-irc/
23:38:46AaronShorttheres some documentation for plugin writing on the wiki
23:39:12scorche|shMunkie: there are wiki pages such as HowToWritePlugins, but please dont use such "words" as "any1"...we require real english in here
23:39:25 Quit AaronShort ("CGI:IRC")
23:41:04markunMunkie: it's also a good idea to look at the code of existing plugins
23:41:04Munkiesorry man. i understand that was a silly question but all i kept finding was "rockbox programming", not plugin programming
23:41:24markunMunkie: any plugin in particular you want to work on?
23:42:26saratogaregarding efficient resizing mentioned above, for linear interpolation, the best way to do it is to:
23:42:35MunkieI want to try create my own. the problem is i only know java, so i have to teach myself C
23:42:38saratoga1) calculate how many source image rows are actually needed
23:42:48 Quit {-phoenix-} (Remote closed the connection)
23:43:05saratoga2) step through the source image 1 row at a time skipping unneeded rows performing bilinear interpolation left to right
23:43:22 Quit UncleRemus ("leaving")
23:43:35 Quit webguest23 ("CGI:IRC (Ping timeout)")
23:44:14saratogafor example, to rescale an 800 by 600 image to 200 by 150, only every other row should be actually needed for bilinear
23:44:40saratogaso you would read in the first and and third row, and use them to compute the first output row of pixels
23:44:43 Join Dan89 [0] (n=598b3ce8@gateway/web/cgi-irc/
23:44:52mevenhy denes
23:45:33Dan89hi there, I have a question about my sansa e250 can you help me?
23:45:47saratogadoing this efficiently is actually not very hard, for a bilinear system the optimal algorithm is both simple and relatively computationally efficient
23:46:11 Quit maddler ("Lost terminal")
23:46:41LloreanDan89: If someone can help you, they'll answer.
23:47:00LloreanOr did you want all 150 other people here to say "no, I don't have any time right now" etc?
23:47:13 Join mmadia [0] (
23:47:30mevendenes: markun me i own a m6sl
23:47:41Dan89i want to transfer some movie to my sansa buy i cant see it in the player
23:47:42Dan89what can i do?
23:48:09kugelsaratoga: Did you already upload a patch? :)
23:48:23saratogano i'm not very interested in it
23:48:40LloreanDan89: Rockbox uses the original firmware's USB right now on that player, so if you can't see it you're having a problem between your computer and the original firmware.
23:48:46denesmeven: hi
23:50:01Dan89no, I meant that i cant see it inside the "file" folder when i open my player.
23:50:06Dan89did you understand?
23:50:14markunmeven: ah, my bad
23:50:20markunhi btw :)
23:50:44LloreanDan89: What's your "show files" setting set to?
23:51:04Dan89what do you mean?
23:51:22Dan89how can i set it?
23:51:45LloreanDan89: It's a setting in Rockbox, described by the manual.
23:53:13Dan89yes but how can i set the option for a folder?
23:53:37 Join webguest23 [0] (n=57bd56da@gateway/web/cgi-irc/
23:54:05LloreanDan89: You set it in general. It's in the manual.
23:54:26Dan89file wiew?
23:55:39LloreanWhere did you put the file?
23:55:46LloreanAnd did you transfer it while in MSC mode?
23:56:01Dan89now i can see it
23:56:22Dan89but how can i play the movie?
23:56:29LloreanClick on it.
23:56:32LloreanJust like songs.
23:56:49Dan89but it dosent work
23:56:50 Quit reacocard (".")
23:57:02LloreanDan89: Did you follow the instructions for converting it?
23:58:01Dan89there is a diffrent manual for this?
23:58:26 Quit herrwaldo ("Konversation terminated!")
23:58:32saratogain the closed resizing patch, where did resize_bitmap get called from?
23:58:40saratogaDan89: no should be in the manual
23:58:44saratogatry searching
23:59:01Dan89so how should i convert it?

Previous day | Next day