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

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

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

#rockbox log for 2009-05-21

00:03:33mcuelenaerekugel: you know of the red?
00:05:14 Quit jgarvey ("Leaving")
00:08:14 Join renke [0] (
00:10:02Strife89AlexP, pixelma, LambdaCalculus37: Okay, I'm back... Can I get a recap? What needs to be done with the patch now?
00:10:44 Quit domonoky (Read error: 104 (Connection reset by peer))
00:10:46pixelmacan someone tell after a quick look at rockpaint.c what the "Toolbar2" is good for? After trying on my c200 (unaltered source), it looks like it jumps to a certain position in the toolbar when that is displayed. The chosen combo together with the simple "Toolbar" gives me a hard time in timing the combo corectly so I'm not sure
00:11:29AlexPStrife89: From what I said, you need to change all the "color" to "colour" not just one or two, replace NW with something clearer, have the second .bmp in a \fname as well, and replace the \fname{.bmp}s which I think will look weird with something like "bitmaps"
00:12:00AlexPStrife89: I did all of this and uploaded a version 11a, but of course feel free to do yourself too if easier
00:12:09pixelmaStrife89: I'm looking at the button table currently
00:12:15Strife89AlexP: Thank you. :)
00:14:03pixelmaand start thinking that the keymaps need some improvement :\
00:14:28 Quit ricky26 (Remote closed the connection)
00:16:20Strife89pixelma, AlexP: So for now, is that everything?
00:17:06Strife89I never would've thought it would take so long to work on one subsection of the manual.....
00:17:15AlexPStrife89: The M5 and X5 both use IAUDIO_X5_PAD
00:17:32pixelmawanted to help out with that
00:17:57AlexPStrife89: Yeah, it takes a while to get it all wrapped up :)
00:18:34Strife89AlexP: Okay, I've made that change.
00:18:38pixelmathe combos are all hard timing wise, no precondition checks etc. :/
00:19:09Strife89pixelma: That's part of why I omitted the "Toolbar2" from the keymaps.
00:19:27mcuelenaeretomers (logs): the repeating USB HID commands are definitely target-related, so no need for you to check the code as you won't find anything useful without the actual target :)
00:20:06Strife89AlexP, pixelma: Wanna commit this then?
00:20:24pixelmaStrife89: do you know what it is supposed to do? Looks to me that it jumps to a certain position in the toolbar (the menu) but am not sure
00:20:27 Join _fml [0] (n=4fd3ca87@gateway/web/cgi-irc/
00:20:42AlexPStrife89: I think pixelma has some outstanding issues with keymaps to deal with
00:20:48Strife89pixelma: I've never managed it.
00:21:00pixelmae.g. unknown keypads
00:21:53Strife89pixelma: So what needs doing, exactly?
00:22:04froggymanis Interix generally faster than Cygwin?
00:22:42AlexPfroggyman: I think the only person that has tried Interix is amiconn - and he had to go to some effort to get it all running with cross-compilers etc.
00:23:12AlexPfroggyman: If you want a faster development environment, then a VM (or native linux of course) is much quicker than cygwin
00:23:17pixelmaStrife89: I'm doing the changes myself currently, and then pastebin the button table part or so
00:23:39Strife89pixelma: Okay; thanks for that. :)
00:23:59Strife89I'll paste it and make the final patch.
00:23:59_fmlHello. I'd like to close FS #7816 (reject) since it's old and not needed anymore and to commit FS #10073, even if it might contain some wrong mappings. But I think if it's in the manual then the errors will be noticed sooner. And it will be easier to correct them than to maintain a huge patch. Any objections?
00:24:43_fmlkugel: have you seen my question in the beginning of the yesterday's log?
00:24:51froggymanAlexP: ok thanks, I thought that it might have been considered as fast as native Linux, but it doesnt seem like it; i'll just stick with Linux for compiling
00:25:27pixelmathe Toolbar2 really calls the Toolbar with the cursor at a certain position. The x value for the cursor there seems to be hard coded - it's in the middle of the H300 display (so I guess it's 110 which is why it's to the right on the c200's display
00:26:22AlexP_fml: I have some time tomorrow, so I'll go through 10073 and check it
00:26:59 Quit stripwax (Read error: 104 (Connection reset by peer))
00:27:00pixelmaStrife89: your leaving out of that one seems to be the easiest solution for now, it's not really important anyway
00:27:25AlexP_fml: Also (out of interest), why is 7816 not needed any more?
00:28:00 Join {-phoenix-} [0] (
00:28:03 Quit robin0800 (Read error: 110 (Connection timed out))
00:28:08Strife89pixelma: Okay. Just pastebin whatever you changed and I'll make one last patch. :)
00:28:12_fmlAlexP: I thought it describes menues and they are already in the manual. Or am I mistaken?
00:28:40AlexP_fml: It is a plan IIUC - so it shows the full structure
00:29:17AlexPExcept his screen shots seem not the be there anymore
00:29:26 Quit petur ("Zzzz")
00:29:27_fmlAlexP: the problem with FS #10073 is that it describes keymappings for a device I don't own. So it's hard for me to verify. But once it's in the 'official' manual users will notice the errors and report them.
00:29:44AlexP_fml: I do own an S, so I'll check it tomorrow
00:30:27AlexPSeeing as I have the time, it might as well go in correct in the first place, and it'll be easier for me to check before it goes in :)
00:32:08_fmlAlexP: ah, re FS #7816: I misunderstood its purpose. I won't touch it then. Thanks for clarifying!
00:32:11AlexP_fml: If I understand it, FS #7816 would fill in and give a tree view of the menu structure to make it easier to find an option
00:32:12 Quit {-phoenix-} (Remote closed the connection)
00:32:40AlexP_fml: I think it is that - I haven't tried the patch :)
00:34:25 Join firedix [0] (n=firedix@
00:34:26_fmlAlexP: ok. But I think that all options are already listed in the table of contents. So to find an option you just have to do a search in the PDF. A tree of options would take several pages.
00:35:08_fmlBt I won't touch the task since it does something different than what I thought.
00:35:13AlexP_fml: Yes - I'm unsure myself. I like the idea, but am not sure how it would fit - I think in an appendix it would be quite good
00:35:23AlexP_fml: I'll assign myself FS #10073 to remind me to do it tomorrow :)
00:35:42_fmlAlexP: ok. Can you commit it then?
00:36:02 Join Horschti [0] (n=Horscht@xbmc/user/horscht)
00:36:16AlexPChecking all those plugins will nicely waste some time :)
00:36:24_fmlAlexP: will a tree of options be helpful? I'm not sure (my personal view)
00:36:34AlexP_fml: I'm not sure either :)
00:36:45 Quit renke ("leaving")
00:37:00_fmlAlexP: that's why I thought the checking is not feasible for a single person and should be distributed to all users :-)
00:37:36AlexP_fml: hehe :) I have the day off tomorrow though, so it'll be something to do. If I get overwhelmed we can fall back to that :)
00:37:40pixelmaStrife89: currently doing some test compiles. Would it be ok to put the quit button thing at the top (because the exclusion of the complete line doesn't work for some stupid reason when it's at the end - it's breaking the build there completely - and it doesn't really fit somewhere else)
00:38:14Strife89pixelma: Fine by me. :)
00:38:21 Join BHSPitMonkey [0] (n=stephen@unaffiliated/bhspitmonkey)
00:38:42_fmlAlexP: ok. Wouldn't it be easier to check the patch against the source code? I.e. not on the device.
00:39:00AlexP_fml: Either or
00:39:26AlexPBut I can do both if needs be, so not a problem
00:40:09_fmlAlexP: ok. Then I forget it! :-)
00:40:20 Quit {phoenix} (Read error: 110 (Connection timed out))
00:40:52 Join Dillizar [0] (n=Elive_us@
00:40:52_fmlI just looked what open manual patches we have. One is for viewport WPS tag. But it has a strange TeX bug that I can't fix.
00:41:11Dillizartoffe82, are ya here
00:41:41pixelma_fml: which one is this?
00:42:23_fmlpixelma: FS #9934
00:42:31mtlinuxstb : This line (from flac.c) : "bit_buffer = ci->request_buffer(&buff_size, MAX_FRAMESIZE+16);" , to imitate that in rm2wav.c : memcpy(bit_buffer, file_buffer, MAX_FRAMESIZE+16) ?
00:43:46pixelma_fml: and what happens?
00:43:52linuxstbmt: If I was you I would read the entire file into a buffer, and then implement your own "request_buffer" and "advance_buffer" functions,
00:45:19mtlinuxstb : That's what I'm doing. buf = request_buffer(.., reqsize) mallocs reqsize of data to buf, right ?
00:45:59linuxstbNo, it doesn't malloc. It gives a pointer to data that is already in the buffer.
00:46:44_fmlpixelma: I don't remember exactly but it's described in the task. IIRC the problem is that there are two nested \opt's or \nopt's which are in a table. Or something like that.
00:47:26pixelmaI see it now, there's a fixme in the patch itself
00:49:24pixelmathe whole patch looks a bit confusing to me at the moment though
00:50:00Strife89pixelma: Sorry to nag, but how's it coming? :)(
00:50:24_fmlpixelma: what's wrong about it?
00:50:28mtlinuxstb : OK. But, I'm a bit confused by this line in codecs.h " ...<reqsize> tells the buffer system how much data it should try to _allocate_. " That's why I assumed there were some sort of memory allocation.
00:51:19pixelmaStrife89: good, just doing one more test compile (done: X5, Ipod Video, small H10)
00:51:30linuxstbmt: Yes, that comment is misleading.
00:51:50pixelma_fml: can't figure out currently which section goes where. Maybe it's a bit late...
00:52:17_fmlpixelma: yes, it definitely is. I'll leave now. Bye!
00:52:20 Quit merbanan (Read error: 110 (Connection timed out))
00:52:31 Quit _fml ("CGI:IRC 0.5.9 (2006/06/06)")
00:52:33 Quit matsl (Read error: 110 (Connection timed out))
00:53:25 Quit Horscht (Read error: 110 (Connection timed out))
00:55:22toffe82Dillizar: yes but busy...
00:55:43Dillizarok toffe82 will ya be free in 1h or so??
00:55:59mtlinuxstb : er- Sorry, I'm still a bit confused. :) - Why is reqsize necessary then ? Why not just deal with the pointer to the buffer directly, keep advancing the pointer and checking not to go past EOF or realsize ?
00:56:20toffe82Dillizar: not before 3 or 4 hours... sorry
00:56:41Dillizarwhats the time there btw cuz here is 1am
00:56:42linuxstbmt: Because the buffer is circular - you need to deal with where it wraps.
00:56:50Dillizartoffe82, ?
00:57:04linuxstbmt: And the entire file may not be buffered.
00:57:22toffe82Dillizar: 4pm
00:59:16 Quit Thundercloud (Remote closed the connection)
00:59:42mtlinuxstb : So shouldn't a check be done after every request_buffer to see whether realsize >= reqsize (still didn't reach end of stream)?
00:59:47toffe82Dillizar: try tomorrow morning, I should be more available
01:00:19Dillizarmy tomorrow morning toffe82? :)
01:00:33linuxstbmt: Yes, you need to check for end of stream.
01:00:44toffe82so afterenoon for you
01:02:04 Quit BHSPitMonkey (Remote closed the connection)
01:04:17mtlinuxstb : Where is this check done in flac.c ? (Couldn't find it in flac_decode_frame() ).
01:06:28Unhelpfulmcuelenaere: did you try the "rebuild cache" menu option? ;)
01:06:35linuxstbmt: The "while (bytesleft) {" loop.
01:06:53linuxstbmt: i.e. it keeps looping until there are no bytes left.
01:08:12 Join itcheg [0] (i=62db4c46@gateway/web/ajax/
01:08:21mtAha ... I was looking at buff_size. :)
01:08:52Unhelpfulpictureflow rebuilds the cache if the "cache version" config option is not set correctly. rebuild cache clears this value, and the expected value is also incremented whenever the format changes
01:11:35mcuelenaereUnhelpful: thanks, didn't know how to go to the menu :) Rebuilding the cache didn't change a thing though..
01:12:05mcuelenaereUnhelpful: another weird thing is when I 'click' on a cover, it flips but doesn't show a track list (it also seems to freeze my player)
01:12:51Unhelpfulmcuelenaere: strange. what revision are you using? i changed the track list code last week or so, but the new code worked quite well on my targets and sim...
01:14:04mcuelenaereUnhelpful: r21001
01:15:52pixelmaStrife89: button table is here . Cleaned up a bit (putting more XY_PAD into one opt when the buttons were the same, got rid of the Fuze pad etc., excluded the quit button on the Ipods - already told you)
01:15:54 Quit froggyman (" ajax IRC Client")
01:16:08mcuelenaereUnhelpful: one more thing (a bit unrelated): is it normal that the FPS is pretty low (about 13-14) when just 'viewing' the cover, but when you select other covers it rises to about 60?
01:16:35kugelmcuelenaere: weird ones
01:16:36Unhelpfulthe in-memory cover cache and the album and track lists all come from the plugin buffer. the track list builds first, then the rest of the space is managed by buflib for the cover cache
01:17:02Unhelpfulmcuelenaere: that's normal, it raises the timeout for input events when it's not animating.
01:17:35pixelmaStrife89: as I said earlier the IPOD_1G2G_PAD aren't really needed as they are greyscale targets, but they don't hurt and maybe one day someone will port rockpaint to greyscale..
01:17:35Strife89pixelma: Pasted. Doing a quick test over here...
01:17:49kugelmcuelenaere: I only changed to #include "cpu.h", could that cause the reds?
01:18:33 Join JdGordon| [0] (i=836b0070@gateway/web/ajax/
01:18:51kugel"apps/plugins/ syntax error "
01:20:12mcuelenaerekugel: perhaps the linker can't handle C function definitions?
01:20:42mtlinuxstb : One more thing , For audio packets reordering there needs to be some sort of data copying; to be able to have one scrambling unit for the decoder.
01:21:03kugelI thought the soc specific .h's only contain #defines
01:21:21mcuelenaerekugel: at least jz4740.h doesn't, it also contains static (inline) functions
01:23:08mcuelenaerekugel: can't you only include your soc defines (perhaps in an #ifdef)?
01:23:22Strife89FS #10202 updated one last time (hopefully).
01:23:26 Quit JdGordon| (Client Quit)
01:23:39 Quit Dillizar ("I ♥ Elive")
01:23:48Strife89pixelma, AlexP: Think it's commit ready? :)
01:24:05AlexPStrife89: Just reading it now :)
01:24:29kugelmcuelenaere: sure
01:24:34Unhelpfulmcuelenaere: just guessing, but none of this stuff happens on the sim, right? :)
01:24:53mtlinuxstb : Unless, cook_decode_frame() is called from the descrambling loop inside rm_get_packet() (or something like that), so that get_packet() would send decode_frame() a pointer to the next frame ?
01:24:54mcuelenaereUnhelpful: I think not :) (haven't tried yet though)
01:25:14AlexPStrife89: The \fname I added to the second .bmp is missing
01:25:17*mcuelenaere guesses this has something to do with MIPS (+GCC perhaps)
01:25:47Strife89AlexP: Whoops, I didn't copy/paste your edits....
01:26:17pixelmaAlexP, Strife89: it might be nice to have a bit of indentation in the \item part
01:26:27linuxstbmt: Avoiding those copies would be best...
01:27:02AlexPStrife89: You got all the others :)
01:27:21AlexPpixelma, Strife89: yes, wouldhelp readability a bit
01:27:26Unhelpfulmcuelenaere: hrm, what happens if you convert that cover at the same size to a bmp, and rename the jpg (to something AA search won't find)?
01:27:26Unhelpfuland you'll need to rebuild cache, too.
01:29:19*pixelma isn't so sure about the \\nopt in the button table anymore - while they can make a really long list quite sure, they are also automatically true for a new keypad :\\
01:30:11mtlinuxstb : I'm thinking now of 2 solutions : writing a rm_decode_packet(), which would do both the data reordering and decoding according to the codec. Or leave rm_get_packet() as it is but modify it to return a pointer to the next frame which could then be sent to the decoding function.
01:30:15AlexPpixelma: A new keypad ought to be checked anyway - and that is true of any \nopt
01:30:29CIA-38New commit by kugel (r21002): Fix reds and yellows. The targets that showed have more that just #define (like inline functions) in their soc specific header, which the linker ...
01:31:11kugel+reds :/
01:31:13mcuelenaereUnhelpful: gives the same corruption
01:31:19AlexPpixelma: I fel indifferent towards which way round it is done :)
01:31:24Strife89pixelma, AlexP: Updated. :)
01:33:06***Saving seen data "./dancer.seen"
01:33:16AlexPStrife89: I think pixelma more meant indent the description under each \item, but I'll let her comment
01:33:27mtlinuxstb : Which boils down to : Any codec just calls rm_decode_packet, or gets a frame pointer from rm_get_packet and then call its own decoding function. Which is better (Assuming I'm capable of writing comprehensible sentences right now !)
01:35:08Unhelpfulmcuelenaere: hrm. what size was the AA when you tested that?
01:35:18 Join kkurbjun [0] (n=kkurbjun@rockbox/developer/kkurbjun)
01:36:05 Join wincent [0] (
01:37:17mcuelenaereUnhelpful: that particular image is 200x196
01:38:50linuxstbmt: I like the sound of the second idea - rm_get_packet returning a pointer to a frame. But I don't fully understand what the parser is doing...
01:38:56Strife89AlexP: I went ahead and added some further indentation.
01:39:30Strife89AlexP, pixelma: Okay, I've got revision 14 up, and I reviewed it one last time and didn't spot any other problems.
01:41:24Unhelpfulmcuelenaere: but what size did you have it displayed at? if you used the test_core_jpeg plugin, it would scale it to fit the screen - that should be linear upscaling, based on the dimensions of the screenshot. if you used a WPS, it might have used a size smaller than the image, and scaled it down.
01:41:33Unhelpfulonly parts of the two scalers are shared :)
01:42:08*mcuelenaere builds test_core_jpeg
01:43:43martian67is there a driver available for windows XP that will allow rockbox to charge while playing?
01:44:49AlexPIt needs something like HID to be implemented in Rockbox USB as I understand it
01:45:01Unhelpfulmartian67: i believe work is in progress on faking rockbox as an HID device. the trouble as i understand it is that windows won't complete device enumeration if it can't match the device to a driver, and then the device can't get full power.
01:45:09pixelmaStrife89: looks really nice now. Unfortunately it's quite late here and so I don't really want to commit anything now (also haven't added the screenshots locally yet)
01:45:30martian67Unhelpful, can i just modify a dummy driver to match the IDs rockbox provides?
01:45:44Unhelpfulmartian67: maybe.
01:45:46 Quit dfkt ("-= SysReset 2.53=- Ph'nglui mglw'nafh Cthulhu R'lyeh wgah'nagl fhtagn.")
01:45:58martian67like there are usb lamp drivers
01:46:07AlexPStrife89: I'll have a look :)
01:46:18martian67should be a simple matter to add the ids rockbox provides to the infs
01:46:59Unhelpfulwhy why why would a USB lamp have a driver?
01:47:12martian67because of the aformentioned problem :)
01:47:22*Unhelpful facepalms
01:47:26gevaertsusb lamps tend not to bother about spec compliance
01:47:32mcuelenaerehrm I suppose I need to edit viewers.config to get test_core_jpeg to work?
01:48:33Unhelpfulmcuelenaere: it should be in there already, i added the test/benchmark viewers to it already, since rockbox seems to be smart enough about ignoring entries that map to nonexistant viewers
01:50:10 Join JdGordon| [0] (i=460130e7@gateway/web/ajax/
01:50:51mcuelenaereUnhelpful: you're right, it's already there. Then why doesn't it show up in the open with.. list? (or perhaps I'm too tired to spot it :))
01:51:58 Quit Zarggg ()
01:52:10Unhelpfulit's pretty low down. i usually find it faster to get there from the end of the list than from the beginning.
01:53:09 Quit n1s ("Lämnar")
01:53:32mcuelenaereah had to reload the Rockbox binary to get it to show up
01:53:51mcuelenaereUnhelpful: the image shows up uncorruptedly
01:54:38Unhelpfulmcuelenaere: i think the list is stored in memory at startup. if i build too many viewers on recorder sim, it complains about the list being full.
01:54:47 Join tessarakt [0] (
01:57:13 Quit itcheg (" ajax IRC Client")
01:58:32AlexPStrife89, pixelma: So shall I press go? :)
01:59:30Strife89AlexP: I vote "yes". :)
02:03:12 Quit JdGordon| (" ajax IRC Client")
02:03:50AlexPpixelma: Any further comments? It looks OK to me.
02:04:19pixelmanone... go ahead :)
02:05:03CIA-38New commit by alex (r21003): FS #10202 - Add Rockpaint to the manual by Michael Carr.
02:05:09jordan`)/go pixelma
02:06:32 Join Zarggg [0] (
02:06:58 Join simjoko [0] (
02:07:44Unhelpfulmcuelenaere: i think i have your solution. pictureflow's output_row_8_transposed eroneously uses RGB888 values in LCD_RGBPACK_LCD :)
02:07:48AlexPStrife89: Ooops, I forgot CREDITS (unless you are already there?)
02:07:57Strife89AlexP: I'm not.
02:08:06AlexPOK, you soon will be
02:08:11 Join itcheg [0] (i=62db4c46@gateway/web/ajax/
02:08:20Strife89AlexP: Appreciated. :)
02:08:48Unhelpful(and that particular cover happens to fit PF's cover bounding box without any scaling)
02:09:22CIA-38New commit by alex (r21004): Ooops, forgot to add Michael Carr to CREDITS.
02:10:01*mcuelenaere didn't know Rockbox supported RGB888
02:10:36simjokotoday my touch wheel arrived from hong kong \o/
02:11:33Unhelpfulif i'm right, i think i can reproduce it on sim, and i'll have a fix in a few. :)
02:13:05 Quit pixelma (Nick collision from services.)
02:13:05 Quit amiconn (Nick collision from services.)
02:13:06 Join pixelma_ [50] (n=pixelma@rockbox/staff/pixelma)
02:13:08 Join amiconn_ [50] (n=jens@rockbox/developer/amiconn)
02:13:26 Nick pixelma_ is now known as pixelma (n=pixelma@rockbox/staff/pixelma)
02:13:28 Nick amiconn_ is now known as amiconn (n=jens@rockbox/developer/amiconn)
02:17:33 Part toffe82
02:17:43mcuelenaereUnhelpful: hmm I can't reproduce it on sim here
02:18:18 Quit itcheg (" ajax IRC Client")
02:18:22Unhelpfulmcuelenaere: that's odd, i would expect you should be able to, as we sim the LCD pixel formats..
02:19:27 Quit simjoko ("When two people dream the same dream, it ceases to be an illusion. KVIrc 3.4.0 Virgo")
02:19:55 Quit miepchen^schla (Read error: 60 (Operation timed out))
02:20:03mcuelenaereUnhelpful: you say LCD_RGBPACK_LCD uses RGB888 but I can't seem to find any reference to RB888 in lcd.h?
02:20:08 Join evilnick_bs [0] (i=620ec27e@gateway/web/ajax/
02:20:55 Quit blithe ("Lost terminal")
02:21:06 Join blithe [0] (
02:21:11Unhelpfulsorry if i was unclear, what i'm trying to say is that LCD_RGBPACK_LCD expects R/G/B to be respectively 5-, 6-, and 5-bit values, and pictureflow is passing it 8-bit values :)
02:22:05mcuelenaereuh? then how did this work in other targets/sims? :)
02:22:21 Quit Hillshum ("ChatZilla 0.9.83 [Firefox 3.0.3/2008092417]")
02:24:25Unhelpfulmcuelenaere: it wouldn't have been a problem until quite recently, when direct output to custom formats with RGB888 input was added. and it will only happen when a cover is loaded in pictureflow that is the desired size without any scaling.
02:26:28Unhelpfuland i can't imagine why it would not do the same thing on a sim, *if* you're using the same cover on a sim with the same LCD dimensions
02:27:27 Join itcheg [0] (i=62db4c46@gateway/web/ajax/
02:28:49mcuelenaereUnhelpful: \o/ you seem to be right, if I use a cover of >1200 image width it doesn't give image corruption :)
02:29:10mcuelenaere(on target)
02:29:34Unhelpfulmcuelenaere: 199x196 or 201x196 would probably have been fine as well
02:30:06 Join miepchen^schla [0] (
02:30:08Unhelpfulbut a bmp that is 200 wide, or a jpeg that is 200, 400, 800, or 1600 wide will probably all have that problem
02:31:04*mcuelenaere wonders why no-one else spotted this before
02:31:24*mcuelenaere decides to go to bed
02:31:54 Quit mcuelenaere (" - Chat comfortably. Anywhere.")
02:32:17Unhelpfulprobably they didn't happen to load any image of *exactly* the wrong size, in pictureflow :/
02:32:32Unhelpfulor they had a cached pfraw of that image from before the bug.
02:35:46 Quit JdGordon ("Leaving.")
02:39:02 Join BHSPitMonkey [0] (n=stephen@unaffiliated/bhspitmonkey)
02:40:26 Join JdGordon [0] (
02:40:35 Join Hillshum [0] (n=chatzill@unaffiliated/hillshum)
02:48:35mtlinuxstb : I'll try again sometime later with it, but, what if the data copying is inevitable to construct a scrambling unit?
02:52:34mtBecause for example, in one sample the scrambling unit contains 16 consecutive frames while each rm packet contains 5 frames plus the packet header of course, so adapting the descrambling formula to this seems to be almost impossible. (packet headers have to be stripped at least)
02:52:49 Part kugel
02:56:13 Quit HellDragon (Read error: 54 (Connection reset by peer))
02:56:38 Join HellDragon [0] (
03:08:07 Join dberg918 [0] (
03:09:01CIA-38New commit by unhelpful (r21005): Fix cover corruption reported in PictureFlow by Maurus Cuelenaere, bump CACHE_VERSION to force cache rebuilds.
03:09:11dberg918I'm getting a weird message when I try to run the configure script to build rockbox, but there's probably a very simple solution
03:09:30Unhelpfulwhat's the message?
03:09:44dberg918it says "
03:09:46dberg918You're running this script in a path that contains space. The build
03:09:47dberg918system is unfortunately not clever enough to deal with this. Please
03:09:49dberg918run the script from a different path, rename the path or fix the build
03:10:36Unhelpfulright, so, move your checkout of the source, and your build directory, so that they're not under any directories with a space in the name.
03:10:36dberg918I'm running Arch on my Dell laptop
03:11:00dberg918OH, that's what "space" means...
03:11:32dberg918that must be a rockbox error message then
03:11:38dberg918from the configure script
03:11:40krazykitit is.
03:11:51dberg918probably would've helped to look at it
03:22:11 Join cool_walking_ [0] (i=cb3b81c3@gateway/web/ajax/
03:26:58 Quit Strife89 ("Gone to bed.")
03:32:15dberg918I'm working on a theme at the moment, and I'm getting an error when it tries to load one of the bitmaps
03:32:26 Quit efyx_ (Remote closed the connection)
03:32:37 Join efyx_ [0] (
03:32:38dberg918it says "read_bmp_fd: Bitmap too large for buffer: 28350 bytes."
03:33:08***Saving seen data "./dancer.seen"
03:33:15dberg918and then it throws the error that it fails to load my volume bitmap strip
03:33:36evilnick_bsEr... try reducing the size of the bitmap image
03:34:18dberg918well, I can split the strips into seperate files I guess
03:34:36dberg918the thing is, the volume bar is vertical in my WPS, so the bitmap is quite long
03:34:53 Join fdinel [0] (
03:34:53dberg918675 pixels high
03:35:06dberg918but the bitmap is only 50 KB
03:37:27 Join heardcameron0000 [0] (n=480f2be5@gateway/web/cgi-irc/
03:38:29evilnick_bsI meant size as in filesize, not pixels
03:38:56evilnick_bsIt's possible that the buffer for bitmap strips is quite low
03:40:40 Quit heardcameron0000 (Client Quit)
03:42:09Unhelpfulthe bitmap reader doesn't really care about filesize. this message is likely occurring because whatever buffer the WPS code allocates for that load is not large enough.
03:42:42 Quit itcheg (" ajax IRC Client")
03:43:52dberg918yeah, I just commented out a viewport and it loaded just fine
03:44:10dberg918hmm, back to the drawing board
03:44:35dberg918never thought I'd run into this problem, given I'm making a simplified WPS
03:45:07dberg918just album art and artist/track info
03:45:23dberg918then volume and battery bars on each side when you change the volume
03:47:04 Join CaptainKwel [0] (
03:50:24 Join froggyman [0] (n=47ba0b80@gateway/web/cgi-irc/
03:52:46 Quit evilnick_bs (" ajax IRC Client")
03:53:04 Join itcheg [0] (i=62db4c46@gateway/web/ajax/
03:58:25 Nick fxb is now known as fxb__ (
04:00:06Unhelpfuldberg918: it's intended that the items in a bitmap strip will be fairly small, i believe..
04:00:36Unhelpfulrasher: didn't you push bitmap strips to the size limit in your cabbie wide?
04:02:12 Quit Hillshum ("ChatZilla 0.9.83 [Firefox 3.0.3/2008092417]")
04:06:04lymecaCan rbutil take an HFS+ iPod and automatically re-do the partitions so it is FAT32?
04:09:04krazykitat least, not right now. it'd be possible, probably
04:10:01 Join LambdaCalculus37 [0] (n=rmenes@rockbox/staff/LambdaCalculus37)
04:11:16LambdaCalculus37So I'm going to figure that FS #10202, has been tested, retested, fixed, refixed, gone over with a fine tooth comb, and given lots of tender loving care?
04:12:40LambdaCalculus37Because I'm ready to commit it.
04:13:39LambdaCalculus37Unhelpful: FS #10202 is Strife89's manual entry for Rockpaint. I was helping him fix it up earlier this afternoon along with AlexP and pixelma.
04:13:44krazykitLambdaCalculus37, perhaps you should refresh the flyspray page
04:13:44Unhelpfullooks like it was committed as r21003. :)
04:14:10*LambdaCalculus37 refreshes
04:14:21LambdaCalculus37AlexP beat me to it!
04:18:53 Quit itcheg (" ajax IRC Client")
04:20:25 Join lee321987 [0] (
04:20:47LambdaCalculus37Meh... guess I'll get something else to work on, then.
04:20:51 Quit LambdaCalculus37 ("Fwump")
04:21:46lee321987I'm using the VMware Debian image from RBs' site. I've used Ubuntu before, but not Debian. Is there a file manager in there anywhere?
04:23:36krazykityou have a command line ;)
04:26:17cool_walking_If you really want a file manager you can just "sudo apt-get install <filemanagername>". There may not be enough space on the VM's disk though.
04:26:36Unhelpfuldoes the vmware image even include X?
04:26:45cool_walking_The wiki says it does.
04:30:31 Quit spyder ("leaving")
04:31:00 Join spyder [0] (
04:32:01lee321987Do I have to have a DHCP server running on my network to access the Debian environment from Windows? All I want to do is copy my RB source, to it so I can compile −− is there an easy way to do this? I tried telling VMware to use an ISO image for the virtual CDROM, but when I navigate to "cdrom" or "cdrom0", the command "ls" does nothing.
04:33:20cool_walking_No you don't need a DHCP server. You should be able to access debian through windows networking. I think you type something like "\\debian\user" in Windows Explorer window.
04:33:25lee321987BTW I have no network (only one computer).
04:33:54cool_walking_You don't need a real network, it's a virtual one created by the VM.
04:35:07cool_walking_The CD image not working.. did you mount the drive within Linux?
04:35:22lee321987no, how?
04:35:58lee321987you were right −− Share name: \\debian\user −−- but that doesn't work
04:36:31lee321987Do I type anything before "\\debian"?
04:36:50cool_walking_Getting kind of off-topic, this is general Linux stuff... but type "mkdir -p /mnt/cdrom; mount /dev/cdrom /mnt/cdrom"
04:37:02cool_walking_no, nothing before "\\debian"
04:40:19lee321987"mount /dev/cdrom". Thank you. Sorry for off-topic. I'll figure the rest out.
04:46:09 Quit spyder ("leaving")
04:46:19 Join spyder [0] (
04:46:29 Quit BHSPitMonkey (Remote closed the connection)
04:46:49 Join spyder_ [0] (
04:47:08 Join BHSPitMonkey [0] (n=stephen@unaffiliated/bhspitmonkey)
04:47:34cool_walking_Type "/sbin/ifconfig" in Linux, and, using the "inet addr" that comes up, try e.g. "\\\user" instead of "\\debian\user".
04:47:37 Quit spyder_ (Client Quit)
04:50:18lee321987cool_walking_: inet addr reports −−- didn't work. Does that number seem ok?
04:50:30BHSPitMonkeyThere's no place like
04:50:54lee321987oops. one sec...
04:51:06BHSPitMonkeylee321987, that address refers to yourself
04:51:13BHSPitMonkeyjust like localhost
04:51:53lee321987The Ethernet settings for the VM image is set to "Bridged" −− this is right?
04:54:02lee321987I changed it to NAT. −− restarting...
04:55:06Unhelpfuli doubt that was the problem. did ifconfig report any interfaces besides lo?
04:55:15cool_walking_You should have a second interface (other than the "lo" or "loop" or "loopback" one).
04:56:13lee321987I just noticed "ifconifg" results have _two_ "inet addr" feilds. −− trying the other one...
04:57:43lee321987That did it.
04:58:14lee321987Sorry −− I read backwards sometimes.
05:00:07cool_walking_if you don't want to have to remember than number, open your c:\windows\system32\drivers\etc\hosts file (it's a text file - use notepad) and add a line to the bottom like "<whatever the address was> debian".
05:00:43lymecaI have a 4GB 2nd gen Mini and it was HFS+ so I put a new DOS partition table on with fdisk but how big should I make the first partition?
05:02:07lee321987cool_walking_, Unhelpful, BHSPitMonkey, Awesome. Thanks a lot for all your help!
05:04:09 Quit miepchen^schla (Read error: 101 (Network is unreachable))
05:23:02 Join itcheg [0] (i=62db4c46@gateway/web/ajax/
05:23:51lee321987Rockbox build time on Debian VM = 2min 30sec −−- awesomeface.jpg
05:32:58evilnick_homelymeca: Have you tried reading this:
05:33:09***Saving seen data "./dancer.seen"
05:33:54evilnick_homecool_walking_: Sorry, just saw your reply!
05:35:34 Quit froggyman ("CGI:IRC (EOF)")
05:36:33 Quit fdinel ("Miranda IM! Smaller, Faster, Easier.")
06:02:56 Quit tomers ("ChatZilla 0.9.84 [Firefox 3.0.10/2009042523]")
06:03:54 Quit efyx_ (Client Quit)
06:07:21 Quit dberg918 (Remote closed the connection)
06:18:47 Quit firedix ("Ex-Chat")
06:21:26lee321987Good day, I hope you have.
06:21:29 Quit lee321987 ("ChatZilla 0.9.84 [Firefox 3.0.10/2009042316]")
06:22:27 Quit CaptainKwel (Read error: 110 (Connection timed out))
06:23:04 Quit DataGhost (Nick collision from services.)
06:23:12 Join DataGhost [0] (i=dataghos@unaffiliated/dataghost)
06:31:54 Join evilnick_home1 [0] (
06:32:24 Join |ahmina| [0] (
06:47:39 Join roomtemperature [0] (n=blayde@
06:49:49 Quit evilnick_home (Read error: 110 (Connection timed out))
06:50:43 Quit lymeca (Connection timed out)
07:04:28 Join jmillikin [0] (
07:25:05 Quit kkurbjun (Read error: 110 (Connection timed out))
07:33:10***Saving seen data "./dancer.seen"
07:38:37 Join roomtemperature1 [0] (
07:42:45 Join Horscht86 [0] (
07:45:57 Quit roomtemperature1 ("Leaving.")
07:46:15 Join roomtemperature1 [0] (
07:47:57 Quit jmillikin (Read error: 110 (Connection timed out))
07:49:08 Quit roomtemperature (Read error: 110 (Connection timed out))
07:49:31 Nick roomtemperature1 is now known as roomtemperature (
07:57:23 Quit JdGordon (Read error: 104 (Connection reset by peer))
07:59:27 Quit Horschti (Read error: 110 (Connection timed out))
08:00:20 Join JdGordon [0] (n=jonno@rockbox/developer/JdGordon)
08:10:09 Quit tessarakt ("Client exiting")
08:15:32 Quit BHSPitMonkey (Read error: 110 (Connection timed out))
08:19:32 Join timc [0] (n=aoeu@
08:19:50 Join ender` [0] (
08:27:02 Join n1s [0] (n=n1s@rockbox/developer/n1s)
08:32:31 Quit timc (Read error: 104 (Connection reset by peer))
08:34:52 Join timc [0] (n=aoeu@
08:42:26 Join Rob2222 [0] (
08:48:34 Quit |ahmina| ("KVIrc 3.4.0 Virgo")
08:48:55 Join psycho_maniac [0] (i=psycho_m@
08:52:18 Quit GodEater (Read error: 110 (Connection timed out))
08:56:01 Join robin0800 [0] (
08:56:55 Join bertrik [0] (
08:58:08 Join lymeca [0] (
08:58:58 Quit roomtemperature ("Leaving.")
09:00:44 Quit Rob2223 (Read error: 110 (Connection timed out))
09:04:53psycho_maniacanybody know how to delete alltel shop from my stock alltel rom? i found the .link file but it links to "\windows\InHandStarter.exe" would that effect anything else if i deleted it?
09:05:31linuxstbpsycho_maniac: #rockbox?
09:06:04psycho_maniacoops wrong window
09:06:18 Quit robin0800 (Remote closed the connection)
09:20:03 Join flydutch [0] (
09:27:46 Nick Horscht86 is now known as Horscht (
09:29:20Unhelpfulwould somebody who knows sh-1 asm mind taking a look at this? it's an attempt at a 16-bit * 32-bit -> 32-bit multiply.
09:30:48 Join Thundercloud [0] (
09:31:03 Join petur [0] (
09:31:42 Quit psycho_maniac (" HydraIRC -> <- *I* use it, so it must be good!")
09:32:31 Quit _lifeless (Remote closed the connection)
09:32:48 Join _lifeless [0] (n=lifeless@
09:33:13***Saving seen data "./dancer.seen"
09:42:12linuxstbUnhelpful: I think that's pretty much just amiconn
09:42:57Unhelpfuli kind of figured, i just hate to always be bugging one dev with these questions :)
09:47:12 Join matsl [0] (
09:54:04 Quit Thundercloud (Remote closed the connection)
09:55:26 Join mehrab [0] (n=quassel@
10:06:24 Quit _Auron_ ("Infinity repeatedly denies rumours of plotting with zero to bring down the Universe.")
10:10:33 Quit mehrab (Read error: 54 (Connection reset by peer))
10:14:11 Quit cool_walking_ (" ajax IRC Client")
10:20:40 Join Grahack [0] (
10:23:14 Join robin0800 [0] (
10:32:42 Join GodEater [0] (n=nnyeahri@rockbox/staff/GodEater)
10:35:41 Quit matsl (Read error: 110 (Connection timed out))
10:36:50 Join stoffel [0] (
10:41:48 Join Jaykay [0] (
10:47:02 Quit bertrik (Remote closed the connection)
10:47:36 Join bertrik [0] (
10:48:13 Quit Grahack ("Leaving.")
10:48:47 Join Grahack [0] (
10:50:30 Quit Grahack (Client Quit)
10:52:02Jaykayi suggest closing of FS #8982 - i already wrote a comment there some months ago but nobody did something
10:58:52Unhelpfulamiconn: if it's not too much trouble, i have an attempt at a 16x32->32 mutliply for sh-1 that i could use some input on.
10:59:25linuxstbgevaerts: Do I remember correctly that you were looking at this bug (mis-detection of sector size on ipod 5.5g) ?
11:08:34 Join renke [0] (
11:08:54mtlinuxstb : The descrambling formula determines the when a frame should be decoded, so when reading from the file, the first frame could be decoded the 300th for example. So in order to avoid copying, we have to keep looping through the whole DATA chunk, searching for a frame whose decoding position according to the formula is 1 greater than the previous value, and then return a pointer to...
11:08:56mt...this frame. Is this still going to better than data copying ?
11:10:36linuxstbThat seems a very strange "design"... Can it really be 300?
11:11:34markunI don't know anything about real audio, but this sounds really weird for something that was originally used for streaming audio
11:12:24Unhelpfulmaybe that's why realplayer spends so long "Buffering..." ;)
11:12:58mtI tried displaying the position each time the formula changes, and I got values like 100+ for a really small sample.
11:13:07mtLet me recheck.
11:14:08 Join {phoenix} [0] (
11:14:37linuxstbmt: If this is really the case, then we may have some problems - the buffering code isn't designed to allow you to seek backwards and forwards in the file...
11:14:48Unhelpfultry logging the actual order in stream and the sequence order per the descrambling formula... surely there's *some* bound to how far away a frame is stored from where it's played.
11:16:00Jaykayan article from says that with installing rockbox on an ipod you lose any warranty of the ipod - is that true?
11:16:46GodEateryou'd have to check with apple to be sure
11:17:05GodEaterto be honest though, all the ipods that Rockbox runs on are now so old they're probably out of warranty anyway
11:18:04 Quit robin0800 (Read error: 110 (Connection timed out))
11:19:53mtsorry, it seems 300 was a bit exaggerated (I forgot to divide on frame_size !). : this is a sample of the decoding order.
11:23:05mtlinuxstb : If it's OK to have a temp memory to hold the scrambling, there should be no problem in seeking. A rm packet (the one with the packet header and containing multiple frames) which determines the start of a new scrambling unit is flagged as a keyframe, and has the timestamp updated. The time stamp then remains constant for the length of one scrambling unit.
11:24:53mtThe formula mainly depends on the position of a frame in its parent packet and the position of the packet in the length of a scrambling unit in the original file.
11:25:46Unhelpfulthe upper bound would be the length of the one scrambling unit?
11:26:51mtSo when seeking, the code should look for packets flagged as keyframes, and when seeking is done, return the time stamp of the current packet.
11:27:23mtUnhelpful : yes. More specifically, the number of frames per scrambling unit, which in this sample was 90.
11:27:35 Quit petur ("reboot")
11:32:45 Quit stoffel (Read error: 104 (Connection reset by peer))
11:33:17***Saving seen data "./dancer.seen"
11:36:05 Join stoffel [0] (
11:44:29 Join bmbl [0] (n=Miranda@unaffiliated/bmbl)
11:49:39mtlinuxstb : Another thing for seeking, the code should look for the key frame whose decoding position is 0 according to the descrambling formula. (That's all I can think of regarding seeking for now.)
11:54:16linuxstbmt: How big (in bytes) is a frame?
11:54:27linuxstb(in the worst case, if you know)
11:55:04linuxstbmt: Also, do you know if this is specific to cook, or if all realaudio codecs are this strange?
11:55:58mtThe descrambling formula is specific to cook.
11:56:27mtIn my samples I got 280 bytes so far for a frame size.
11:57:37 Join robin0800 [0] (
11:58:13 Quit robin0800 (Remote closed the connection)
11:58:33 Join robin0800 [0] (
12:03:07bertrikThere's a lot of minor duplicate weirdness in rockbox. I'm tempted to fix it, but on the other it doesn't really fix a specific bug.
12:03:24bertrikLike the function tuner_power is defined in both tuner.h and in power.h :P
12:04:01bertriktree.h declares extern struct gui_syncstatusbar statusbars; :S
12:04:09mtlinuxstb : the current descrambling formula is specific to both cook and AC3.
12:04:21mtThere's a different one for ra_28.
12:05:45n1sbertrik: nothing wrong with cleanups even if they don't fix a visible bug
12:08:32 Quit robin0800 (Remote closed the connection)
12:08:52 Join robin0800 [0] (
12:13:12linuxstbmt: So it's possible that the entire scrambling unit is within 32KB? e.g. 90 * 280 = 25200 bytes.
12:14:17linuxstbbertrik: I would go further than n1s and say such cleanups are a good thing... They sound like they could cause bugs (or at least, programmer confusion) in the future.
12:15:15bertrikn1s, linuxstb thanksfor your support :) I may have a go at some cleanups the coming days
12:18:30mtlinuxstb : I think so. (by the way for the 280-byte sample number of frames per one unit = 80 so 80*280 = 22400 bytes - a little less :) ).
12:21:05mtlinuxstb : The memory I used for holding one scrambling unit was 30K and it worked for all the samples I got. And it could be reduced to ~25k.
12:21:49linuxstbmt: Can I just check I understand how a realaudio file is structured? i.e. the stream consists of a number of "packets", and a packet will contain a number of frames. The cook codec itself requires a "scrambling unit", and that is contiguous sequence of frames, but not in the correct order?
12:22:05linuxstbAnd the frames for a scrambling unit can be split across packets?
12:23:02mt"not in the correct order " - If you mean in the original file, then yes that's how it's structured.
12:24:07linuxstbSo for example, if a scrambling unit is 4 frames (for simplicity), then the frames would be something like 1c, 1b, 1d, 1a, 2d, 2a, 2c, 2b, etc
12:24:16linuxstb(in the original file)
12:26:31mtlinuxstb : Yes. Similar to what you said.
12:27:31mtThe frames for a scrambling are always split across multiple packets. (the number of those packets is determined by sub_packet_h)
12:29:28linuxstbmt: But all frames for one scrambling unit are together in the file? i.e. you don't get something like "1c, 1b, 2c, 1a, 1d, 2b, 3d, 2a, 2d, 2c" ?
12:31:48mtAll frames for one scrambling unit are packed together yes.
12:32:13mta packet which is flagged as a "key-frame" marks the beginning of a new unit.
12:32:40mtand the first frame in this packet is always decoded first.
12:33:51linuxstbWhere is the order of the frames stored? Is that in the first packet?
12:34:32linuxstbOr is it always the same order?
12:34:42mtYou mean the order of the frames after descrambling ?
12:35:42linuxstbI mean the order of the frames in the original stream.
12:37:34 Quit SUSaiyan` (Read error: 104 (Connection reset by peer))
12:37:56mtIt's not stored. The descrambling formula calculates the order of decoding for each frame. And the frame ordering is the same for every scrambling unit (since the order is dependent on the frame's position in the parent packet and the position of this packet in the scrambling unit).
12:43:29 Join SUSaiyan [0] (
12:45:11 Join funman [0] (n=fun@rockbox/developer/funman)
12:47:35linuxstbmt: So could the parser request 32KB of data using request_buffer(), and then find the starting locations of all the packets required for the next scrambling unit, and pass those pointers to the cook codec?
12:52:29mtMaybe have an array of pointers to the correct locations ? For example, the array would have a size of 90-100 (for the worst case). And, using the descrambling formula, each frames location could be passed to its proper location in the array, so a frame which is to be decoded the 64th is passed to pointer number 64 of the array.
12:55:08mtA decoding loop similar to the current one in main.c, would loop to an upper limit of (nb_frames_per_unit), and decode a sub_packet_size from the location determined by the current pointer. Sounds possible.
12:59:20 Join domonoky [0] (n=Domonoky@rockbox/developer/domonoky)
13:00:47 Nick fxb__ is now known as fxb (
13:10:52 Join mrkiko [0] (
13:15:44 Join spoekes [0] (n=51ad8e8b@gateway/web/cgi-irc/
13:17:26 Join webguest93 [0] (n=51ad8e8b@gateway/web/cgi-irc/
13:19:39bertrikhm, both stdio.h and sprintf.h declare snprintf and vsnprintf
13:20:12mtlinuxstb : Update - Still working with a file descriptor, the code loads the addresses of the frames as you suggested, and for decoding it seeks to the proper location, reads a frame into a buffer, decodes it and it works. :)
13:20:17 Quit webguest93 (Client Quit)
13:20:27 Join webguest75 [0] (n=51ad8e8b@gateway/web/cgi-irc/
13:20:46 Quit spoekes ("CGI:IRC (Ping timeout)")
13:25:12 Quit stoffel (Read error: 113 (No route to host))
13:26:49 Join moos [0] (n=mustapha@rockbox/staff/moos)
13:30:40 Quit flydutch (Remote closed the connection)
13:30:54 Join flydutch [0] (
13:31:47 Quit webguest75 ("CGI:IRC (EOF)")
13:33:19***Saving seen data "./dancer.seen"
13:37:37 Join BXCracer [0] (
13:37:50 Join tessarakt [0] (
13:39:25CIA-38New commit by alex (r21006): Add plugin keys to the manual for the Gigabeat S, plus a couple of missing ones for the Gigabeat F. Also add missing equalizer keys for the S. Vast ...
13:39:40amiconnUnhelpful: That looks ok-ish - there are some unclear details:
13:40:28amiconn(1) Why do you define [t2]? It's never used. (2) [t2] could be defined without '&', allowing better register reuse
13:40:51amiconnErr, I mean [t1] in (2) of course
13:41:41amiconn(3) Imho it is better to define such things as a static inline function, using a macro to call it as necessary
13:46:39CIA-38New commit by funman (r21007): Sansa AMS: Partly revert r20923 (reset IDE maximal freq to 90MHz to fix problems with some players)
13:46:54***Alert Mode level 1
13:46:54CIA-38New commit by funman (r21008): Sansa AMS: move OF size declaration in model specific config
13:49:47funmanbertrik: when using caches on the Sansa AMS, i2c fails at 400kHz because the "check if still busy" at the beginning of ascodec_read() & ascodec_write() fails : i2c is still busy. If I add 100 loops of 5 consecutive nops just before the i2c_busy() check , i2c works fine. But I wonder if we couldn't just put a loop waiting for readiness, like at the exit of these 2 functions?
13:51:25bertrikhm, yes I guess we could
13:52:08bertrikstill weird how the i2c ready bit acts
13:52:56funmanreally a doc one this i2c module would help, and i don't think this is a sensitive document for AMS
13:53:17bertrikfunman, indeed, all of this would be trivial with the i2c docs
13:53:46funmandid you see a reference number in the as3525 datasheet ?
13:54:25bertrikI looked a bit closer at the OF: they do use a nice interrupt driven i2c, but in the end the codec read/write routines still busy-wait for the i2c result
13:54:58bertrikAlso I think I have a bit more understanding of the various bits in the i2c control and status registers, I'm considering adding it as a comment in the source code
13:55:18bertrikfunman, no I didn't notice one, but I'll have a look right away
13:55:19funmanthat can't harm :)
13:55:44funmanwell there doesn't seem to be a reference code or anything, and googling the register names bring me to rockbox
13:56:04bertrikyeah, tried google too with the same result :P
13:56:55***Alert Mode OFF
13:58:49 Join MarcGuay [0] (
14:00:52 Quit robin0800 (Remote closed the connection)
14:01:21MarcGuayIdea: Move the rest of the AlbumArt page to the manual and shift the album album programs to the UsefulTools page.
14:01:43MarcGuay"album album" = "album art";
14:07:20 Join froggyman [0] (n=187b533e@gateway/web/cgi-irc/
14:16:29 Join stoffel [0] (
14:22:07 Quit froggyman ("CGI:IRC (EOF)")
14:24:46 Join dfkt [0] (i=dfkt@unaffiliated/dfkt)
14:29:54CIA-38New commit by mcuelenaere (r21009): Replace some TABs with spaces (no functional changes)
14:32:28CIA-38New commit by bertrik (r21010): Remove some unused declarations
14:33:45funmanbertrik: if you don't want to commit your i2c registers descriptions now, I'd like to have it personally
14:34:17 Join LambdaCalculus37 [0] (i=44a0430d@rockbox/staff/LambdaCalculus37)
14:34:23 Join merbanan [0] (
14:36:27bertrikfunman, I'd like to put in a little more analysis effort in it, a few things are still just educated guesses
14:36:59funmanI had given a try to it when looking for headphones detection but I gave up and dumped my results when I saw that only headphones _insertion_ could be detected
14:40:04MarcGuayUSB transfers on the ipod nano with RB are as slow as emergency disk mode. Normal?
14:40:39 Quit itcheg (" ajax IRC Client")
14:40:46LambdaCalculus37MarcGuay: I got the same thing with a friend's nano. It's worse with writes than it is with reads.
14:41:51MarcGuayI feel like I'm downloading with a modem in 1991.
14:41:55LambdaCalculus37Hehe. :)
14:42:02LambdaCalculus37That is dog slow, though.
14:50:13renke13KB/s −− I find that quite fast
14:52:07 Join cobrajs [0] (n=4856323d@gateway/web/cgi-irc/
14:52:50 Quit cobrajs (Client Quit)
14:52:53 Join cobrajs [0] (n=4856323d@gateway/web/cgi-irc/
14:53:12 Quit funman ("leaving")
14:55:50 Quit Beta2K (Read error: 104 (Connection reset by peer))
14:55:56 Quit {phoenix} (Remote closed the connection)
14:57:16 Quit stoffel (Read error: 113 (No route to host))
14:57:47gevaertsMarcGuay: you could try the dma patch
14:59:28 Join Beta2K [0] (
15:00:20LambdaCalculus37Question... the iPod video 30GB use a LSS of 512 or 2048? In connection with
15:00:31 Quit cobrajs ("CGI:IRC (EOF)")
15:00:46 Join itcheg [0] (i=41d59de2@gateway/web/ajax/
15:02:11 Part wincent ("Kopete 0.12.7 :")
15:05:28 Join kugel [0] (i=kugel@rockbox/developer/kugel)
15:09:17gevaertsLambdaCalculus37, linuxstb: not sure, but it the line that prints out the size assumes 2048, so that doesn't really mean much
15:10:41gevaertsi.e. a 30GB 2048-byte-sector partition will always be shown as 7GB by the bootloader
15:10:58CIA-38New commit by mcuelenaere (r21011): Touchscreen targets: fix time-set screen
15:11:00 Quit Lss (Read error: 104 (Connection reset by peer))
15:11:17 Join Lss [0] (
15:11:31amiconnLambdaCalculus37: The disk size doesn't matter, what matters is whether its a G5 or a G5.5
15:11:46 Join stoffel [0] (
15:12:21amiconnG5 uses 512 bytes / logical sector (both 30GB and 60GB), G5.5 uses 2048 bytes / logical sector (both 30GB and 80GB)
15:18:56LambdaCalculus37amiconn: Thanks for the info.
15:19:19LambdaCalculus37So would any devs like a Sansa Clip? I have three coming in that I got on Sellout.Woot.
15:19:50bertrikLambdaCalculus37, do you know if they're v1 or v2 clips?
15:21:14LambdaCalculus37bertrik: They're all 2GB Clips. But I guess I won't know until they arrive.
15:21:25*LambdaCalculus37 is betting they may all be v1's, though
15:21:36 Join evilnick_7 [0] (i=0c140464@gateway/web/ajax/
15:22:39 Quit renke (Read error: 110 (Connection timed out))
15:25:30CIA-38New commit by bertrik (r21012): Fix tuner_power warning on ondio FM
15:29:45 Quit stoffel ("leaving")
15:30:41 Quit kugel (Read error: 104 (Connection reset by peer))
15:30:47MarcGuayDefault behavior on USB plug from off-state: nano->load rockbox, make USB connection e200->load OF, no connection. expected?
15:31:05 Join kugel [0] (i=kugel@rockbox/developer/kugel)
15:31:21 Join robin0800 [0] (
15:32:17gevaertsMarcGuay: the "no connection" bit is a bit weird, but there are several possible reasons for it. I've seen the sansa OF not connect a bit too often to be surprised...
15:32:29MarcGuayscratch that last "no connection", must be a cable or OF prob..
15:32:39 Quit kugel (Nick collision from services.)
15:32:42 Join kugel_ [0] (
15:32:50 Nick kugel_ is now known as kugel (
15:33:00gevaertsbut yes, booting the OF is still expected with the official bootloader
15:33:12MarcGuaygevaerts: Why the e200 and not the nano?
15:33:22***Saving seen data "./dancer.seen"
15:33:40robin0800gevaerts: only from off
15:34:04gevaertsthe ipod bootloader never started the OF when booted with USB plugged in
15:34:35MarcGuayIt went to disk mode...
15:34:44gevaertsnot from initial boot :)
15:35:04gevaertsit detects USB when booted, and then reboots to disk mode
15:35:12MarcGuayAh, righto.
15:35:31 Join miepchen^schla [0] (
15:35:36MarcGuayWhy not have it default to RB?
15:35:42gevaertsso yes, the end result looks a bit weird, but there's nothing unexpected...
15:36:01MarcGuayThe e200 boot, that is.
15:36:12gevaertsMarcGuay: there's no rockbox release with native USB for them...
15:36:55 Join at0m [0] (
15:37:03MarcGuaygevaerts: But the current builds have it enabled. I don't see why it should admit to being usable when you connect with rockbox running but not from the off-state.
15:37:11robin0800MarcGuay: with the experimetal bootloader thats what happens
15:37:31MarcGuayAh, righto. bootloader!=build.
15:37:57gevaertsMarcGuay: if you build a bootloader yourself, you'll get one that doesn't go to the OF from usb
15:37:58robin0800MarcGuay: its in flyspray
15:39:54robin0800MarcGuay: FS #9955
15:40:10MarcGuayrobin0800: Got it, thanks.
15:41:20AlexPgevaerts: Have the problems that prevented it last time been solved to the extent of allowing USB to be enabled in 3.3?
15:42:26 Join matsl [0] (
15:42:50gevaertsAlexP: the immediate ones, yes. There are some remaining related problems that may still block it though. The ones I can think of are charging (maybe not a problem on the sansas), and the start screen issue
15:43:08AlexPstart screen issues?
15:43:24 Join efyx_ [0] (
15:43:34gevaertsset the startup screen to any of a pretty large list, and things break when you boot with usb connected
15:43:38*gevaerts looks for the bug number
15:45:01gevaertsFS #9957
15:46:10*gevaerts thinks he should probably start looking into this...
15:46:20 Quit ufoman (Remote closed the connection)
15:46:59AlexPgevaerts: A fine plan :)
15:47:25gevaertsAlexP: you're welcome to join :)
15:47:38AlexPThat backfired :)
15:48:02*gevaerts starts by trying to find out where that start screen thing is done
15:48:04MarcGuayAnyone else having problems with "attach a file" in flyspray? ubuntu 8.10 + firefox 3.0.8.
15:48:24MarcGuayJavascript is failing?
15:48:47robin0800gevaerts: See also FS #9955
15:49:29gevaertsrobin0800: yes, I've noticed that people use that one as a dumping ground for all sorts of usb related issues and speculation that are not related to the bootloader at all...
15:50:39*gevaerts doesn't think that anyone should be expected to look in all bug reports to see if there happens to be a comment somewhere in it about a bit of code he knows about
15:52:15Tornegevaerts: that seems perhaps optimistic :)
15:52:42 Quit TheDJACR (Read error: 60 (Operation timed out))
15:54:28gevaertsTorne: maybe it is, but that won't stop me grumbling :)
15:54:39*Torne grins.
15:54:57gevaertsMarcGuay: those tasks aren't really related...
15:55:52MarcGuaygevaerts: Isn't 9957 reporting problems related to 9955?
15:55:53robin0800gevaerts: And any way a lot are to do with the bootloader and prompted the production of some more test bootloaders
15:57:28robin0800MarcGuay: Yes and some of the problems are in 9955
15:57:53gevaertsrobin0800: I don't deny that there are some comments (maybe even most) that are related to the task, but wrong behaviour of the rockbox binary itself has nothing at all to do with whether or not the bootloader works. Would you add a comment there saying that flac files don't play properly anymore if that happened to start around the same time?
15:59:21gevaertsMarcGuay: related to in the sense that they're both about rockbox, yes
16:00:05robin0800gevaerts: No of course not but in this case untill this bootloader the players had never started in rockbox when connected to usb
16:00:29gevaertsrobin0800: the question is, *why* are those comments in 9955?
16:01:11 Join TheDJACR [0] (n=TDJACR@Wikipedia/Thedjatclubrock)
16:01:12gevaertsrobin0800: so? There was a bug report about this (9957), yet you decided that 9955 was more suited for your comments
16:01:31MarcGuaygevaerts: I presume it will be of interest to someone working on one of those tasks to have a handy reference to the other - hence, related.
16:02:14gevaertsMarcGuay: someone working on bootloader issues doesn't need to know or care about what happens afterwards, and someone who works on 9957 doesn't need to care about specific bootloader issues
16:02:39robin0800gevaerts: In my case I watched 9957 for an updated bootloader to fix the start up problems and missed the mention of 9957 until today
16:03:17gevaertsso you don't search for related bugs before submitting problem reports?
16:03:18robin0800first 9957 shouldbe 9955
16:04:00 Quit efyx_ (Remote closed the connection)
16:04:06robin0800gevaerts: Mine wasn't the first
16:04:23 Join efyx_ [0] (
16:04:53MarcGuaygevaerts: I just read over 9957 more carefully and, unfortunately for my pride, have to agree.
16:05:44 Quit DataGhost (Nick collision from services.)
16:05:45gevaertsrobin0800: indeed. The first mention of start screen issues starts with "My bug is a different bug that is not a part of this"
16:05:52 Join DataGhost [0] (i=dataghos@unaffiliated/dataghost)
16:07:39robin0800gevaerts: I don't know whether thats true is it a new bug or a bootloader bug?
16:08:47gevaertsrobin0800: FS #9957 was submitted two minutes after this, bu a committer
16:08:49evilnick_7robin0800: If you test with the test bootloader and then again with the released bootloader then you'd be able to narrow it down to bootloader vs. build
16:10:15robin0800evilnick_7: No you wouldn't because untill this bootloader you would not have booted to rockbox
16:15:28CIA-38New commit by bertrik (r21013): Simplify some redundant boolean expressions
16:15:50AlexPEither way, the problems with USB in 9957 are not caused by the bootloaders from 9955, and comments on the USB problem do not belong there
16:16:20AlexPPutting comments in the wrong place just means that people than can fix the issue are less likely to see them
16:17:37robin0800AlexP: Your certain its a USB problem?
16:17:44AlexProbin0800: yes
16:17:57gevaertswell, it's a startup order issue in apps/main.c
16:18:06AlexPWell yes
16:18:15AlexPBut not bootloader is what I meant :)
16:22:21 Quit thegeek (Read error: 104 (Connection reset by peer))
16:22:41robin0800gevaerts: AlexP I'll hapily test it when its fixed but still waiting for gnome to fix their usb problem
16:23:25AlexPIf you are talking about ubuntu 9.04 then that a) Isn't a Gnome problem and b) There has been a work around for ages
16:23:39AlexPAnd that work around has been discussed in here on a few occasions
16:24:01robin0800AlexP: DetailsPlease
16:24:13AlexPRob2222: Search the logs
16:24:19AlexPRob2222: Sorry
16:24:25AlexProbin0800: It is in the logs
16:24:27 Join mrkiko_ [0] (
16:25:36robin0800AlexP: I don't think theres a work around for c200 only e200
16:25:50AlexProbin0800: why do you thing that?
16:26:16AlexPHave you tried the same thing as suggested for the e200, and gigabeat S, and...
16:26:55robin0800AlexP:No thats why I asked for details
16:27:07AlexPAnd that's why I said to search the logs
16:27:22AlexPYou are just as capable of searching them as me
16:28:18MarcGuayIs there anything holding those bootloaders back from the 3.2 release? Seems the only problems being reported on the task are actually related to 9957.
16:28:25MarcGuayMake that 3.3.
16:28:41AlexPMarcGuay: It depends whether USB is enabled in 3.3
16:28:56AlexPBut there is nothing wrong with the bootloaders themselves I don't think
16:29:19gevaertsMarcGuay: there are a lot of devices that still don't charge properly in rockbox
16:30:03*gevaerts wouldn't mind some testing of the FS #9957 patch he just posted
16:30:11MarcGuaygevaerts: A reason to default to the OF?
16:30:15AlexPThat too - but the bootloaders are OK - it just depends if the behaviour is wanted
16:31:01gevaertsMarcGuay: maybe a reason to switch to the OF for USB, yes. People are used to have their player charge while it's connected
16:31:50CIA-38New commit by bertrik (r21014): Move private #defines from i2c-pp.h to i2c-pp.c
16:35:13Horschtmay I say that the behaviour mentioned in FS #9957 does not happen on ipod 5.5G 80GB r20988
16:36:01n1sMarcGuay: i committed a fix for the c200 bootloader (and maybe e200 too) after the test bootloaders in that task were built so those two should be rebuilt before a release
16:37:09 Quit mrkiko_ ("leaving")
16:39:26 Join kkurbjun [0] (n=kkurbjun@rockbox/developer/kkurbjun)
16:39:31gevaertsHorscht: is there actually something to resume?
16:39:46Horschtit resumes without the cable plugged in
16:40:00gevaertshm. I could reproduce it on my e200
16:40:18gevaertsanyway, the patch I added should fix it for everyone
16:40:32CIA-38New commit by alex (r21015): Add screenshots for Chopper to the manual.
16:40:34Horschtthe screen says "loading", then stops and goes into USB mode (and actualy shows up as a drive)
16:41:16Horschtallthough it does not resume to wps after disconnecting
16:41:37gevaertsthere could be timing issues involved as well. Your ipod is very different from sansa in that respect, as it has a seriously different disk spin up time
16:42:17gevaertsI don't think it's expected to resume after disconnecting. That may be a different feature request, but I don't see anything in the code that would do that
16:42:31Horschtyeah, I was just thinking i should try as the Ipod is a PP target as well, isn't it?
16:43:00gevaertsit is, yes
16:43:47gevaertsthis isn't really PP specific, but none of the other software usb stack players are actually usable enough yet for this issue to come up
16:44:25Horschtah, i just thought because it said so in the initial FS comment
16:45:09gevaertsactually which hardware USB players don't have bootloader USB?
16:47:19*gevaerts would like to test his patch on a hardware USB target, but he isn't sure which one to try
16:47:20 Quit mrkiko (Read error: 110 (Connection timed out))
16:47:36LambdaCalculus37gevaerts: H300!
16:48:03gevaertsdoesn't that one have bootloader USB?
16:48:16LambdaCalculus37It does.
16:48:26AlexPSo it won't help :)
16:48:27LambdaCalculus37What about your iAudio?
16:48:32gevaertsthat makes it pretty hard to test early-boot USB connections
16:49:08gevaertsI never could get a usb connection within rockbox on my x5. Only rom/OF USB works
16:49:34*gevaerts suspects that there's something wrong with the hardware on it
16:50:14LambdaCalculus37gevaerts: It's still in a better shape than mine. ;)
16:50:38gevaertsyours? You never paid for trhat heap of junk! ;)
16:50:58*gevaerts is happy to have found a free garbage disposal system though :)
16:51:07LambdaCalculus37gevaerts: :P
16:54:03CIA-38New commit by gevaerts (r21016): Re-order usb init a bit, so software and hardware usb are the same. This fixes FS #9957
17:05:22 Quit robin0800 (Remote closed the connection)
17:05:24 Join petur [0] (
17:05:43 Join jgarvey [0] (
17:11:31AlexPdamn, he just left
17:12:04AlexProbin0800: On the odd chance that you read this - sorry, it wasn't discussed in here as much as I thought
17:32:38 Join _Auron_ [0] (
17:33:26***Saving seen data "./dancer.seen"
17:35:06 Join _2M1R_ [0] (
17:36:41_2M1R_someone use rockbox on sansa fuze?
17:39:52_2M1R_firmware update to rocksbox on fuze v1 is safe?
17:40:13 Join saratoga [0] (i=9803c6dd@rockbox/developer/saratoga)
17:41:15Jaykay_2MIR_: the fuze is not supported yet
17:41:20LambdaCalculus37Rockbox on the Sansa Fuze is only meant for developers at this point. It's not ready for everyday use.
17:42:56 Quit merbanan (Read error: 60 (Operation timed out))
17:43:06_2M1R_i know
17:43:35_2M1R_ok i will wait
17:43:56_2M1R_thanks for help
17:44:52 Join bagawk_ [0] (
17:58:11MarcGuaypixelma, Llorean: If you've got a spare minute could you review FS #10207?
17:58:50 Quit bagawk (Read error: 110 (Connection timed out))
18:03:46MarcGuayAlexP: You succesfully installed a single-bootloader on your beast, yes? Or am I thinking of GodEater?
18:04:14AlexPMarcGuay: I did, yes
18:04:47MarcGuayAlexP: Did you use sendfirm or simply copy an nk.bin over to the first partition and replace the existing one?
18:05:41AlexPI *think* (it was a long time ago) that when charging went in I replaced the existing dual-boot one with a single boot one directly (i.e. just copied it to the firmware partition)
18:07:55MarcGuayWhat's the command to convert the bootloader.bin to an nk.bin?
18:08:45AlexPyou should get the nk.bin directly when you buld the bootloader?
18:09:20 Quit feisar-- (Remote closed the connection)
18:09:23 Join feisar-- [0] (
18:09:50MarcGuayYou're right.
18:09:55*MarcGuay is blind
18:15:54 Join mcuelenaere [0] (
18:16:37mcuelenaerekugel: RE FS #9174: why would you want floating support when there isn't any available?
18:16:57kugelnot in hardware, no
18:17:25kugelbut if it can be emulated, why not (unless that would be major additional work)
18:17:33kugelI thought it was supported in the previous patches
18:19:08mcuelenaereI believe software floating point is quite an overhead
18:19:41domonokykugel, you know that emulated fp is slooooow ? so its probably best to not add it.
18:20:01kugelI know yes
18:20:03pixelmaMarcGuay: I just think it's a bit dangerous to have "start a new file/recording" and menu on the same button with the difference only being if it's a short or a long press (just remember thinking the same while trying to figure out a c200 keymap). I'm not sure if it's already a help if menu is on the short press as accidentally going to the menu might not hurt as much as accidentally splitting a file, assuming you are more likely to trigger the
18:20:03pixelma short press action unintentionally - I guess there is a reason this action doesn't exist in the Ondio recording screen :\
18:20:06kugelcalc has it too though
18:20:51kugelthis is about a plugin and scripts in the end, no core functionality
18:21:54mcuelenaereIMO it should just be left out
18:23:11kugelIMO not
18:23:24*domonoky is with mcuelenaere :-)
18:23:40AlexPkugel: What advantage does it bring (I'm interested)
18:24:28 Join barrywardell [0] (
18:24:33kugelare the scripts time critical code or something? I think not
18:25:01kugelsupporting floats makes it easy for script people to adapt existing usefull scripts
18:25:31AlexPDoes their presence always slow it down, or just when they are used?
18:25:42*mcuelenaere looks for softfloat vs hardware FP numbers
18:25:51saratogathey'll waste memory always but shouldn't be slower unless you actually do computation with them
18:26:01saratogaof course for a plugin the memory thing may or may not matter
18:26:14MarcGuaypixelma: The menu isn't available while recording is happening so presumably if someone is pressing SELECT while recording, it's because they want to split?
18:26:42MarcGuayOr they don't know that the menu isn't available while recording. Maybe a note to add to the manual.
18:26:43kugelAlexP: I expect it not to slow anything down unless you use them
18:27:04domonokyif i understand lua correctly, we can only choose what "Number" is, so either all numbers are float, or all are integers...
18:27:31domonokyif all numbers are handled as float, it probably hurts badly in performance.
18:27:48pixelmaMarcGuay: hmm... could be remembering wrong but isn't this button also used as an alternative start recording button?
18:28:15MarcGuaypixelma: The "split" button also starts recording... Right...
18:28:25MarcGuayMissed that.
18:28:33AlexPthat could be disabled
18:28:42pixelmamaybe that could be revised
18:28:43AlexPif there is already a start recoriding button
18:28:49MarcGuayIt seems redundant.
18:28:56AlexPIn fact, that would make it a little less confusing in my mind
18:29:17kugeldomonoky: well, if that's true, then it should be left out yes
18:30:08kugelbut all modern scripting languages I know of differentiate between int and float
18:30:25MarcGuayAlexP, pixelma: If there's a reason it exists we should find out now before changing anything. Maybe a low-button target needs it?
18:30:31mcuelenaerekugel: I don't think lua does
18:30:37AlexPMarcGuay: Possible
18:31:03AlexPMarcGuay: A button vs function vs target table would be useful :)
18:31:23pixelmaMarcGuay: I doubt it's because of a low-button target...
18:31:32domonokykugel: take a look here:
18:37:51kugeloh, I see
18:37:56kugelno fp then :)
18:38:01*mcuelenaere isn't sure yet how to get BUTTON_* available in lua
18:39:58 Nick fxb is now known as fxb__ (
18:40:03kugelmcuelenaere: maybe not at all? making ACTION_* available seems easier and less target dependent
18:40:32*mcuelenaere didn't think of that
18:40:54mcuelenaereI was thinking of some kind of hash table where BUTTON_* strings are mapped to their int equivalents
18:42:48 Join JdGordon| [0] (i=836b0070@gateway/web/ajax/
18:44:09kugelcan get_action exported as is?
18:47:17mcuelenaerekugel: I don't see why it couldn't
18:47:31mcuelenaerebut then you would also need to export all these ACTION_*
18:47:42kugelthat's a single enum afaik
18:47:53mcuelenaere(or generate some kind of 'helper' lua file which contains definitions for them)
18:48:01 Join toffe82 [0] (n=chatzill@
18:48:02mcuelenaerekugel: int (*get_action)(int context, int timeout);
18:48:15mcuelenaere(according to plugin.h)
18:52:15 Join robin0800 [0] (
18:52:21mcuelenaerekugel: get_action is already exported to lua
18:53:09kugelthe ACTION_* enum too?
18:54:30robin0800shall I raise a new bug just tested r21016 and it still dosen.t appear to work on my C200
18:56:45mcuelenaerekugel: no
18:58:31pixelmarobin0800: "it"
18:59:39robin0800gevaerts: my c200 from off boots to rockbox but dosn't go to usb
19:01:10*gevaerts doesn't give an opinion until he's seen some details
19:02:01robin0800gevaerts: What details would you like
19:02:06 Quit bertrik (Remote closed the connection)
19:03:00pixelmaMarcGuay: thinking about it - wouldn't it make things easier, especially on targets with few buttons, if the recording screen was "aware" if a recording is running? I imagine it like this: one button to start the recording which will then act as "split" button while recording - and another to stop recording which could also leave the screen when recording is stopped. Other targets that have enough buttons could still have additional buttons for it
19:03:01gevaertswhat OS, what start screen, does it work when plugged in later, what exact build, which bootloader
19:03:11pixelmaMarcGuay: just a bit brainstorming here
19:04:00 Join bertrik [0] (
19:05:38*mcuelenaere thinks some sort of script should generate the ACTION_* defines and export it to an actions.lua file
19:05:56robin0800gevaerts: Linux ,WPS ,Yes, R21016 Fs#9955
19:06:34gevaertswhat's the dmesg output?
19:06:44AlexProbin0800: And have you done the work around for the ubuntu bug?
19:08:01gevaertsis this an official build?
19:09:28 Join funman [0] (n=fun@rockbox/developer/funman)
19:11:18robin0800gevaerts: Yes and Yes
19:12:09AlexProbin0800: could you specify which questions your yes answers?
19:12:31AlexProbin0800: and then pastebin the dmesg output that gevaerts asked for ?
19:13:31robin0800gevaerts: Nothing in dmsg Kernal log
19:13:49MarcGuaypixelma: That's a neat idea.
19:14:06MarcGuayTricky with pausing, though.
19:14:22pixelmahmm... true
19:14:23MarcGuayStart and pause seem natural on the same key, rather than start and split.
19:14:40AlexProbin0800: really nothing? I see USB related info for a Sansa including a failure to enumerate
19:14:49MarcGuayMaybe it's best to remove the "start recording" functionality from the split key.
19:16:49pixelmaI can imagine that some owners (of e.g. the H300) will complain if they are used to it. We should get someone who records often enough to comment... petur?
19:16:50robin0800AlexP: Thats all you get Earlier is proof of usb connection then disconect turn player off and then connect and start
19:17:15AlexProbin0800: That isn't nothing
19:18:05AlexProbin0800: So what did you do to fix the ubuntu problem?
19:18:37pixelmaMarcGuay: but on targets with few buttons that could already help, I imagine.
19:19:31pixelmas/few buttons/few intuitively usable buttons
19:20:58MarcGuaypixelma: Do you think all of this talk prevents my patch from being an improvement? It's a step in the right direction, at least.
19:24:47pixelmaprobably better than non-working things. Wasn't there someone else working on the Ipod recscreen mapping too? I remember some talk about it here the last week or so but don't know if a patch was submitted...
19:30:01robin0800AlexP: changed 20-lbgphoto2.fdi
19:31:05MarcGuaypixelma: I think it was whoever opened the bug report.
19:31:47FlynDicefunman: Hi funman, I see you think the MCI_CLOCK numbers are not giving us good information
19:32:22MarcGuaypixelma: The first patch might do the job more cleanly.
19:32:37funmanyes see sd_enable() in ata_sd_as3525.c : I think when sd is disabled the regs read as 0
19:33:29***Saving seen data "./dancer.seen"
19:35:33FlynDiceThat's consistent with what I have seen. When my playback stops I have the value0x54f for the register and CGU_PERI MCI and NAF enable values stiil set as if sd_enable wasn't being called to unenable or not making it through that function
19:35:51pixelmaMarcGuay: it's just removing the linking to the next context if I see correctly - can you still stop the recording then? I don't have all the details at hand but thought the recscreen was using some action from the standard context to do that (?)
19:35:59CIA-38New commit by funman (r21017): Sansa AMS: show correct clock speeds for NAND & SD card
19:36:21funmanFlynDice: have a look at r21017
19:36:40MarcGuaypixelma: There was just some annoying overlap. Stop/cancel/menu is the MENU button.
19:36:42pixelmaMarcGuay: maybe I should also read the comments...
19:37:27MarcGuayErr, forgot about the lockup if you try to shutdown from the rec. screen.
19:37:33MarcGuayThat's a dealbreaker.
19:38:33FlynDicefunman: same code without the mmu enabled works just fine...
19:39:24MarcGuayAlexP: Is there any way you could send me your single-boot nk.bin so I could see if it works? If it doesn't, it might point to a hardware difference, or permanent changes caused by the v1.2 updater.
19:40:18 Part toffe82
19:40:43funmanFlynDice: so what is the bug : 405 instead of 400kHz ?
19:42:08 Join tvelocity[away] [0] (
19:42:15funmanby the way, did you have time to look at FS #10216 (SD on 8GB models)?
19:43:26 Join BryanJacobs [0] (
19:43:33FlynDicefunman: sorry 405 was with 62MHz PCLK, right now I'm showing 402 with a 64Mhz PCLK and with the divider at 4f the math works out right...
19:45:40funman62M / ((0x4f+1)*2) = 387.5kHz
19:45:51FlynDicewell I started to look at FS #10216 and thought I amy be getting in over my head... ;)
19:46:18FlynDiceI got 402.5 on my calc let me look again
19:47:06 Join CaptainKwel [0] (i=2669ecc2@gateway/web/ajax/
19:47:08FlynDiceah... 64M
19:47:10funmanyou have the pl180 pdf in your mailbox
19:47:12funmanoh sorry
19:47:16peturpixelma: I don't use split, but I can imagine that several people use it and will want it to stay the same. Have to scroll back more to get the whole discussion though...
19:47:24 Join toffe82 [0] (n=chatzill@
19:47:24funmanwell it gives exactly 400kHz then
19:47:25FlynDiceexcellent Thank you
19:47:32AlexPMarcGuay: Sure, give me a moment
19:48:07MarcGuayAlexP: Unless it's the one from the wiki, which doesn't work for me. I just remembered there's a task for this: - catching up on it now...
19:48:32AlexPMarcGuay: I don't know - I'll send it anyway and you can try :)
19:48:52MarcGuayAlexP: Sounds good.
19:49:10pixelmapetur: it's not about removing split itself, just the feature that it also starts a recording (meant to solve the issue that if you use e.g. a short select for split and a long select for the menu, it is possible to accidentally start a recording
19:49:50JdGordon|why is that a probleem though?
19:49:57pixelmato solve some keymapping issues
19:50:04funman__udivsi3 symbol comes from gcc so i assume it's bugfree
19:50:15BryanJacobsthought I should ask before I code one: is there some way to run a codec from outside of rockbox? I already know about simulator builds but that's not the dev cycle I'm looking for
19:50:40JdGordon|i belive thats something on our wish list...
19:50:53pixelmaJdGordon|: me?
19:50:59JdGordon|no, BryanJacobs
19:50:59BryanJacobsJdGordon: the "rockbox as an application" thing?
19:51:11pixelmaah, ok
19:51:27funmanFlynDice: when using an uncached buffer for sd DMA transfers, there is no more disk problems, so I'm not sure if it's worth looking into that
19:51:28JdGordon|no, being able to run the playback engine or even just a codec like you want...
19:51:28peturpixelma: having only one key start recording shouldn't be an issue I think
19:51:34BryanJacobswell, I was just planning to write a harness to ensure decompression is correct
19:51:45BryanJacobsif it would be helpful I'll post it up when it's done
19:51:56BryanJacobsshouldn't be too hard to get working, looking at the code
19:52:05FlynDicefunman: did you notice the clock config I was running to get my results, the biggest differences being the lower FCLK and lower IDE_CLK. I basically ran the same code before and could not play for more than a second or 2.
19:52:20pixelmapetur: thanks for the opinion, you were the only one I knew using the recscreen regularly... :)
19:52:52MarcGuayAlexP: Your bootloader works.
19:53:01Jaykayis the buttontable in general settings−−>bookmarking in the manual really needed? it looks like another setting and is like the file/playlist browser
19:53:07funmanok but it still crashes in the end, so is this a real progress?
19:54:09AlexPMarcGuay: Interesting
19:54:18AlexPMarcGuay: I wonder what the difference is?
19:54:27MarcGuayAlexP: Not entirely sure what this reveals.
19:54:53AlexPMarcGuay: Unless we know what revision this working bootloader is, not much
19:54:54MarcGuayAlexP: Is your file timestamped?
19:54:58FlynDicefunman: well, it's better than before so I thought it might be worth it to look at what changed and see if we could figure out why
19:55:05AlexPMarcGuay: Apart from that it is a Rockbox change
19:56:00gevaertsrobin0800: does it fail every time? I've just installed an official build (r21017, I was too late to download 21016) and the bootloader from FS #9955 on my c250, and it seems to work properly (I've tried 5..10 times so far)
19:56:02funmanFlynDice: since i have deadlocks with current svn code (on clip & less frequently on fuze), i suppose that enabling caching just make these bugs more visible
19:56:09 Quit robin0800 (Remote closed the connection)
19:56:34AlexPMarcGuay: 24/02/2009 00:41 - but I don't remember updating it anything like that recently...
19:56:36 Join robin0800 [0] (
19:56:54FlynDicefunman: on another note I have no problem copying a file from msd to internal sd with mmu and cacheing enabled
19:56:56gevaertsrobin0800: (in case you missed this) does it fail every time? I've just installed an official build (r21017, I was too late to download 21016) and the bootloader from FS #9955 on my c250, and it seems to work properly (I've tried 5..10 times so far)
19:57:15funmanFlynDice: by the way since we identified some time ago that fclk was incorrect (lower than pclk), I think that part should be committed
19:57:56robin0800gevaerts: I'll reinstall bootloader
19:58:10FlynDicefunman: That makes sense to me
19:58:21 Quit robin0800 (Remote closed the connection)
19:58:29 Quit tvelocity[a] (Read error: 110 (Connection timed out))
19:58:30funmanand also run some battery test with that change
19:58:43funmani'll look into that
19:59:03 Quit funman ("leaving")
20:00:28*AlexP spots an alive jhMikeS on the mailing list!
20:01:06*MarcGuay spots AlexP testing beastpatcher or February 24th! :)
20:01:27AlexPMarcGuay: Ooooh
20:01:28MarcGuayI guess that's when you copied it to the drive.
20:01:36AlexPMarcGuay: Good thinking
20:01:37MarcGuayDoesn't tell us when it was built, though.
20:01:48AlexPIsn't it built into beastpatcher?
20:01:55MarcGuayNot sure.
20:01:57AlexPI can't remember
20:02:10MarcGuayThat would explain the "v1", I guess.
20:02:33AlexPAnd would also give a good idea of when it was built
20:02:35MarcGuayWho's the beastpatcher master?
20:02:42AlexPlinuxstb wrote it
20:05:42Jaykayanother try: is the buttontable in general settings−−>bookmarking in the manual really needed? it looks like another setting and is like the file/playlist browser
20:06:05MarcGuayAlexP: When you build beastpatcher you pass it a bootloader.bin.
20:06:12MarcGuayDid you build your own?
20:06:26AlexPI don't think so, let me look at the logs
20:07:10MarcGuayJaykay: The delete button is unique.
20:07:29MarcGuayJaykay: What do you mean by "it looks like another setting"?
20:07:46AlexPMarcGuay: No, I'm pretty sure linuxstb gave it to me
20:08:06MarcGuayAlexP: Okay. I might try to roll one from that date and see what happens.
20:08:27JaykayMarcGuay: in the manual it is written like a 4th setting - maybe the characters are a bit smaller but that's not noticeable
20:08:50MarcGuayJaykay: I agree that the general stuff (scroll up/down) is pretty redundant.
20:09:09MarcGuayJaykay: HTML or PDF?
20:09:13 Quit _2M1R_ (Remote closed the connection)
20:09:18 Join Thundercloud [0] (
20:09:59Jaykaybut in html it doesn't look different too
20:10:05 Join robin0800 [0] (
20:10:19MarcGuayJaykay: I see what you're saying.
20:11:04 Quit HellDragon (Client Quit)
20:11:19MarcGuayJaykay: Any suggestions on how to improve it?
20:11:25Jaykayi would remove the rec-key functionality from rockbox and the buttontable from the manual ;)
20:11:29robin0800gevaerts: Mine still dosn't work if now playing is selected as start screen
20:12:08MarcGuayJaykay: I agree.
20:12:51gevaertsrobin0800: can you double-check the revision from System->Rockbox Info?
20:13:22 Quit kugel (Read error: 110 (Connection timed out))
20:13:45 Join {phoenix} [0] (
20:14:29robin0800gevaerts: R21016-090521
20:14:33mcuelenaereanyone object against committing FS #9174?
20:15:00 Join HellDragon [0] (
20:15:45 Quit flydutch ("/* empty */")
20:17:33gevaertsrobin0800: could you try r21017? I don't have any real idea about what's going on, and I'd like to exclude a build issue (as I haven't tested the official r21016 build)
20:21:15 Part BryanJacobs
20:21:57gevaertshm, it doesn't work properly with the database screen here
20:23:45*gevaerts wants to know if this works properly on a hardware-usb target
20:23:54robin0800gevaerts: Works with Main menu
20:24:33Horschtgevaerts, hardware usb?
20:24:50gevaertsHorscht: those that have a hardware ata-usb bridge
20:25:37Horschtwhich ones do? can I be of help? I guess not since I can not recreate this "wrong behaviour", rgiht?
20:26:12moosmcuelenaere: go go go!
20:26:25gevaertslots do, but most (or all?) of them also have bootloader usb, which makes this somewhat untestable
20:26:34pixelmagevaerts: I could try on my Ondio or M5
20:27:26pixelmawould should I test exactly?
20:28:24gevaertspixelma: set the start screen to wps or database (or both in different tests), and see what happens if you boot by plugging in USB.
20:28:51pixelmathen I can't test
20:29:29pixelmathe Ondio doesn't boot automatically by plugging USB and the M5 boots into the Cowon loader screen from off-state
20:30:40gevaertscan you boot the ondio while it's plugged in?
20:31:10gevaertswhat matters is that usb is plugged in at the moment the main firmware starts to run
20:31:30pixelmayes, holding the USB power button
20:31:34MarcGuayAlexP: Booya. Building a single-bootloader from Feb 24th works fine (it's the same size as yours - probably the same revision).
20:31:48AlexPMarcGuay: Wow, that's good
20:31:53CIA-38New commit by bertrik (r21018): Update as3525 i2c register description
20:31:59AlexPMarcGuay: So next thing is to track down when it breaks :)
20:32:21pixelmagevaerts: shall I test a certain revision? Got to update before, probably...
20:32:32MarcGuayAlexP: Yay! Err. Nay?
20:32:38AlexPhehe :)
20:32:56AlexPI would, but I don't want to have to spend hours copying 120 GB back over
20:33:17AlexPI could do this weekend, but not tonight
20:33:53MarcGuayAlexP: I'm going to take a look at it.
20:34:01AlexPMarcGuay: Good man :)
20:34:22gevaertspixelma: not really. I don't think anything changed in relevant code in months
20:34:51JaykayMarcGuay: could you tell me how to delete this rec-functionality )or do it yourself :P )?
20:36:19MarcGuayJaykay: I'm not sure where to look. I brought this up a while ago and there were people who wanted to leave it. JdGordon? Llorean? I seem to remember one of them having a strong opinion about it's usefulness.
20:41:24pixelmagevaerts: sorry, just understood now that you wanted to test the USB connection... so just simply boot after plugging the Ondio.
20:43:24 Quit jfc (Read error: 104 (Connection reset by peer))
20:43:46 Join jfc [0] (
20:44:06 Join dfkt_ [0] (
20:45:48Jaykaycould someone point me to where to search for the bookmark list in the code? and where... button-actions are defined
20:46:02Jaykayfor the bookmark list
20:46:22 Join thegeek [0] (
20:46:30Jaykaywell ok, maybe i get the price for the most incomprehensible question again...
20:48:29pixelmaseems to work correctly (for start screen WPS and database), Rockbox boots into the USB screen and I get the data connection. Build is r20614M-090404, M for backlight modded build, on XP and I did the test twice, one time with MMC plugged the other without as then the USB bridge can only connect either or. All 4 tests were ok
20:48:36pixelmagevaerts ^
20:48:49gevaertsok. Thanks
20:48:53AlexPJaykay: kaymaps are in apps/keymaps
20:49:23*pixelma updates regardless
20:50:21robin0800pixelma: That was from off I take it?
20:52:01pixelmaheh, I already put on a build last saturday and just didn't flash it....
20:53:35JaykayAlexP: REC in the bookmark list deletes the bookmark - i can't find this in the keymap, where is it?
20:53:51AlexPfor which target?
20:55:02AlexPJaykay: apps/keymaps/keymap-e200.c:269
20:55:37JaykayAlexP: thanks, i searched only for bookmark
20:57:12Jaykaywhat does LAST_ITEM_IN_LIST__NEXTLIST(CONTEXT_LIST) mean/do?
20:59:26pixelmaspeaking of which... I edited tools/configure a bit to make the Backlight mod build an advanced build option for the Ondios (largely copy-paste-adjusted from the H100's RTC mod change). Now: there are two places which check whether an Ondio was chosen and I do that currently by checking modelname ondiosp or ondiofm. It works but is there a "better" solution?
20:59:35gevaertsI'm pretty sure that it is a timing issue
20:59:57 Quit perrikwp (" ajax IRC Client")
21:00:01 Join perrikwp [0] (i=4aa794a0@gateway/web/ajax/
21:00:46pixelmaUSB is pretty slow on Ondio if that info helps
21:01:09CIA-38New commit by gevaerts (r21019): revert r21016, as it doesn't solve anything
21:01:30gevaertsI can make things work on my c200 by adding sleep()s here and there
21:01:31 Quit dfkt (Read error: 110 (Connection timed out))
21:01:37MarcGuayJaykay: If a button that isn't defined in that section it will look in the context_list definitions.
21:01:45CIA-38New commit by mcuelenaere (r21020): Commit FS #9174: Lua scripting language by Dan Everton
21:02:25MarcGuayAlexP: Nothing is easy. Your the Feb24th bootloader only works _sometimes_. The beast is earning it's name,
21:03:22 Join {-phoenix-} [0] (
21:03:35AlexPIt really is odd this :/
21:05:11JaykayMarcGuay: so i can just remove "{ ACTION_BMS_DELETE, BUTTON_REC, BUTTON_NONE },"?
21:06:20MarcGuayJaykay: You should probably also look for whatever is listening to that action. Somewhere in the bookmarking menu code itself.
21:06:41JdGordon|there was discussion of removing that button...
21:06:49JdGordon|I dont remember what the outcome was
21:07:02*JdGordon| cries and remembers he still wants to fix up bookmarking
21:07:22*LambdaCalculus37 doesn't seem to have m68k-elf-gcc in /usr/local/m68k-elf/bin on his Mac
21:07:41JdGordon|did the compile finish?
21:08:05LambdaCalculus37JdGordon|: Yeah, I was able to build successfully.
21:08:41 Quit {phoenix} (Success)
21:08:52LambdaCalculus37Actually, it seems I might be missing a few other m68k-elf binaries. :/
21:10:00 Quit perrikwp (" ajax IRC Client")
21:10:05LambdaCalculus37I don't have c++filt, cpp, gcc, gcc-4.0.3, gccbug, or gcov in my m68k-elf/bin directory.
21:10:14 Join {phoenix} [0] (
21:10:21*LambdaCalculus37 wonders why building the m68k tools is giving him so many issues
21:18:37JdGordon|is this the same problem we were getting? with the -r arg?
21:21:47 Quit {-phoenix-} (Success)
21:22:35CIA-38New commit by mcuelenaere (r21021): Forgot to remove this testing code..
21:22:40 Quit Thundercloud (Remote closed the connection)
21:23:25LambdaCalculus37JdGordon|: That was reverted by linuxstb IIRC.
21:24:11AlexPLambdaCalculus37: Only partially
21:24:44AlexPLambdaCalculus37: It was binutils that needed it I think, using it for gcc too broke m68k gcc compilation
21:24:52AlexPon linux anyway
21:24:57 Quit moos ("Rockbox rules the DAP world")
21:25:04AlexPand he tested it on Mac OS X
21:25:11AlexPI'm pretty sure
21:25:16LambdaCalculus37AlexP: Yes, according to;revision=20899
21:26:00LambdaCalculus37Great, so I have to either get the bins from someone or just rebuild on my Mac.
21:27:14 Join perrikwp [0] (i=4aa794a0@gateway/web/ajax/
21:27:18 Quit jon-kha (Read error: 145 (Connection timed out))
21:28:57*Horscht goes searching flyspray foor cool patches/addons
21:29:03Jaykayi get an error while compiling r21020 for the e200: in menu.c line 455 and 462
21:29:30Jaykayind function "do_menu": LANG_RESET_SETTING undeclared
21:29:30 Join arohtar [0] (
21:29:52Jaykayand 7 others
21:29:59linuxstbJaykay: Did you "make clean" ?
21:30:32gevaertsand was it clean after that?
21:31:32gevaertswhat did your build directory contain after make clean?
21:31:40 Join moredhel [0] (
21:32:38 Quit faemir (Read error: 104 (Connection reset by peer))
21:33:20Jaykayi just deleted the directory and created it again, now it works...
21:33:33***Saving seen data "./dancer.seen"
21:33:42 Join faemir [0] (
21:33:53Jaykayanyway, thanks for your help :)
21:34:19 Quit HellDragon (Read error: 104 (Connection reset by peer))
21:34:31 Join HellDragon [0] (
21:34:31 Quit arohtar (Connection reset by peer)
21:35:26 Join arohtar [0] (
21:36:54 Quit moredhel (Read error: 54 (Connection reset by peer))
21:37:36 Join moredhel [0] (
21:38:27 Quit faemir (Read error: 54 (Connection reset by peer))
21:40:29 Quit arohtar (Connection reset by peer)
21:41:43 Join Ubuntuxer [0] (
21:42:16CIA-38New commit by mcuelenaere (r21022): Lua: fix some issues with rocklib
21:42:45amiconngevaerts: What did you try to solve with r21016?
21:43:21 Quit mcuelenaere (Remote closed the connection)
21:43:47 Join wodz [0] (n=594b9d7e@gateway/web/cgi-irc/
21:44:19amiconnAh, hmm
21:45:31wodzhello, Is there someone experienced in interpreting xray PCB photos?
21:46:10wodzI'm trying to find some traces in MPIO HD200 PCB
21:47:34amiconngevaerts: The block in init() (for hardware usb) is known as "early usb". It is necessary to do it there on the Ondio in order to allow USB on power up with an MMC inserted
21:48:44amiconn(USB must be enabled before rockbox itself tries to access the MMC, because an MMC can only switched from MMC to SPI protocol but not back without power cycling)
21:49:11 Join hd [0] (
21:49:11amiconnIf this block doesn't work for software USB, the usb_detect() function is too slow
21:49:13 Quit hd (Read error: 104 (Connection reset by peer))
21:51:05 Quit thegeek (Read error: 104 (Connection reset by peer))
21:51:20gevaertsamiconn: The problem is that (sometimes?) on software USB, when booting with usb plugged in and the start screen not being the main menu, the usb connection fails. r21016 was clearly not the right thing to do. I'm now sprinkling logf()s in usb.c to see what exactly goes wrong
21:53:49 Quit Ubuntuxer ("Leaving.")
21:54:44amiconnI think that r21016 should solve the problem if you can make usb_detect() return true immediately after usb_start_monitoring() is called when usb is plugged
21:55:28 Quit _Auron_ ("Infinity repeatedly denies rumours of plotting with zero to bring down the Universe.")
21:56:01 Join thegeek [0] (
21:56:37 Quit wodz ("CGI:IRC (EOF)")
21:56:47 Join froggyman [0] (i=47ba0b80@gateway/web/ajax/
21:57:02 Quit LambdaCalculus37 (" ajax IRC Client")
21:57:26froggymanwhat is faster (for compiling) colinux or VMware?
21:57:37gevaertsthe arc driver is fully interrupt-driven, so usb_detect() should be fast enough
21:57:49JaykayMarcGuay: is enough for e200?
21:58:07gevaertsbut actually the problem seems to be that there's a thread that doesn't send SYS_USB_CONNECTED_ACK
21:59:39MarcGuayJaykay: What happens if you try to delete a bookmark from the context menu? Does it use ACTION_BMS_DELETE?
22:00:20 Join grndslm [0] (n=grndslm@
22:00:49Jaykayapparently yes - i just tried it and rockbox did nothing when i tried to delete it.
22:01:43Jaykayso if i just leave out the removal of the case it should work
22:01:56 Quit evilnick_7 (" ajax IRC Client")
22:01:59MarcGuayJaykay: I may have fed you a false lead earlier.
22:02:47MarcGuayIf you do "grep -r 'ACTION_BMS_DELETE' in the keymaps folder you can find all instances of it and deal with other targets as well, if you like.
22:04:34Horschton the wps, pressing the center button brings up the file browser. Is it possible to change that so it brings up the database instead?
22:08:00 Join antitrons [0] (i=antil33t@
22:08:06 Quit antil33t (Nick collision from services.)
22:09:20froggymanso is Colinux considerably faster than VMware? or is it not to noticible?
22:13:23JdGordon|Horscht: if you go into the wps from the database it will go back to the db i think
22:14:47JdGordon|froggyman: if your computer has hardware virtualization... vmware might be faster
22:15:06Horschtbut what if i didn't start the wps from the databse? If i start my player and simply resume. Pressing the center/select button still throws me to fileviewer mode, but I am a person who prefers database over fileview
22:15:21 Join p3tur [50] (n=petur@rockbox/developer/petur)
22:15:24JdGordon|then you'll need to fix the code
22:16:03JdGordon|and dont ever expect a setting patch to get accepted, because (STUPIDLY) there is too much oposition to any idea of customisable buttons...
22:16:21froggymanJdGordon|: so overall its slower than CoLinux, but a lot faster than Cygwin?
22:16:48JdGordon|I dont know.... the only way to know for sure is set all 3 up and do a trial one with each
22:18:04froggymanohh ok, well it says on the wiki that it is faster than cygwin; I'll try to get CoLinux setup as well and post some times for all 3
22:18:32amiconnWell, either SELECT (or whatever the browse button is on other targets) should go to the browser selected by the startup screen (if that tis set to a browser), or .playlist_control should also store which browser it was built from
22:19:11 Join firedix [0] (n=firedix@
22:19:42 Quit firedix (Read error: 104 (Connection reset by peer))
22:19:49amiconnThe latter could be difficult as you can add things to a playlist from either broswer
22:20:26Horschtok, then i'll try to find out where in the source this behaviour is located. I myself simply have no use for filebrowser
22:21:48 Quit itcheg (" ajax IRC Client")
22:21:59 Join itcheg [0] (i=41d59de2@gateway/web/ajax/
22:24:47JdGordon|Horscht: I would personally be in favour of having a setting like "primary track browser" which would handle the select button, and the order in the main menu.... but yeah, never going to happen... mever going to even get a decent discussion either
22:25:02Horschttoo bad
22:25:33gevaertsI don't see why a solution for this wouldn't be accepted
22:25:53JdGordon|customisable buttons
22:25:56JdGordon|customisable menus
22:26:17gevaertsI don't think this is either of those
22:26:49 Quit itcheg (Client Quit)
22:26:57JdGordon|then you are in the select minority :(
22:27:00AlexPI don't see the need to change the order in the menu though
22:27:23JdGordon|well, if you never use the file browser, why should it get the top spot?
22:27:52AlexPIt is a case of going one down the list - I don't see the benefit
22:28:07AlexPI wouldn't care if dataase were first even though I never use it
22:28:25pixelmaI wouldn't object to the - go back to the browse start screen solution
22:28:35JdGordon|quick access... i.e if you are in the middle of the list you know you can just scroll up or hold up (depending on target) and you will get there without looking
22:28:45AlexPAnyway, for the button it would be good to go back to where you came from, across reboots too
22:29:11JdGordon|pixelma: that wont work... what if your start screen is the WPS?
22:30:36pixelmaok, maybe the go back to whichever browser you came from is better. But isn't there still the problem that you can't go back to the right spot in the database?
22:31:10gevaertsas long as that isn't solved, going back to the database doesn't make that much sense
22:31:33pixelmabtw. the file browser doesn't have the full top spot ;)
22:31:44Horschtwhy not? I'd probably want to go somewhere else in the database anyways...
22:32:09JdGordon|pixelma: haha yes you are right... but who ever uses the bookmarks :)
22:32:30pixelmaI don't use them but still have them enabled
22:32:34AlexPRecent bookmarks is my top menu item :)
22:32:39AlexPI use them all the time
22:33:00 Join itcheg [0] (i=41d59de2@gateway/web/ajax/
22:33:19*gevaerts uses bootmarks :)
22:33:54gevaertsalso bookmarks actually
22:33:59JdGordon|Horscht: well, the point is that the best you could do sometimes with going back to the DB is being put in the top menu(?).. whereas with the filebrowser you will always get to the file
22:34:38gevaertsof course with the db there are several ways to get to a file, so which one do you use?
22:34:43Horschtyeah, but why would I want to get to the file? It's currently playing... I want to play some different file, most likely from a different album or even artist
22:35:10AlexPThat isn't true for everyone though
22:35:17JdGordon|that is another discussion entirely... thats why we have the follow playlist setting...
22:35:23AlexPand it is much quicker to back out than go back in
22:35:27gevaertsah, but you're talking about what select should do then, not about it not working for the database
22:36:02AlexPbut yes, that is follow playlist, which is a problem for the database
22:37:05*JdGordon| wishes the db browser could use a path based system so we could go back to a track/position
22:37:32Horschtgevaerts, yes. i am talking about select being able to go to the database when your start screen is the WPS. currently, it goes to the filebrowser showing me the root of the player
22:38:25JdGordon|I'm not really sure why it cant though... anrt all the labels unique in a screen? so we could do something like "\albums by artist\blaa\track name" ad have it run through the searches
22:39:02froggymanwith VMware, on the quick start guide ( ) it says to extract the debian image, but doesnt say where; so what directory should it get extracted to?
22:39:05gevaertswe'd have to store that info somewhere
22:39:42JdGordon|the path? I dbout thats the biggest issue
22:40:59amiconnJdGordon|: Well db paths are dynamic, so that would be difficult, if it is possible at all
22:41:28froggymanis it just "played" then?
22:42:39JdGordon|amiconn: sure, it would have to be stored as each filter gets slected (r howver it happens in the db terminology)
22:43:07 Part CaptainKwel
22:43:25*froggyman reads the guide again, and realises that he miss read it; and now understands'
22:43:34JdGordon|the point is that if you go A -> B -> C through the db browser, and then store it as /A/B/C then the code shuold be able to go and redo the stepping magically
22:44:48amiconnThe database also offers string search
22:45:33 Join flydutch [0] (
22:45:33*gevaerts thinks that the easiest solution is to drop the database ;)
22:45:52JdGordon| /A/%artist=blaa/C .... it *could* be done...
22:46:33JdGordon|well, the databsase is fine.. its the browser frontend for it which is difficult :)
22:48:47 Quit JdGordon| (" ajax IRC Client")
22:50:12 Quit dfkt_ ("-= SysReset 2.53=- Ph'nglui mglw'nafh Cthulhu R'lyeh wgah'nagl fhtagn.")
22:53:45gevaertswhere can I find the database ui code?
22:54:02amiconnWhat about things like playcount or rating? Those can change even if the database "folder" is restored
22:57:01gevaertsThe usb-on-boot issue seems to be caused by the SYS_USB_CONNECTED event on button_queue not being acked
22:57:31gevaertsso something somewhere is reading the button queue and not handling those
23:00:21Horschtso, in gwps.c i change both instances of GO_TO_PREVIOUS_BROWSER to what? GO_TO_DATABASE?
23:03:47Horschtline 486 and 854
23:04:58 Join earworm [0] (
23:05:49earwormhey guys... i was looking at this manual page:
23:06:16*amiconn thinks that the previous browser could be an nvram bit
23:06:22earwormhowever, i am unclear how should i name each of the covers to correspond with each album
23:06:46earwormsince i'd like to put all the covers in one folder
23:07:00Horscht# /.rockbox/albumart/artist-albumtitle.{jpeg,jpg,bmp}
23:07:05 Part high-rez
23:07:08Horschtwould be the one to use.
23:07:24*gevaerts isn't sure how to debug this further
23:07:49earwormi see, thanks for the hint
23:08:16Horschtyou're welcome
23:13:31*gevaerts found the bug
23:14:34gevaertsThe problem is in action_userabort() in apps/action.c, which throws away all events
23:15:53gevaertsWhen booting with a start screen that does this fairly soon, it means the event gets eaten
23:16:23amiconn(1) Why is this called there? (2) An action loop must not throw away events like usb connection (well, except for debug stuff...)
23:16:53gevaertsIt's called somewhere in the database screen startup
23:17:34 Quit p3tur (Remote closed the connection)
23:18:04 Quit itcheg (" ajax IRC Client")
23:18:29 Quit petur ("Zzzzz")
23:18:33gevaertsmy bet is apps/root_menu.c:128
23:19:15 Quit earworm ("ChatZilla 0.9.84 [Firefox 3.0.10/2009042316]")
23:19:18amiconnWhy doesn't that hit hwcodec?
23:20:27 Quit bmbl ("Woah!")
23:21:41gevaertshardware usb targets use the early usb bit
23:21:51 Join earworm [0] (
23:22:18amiconnYeah, but why doesn't that work on swcodec then?
23:22:40 Quit Xerion (Read error: 104 (Connection reset by peer))
23:23:06*gevaerts reproduces the bug in a non-boot context on his gigabeat F
23:23:08 Join dberg918 [0] (
23:24:02dberg918in the midst of building a new wps, i have stumbled upon a bug with the %mv tag
23:24:21gevaertsamiconn: if you make a recursive playlist of a big directory and plug in while it's busy, it won't connect either. Same bug
23:24:33dberg918I wanted to see if anyone knew about it before I went ahead and filed the bug report
23:25:35dberg918the bug occurs when you use a conditional %mv tag and use the timing tag (%t#), the parser reads it as if the conditional is left open
23:25:38Horschtok, what would I have to alter to change the behaviour of the center button on the wps screen to go to the root of the database instead?
23:26:38pixelmadberg918: you can't put the %t inside the conditional
23:26:56 Join mcuelenaere [0] (
23:27:01pixelmathat's valid for every tag by the way
23:27:30gevaertsamiconn: the ARC controller doesn't go to USB_INSERTED straight away. It starts out with USB_POWERED, and only gets to USB_INSERTED when it detects a real USB bus (i.e. not just a charger)
23:28:48dberg918is the %mv tag useless without the conditional brackets?
23:29:19dberg918or will the line timeout without them?
23:29:58pixelmadberg918: I'm not sure I understand what you want to do
23:31:02pixelma(and I admit, I haven't used %mv yet myself - just reading the CustomWPS page myself)
23:31:21amiconngevaerts: That means the logic is almost reverted compared to hw usb
23:31:59dberg918I was hoping to have a line at the bottom that would alternate between the track title and artist name every 2.5 seconds when the volume is changed, and then disappear after 7.5 seconds
23:33:14gevaertsit's pretty different, yes. This is part of the overhaul jhMikeS did to make usb and charging work together properly
23:33:36***Saving seen data "./dancer.seen"
23:33:46 Quit Jaykay (Read error: 110 (Connection timed out))
23:36:02pixelmadberg918: you can do it like this: "%t2.5%?mv<%ia|whatever you want to display otherwise>;%t2.5%?mv<%it|whatever you want to display otherwise>" , although you might notice if the "whatever you want to display otherwise" part is scrolling that it will reset every 2.5s. It's a bit of a workaround
23:36:46dberg918I'll give it a shot
23:36:56dberg918thanks for the tip, I had a feeling there had to be a way to get around it
23:37:51*pixelma wonders what the %mv tag does when used unconditionally
23:37:55 Quit flydutch ("/* empty */")
23:38:54 Join d3v14710n [0] (i=d3v14710@unaffiliated/d3v14710n)
23:39:46 Quit BXCracer (Remote closed the connection)
23:39:53dberg918from what I've tried pixelma, it does nothing
23:39:58dberg918nothing at all
23:40:52amiconngevaerts: I think this could be fixed by having yet another state
23:41:27gevaertsamiconn: something like "connected, but not in any specific use mode"?
23:42:16amiconnThe early usb loop exits when it find usb_powered mode, in order to avoid locking up when the user wants usb power right from the start
23:42:40dberg918pixelma: that worked! although I still can't use the scroll tag (%s) for some reason...
23:43:15amiconngevaerts: Yes, something like an initial state for usb connection, which would then switch either to connected state, or to power/charging state
23:43:51amiconnaction_userabort() needs fixing independently of that
23:44:56gevaertstrue. Early usb not working is not very important if it connects soon after anyway. Events getting lost can cause lots of interesting issues
23:45:26earwormhey Horscht, i tried your suggestion about naming the cover files in this manner # /.rockbox/albumart/artist-albumtitle.{jpeg,jpg,bmp} and it works. my question is, which way is more battery efficient, having all the covers in one folder, or placing them in individual directories of each album?
23:45:59pixelmadberg918: you should be able to but "resetting" every 2.5s might be a bit quick to see. One restriction I'm aware of with the scrolling is that things which change dynamically (like playtime or even volume when adjusted) don't work correctly on a scrolling line, track info should
23:46:14Horschti don't think that'll have a significant difference
23:46:53Horschtafaik there's might be slight difference between jpg and bmp, though
23:47:06dberg918pixelma: that's true...and I think I have enough space at the bottom that scrolling shouldn't be necessary
23:47:30dberg918now if only there was a way to get the song info to pop up when I change tracks...
23:47:32earwormjpgs are better?
23:47:43dberg918and timeout after 7.5 seconds
23:47:57HorschtI am not sure, but if at all, jpgs should be slower
23:48:28Horscht*jpgs could slightly lower battery life
23:48:51Horschtas they need to be decompressed and scaled, bmps only need to be scaled
23:49:42earwormHorscht: ah i see. :D so I take it, bmps are a preferred format ?
23:50:26amiconngevaerts: Well, early usb *is* important for (flashed) Ondios. It's the only way to conveniently connect usb on startup with an MMC plugged
23:50:27 Join itcheg [0] (i=62db4c46@gateway/web/ajax/
23:50:36pixelmadberg918: maybe you can get a similar effect with the %px tag - only specifying something for the beginning of the track. It will be percentage of the song though and not a given amount of time
23:50:51gevaertsamiconn: yes, but they don't suffer from this bug
23:50:52Horschtearworm, as I said, I don't know wether or not this is the case, but I realy doubt it will have any significant effect at all. It's a matter of choice, i'd say.
23:50:55 Quit JdGordon ("Leaving.")
23:51:09HorschtI myself prefer jpg, as i can use my existing folder.jpg art
23:51:24gevaertsI mean that I don't care too much whether early usb works for software usb
23:51:33amiconngevaerts: They do e.g. in the case you described
23:51:56earwormHorscht: good, that seems reasonable, so thanks again
23:51:58amiconn(whenever action_userabort() is active when plugging usb)
23:52:29 Join JdGordon [0] (
23:52:30 Quit {phoenix} (Read error: 104 (Connection reset by peer))
23:53:22pixelmagevaerts, amiconn: do I need a long list to try (only following a bit)?
23:55:30gevaertspixelma: no. It's pretty clear what's going on
23:56:12gevaertsit's not clear to me what the best way to fix action_userabort() is though
23:58:22CIA-38New commit by mcuelenaere (r21023): Lua: add actions.lua to the zip file

Previous day | Next day