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

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

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

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

#rockbox log for 2010-04-21

00:03:24CIA-5New commit by 03pixelma (r25692): Fix broken Iaudio M3 manual and missing buttons in the M:Robe 100 manual, caused by the same weirdness of tex: breaking a long opt list needs escaping ...
00:06:32pixelmaB4gder: let's see what happens with tomorrow's manuals
00:08:35 Quit robin0800 (Ping timeout: 248 seconds)
00:09:16 Quit efyx (Remote host closed the connection)
00:15:57 Quit DerPapst (Quit: Leaving.)
00:27:24 Quit domonoky (Read error: Connection reset by peer)
00:27:31 Quit liar (Ping timeout: 268 seconds)
00:31:07 Quit ender` (Quit: Female software engineers become sexually irresistible at the age of consent, and remain that way until about thirty minutes after clinical death. Longer if it's a warm day.-- Scott Adams)
00:31:26 Join b0hoon [0] (
00:32:09b0hoonpixelma: ping
00:33:01b0hoonso it looks like after my last commits...
00:34:03b0hoonpixelma: but it builded for me locally for 5 different targets
00:35:21pixelmayeah for me too, just the M3 manual didn't build of the ones I tried. It's still weird, we'll see what happens tomorrow
00:35:49b0hoonpixelma: i wasn't aware of this, should i change the way of breaking the lines or wait?
00:36:02b0hoonpixelma: ah, ok
00:37:07 Quit phanboy4 (Ping timeout: 265 seconds)
00:37:26pixelmaI think I caught all the problematic ones. It's most important if you break \opt{pad,pad,pad,pad}
00:41:22b0hoonok, i'll pay more attension to this next time...
00:41:28pixelmayou'll sometimes see the % (actually a comment that helps "commenting out" the linebrak) at the end of \opt{bla}{blub} too because tex *sometimes* adds a space for the targets where this is not true. Then you'll sometimes see more than one space in the resulting PDF for all targets that are not included. Sometimes it works correctly though and I didn't find a pattern yet
00:43:10b0hoonyes. i was wondering what for it is.
00:43:30pixelmab0hoon: nice to hear :)
00:43:49 Quit Kitar|st (Ping timeout: 260 seconds)
00:45:03 Join kerwood [0] (
00:45:23pixelmathere is a page in the wiki that explains some of the things, including these two points. It's called LatexGuidelines if I remember correctly
00:47:23b0hoonpixelma: because in some of the cases it was there and is some it wasn't
00:47:51 Join Kitar|st [0] (~Kitar_st@
00:48:15pixelmayeah, probably added when someone saw additional space in a specific line
00:52:03b0hoonok, loud and clear in Include/Exclude section on the wiki :/
00:52:46*b0hoon should RTFM first
00:53:29pixelmadon't be too hard on yourself, everyone has to learn and makes mistakes :)
00:54:24pixelmaespecially when you are new to it
00:56:09b0hoonyes, but...i've red it before, unfortunately without deep understanding, now i have to do it carefuly.
00:57:51pixelmayou'll keep it better in mind now ;) :)
00:58:01b0hoonyeah :D
01:00:09pixelmaand yes, tex isn't helpful there and sometimes does unexpected things
01:04:29 Quit MethoS- (Read error: Connection reset by peer)
01:06:33 Join MethoS- [0] (~clemens@
01:07:18 Quit MethoS- (Remote host closed the connection)
01:09:51b0hooni've spotted the lack of coma in the calculator.tex in line 17, but it was introduced before, so i left it during my changes.
01:10:10 Join Soap_Hotel [0] (
01:10:53***Saving seen data "./dancer.seen"
01:11:20b0hoonso it had to compile before...
01:11:53 Join MethoS- [0] (~clemens@
01:12:35b0hoonpixelma: GTG, thanks.
01:13:18pixelmanight, I'll have a look
01:13:31 Part b0hoon ("GTG. Bye.")
01:15:32 Quit xiainx (Quit: Good Bye)
01:20:05 Join xiainx [0] (
01:20:44 Quit Schmogel (Read error: Connection reset by peer)
01:28:43 Quit dfkt (Quit: -= SysReset 2.53=- Ph'nglui mglw'nafh Cthulhu R'lyeh wgah'nagl fhtagn.)
01:28:53 Quit Soap_Hotel (Quit: CGI:IRC (Ping timeout))
01:30:50 Quit MethoS- (Remote host closed the connection)
02:00:16 Join Strife89|PalmTX [0] (
02:04:54 Quit jd (Quit: Ω)
02:11:13 Join jd [0] (
02:11:13 Quit jd (Changing host)
02:11:13 Join jd [0] (~jd@Wikipedia/HellDragon)
02:13:43 Quit panni_ (Quit: ( :: NoNameScript 3.81 :: ))
02:13:55 Join panni_ [0] (
02:17:33 Join gusto [0] (~gusto@
02:20:32 Quit kugel (Remote host closed the connection)
02:21:48gustoso i read that you have support for 2nd generation of ipod nano, i have the 3rd ipod nano and i am now using linux and have problems w/ synchronizing, the files are there but ipod doesnt know them any more
02:22:27gustoi think staying w/ this firmware that needs to be indexed in order to know whats on it isnt the way to go
02:23:32gustowhat are the differences between 2nd and 3rd gen of ipod nano? so do you expect to take a long time till support of the 3rd gen ipod nano as well?
02:25:35gustowill i be able then to just copy some files over and be able to play it w/o any sync or special sw?
02:26:07gustoi mean just to mount the dev and copy some mp3s there and play it there
02:31:37 Join n1s [0] (~n1s@rockbox/developer/n1s)
02:31:47 Nick fxb is now known as fxb__ (
02:34:03gustook, i found this
02:35:45gustomaybe i could send my 3rd nano to you? some developers?
02:36:39 Join Strife89 [0] (
02:37:32 Join robin0800 [0] (
02:38:57 Join saratogalab [0] (~9803c20d@gateway/web/freenode/x-akchumzzebxnxhts)
02:41:06 Quit Strife89|PalmTX (Quit: My laptop started just in time, though. My Palm's battery needs charging.)
02:54:00 Quit n1s (Ping timeout: 276 seconds)
02:54:48 Quit komputes (Remote host closed the connection)
03:05:22 Join n1s [0] (~n1s@rockbox/developer/n1s)
03:09:54 Quit panni_ (Ping timeout: 248 seconds)
03:10:24hobbsUnhelpful: realized that you sent me something yesterday that I just pulled out of logs now. Yeah, I'm aware that voltage factors in as well. Although it already runs quite close to min "spec" voltage at 264.
03:10:56***Saving seen data "./dancer.seen"
03:14:39hobbsUnhelpful: and I realize I'm probably not going to be all that productive and none of my work is likely to be included but I just felt like seeing what I could do
03:16:19Unhelpfulbut i haven't said anything in this channel in a few days?
03:17:50 Quit RadicalR (Quit: Nettalk6 -
03:18:14 Join Antibuddha [0] (
03:18:35Antibuddhais the screen brightness difference between 5g and 5.5g ipod still there when rockboxed?
03:18:40hobbshmm, as it turns out you're right. That's what I get for having this thing log unix timestamps
03:19:27hobbsso I was actually answering something you said to me 10 months ago ;)
03:21:02 Join RadicalR [0] (
03:21:15 Join panni_ [0] (
03:22:17Unhelpfulhah... if you know about when maybe i could find it in the web logs?
03:22:40hobbsnah, the only missing context I need to provide is: Gigabeat S, frequency/voltage :)
03:23:16Unhelpfulah... i'v been putting off for a good while having a go at enabling the automatic scaling
03:23:27 Quit jd (Quit: Ω)
03:27:35hobbsyeah, I couldn't follow how to do that −− there seems to be a lot more info required than what's actually documented anywhere
03:27:39 Join jd [0] (
03:27:39 Quit jd (Changing host)
03:27:39 Join jd [0] (~jd@Wikipedia/HellDragon)
03:29:04Unhelpfuli'm not sure that's the case... but i haven't got it entirely figured out :/
03:30:20 Join fdinel [0] (
03:31:22hobbsso anyway, I was just using fixed setpoints for frequency/voltage. Got it flipping between 132 and 264 nicely enough, but 66 gives problems
03:32:18saratogalab33 MHz or even lower would be best on the beast for unboosted
03:33:30 Quit n1s (Ping timeout: 265 seconds)
03:34:27hobbsyeah, I realize the bench says it can keep up with MP3 around 30
03:44:16hobbsanyway, I shall hack more. And like I say, my work will probably get stepped on completely, but if that's the case I won't be sad
03:46:26 Join n1s [0] (~n1s@rockbox/developer/n1s)
03:47:40Unhelpfulit doesn't have to, if you get it to save some runtime ;)
03:48:21Unhelpfuli believe auto voltage scaling can be enabled separately... so we could just enable auto voltage and tweak frequency manually.
03:48:51Unhelpfuland somebody else must've told you some of it because it's news to me that we're nearly at minimum voltage at 264MHz
03:49:43hobbsUnhelpful: patch from jethead71 from a week or so ago, seeing it in the changelog was what got me thinking
03:49:56 Quit adnyxo (Ping timeout: 245 seconds)
03:50:00hobbsdropped it down from whatever unreasonable thing it boots up with (1.6 or 1.65?) to 1.35
03:51:07hobbsbut I agree, auto-voltage and manual CPU would be a good trick
03:54:13n1sUnhelpful: i think it's near the minimum for that freq
03:54:37n1slower freq could use even lower voltage IIUC
03:56:29n1sgusto: Ports are done by interested developers, and noone seems to be working on the nano3g at the moment, so sending someone your ipod might work if you get them interested in the port but the effort will likely take a long time
03:58:06LloreanI thought jhMikeS said that what we're setting it to is good for up to 399mhz still
03:58:33n1shobbs: have you measured any runtime differences?
03:59:07guston1s, ok
03:59:23 Quit Antibuddha (Quit: ChatZilla 0.9.86 [Firefox 3.6.3/20100401080539])
03:59:23hobbsn1s: not yet. Honestly I'm not sure what shape my battery is in either ;)
03:59:26guston1s, so, i have to get around by myself
03:59:57n1sgusto: looks like it at the moment, unless you can get other people interested
04:00:26hobbsn1s: I did a run last night, at 132MHz & 1.25V, runtime was 5h10m, but I didn't do a baseline run ;)
04:00:47hobbsI need to get lower freqs working if I can :)
04:04:37 Quit dys (Ping timeout: 276 seconds)
04:05:02 Join dys [0] (
04:05:56n1swow, that is a pretty bad time
04:06:28hobbsit lives in my car
04:06:32hobbsand it was used when I bought it :)
04:06:33n1si think i got 6-7h with the much higher voltage and 264MHz when i last ran abench
04:06:44n1smine was used too
04:06:50hobbsI think a baseline run would be just as bad
04:07:58n1sprobably, but the really interesting thing is mprovement :)
04:08:20hobbsso I'll go back to stock and do one tonight and make sure that I didn't make things worse :)
04:08:41 Join anewuser [0] (anewuser@unaffiliated/anewuser)
04:09:13guston1s, and does it sound promising? i mean, can i remove the apple firmware after completely and use like described on the site (drag and drop files and play on the move)?
04:10:22n1sif rockbox is ported, you can drag and drop like to any other usb mass storage device
04:10:41gustoand what about the apple FW?
04:10:45n1sif you cna get rid of the original firmware completely is a detail of the port
04:10:57gustothat means?
04:11:15RadicalRBasically, the FW becomes optional.
04:11:18n1sso i can't tell you that now, it depends on how the bootloader etc is designed
04:11:25RadicalRThat too.
04:11:35RadicalRHowever, given the pattern of the Nano2g
04:11:53n1ssome of our ports have optional dualboot with the OF, some mandatory and some have only single boot
04:11:53RadicalRI think it's safe to say that it'll probably follow suit.
04:12:08n1syes, probably
04:12:48gustoand on the nano2g you can get rid of the original fw completely?
04:14:36RadicalRYes, I just keep it on just in case.
04:14:49RadicalRWell, hold on. Let me double check that.
04:14:59RadicalRIt goes to the iLoader
04:15:06RadicalRwhich allows you to pick what you want to boot.
04:15:33 Join Rob2223 [0] (
04:15:36gustoso you can delete the first partition where the fw is then, right?
04:16:30gustoand other thing that am interested in pretty much about this topic is what filesystems can you use then? only FAT32 or also ext2 and other?
04:16:52n1sonly fat
04:16:57RadicalRNix on the first one, and rockbox can only read fat
04:17:52gustois there a practical reason, or is it just because there is not enough ppl to write code for it?
04:18:54scorcherockbox supports FAT32 as well...
04:19:05 Quit Rob2222 (Ping timeout: 265 seconds)
04:19:17scorchegusto: the reasoning is mentioned here:
04:19:26gustobecause i think on such big storages like nano3g has with 8 GB and newer ones have even more a better FS would show up
04:19:52scorcheit isnt about "better" is about "compatibility"
04:21:35gustoah on that site the reason is, because these devices are too weak, have less RAM and other ressourcen, maybe that isnt true for modern ones any more, or will not be
04:21:50gustoyes, compatibility as well
04:22:12gustobut not everybody cares about compatibility, the oroginal firmware cares the less
04:23:00scorcheram is always a major concern...the more RAM we have, the more audio buffer can fit into there...the more audio buffer, the less we have to spin up or access the disk/storage...the less we have to access, the more battery life...
04:23:13gustoforcin ppl to use itunes to synchronize their ipod isnt exactly copmatibility improving
04:23:50gustoi understand that
04:24:38gustois on routers the same
04:25:05gustothe more free ram you have, the more nat connections you can route through
04:30:16n1sthere also is not a big gain in a "better" fs for most people, as fat* works pretty well
04:32:12LloreanThere's really not a very strong reason for moving them to more 'advanced' filesystems anyway. Most of the better features don't relate to the use of it as a portable music player, while FAT32 as a filesystem is pretty much understood by any modern host computer making Rockbox as a UMS device pretty much universally compatible (or as close as you really can be)
04:34:13 Quit fdinel (Quit: Miranda IM! Smaller, Faster, Easier.
04:36:00 Quit n1s (Ping timeout: 265 seconds)
04:38:19hobbseven when it comes to filesystem writes, FAT32 is less likely to trash things in a crash than FAT16 was :)
04:38:34hobbsand people probably wouldn't want to pay the price for journalling on a portable device anyway
04:45:26 Quit Barahir_ (Ping timeout: 260 seconds)
04:47:03 Join Barahir [0] (
04:47:25 Join n1s [0] (~n1s@rockbox/developer/n1s)
04:48:08 Join CGL [0] (~CGL@
04:52:32Lloreanhobbs: Considering a large amount of the content on a Rockbox device (especially large storage ones) is written once, anyway.
04:59:06 Quit robin0800 (Remote host closed the connection)
04:59:28 Quit Strife89 (Quit: Moving to my Palm.)
05:00:02 Join dys` [0] (
05:00:40 Join phanboy4 [0] (
05:02:28 Quit dys (Ping timeout: 276 seconds)
05:04:13 Join BHSPitMonkey [0] (~stephen@unaffiliated/bhspitmonkey)
05:05:01 Quit CGL (Remote host closed the connection)
05:11:00***Saving seen data "./dancer.seen"
05:17:11 Quit n1s (Ping timeout: 276 seconds)
05:19:12 Join n1s [0] (~n1s@rockbox/developer/n1s)
05:31:42 Quit xiainx (Quit: Good Bye!)
05:31:49 Quit Horscht (Quit: Verlassend)
05:36:15 Join CGL [0] (~CGL@
06:01:19hobbsLlorean: right, and even if it happens that you nuke the database or your scrobbler log... oh well. The way FAT works it's exceptionally unlikely to take your music with it :)
06:01:32 Quit n1s (Ping timeout: 264 seconds)
06:12:02 Quit FlynDice (Quit: Time to go Fly....!)
06:12:21 Join FlynDice [0] (
06:13:02 Join CaptainKewl [0] (
06:13:33 Quit panni_ (Read error: Connection reset by peer)
06:15:05 Join n1s [0] (~n1s@rockbox/developer/n1s)
06:15:09 Quit FlynDice (Client Quit)
06:15:29 Join FlynDice [0] (
06:16:19 Join xiainx [0] (
06:34:40 Join LinusN [0] (~linus@rockbox/developer/LinusN)
06:46:53 Quit anewuser (Ping timeout: 268 seconds)
06:48:23 Quit n1s (Ping timeout: 246 seconds)
07:05:00 Quit CaptainKewl (Remote host closed the connection)
07:10:03 Join RandomInsano [0] (
07:11:04***Saving seen data "./dancer.seen"
07:14:49 Quit detaos (Ping timeout: 276 seconds)
07:17:55 Quit RandomInsano (Quit: CGI:IRC (EOF))
07:18:18 Join RandomInsano [0] (
07:20:25RandomInsanoI'm creating a build target for the irivier x20. They use the Telechips 8200 chip with an arm946es core. Am I fine specifying arm946cc for the core type?
07:21:35RandomInsanoHmm... all cores seem to end in cc. I'm going to assume I'm good :P
07:22:44pixelmaall daily manuals are there again. The one manual commit yesterday fixed breakage that only affected the M3 manual, yesterday all manuals "before" the M3 one in the table existed (including the D2), the M3 and everything behind it showed the previous day's manual. This is very different to what usually happened, so while I'm glad it's fixed I'm also curious what exactly happened
07:23:07saratogalabRandomInsano: I think for now all that matters is that you get the basic arm instruction set right, so that should be fine
07:24:02RandomInsanosaratogalab: thanks for the advice.
07:24:06 Join detaos [0] (
07:25:05 Quit RandomInsano (Quit: CGI:IRC)
07:25:06 Join RandomInsano1 [0] (
07:25:52 Quit n17ikh (Ping timeout: 240 seconds)
07:30:59 Join n17ikh [0] (
07:33:36 Join n1s [0] (~n1s@rockbox/developer/n1s)
07:33:51CIA-5New commit by 03pixelma (r25693): Add missing comma (pointed out by Szymon Dziok). Without it the MROBE100_PAD was ignored and the manual didn't mention the 'Up' and 'Down' buttons in ...
07:45:53 Join einhirn [0] (
07:46:47 Join esperegu [0] (
07:47:12 Quit detaos (Read error: Operation timed out)
07:47:54 Quit esperegu (Remote host closed the connection)
07:48:25 Quit Farthen (Remote host closed the connection)
07:49:27RandomInsano1Anyone in the room have experience porting to Telechips-based devices? I'm a little stumped. Do I need to define special stuff for the TCC8200 or can I piggyback off a different chip's config for now?
07:49:53RandomInsano1I'm editing firmware/export/config.h
07:53:33 Quit einhirn (Read error: Connection reset by peer)
07:55:17JdGordonRandomInsano1: depends on how different it is to another chip
07:55:29RandomInsano1Drat. And I have no idea.
07:56:41JdGordonhave you got datasheets?
07:57:05RandomInsano1Telechips locked them out awhile ago. There was a forum post on where to find them, but the links are dead
07:57:08 Join Farthen [0] (
07:58:41 Join mikroflops [0] (
07:59:20RandomInsano1I'll do some digging.
08:00:01*pixelma just remembered something
08:00:25 Quit n17ikh (Ping timeout: 265 seconds)
08:00:47pixelmaB4gder: could it be that the build script for the daily manuals is very simple and does not advance if building one fails?
08:00:53 Join naag [0] (~harish@
08:01:28 Part naag
08:01:36 Join Buschel [0] (
08:02:12 Quit mikroflops_ (Ping timeout: 240 seconds)
08:02:36 Join n17ikh [0] (
08:03:10pixelmaI just remembered that there is a non-stop mode for tex which means "go ahead even when there are errors" and I think now that this was used before but isn't anymore (maybe on purpose). If the script is this simple it was no problem when compiling with non-stop mode but would be now. Unfortunately I don't remember how the mode is called exactly as I don't use it
08:04:36pixelmaI would also prefer it if it's not used anymore because you won't get semi-broken manuals but then the script would need to be a bit more clever
08:05:03RandomInsano1At least the datasheets exist somewhere :(
08:06:24JdGordonI'm sure *someone* has them backed up somewhere...
08:07:14RandomInsano1Oh, I'm sure they do. Maybe I'll PM other Telechips porters
08:07:50JdGordonyeah, pm shotofadds would be a good bet
08:09:12 Quit Lynx_ (Quit: Want to be different? Try HydraIRC -> <-)
08:11:11 Quit n1s (Quit: Lmnar)
08:17:59 Quit CGL (Quit: Saliendo)
08:20:31 Join Bagder [0] (~daniel@rockbox/developer/bagder)
08:21:45 Join Zagor [0] (~bjst@rockbox/developer/Zagor)
08:25:41 Join ender` [0] (
08:35:28 Quit avn (Ping timeout: 245 seconds)
08:37:08 Quit saratogalab (Quit: Page closed)
08:37:20 Join avn [0] (~avn@
08:39:08 Join esperegu [0] (
08:39:45linuxstbRandomInsano1: Which target are you working with? The first thing to do is to get tcctool working, so you can upload and test your own code safely.
08:43:31 Quit FOAD (Ping timeout: 264 seconds)
08:43:56 Join flydutch [0] (
08:44:09 Join FOAD [0] (
08:44:54linuxstbRandomInsano1: This forum page seems to have some interesting links -
08:46:03 Quit Buschel ()
08:48:30 Quit amiconn (Disconnected by services)
08:48:32 Join amiconn_ [0] (quassel@rockbox/developer/amiconn)
08:48:35 Join einhirn [0] (
08:48:54 Nick amiconn_ is now known as amiconn (quassel@rockbox/developer/amiconn)
08:50:00 Quit pixelma (Disconnected by services)
08:50:03 Join pixelma_ [0] (quassel@rockbox/staff/pixelma)
08:50:25 Nick pixelma_ is now known as pixelma (quassel@rockbox/staff/pixelma)
08:51:24 Join petur [0] (~petur@rockbox/developer/petur)
08:53:02RandomInsano1linuxstb: tcctool is compiled, have some firmware images. I just need to open one of my devices and configure it for usb boot mode
08:53:48linuxstbRandomInsano1: A lot of devices enter usb boot mode via a button combination - yours don't have that?
08:53:50RandomInsano1As far as target, TCC8200?
08:53:58linuxstbI meant which MP3 player?
08:54:31RandomInsano1Insignia NS-DV2G specifically. There's a solder jumper on the board.
08:55:28RandomInsano1It looks to be nearly identical to the iriver x20
08:55:41linuxstbThe main problem with the telechips targets is the software flash translation layer (FTL). We don't have a reliable read-write driver for that (and it is also slightly different on different telechips devices).
08:55:48RandomInsano1Hardware wise at least. Boards are layed out differently
08:56:09RandomInsano1I have a MicroSD at least
08:57:02RandomInsano1I'm interested in getting as far as I can with it. When I get bored or hit an impossible wall, I'll give up.
08:57:17linuxstbThat's what we all do ;)
08:57:36 Nick dys` is now known as dys (
08:58:30RandomInsano1It's kind of interesting, someone who owned an iriver was able to use my firmware on it nearly perfectly, but without sound. Also, the images if the development board show a lot of the same chips as the final product.
09:01:23RandomInsano1Is the CPU_FREQ define in decimal? I'm stealing the Cowan D2 config file and it's set to 48000000
09:03:35linuxstbYes, but that define may not actually be used to set the CPU frequency - it depends on the driver. It's used to display the CPU speed in the debug menus. (IIRC). So don't worry too much about it for now.
09:03:58RandomInsano148000000 it is then.
09:09:11*pixelma wants the "simplify PLA" patch in - :\\
09:09:23linuxstbRandomInsano1: If I was you I would start by disassembling the original firmware, and find the code that does the initial hardware setup - it should be the first functions being called. You can then compare that to the Cowon D2 code (7800) and see how similar the 8200 is.
09:10:36RandomInsano1I don't know if my one Comp Sci class on assembly language has given me a firm grasp of things, but I'll give it a shot.
09:11:07***Saving seen data "./dancer.seen"
09:12:46 Quit dys (Remote host closed the connection)
09:13:32linuxstbGoogle for the "arm quick reference guide" - that's all I used to understand ARM.
09:14:16 Quit avn (Ping timeout: 248 seconds)
09:14:57 Join lpereira [0] (
09:15:51Zagorpixelma: I'll take a look at the manual build script
09:16:15 Join avn [0] (~avn@
09:16:26amiconnhobbs: FAT16 and FAT32 work practically the same; there are only two differences: (1) the size of the FAT entries. (2) FAT16 has a dedicated (and size limited) root dir, while on FAT32 it's an ordinary directory
09:16:58amiconnRockbox also supports FAT16 on devices where it's needed (those supporting memory cards and/or having small internal storage)
09:18:32 Join DerPapst [0] (
09:20:54 Quit avn (Ping timeout: 276 seconds)
09:21:30 Join efyx [0] (
09:21:31 Join avn [0] (~avn@
09:21:39Zagorpixelma: btw the daily build scripts are available in
09:22:07Zagorthe manual script is quite stupid but doesn't care about the result of each build. it always loops through every one.
09:22:41pixelmathen I still don't understand what happened yesterday :( but thanks for looking
09:34:18 Join TheSeven [0] (TheSeven@rockbox/developer/TheSeven)
09:36:52 Join pamaury [0] (~pamaury@rockbox/developer/pamaury)
09:41:56 Quit avn (Ping timeout: 276 seconds)
09:43:54 Quit RandomInsano1 (Quit: Leaving.)
09:44:40 Quit slck (Ping timeout: 248 seconds)
09:44:46 Quit arbingordon (Quit: `)
09:45:28 Join slck [0] (Gforce@Slackware.SlackPix.Com)
09:48:12 Join avn [0] (~avn@
09:51:55 Join mitk [0] (~mitk@
09:54:27 Quit phanboy4 (Quit: Leaving)
10:00:52 Quit avn (Ping timeout: 260 seconds)
10:02:42 Join avn [0] (~avn@
10:05:44 Part LinusN
10:06:09 Quit Battousai (Read error: Operation timed out)
10:07:04 Quit slck (Ping timeout: 248 seconds)
10:08:46 Join Battousai [0] (~bryan@gentoo/developer/battousai)
10:09:36 Quit parafin (Ping timeout: 276 seconds)
10:10:14 Quit ps-auxw (Remote host closed the connection)
10:10:23 Quit chaos (Ping timeout: 268 seconds)
10:10:32 Join n1s [0] (~n1s@rockbox/developer/n1s)
10:13:43 Join parafin [0] (
10:14:14 Join kugel [0] (~kugel@rockbox/developer/kugel)
10:20:02 Join ps-auxw [0] (
10:21:57 Quit n1s (Ping timeout: 276 seconds)
10:24:21 Join funman [0] (~fun@rockbox/developer/funman)
10:25:59funmanranma: re: power usage on c200v2, can you dump the ascodec registers when running OF with your JTAG setup ?
10:27:03 Join chaos [0] (~chaos@gentoo/user/ch4os)
10:39:39 Join Lynx_ [0] (~bayer@
10:49:33 Quit jfc^2 (Ping timeout: 258 seconds)
10:52:27 Join mt [0] (~mtee@rockbox/developer/mt)
10:52:52 Quit BHSPitMonkey (Remote host closed the connection)
11:02:36 Quit antil33t (Read error: Connection reset by peer)
11:07:42 Join antil33t [0] (
11:07:58ranmafunman: Hmm, maybe. If the i2c and ascodec are still clocked when the processor is stopped it shouldn't be that difficult
11:09:05funmani hope we can find why there is so much difference between OF and us
11:11:08***Saving seen data "./dancer.seen"
11:14:31ranmaUh, but not today. I soldered connectors into the JTAG setup so the player is more usable as an mp3 player, but apparently one of the cables came of, so that will have to wait until I can fix that...
11:15:38funmani'm getting in touch with a good solderer so i can use JTAG on Clipv2
11:21:05 Quit petur (Quit: reboot)
11:24:26 Join petur [0] (~petur@rockbox/developer/petur)
11:30:42 Quit TheSeven (Ping timeout: 260 seconds)
11:34:04 Join |DaMaGeD| [0] (~dam@
11:35:29 Quit |DaMaGeD| (Client Quit)
11:41:14hobbshumma humma humma
11:51:09kugelpixelma: I saw your wish and resyncd the pla rework patch
12:07:02 Join wodz [0] (
12:08:25JdGordonkugel: speaking of old patches.. whats the story with resizing skin buf?
12:08:29 Nick fxb__ is now known as fxb (
12:08:39kugelit's rotting..
12:08:56JdGordonwhat else is needed for it?
12:09:09kugelI don't really like the patch. it has to work with that strange buffer layout which makes it a hack in many places and very error prone
12:10:16JdGordonyou know that is very simple to change if the logic to check buf size requirements on images is there right?
12:11:00kugelthere's also the tokens array
12:12:34jhMikeSwodz: any luck? I have another thought. perhaps starting a new AD conversion on its own int is producing an int flood? a tick task may be more appropriate.
12:12:48JdGordonalso, why do you keep saying it is error prone?
12:13:30 Join n1s [0] (~n1s@rockbox/developer/n1s)
12:13:55kugelI've seen so many strange things when working on it, and every tag needs to be reconsidered whether to load it immediately or in the second run, and the logic to decide whether a theme fits isn't very easy too
12:14:38wodzjhMikeS: no luck
12:15:07hobbshey look at that
12:15:42wodzjhMikeS: The problem is ADC conversion time. It takes 4096 adc clock ticks to complete. adc clock can be at best Fcpu/2 and only in theory
12:16:16 Join M3DLG [0] (~M3DLG@
12:16:38wodzjhMikeS: What I did is setup ADC clock such as ADC interrupt is fired with frequency ~400Hz. This gives ~100Hz update frequency per channel
12:18:42jhMikeSwodz: hmmm. I might suggest building without HAVE_PRIORITY_SCHEDULING, just to build things as simply as possible. There *could* be something and it should be found out if so.
12:18:55 Join LinusN [0] (~linus@rockbox/developer/LinusN)
12:18:55kugelJdGordon: I never actually had a version which worked correctly
12:19:05JdGordon:) ok
12:19:47wodzjhMikeS: Ok I give it a try
12:21:19wodzto comment out huge block of code it is #if 1 #endif or #if 0 #endif ?
12:21:26jhMikeSnoone in particular: Is something up with "Repeat All"? It seems that setting that repeats only one directory if starting playback from the file, not all of them, which it seemed to do before.
12:22:16jhMikeSwodz: #if 0. go to system.h and "0" the HAVE_PRIORITY_SCHEDULING stuff
12:23:27jhMikeSummm...wait, lol, where the heck was that #define used
12:23:53jhMikeSah, config.h
12:24:51wodzwhat about HAVE_SCHEDULER_BOOSTCTRL
12:25:02wodzcomment it out also?
12:25:55jhMikeSsure, shouldn't hurt
12:26:09*wodz commenting out and building
12:27:02 Join dfkt [0] (~dfkt@unaffiliated/dfkt)
12:29:57n1spixelma, Zagor : it would be quite nice if the manual build script could post its output somewhere in case of an error, something like the regular build table
12:33:10kugeldidn't we want to add it to the build system at some point?
12:33:21wodzjhMikeS: I still observe the delay
12:36:00n1skugel: hmm, yeah, that might be simpler and more elegant, just adding it as a separate kind of build for each target
12:36:20n1sand getting build server people to install the tex stuff
12:37:13*n1s wonders what Bagder thinks about this
12:38:16Bagderwe always wanted to get that added, so yes it'd be nice
12:41:39hobbsjhMikeS: although I admit it probably counts as "barking up the wrong tree" territory, I've got a gigS sitting in front of me doing CPU scaling
12:41:59jhMikeShobbs: me too
12:42:45hobbswhat a coincidence ;)
12:42:47jhMikeShobbs: throught the DVFS system or straight register setting?
12:42:57hobbsthe latter
12:45:11jhMikeSI've got voltage scaling, confirmed at the test points by a multimeter. The DPTC, there's latency issues with updating regulator settings. I may commit without that part enabled which would fix it at 1.6V for 528MHz and 1.35V for lower frequencies.
12:45:33 Quit n1s (Ping timeout: 265 seconds)
12:46:10wodzjhMikeS: Play button is plain GPIO in MPIO. I might change buttons mapping play->select and comment out whole ADC stuff. This way we can eliminate (or confirm) that problem origin is ADC interrupt
12:46:29hobbsjhMikeS: what range of speeds?
12:47:50jhMikeShobbs: 528, 264, 132, divider only. There is a bug with PLL switching, fixed in r2.0 that forces DVFS to reset the PLL. Perhaps because audio uses the USBPLL, 399 may be workable since that requires using SRPLL.
12:49:53hobbswhat about dropping the PLL once at boot and using 66-264?
12:50:33hobbsthat's essentially where I've ended up with my manual switching
12:50:51 Join n1s [0] (~n1s@rockbox/developer/n1s)
12:53:09 Quit M3DLG (Ping timeout: 240 seconds)
12:54:19 Quit kugel (Ping timeout: 276 seconds)
12:54:38jhMikeS264 won't always cut it for everything. the automatic switch works pretty well and is basically delay-free. we could divide further for 66, but there's other warnings in the errata manual about that. Also, changing PLL changes the SDMA core frequency.
12:55:46jhMikeSof course, the issues are clocking-relative issues, not absolute frequency ones
12:57:19hobbsyeah, I didn't have luck with 528/8 = 66, but I did with 264/4 = 66. And... well, let's say I haven't run into the stuff that needs more than 264 yet.
13:00:10wodzjhMikeS: I commented out whole stuff related to ADC and this doesn't remove delay between selecting menu item and entering that menu item
13:00:42jhMikeShobbs: 528/8=66 will set it to CPU=AHB, which is what the bug is about (iirc). amiconn uses it at 528 with APE.
13:03:10hobbsbleh ;)
13:03:38hobbsoh well, always possible that 66 is no big win I suppose
13:03:43jhMikeSat least with the DVFS you just set PMIC pins to select a different reg. setting. no SPI writes are needed.
13:05:33jhMikeSit benefits from voltage reduction more than frequency. I tested at 528 and 264 at 1.6V and got the same batt life. in both cases. simply changing F seems to just make it take longer/short roughtly in proportion before reaching an idle state and everything's been done to keep the ARM core as idle as possible.
13:07:29*jhMikeS , besides complaints about the S with OF, noticed people not being too impressed with Zune battery life either, with a 800mAh batt no less.
13:09:08hobbsmine has a badly-abused battery too ;)
13:11:10***Saving seen data "./dancer.seen"
13:11:34n1sjhMikeS: how much lower voltage can we use if we lower the freq lower than 132MHz?
13:14:12 Join kugel [0] (~kugel@rockbox/developer/kugel)
13:16:40jhMikeSn1s: I haven't seen anything saying it should operate less than 1.35V, ever, despite the datasheet saying the min is 1.22V (that low definitely doesn't work :).
13:17:28jhMikeSfreescales BSP never sets it below 1.35V, unless it's rev2, but S seems to be rev 1.15 all around
13:17:40 Quit shaggy-h (Ping timeout: 240 seconds)
13:17:57n1sah, so there would be little to no gain in going lower then
13:19:00hobbsnow you're making me think my voltage setting isn't
13:19:07jhMikeSn1s: Perhaps it could go a touch lower running so slowly. FS never does it though.
13:20:00n1sjhMikeS: do you know what the OF does?
13:20:13jhMikeShobbs: only 528MHz (aka. "turbo" and "overdrive") needs DPTC to find the minimum. <= 400MHz only requires 1.35V.
13:21:01hobbsright, but I'm running lower. Unless I'm not because I did something wrong there
13:23:01jhMikeSn1s: It leaves SW1ADVS set at 1.4V. I haven't tracked that part down in the driver. It's sort of a mess to wade throught. I'll try to find out.
13:23:19pixelmajhMikeS: about your repeat question. To me it doesn't sound like it has to do with repeat, more like playlist settings. Did you check your "recursively insert directory" settings or maybe you had auto-change directory set before or something?
13:23:53hobbsI suppose I can try changing all the values to 0.9V and make sure I get a hang. That should be fairly harmless ;)
13:24:04jhMikeSpixelma: I have auto-change set for sure. I'll check that other setting but I never touched it myself.
13:24:18 Join JohannesSM64 [0] (
13:24:37 Quit n1s (Ping timeout: 268 seconds)
13:24:41jhMikeShobbs: it will hang way before .9V :)
13:25:29jhMikeSpixelma: recusive is ON
13:25:58hobbsjhMikeS: except it didn't, which is why you've got me suspecting my code
13:26:25wodzjhMike: how can I artificially force thread to run in next context switch?
13:26:36jhMikeShobbs: I would too. It won't run.
13:27:14 Join n1s [0] (~n1s@rockbox/developer/n1s)
13:27:15wodzjhMikeS: I mean if I do breake in debugger what should I set?
13:27:28jhMikeSwodz: make the prority 1, no other threads should have that, otherwise it's automatically selected.
13:28:25wodzjhMikeS: so set threds[n].priority = 1? where n is my thread number
13:28:26jhMikeSwodz: well, it's still automatic at "1", but it will select that and not switch away until it sleeps
13:29:03jhMikeSwodz: don't set those directly or the scheduler won't update itself, use thread_set_priority.
13:30:04wodzjhMikeS: thread will be picked up even if in blocked state?
13:31:15hobbsyep, obviously that bit isn't doing its job :(
13:31:16jhMikeSwodz: it's ignored if blocked but it will run next if unblocked
13:32:07wodzjhMikeS: next question how can I force thread ublock
13:32:29jhMikeShobbs: I think 1.3V would be pushing it. I tried it here without a problem, but still.
13:32:53jhMikeSwodz: if it's waiting for a button, post to the queue
13:33:20wodzok time to poke around
13:34:06 Join shaggy-h [0] (
13:34:12kugelcan't you force a switch_thread() call with gdb?
13:35:24wodzI can but the thread is in blocking state and gets ignored
13:36:50jhMikeSwodz: you want the UI thread to guarantee it runs?
13:37:03 Join TheSeven [0] (TheSeven@rockbox/developer/TheSeven)
13:37:15kugelwodz: what thread?
13:37:39jhMikeSthe main one, that normall handles the use input ??
13:37:55 Join robin0800 [0] (
13:38:51jhMikeSwodz: thread.c, lines 3102-3106, change PRIORITY_USER_INTERFACE to "1". That's were the main thread is "created".
13:40:26wodzit hangs because button_queue is empty
13:40:53jhMikeSwodz: no buttons are getting posted?
13:41:00wodzthat is strange
13:41:24wodzBUTTON_NONE should be posted
13:45:36kugelif your button_read_device returns that if no button was pressed everything should be alright
13:45:40jhMikeSwodz: did you build with THREAD_EXTRA_CHECK set to 1 and KERNEL_OBJECT_CHECKS set to 1 (in their respective .c files)?
13:46:37jhMikeSkugel: I made inquiries along these lines yesterday, to make sure the button variable is showing the right buttons. :D
13:47:45*jhMikeS also could have imagined all that and wouldn't be the wiser
13:49:01wodzjhMikeS: ok I'll make build with this settings
13:49:15wodzwait a sec
13:50:06jhMikeSbtw, it's "THREAD_EXTRA_CHECKS" (misspelled). If you get panics, it's hozed for some reason.
13:51:10 Quit n1s (Ping timeout: 240 seconds)
13:51:34*wodz uploading to the device
13:52:28 Quit mc2739 (Ping timeout: 240 seconds)
13:56:38 Join n1s [0] (~n1s@rockbox/developer/n1s)
13:58:55wodzI still have delays and no errors/panics etc
14:00:01jhMikeSare buttons getting posted normally? it really IS reading them and posting them? or, does the button_queue back up with messages?
14:00:43wodzI can navigate with buttons so I guess button events are passed to the queue
14:01:22jhMikeSwodz: but, the delay. I thought you couldn't navigate.
14:01:29jhMikeSnot easily anyway
14:01:45wodzyes navigating is weird but possible
14:02:22jhMikeSbut every single press should post a message, but maybe they're not and the reading is shoddy?
14:02:33kugelup and down works fine, only selecting menu items doesn't?
14:03:03jhMikeSmaybe the ADC thresholds aren't quite right?
14:03:36wodzthe buttons get taken definitely
14:04:46wodzif I conduct the experiment like this: 1)select item 2) force break in gdb 3) call lcd_update() in gdb I see what should be displayed
14:05:16wodzso button is posted, action is taken but main thread is blocked and lcd_update() is postponed
14:05:45jhMikeSwodz: the menu got drawn in the framebuffer but lcd_update didn't get called?
14:06:35jhMikeSdoesn't sound like ints and threads then :\
14:08:10jhMikeSdid lcd_update not get called or did it just not update the display? I would think it's destined to call that and will reach it before blocking again.
14:09:09wodzregarding up/down buttons - this doesn't work quite right. If I hold down/up it skips one line from time to time
14:09:17wodznot get called
14:09:43wodzI set breakpoint on lcd_update() and I have to wait ~4-5s to breakpoint fire
14:10:38jhMikeSwhat does it do in between the queue wait and the lcd_update?
14:10:56wodzjhMikeS: in main? or in general?
14:11:24 Join b0hoon [0] (~quassel@
14:14:30jhMikeSafter waiting for a button up to where lcd_update should be reached. this sounds quite strange.
14:15:15 Join CGL [0] (~CGL@
14:16:02wodzI don't quite understand
14:16:58 Quit gusto (Quit: Leaving)
14:17:00 Join adnyxo [0] (
14:17:07b0hoonwodz: check your lcd_update_rect function line 193
14:17:19jhMikeSyou said that the code is updating the memory framebuffer but not reaching lcd_update IIUC. now, I just wondering how it could miss something it should be destined to do once it redraws.
14:17:40 Join watto [0] (~watto@
14:19:55*jhMikeS doesn't know how it would update *sometimes* if that were false rather than not ever
14:20:15jhMikeSeek, now me sees it heh
14:21:30wodzoh shit
14:22:14jhMikeSwodz: you presumed it should be called lcd_update and not something else? haha.
14:24:05*jhMikeS thinks the hanging problem will now clear itself up *fingers crossed*
14:25:28wodznavigation issue is gone
14:27:08wodzbut... the overall performance issue is still there :-/
14:27:45wodzplaying mp3 still gets over 100MHz
14:28:08 Quit n1s (Ping timeout: 240 seconds)
14:28:40 Quit parafin (Quit: So long and thanks for all the fish)
14:28:42 Join parafin [0] (
14:29:03kugelwodz: awwee :)
14:29:34jhMikeSwodz: no dsp stuff stuck on? :) not that it should bring it to that.
14:30:50 Join n1s [0] (~n1s@rockbox/developer/n1s)
14:32:53wodzwhat dsp stuff?
14:33:52jhMikeSwodz: eq, crossfeed, tone controls, dithering, etc, etc. :) It's playing at the right pitch?
14:34:39wodzI didn't touched any of this settings
14:34:41 Join Schmogel [0] (
14:35:31wodzb0hoon: You are my hero!
14:36:25b0hoonheh, thanks - you are doing great job!
14:36:27jhMikeSindeed, good call to check the lcd driver
14:37:09 Join Blue_Dude [0] (~chatzilla@rockbox/developer/Blue-Dude)
14:37:56kugelit strikes me that we didn't of that :)
14:38:34TorneTheSeven: I'm going to do some new bootlaoders for the PP ipods; is there any need to update the nano2g bootloader as well at the moment? Do you maybe want to add the boot-disk-mode-on-usb-insert feature I added to the PP ipod bootloader?
14:40:04jhMikeSkugel: some misdirection for awhile, thinking certain things were looked at and the whole nature of the problem was fundamentally different than presented
14:40:34 Quit xiainx (Ping timeout: 265 seconds)
14:40:47kugelyea, wodz told that forcing lcd_update works in the very beginning but we got focussed on threads then
14:41:47jhMikeSI started looking there after being told the FB was getting write and b0hoon caught it first
14:43:22funmanso what was the problem ? (i'm looking at fs#11137)
14:43:28b0hooni looked at it yesterday, but i didn't notice anything...
14:44:06 Quit JdGordon (Ping timeout: 260 seconds)
14:45:19 Join JdGordon [0] (~jonno@rockbox/developer/JdGordon)
14:47:25kugelpixelma: if you get a chance test the latest pla rework patch on your ondio/c200 please
14:47:55b0hoonfunman: lcd_update_rect was returning when the lcd was initialized
14:48:55funmanoops :o)
14:51:06pixelmakugel: sure, will do in the evening probably. I'm at work currently
14:52:16 Quit funman (Quit: free(random());)
14:52:42wodzskipping lines bug is gone also :-)
14:52:51 Nick YPSY is now known as Ypsy (
14:54:02 Quit n1s (Ping timeout: 268 seconds)
14:55:31jhMikeSwodz: is lcd_update(_rect) slow or something? what's with always setting the power saving mode?
14:56:53kugeljhMikeS: we ruled that out already
14:57:07kugelbut now it makes sense that 1/4 updates were so insanely fast
14:57:35 Quit saratoga (Ping timeout: 248 seconds)
14:57:42wodzjhMikeS: lcd_updates are not so slow. Regarding power save mode - this is taken from iaudio m3 driver - I made a comment because I think it is plain wrong
14:58:05wodzI am building test plugins now so we see
14:58:59jhMikeSkugel: ah, I guess I missed out on important weirdness like that, nose buried in other stuff and all.
15:00:29*jhMikeS seems to remember making the very same oops one time, long, long ago with x5
15:08:44wodznow I get 1/1 761.5 fps 1/4 2936.0 fps
15:09:29 Join panni_ [0] (
15:09:47 Join esperegu_ [0] (
15:10:09 Quit esperegu (Read error: Connection reset by peer)
15:11:12***Saving seen data "./dancer.seen"
15:11:33 Join n1s [0] (~n1s@rockbox/developer/n1s)
15:17:29 Quit JohannesSM64 (Ping timeout: 258 seconds)
15:18:04n1swodz: about your performance problem; 1)is the sram used 2)is the dram set up correctly 3)is the cpu clocked as high as you think? is there something firing a lot of interrupts or some tick task that takes too much cpu?
15:20:22wodzn1s: 1)yes 2)rather yes 3) don't know I relay on what rockbox says 4) there is ADC interrupt firing at about 400Hz but ISR is very short
15:21:25n1sit doesn't wait for adc conversion to complete in the isr or something?
15:21:36n1sjust a few thoughts, gtg now
15:22:17 Join JohannesSM64 [0] (
15:22:35wodzin ISR it reads value from ADC and changes channel
15:27:09b0hoonwodz: is this button_filter necessary, you could lower the frequency without it, i don't know...
15:27:16 Join evilnick_B [0] (~0c140464@rockbox/staff/evilnick)
15:27:40wodzit is necessary
15:28:00wodzI need some kind of debounce
15:29:51 Join bmbl [0] (
15:29:51 Quit bmbl (Changing host)
15:29:51 Join bmbl [0] (~Miranda@unaffiliated/bmbl)
15:30:14wodzhmm ADC ISR get linked into DRAM not IRAM
15:34:28Torne <- new ipod bootloaders ready for testing
15:34:33Torneany comments on my text? :)
15:37:29 Join emrecelikten [0] (~c18cf902@gateway/web/freenode/x-cxouhrwagvgkhnmw)
15:38:56wodzb0hoon: about the filter - It has nothing to do with frequency it is moving average filter
15:39:35*jhMikeS was hearing strange EQ anomalies on the beast that seemed to be what preglow described long ago
15:40:14TornejhMikeS: you mean the software eq, or the tone controls?
15:40:24Tornei fixed the beast tone controls the other day :)
15:40:24jhMikeSTorne: software EQ
15:40:45jhMikeSit sounds like wrapping noise, but not so severe
15:40:56hobbsjhMikeS: does letting the GPT count during WFI use more power or what? Because timing is so much more consistent with it on
15:41:10jhMikeSperhaps just one band or something, and precut has no effect on getting rid of it
15:41:16TornejhMikeS: btw, my housemate is interested in working out how to get rockbox to use the beast's DAC's EQ directly
15:41:33Tornei suggested he should probably come on irc and chat to people
15:41:48Tornesince it's mostly a UI issue, if i understand correctly
15:42:04Torne(hardware EQs being similar, but not identical, to each other or our software EQ)
15:42:10jhMikeShobbs: I have no info on that. I just expect it's unneeded and if you're busy waiting, it's not in WFI mode.
15:44:08jhMikeSTorne: Personally, I'd say have a "hardware EQ". Llorean was objecting saying such a thing was confusing. The best compromise I could think up was having 5 "tone controls". Really, it was always an interface thing more than a problem to make it work. I don't care to use them half-assed as bass and treble. We should use that EQ.
15:45:18jhMikeSIf we don't we're just jailing the HW a bit and I thought we were supposed to do the opposite of that :)
15:45:35Torneif someone did the effort to use them as bass and treble, we might as well have it, though, no?
15:45:45jhMikeSeven the gain limiter could be hardware
15:46:13jhMikeSbut of course that won't help PCM saturation
15:46:24Torneuh, that's over my head a bit :)
15:46:43Tornei'm not a.. sound guy :)
15:46:55 Quit TheSeven (Ping timeout: 276 seconds)
15:47:30jhMikeSTorne: maybe combine the gains of several bands for bass and treble to give a good rolloff? Just using one band each for b/t would probably sound bad.
15:47:43Tornewell, i don't know what we do on other platforms
15:47:47 Join dfkt_ [0] (dfkt@unaffiliated/dfkt)
15:47:52Tornebut this kind of fakery is already done on some of them, no?
15:48:00Tornethere are other DACs with some kind of hw eq.
15:49:18jhMikeSTorne: Yes there are. I'd rather free them up. I don't recall any combinations, just using one band as bass, the other treble. I could be mistaken or it changed.
15:49:30Tornei think they just use the top and bottom bands, yeah
15:49:46 Quit dfkt (Disconnected by services)
15:49:51 Nick dfkt_ is now known as dfkt (dfkt@unaffiliated/dfkt)
15:49:54Tornebut anyway, yeah. it's a UI issue, so I'll poke him to come discuss it with people
15:50:39Tornes'not something i care much about either wya personally, but I'm happy to help him get familiar with coding on rockbox ;)
15:52:31 Quit emrecelikten (Ping timeout: 248 seconds)
15:52:50jhMikeSTorne: I don't know. There must be a way of presenting it so it's not confusing between sw and hw eqs.
15:53:00 Quit mitk (Quit: Leaving)
15:54:04wodzI am interested too. WM8750 used in HD200 has hw bass/treble and 3Denchance features
15:54:48jhMikeSwodz: the WM8798 has the 3d thing too. most of what it can do isn't being tapped right now.
15:56:21*jhMikeS doesn't know if it's quite exactly the *same* thing though
15:58:05wodzIs there a chance to commit patch from #11137? It is now 174kb diff 122 files changed, 4358 insertions(+), 16 deletions(-) and it will become bigger as I work on port
15:58:50 Join slck [0] (Venci@Slackware.SlackPix.Com)
16:01:23 Join jgarvey [0] (
16:01:49 Quit bmbl (Ping timeout: 260 seconds)
16:01:49pixelmais there a possibility that it affects other targets?
16:02:25kugelwodz: I think so, maybe some cf guru wants to have a quick look at it (specifically at the changes to shared files)
16:02:56wodzmost intrusive change is commite allready
16:03:27 Join MethoS- [0] (~clemens@
16:03:47wodzIt would be nice if CF guru would like to review the code but I am afraid that size of diff is rather limiting factor
16:06:33kugelI don't think anyone will seriously look at the files specific for the port only
16:07:52kugelusually new ports are committed earlier, like before sim/plugins or even main binary work :)
16:08:23kugelwodz: what's the installation procedure?
16:08:40Bagderwe should get at least the main parts of wodz work merged
16:09:47Zagora review of just the modified files shouldn't have to be too much work
16:10:15Zagormost of the patch consists of new files
16:11:23wodzok, I have to go see You later
16:11:27 Quit wodz (Quit: Leaving)
16:11:34kugelwodz: very much like the sansa ams procedure. I think mkmpioboot should be in rbutil/
16:13:49 Join DerPapst1 [0] (
16:14:14 Quit DerPapst (Ping timeout: 246 seconds)
16:23:50 Part LinusN
16:30:59 Quit CGL (Remote host closed the connection)
16:37:14 Quit kugel (Ping timeout: 276 seconds)
16:46:31 Join jfc [0] (
16:47:51 Part b0hoon ("Back to work.")
16:54:15 Quit Xerion (Quit: )
16:54:29 Join M3DLG [0] (~M3DLG@
17:01:37 Quit Zagor (Quit: Leaving)
17:02:59 Join captainewkl [0] (~2669ecc2@gateway/web/freenode/x-ltdtpgcbmfyrylqy)
17:04:56 Quit lpereira (Quit: Leaving.)
17:05:11 Quit robin0800 (Remote host closed the connection)
17:08:36 Join toffe82 [0] (~chatzilla@
17:11:14***Saving seen data "./dancer.seen"
17:14:49 Quit jnss (Ping timeout: 260 seconds)
17:16:48 Part Blue_Dude
17:18:24 Join kugel [0] (~kugel@rockbox/developer/kugel)
17:21:45 Join jnss [0] (janes@gateway/shell/
17:26:57 Quit M3DLG (Ping timeout: 260 seconds)
17:37:01 Join phanboy4 [0] (
17:38:41amiconnjhMikeS: "Repeat all" and auto-change dir together won't work. Auto-change is above playlist handling. Afaik this is documented in teh manual
17:42:18jhMikeSamiconn: hmmm. it always did that and I only used it for a battery bench since there weren't many files on disk at the time. IIRC think the auto change is in the playback engine in playback.c.
17:43:07amiconnAuto-change makes a new playlist from the next directory if the old one ends. Repeat all causes it to never end...
17:43:52amiconnBut then you don't need auto-change for battery bench. Just let it repeat one folder...
17:44:25*amiconn doesn't use auto-change dir at all because it would annoy him a lot
17:45:37 Quit DerPapst1 (Ping timeout: 246 seconds)
17:46:12*gevaerts used to use auto-change directory with random folder advance until he implemented the random folder playlist thing
17:46:36jhMikeSit just seems it should work that way intuitively and repeat everything on the disk if auto change is on. if it's not on, repeat the directory.
17:49:36jhMikeSI'll say, because, it would then repeat what it would play otherwies just one time
17:51:19*jhMikeS should probably actually review the shiny manual himself in more detail one day or pay attention to that part of the code more. :\
17:54:08 Quit petur (Quit: *plop*)
17:54:21 Quit jfc (Ping timeout: 276 seconds)
17:58:19pamaurygevaerts: I commit a change to my github account to distinguish between old and new usb api, I would really appreciate feedback/advise, I feel it's a bit akward but it's because there are several changes at once, not only the api
17:59:31 Quit pamaury (Quit: Quitte)
17:59:46 Quit Bagder (Quit: It is time to say moo)
18:00:11 Join DataGhost [0] (
18:00:11 Quit DataGhost (Changing host)
18:00:11 Join DataGhost [0] (~dataghost@unaffiliated/dataghost)
18:01:26jhMikeSstrange that ACD should even be an issue with the database rather than simply not-applicable, thus ignored if DB is the source
18:03:01pixelmamaybe more like an auto-change directory non-stop mode
18:10:05 Join jfc [0] (
18:11:18 Quit efyx (Remote host closed the connection)
18:11:29jhMikeSpixelma: just enumerating in my head I'd expect: repeat on (acd: off = repeat dir only, on = play all in a loop), repeat none: (acd: off = play dir once, on=play all once), database as source: acd not relevant (since it's sort of its own file tree), repeat one: well, just repeat one, no matter what source (which I think works)
18:11:48 Join efyx [0] (
18:17:42 Quit esperegu_ (Remote host closed the connection)
18:18:31 Quit kugel (Ping timeout: 265 seconds)
18:19:58 Quit toffe82 (Ping timeout: 265 seconds)
18:20:19 Join toffe82 [0] (~chatzilla@
18:21:10 Join toffe82_ [0] (~chatzilla@
18:22:38 Join toffe82__ [0] (~chatzilla@
18:23:55 Join toffe82___ [0] (~chatzilla@
18:23:55***Alert Mode level 1
18:23:55DBUGEnqueued KICK toffe82
18:23:55DBUGEnqueued KICK toffe82_
18:23:55***Alert Mode level 2
18:23:55DBUGEnqueued KICK toffe82__
18:23:55DBUGEnqueued KICK toffe82___
18:23:55***Alert Mode level 3
18:24:16 Join domonoky [0] (~Domonoky@rockbox/developer/domonoky)
18:24:49 Join CGL [0] (~CGL@
18:25:08 Quit toffe82 (Ping timeout: 276 seconds)
18:26:06 Join DerPapst [0] (
18:26:26 Quit toffe82_ (Ping timeout: 276 seconds)
18:27:02 Quit CGL (Remote host closed the connection)
18:27:44 Quit toffe82__ (Ping timeout: 276 seconds)
18:29:21 Quit efyx (Remote host closed the connection)
18:31:23 Join efyx [0] (
18:32:33 Join pamaury [0] (~c2c7a50a@rockbox/developer/pamaury)
18:33:56***Alert Mode OFF
18:35:14 Join CGL [0] (~CGL@
18:40:23 Join liar [0] (
18:42:02 Quit panni_ (Quit: ( :: NoNameScript 3.81 :: ))
18:47:21jhMikeSACD goes into the *.config as "folder navigation". LOL?!\
18:48:59 Join TheSeven [0] (~TheSeven@rockbox/developer/TheSeven)
18:49:17jhMikeSit seems "next_folder" is dealt with in the playlist code afterall...
18:53:23 Nick Ypsy is now known as YPSY (
18:54:20 Join Xerion [0] (~xerion@
19:04:27 Join petur [0] (~petur@rockbox/developer/petur)
19:05:29 Join Boldfilter [0] (
19:07:45 Quit flydutch (Quit: /* empty */)
19:09:59 Join einhirn_ [0] (
19:11:15***Saving seen data "./dancer.seen"
19:12:45 Join Horscht [0] (~Horscht2@xbmc/user/horscht)
19:14:01 Join M3DLG [0] (~M3DLG@
19:19:12 Quit einhirn_ (Read error: Connection reset by peer)
19:22:29 Join Silverwolf [0] (
19:23:30 Quit Silverwolf (Client Quit)
19:31:21n1sjhMikeS: that whole directory playlist handling thing is pretty hacy IMO, i think we should come up with a better way, maybe just make playing files create a regular otg playlist that is handled the same way as any other playlist
19:31:28n1shacky, even
19:34:01 Join xiainx [0] (
19:35:33jhMikeSn1s: some things just don't seem to follow instinct. it one of those options that should mostly explain itself. otoh, "caption backlight" never explained itself; might as well call it "captain blackbeard".
19:36:57n1slots of esotgeric options we do have
19:38:30n1sbut i think that if we had an easy way to select multiple dirs for insertion into a playlist, and made playing a file equivalent to inserting that dir into a playlist we could drop ADC and the whole special handling of dir playlists
19:43:13 Quit M3DLG (Ping timeout: 268 seconds)
19:43:14 Quit phanboy4 (Quit: Leaving)
19:43:21jhMikeSI don't even think we have multiple-selection lists and I'd love that for managing files on the player when away from any other means.
19:45:05n1sno, we don't, it's a feature i would love too
19:45:16Lloreann1s: The thing is, if you click on a file, and expect playback to continue through the folder, some people can (quite reasonable) expect new additions to the folder later (for a podcast folder, for example) to be played.
19:46:36n1sLlorean: but doesn't inserting a dir into a dynamic playlist do that?
19:46:49jhMikeShmmm...but what would dropping ACD do? I don't mind the option to stick with playing an album from the browser or just play everything I've got. No database init = more RAM too.
19:47:29jhMikeSLlorean: you mean if they move a file into a directory that is already playing through?
19:47:40Lloreann1s: If you "insert" the dir, the playlist is a list of filenames isn't it? It's an unsaved playlist, rather than an index in the folder?
19:47:46n1sjhMikeS: from the times i've looked at playback or playlist code, the dir handling seems to need quite a bit of special cases
19:47:53LloreanIf I recall, if you insert a folder with subfolders, you can't bookmark until it's saved.
19:48:41LloreanjhMikeS: Say you click "select" on podcast 3 of 5 in your podcasts folder rather than using "insert" to add it in a playlist.
19:48:52LloreanYou get through 3 and part of 4, then stop, then add a podcast six from your computer, then resume.
19:49:05LloreanI'm pretty sure as long as you don't manipulate the playlist playback will eventually reach #6 right?
19:49:15jhMikeSn1s: I can't fathom why it needs 3581 lines of code! of course, what's most of it dealing with?
19:49:29LloreanBut if you "insert" the folder instead, you'll insert the 5 individual items, and when a sixth is added to the folder, unless you also insert it, it won't show up in your dynamic playlist
19:50:48n1sLlorean: i'm not sure but IIRC as long as the playlist is unsaved it is "dynamic" and then it just keeps info like inserted dir foo, inserted file bar so if a dir is inserted and later changed and the playlist is not saved, the new contents of the dir will be included too, i might be wrong of course
19:51:41Lloreann1s: I'm not necessarily saying the change would be bad, just unexpected for some.
19:51:51n1si think this is the cause of the "deleting a file in the dynamic pl will cause resume to pick the wrong file bug"
19:51:59LloreanFor example, if we got rid of the directory index based playback entirely, you'd never get the whole "it resumed in a different file I stopped" bug
19:52:34n1swell, apparently we have unexpected behaviour either way :)
19:52:42Lloreann1s: I'm pretty sure those are playlists created by simply pressing 'select' on a file, rather than using "insert"
19:52:51n1smight be
19:52:56 Join esperegu [0] (~quassel@
19:53:08jhMikeSLlorean: something tells me you shouldn't index where you don't wholly control the contents
19:53:18LloreanjhMikeS: Sadly, we already do.
19:53:41LloreanAnd it does allow the nice behaviour of being able to have a playlist that continues into a file that wasn't there before, but that's about the only positive I think.
19:53:46n1sLlorean: anyway, multi select is something i would really like
19:54:08LloreanMe too
19:54:56LloreanI'd also like to split "right" and "select" one day, so that one could actually be "invoke what is highlighted" and the other could be either "insert" or "select" or some other list action.
19:56:41pixelma the index way is also a lot faster I believe (I always thought this is the reason)
19:56:59 Quit xiainx (Quit: Good Bye)
19:57:44LloreanYes, an "insert" takes a bit for it to iterate through and insert all the files
19:58:13LloreanBut couldn't playback start based on the index when you 'select' a file (since at that one point, we know it's constant) then the iteration to populate the playlist happen after?
20:00:37pixelmahmm... it could also be from the time resume info and other settings where stored in a hidden sector and space to store something was quite restricted?
20:00:49pixelmacould be I'm confusing things though
20:01:05n1sLlorean: yeah, it could, this would also help the case where people select a file in the database from a view of all their files, which causes al those files to be inserted into the pl before playback starts
20:01:45n1si think we would need a pl worker thread then though
20:02:39 Quit Xerion (Ping timeout: 252 seconds)
20:02:56 Join Xerion [0] (~xerion@
20:08:25jhMikeSn1s: erm, we would? thought we had that hanging around somewhere. :)
20:09:13n1soh, we do? then Mr. Someone should fix that databse thing!
20:10:28jhMikeSit's "playlist cachectrl". funky.
20:11:42n1sah, i should read more code before speaking
20:11:54 Join mikroflops_ [0] (
20:13:38jhMikeSwho's gonna read all this code? it reads like "War and Peace", only more serious.
20:15:51 Quit mikroflops (Ping timeout: 252 seconds)
20:16:24n1syeah, it hasn't been touched a lot in recent years either so liukely noone around knows it well
20:17:18 Quit antil33t (Read error: Connection reset by peer)
20:17:25 Join antil33t [0] (
20:18:39jhMikeSgiven the code size of the whole project, I'm sure it's quite excusable to not know something about something. :)
20:19:45jhMikeSn1s: btw, Mr. Someone is Dr. Who?
20:21:53 Nick scorche|sh is now known as rockbox|scorche (~scorche@rockbox/administrator/scorche)
20:22:08n1sHe seems to change shape and name from time to time so, why not?
20:22:57jhMikeSthat code is quite light on comments :\
20:26:12 Quit Xerion (Ping timeout: 248 seconds)
20:27:52 Join fml [0] (
20:28:40fmlHello.What do you think about ? This is an attempt to fix (to eat) spurious key release events after entering the yesno screen.
20:29:28 Quit bluebrother (Disconnected by services)
20:29:29 Join bluebroth3r [0] (~dom@rockbox/developer/bluebrother)
20:30:39fmlBlue_dude: this ^^ is in particular for you
20:31:48 Part watto
20:34:00jhMikeSn1s: whoever cares to can look into the playlist time vortex. I'm sure not ready for a go on that at this moment. :D
20:35:20n1sneither am i, and my heap of patches to test once i get home is growing :)
20:35:48n1smainly attempts at optimizing tremor and cf asm for the new mdct though
20:36:13n1sand a little armv6 asm where the new instructions provide benefit over v4
20:37:06 Nick fxb is now known as fxb__ (
20:40:44CtcpIgnored 4 channel CTCP requests in 0 seconds at the last flood
20:40:44*jhMikeS has heaps of plans, and other beastly things to accomplish
20:41:16jhMikeSn1s: anything that keeps that imx core quiet as much as possible is good
20:42:40n1syup, i want to play a bit more with the beast hardware too, there must be more we can do to save power :)
20:43:10n1sand maybe try to get the tv out working :)
20:43:47*jhMikeS supposes he'll commit scaling code, with DPTC disabled and DVFS enabled and simplify later when he's confident he know what freescale was trying to do with this poorly undocumented thing and its several BSP driver variations.
20:44:28n1syes, it looks like a very tricky thing
20:45:14 Join xiainx [0] (
20:45:57 Quit Boldfilter (Quit: Boldfilter)
20:49:25jhMikeSwhy they have DPTC tables for frequencies where DTPC isn't necessary is a bit perplexing. one driver version only has tables for overdrive, has checks for imx31 rev 1, sets voltages to 1.25V for dvs yet rev 1 can't use that, supposedly, according to the other drivers. there's even another version I found online.
20:53:33jhMikeSthen there's that SDMA stuff, where it has checks in the BSP for rev 1.00, and code! but elsewhere it says it was never a production part. nonetheless, that was much easier to sort out and I can probably remove that overhead from the compiled binary.
20:56:28n1si hope we can assume all the beasts use the same chip rev?
20:56:47*n1s remembers the crashing refurb h120 bug :)
20:57:18CIA-5New commit by 03bluebrother (r25694): Update year in about box. We are still working on this.
20:57:19jhMikeSn1s: I think, but I haven't assumed that. It might be plain irrelevant if 1.00 was only a prototype.
20:58:43jhMikeSn1s: It was probably no longer made by rev 2.00 release anyway, thought I'm not certain. If the latest fw image doesn't handle the case, it's no issue.
20:58:46n1sanyway, i am impressed that you tackle a mess like this :)
20:59:20jhMikeSbrute-force persistence :)
21:00:19jhMikeSthe reptilians (free-SCALE, hint, hint) will not defeat me
21:03:19 Quit hobbs (Ping timeout: 246 seconds)
21:05:22 Join jordan` [0] (
21:08:56 Join archivator [0] (
21:11:19***Saving seen data "./dancer.seen"
21:12:04 Join TexasRockbox [0] (
21:13:59TexasRockboxHello, Is there any way to display both music and video files in file browsing mode while not displaying .jpg files? (ipod 5.5g 240GB custom build with Rockbox USB disabled)
21:14:00archivatorI got one of these: and it only works with the OF (iPod 5.5G). Anything I can do to help? Debug->FM radio says no hardware detected.
21:14:54TexasRockboxNeed to turn on the accessory power
21:15:03archivatorit's on.
21:16:51TexasRockboxThe Apple iPod remote (discontinued but can be found by Google-ing) works. Can't say for any other adapter.
21:17:17TexasRockboxThe Apple FM remote, that is
21:17:19pixelmaTexasRockbox: store them away in a different folder, if you really need to you could set the folder to hidden and only let it display "supported" files
21:18:19TexasRockboxStore all the cover art in another folder?
21:18:34*jhMikeS wants to show hidden files while only browsing supported ones since hidden ones can be supported, or he's just being too demanding.
21:19:30TexasRockboxAll of my cover art files are call cover.jpg and reside in the folder where each album is located.
21:20:04archivatorTexasRockbox: see the manual, C.3 - it''s possible to have everything in one folder, you'll just need to rename the files.
21:20:44pixelmaTexasRockbox: there is one way you could store it one folder and still have it found. (I believe an "albumart" folder inside .rockbox, the AlbumArt wiki explains it. But you need to rename the cover files to artist-album.jpg or so
21:20:46TexasRockboxYikes! 800+ files
21:21:35TexasRockboxIs it known where in the code where I can add .mpg to the list of supported music files and have the video play instead?
21:22:27TexasRockboxI did try changing the extention of the video file to .mp3 and that didn't work
21:23:09pixelmayou mean that they show up if you set your file view to "Music"
21:23:30pixelmajust for clarification, I had to parse the sentence a few times
21:24:55TexasRockboxIf I change the extention of the video file the file shows but does not play. If I leave the extention to .mpg the file isn't visible unless I select show files "All" or "All Supported".
21:25:19TexasRockboxWhat I want to do is show just music and video files only
21:29:36 Part archivator ("Leaving")
21:32:09 Join new_bee [0] (
21:33:04new_beei own a sandisk clip mp3 player, and wanted to install rockbox on it. but it seems that there is only a version of rockbox in the category "not usable". what does this mean exactly?
21:34:21 Join Strife89 [0] (~michael@
21:34:35TexasRockboxThe Clip has several versions the original clip, clip V2 and clip+. Work is being done on the clip V2 and clip+ but isn't fully supported.
21:34:51Strife89Hmm, can I interest someone in looking at this patch?
21:35:07Strife89It compiles, but has no effect.
21:35:28Strife89At least on my iPod Video (5.5) and Color (4 C/P)
21:35:31 Join pixelma_ [0] (quassel@rockbox/staff/pixelma)
21:35:31 Quit pixelma (Disconnected by services)
21:35:53 Nick pixelma_ is now known as pixelma (quassel@rockbox/staff/pixelma)
21:35:55new_beeTexasRockbox: is there a list or something where i can see, what is supported and what is not? maybe i could contribute, too.
21:35:57 Quit amiconn (Disconnected by services)
21:35:59 Join amiconn_ [0] (quassel@rockbox/developer/amiconn)
21:36:18 Join archivator [0] (
21:36:21 Nick amiconn_ is now known as amiconn (quassel@rockbox/developer/amiconn)
21:37:14 Quit tipi^ (Ping timeout: 276 seconds)
21:37:25TexasRockbox@new_bee see the section "Stable Ports"
21:40:46new_beei already looked at it but didnt found any information of what i can do with the nonusable version and what not. i think i have a sansadisk clip+
21:41:05amiconnLlorean, jhMikeS: The index method is one reason why rockbox' playlist creation is so much faster than e.g. archos'
21:41:44jhMikeSamiconn: I'm not even clear on what that is (as a 'method' per se)
21:42:35amiconnThe archos OF had a limit of 999 files per playlist for that reason, and its playlist generation was so slow that early versions even went into idle poweroff before staring to play if you actually tried to add that many files
21:42:38TexasRockbox@new_bee beneath the section Unusuable Ports" there is a link to "status summary of....."
21:42:46 Quit pamaury (Quit: Page closed)
21:43:12amiconnLater versions fixed it (the idle poweroff, not the slow speed, by not considering playlist creation as idle)
21:44:45amiconnAnd the playlist cachectrl thread only exists on swcodec, and for a very different reason than playlist *building* (iirc)
21:45:19archivatorHmm, can we have the first voltage measurement (after boot) taken after the disk has been stopped? I find the low battery level reported kind of annoying.
21:45:49jhMikeSthis method doesn't bother with filenames, but directory indexes?
21:46:12jhMikeSamiconn: that was @ me?
21:48:25n1sarchivator: if you resume on startup, that could be a few seconds before reading it, and that would be annoying IMO
21:49:33archivatorn1s: I *do* resume on startup and I also have the battery level announced (voiced). It's quite startling hearing that I the battery is at 9% and a few moments later when I look it's at 31% :-/
21:49:37 Join robin0800 [0] (~robin0800@
21:50:08n1swhat player is that?
21:50:25archivatoriPod 5.5 with a replaced battery
21:51:08jhMikeSamiconn: it seems sensible in general however with perhaps a few trouble spots that need smoothing out for it to work
21:51:49 Join Llorphone [0] (~pocketirc@
21:52:06amiconnThis method has been in use for years now, it works and is fast
21:52:24jhMikeSI mean, work without mistaking one thing for another. obviously it works. :)
21:52:31n1sand has some drawbacks
21:53:01n1sjust storing the filename would fix the resume wrong file thing
21:53:13new_beeTexasRockbox: thank you
21:53:22jhMikeSof course anyone sadistic enough can delete a file, give a new one the same name as a that one, and complain about it being wrong. :)
21:53:57 Join geertvdijk [0] (
21:54:41jhMikeSbetter md5sum everything, just to be sure...or not
21:54:47TexasRockbox@new_bee You're welcome. Keep checking back and look for updates. Monitor this topic:
21:55:35pixelmaamiconn: the problem with it which also gets reported sometimes is resuming a different track if one or more tracks get deleted in the meantime
21:55:54new_beeTexasRockbox : thanks againg, i will. i will also read in how to contribute. maybe i can help. i know some c
21:56:44amiconnWell, if you can find a better method while keeping the speed - go ahead
21:56:46TexasRockbox@new_bee Great! Please feel free to contribute, work is ongoing and any help is appreciated
21:57:38jhMikeSn1s: that just it, how to rematch the index, if index method isn't abandoned (and I suspect it won't be so easily).
21:57:44n1sarchivator: you could try increasing the delay before first battery reading in powermgmt.c:648
21:58:16amiconnImo being slightly off in rare cases (how often do you add/remove stuff from your dap *in the current folder*?) is much better than being slow *every time*
21:58:51jhMikeSwe should probably have find voltage reading compensation while spinning, speced for each battery, just like the charging curve
21:58:54archivatorn1s: thanks, I'll take a look.
21:58:54amiconnAnd just storing the filename won't help
21:58:57TexasRockboxAgain, is there a way to configure Rockbox via the interface to show both music and video files? If not, where in the code would it be changed to permit the showing and playing of supported video types?
21:59:02 Join Luca_S [0] (
21:59:38amiconnIf you overwrite one file with another file of the same name, the dir index might change, depending on the tool you use and the host's fat implementation
21:59:51CIA-5New commit by 03alle (r25695): Fix grammar in the comment (make code non offending :-)
22:00:06n1samiconn: now *that* is a rare case
22:00:22amiconnI doub't it's rare - I'd rather expect it to be the norm
22:00:53LlorphoneTexasRockbox: doesn't "supported" include video?
22:01:00n1si can't remember when i deleted a file and replaced it with one with the same name
22:01:22pixelmaI've heard of people storing all their tracks in the root folder and you wont know with e.g. iTunes stored music getting off topic a bit). I'm all for keeping the quickest method and never had a problem with deleted files, just stating why the discussion came up
22:01:34n1sanyway i meant to store the filename in addition to the index so we match index like we do now but then also check the filename
22:01:51TexasRockbox@llorphone yes it does but I use album art in the album subdir and I don't want to see the .jpg
22:02:09 Quit Llorphone (Remote host closed the connection)
22:02:15n1spixelma: people with itunes files are unlikely to use the filebrowser for playing
22:02:18amiconnOverwriting a file with the same name does e.g. happen when editing tags directly on the target's disk
22:02:18pixelman1s: and what do you do if they filenames don't match?
22:02:24n1sgive up
22:02:58n1samiconn: ok, so it isn't perfect but better than what we have
22:03:09pixelman1s: resume info will probably stored from the folder name still?
22:03:09amiconnIf it's id3v2 and the tag editor has to change the tag's size, that essentially means a copy operation, which can only be done using a tempfile
22:03:25TexasRockboxThe idea is just to have only music and video appear
22:03:27n1sthe common case for this seems to be people who listen to podcasts and then deletes them
22:03:50Strife89Is there any work on using the Broadcom chipset in the iPod Videos?
22:04:02Strife89A patch, for instance?
22:04:12amiconnWell, if you delete old entries while the folder is still playing, things *should* keep working as expected
22:04:24n1sStrife89: not much, someone made the tv out work a bit
22:04:26jhMikeShow's this stuff sorted if it's up to the fat which order the files are found? it seems names have to be dealt with in some way.
22:04:43Strife89n1s: Ah.
22:05:18n1samiconn: i think more like listen, stop, connect to computer, add and delete a few podcasts, resume, bang
22:06:27amiconnNever happened to me but then I've never listened to a single podcast on any device other than my computer
22:07:17*amiconn isn't sure how the name -> index assigment actually works
22:07:18n1ssure, it's never happened to me either, but then i don't listen to podcasts
22:07:40jhMikeSme suggests invalidating the playlist after connect
22:07:48amiconnOn my computer I've listened to one or two podcasts so far
22:07:55amiconnjhMikeS: eurgh...
22:08:09amiconnThis would mean my playlist is gone when updating rockbox
22:10:37 Quit JohannesSM64 (Quit: WeeChat 0.3.2-dev)
22:11:35fmlWhat about storing the file names (in a background process?), but only for files near the current position in the playlist.
22:11:43 Quit robin0800 (Remote host closed the connection)
22:12:10 Join robin0800 [0] (
22:12:24jhMikeSamiconn: stuff you added manually I take it? obvious rebuilding under the same rules should produce the same list.
22:12:55jhMikeS*assuming nothing changed of course
22:13:02amiconnYes, but it won't reproduce my current position
22:13:47*amiconn always wonders why people want to change things which are proven to work reasonably well for a long time already
22:14:30n1sit doesn't work well for everyone
22:14:54*jhMikeS refrains from giving some of the many possible retorts to that :P (too political)
22:15:01fmlamiconn: then the playlist creation would still be fast, but playlist manipulations were also better detectable
22:15:10amiconnThen change it in a way that doesn't make it worse for those cases where it works well
22:15:51amiconnThose cases are still the majority... and for the podcast case there is a workaround
22:16:12n1si think storing the filename of the currently playing song is a good compromise, and give up if the name and index doesn't match
22:16:12n1sbut no, i don't care enough to do anything about it
22:16:20amiconnInsert the podcasts individually instead of playing the first track in the folder
22:17:10n1salso if people think about the problem, maybe someone will come up with a way that is both fst and more reliable
22:17:16amiconnTe resulting list will load slower (probably not noticeably provided there aren't too many podcasts in the folder), but position will always be correct
22:17:28jhMikeSamiconn: what do you mean "your current position"? I wouldn't think it would change if a playlist were a perfect duplicate.
22:17:29fmln1s: the background thread could e.g. ensure that two files around the current position have the names stored so that small shifts are recoverable
22:17:51amiconnjhMikeS: My playing postion...
22:18:07amiconnIf you invalidate the list and I have to re-create it, it would play from the beginning
22:18:30amiconnThe background thread does not exist on hwcodec
22:18:50n1sfml, sure, it would need to be a compromise
22:19:16archivatordomonoky, bluebroth3r: what's the status of the concurrency patch? i.e., should I do something more to get it comitted?
22:19:17fmlamiconn: how is the DB created on hwcodec?
22:19:31n1samiconn: that is one thing that would also be easier with a way to multiselect, right now, it's a bit tedious
22:19:38jhMikeSamiconn: why should file and current time index be discarded? I can't think _why_ the playlist couldn't be started from anywhere. ??
22:19:54amiconnMultiselect would be nice. Somebody please find enough buttons for that...
22:20:11pixelmajhMikeS: from the middle of the track?
22:20:17bluebroth3rarchivator: sorry, didn't have time looking at it the last days (was stuck due to volcano ashes for some days ...)
22:20:26domonokyarchivator: i am not sure. i think we said last time, to just wait for after the release. So i think it could go in.
22:20:38pixelmajhMikeS: and do you always remember which track you were at last
22:20:50amiconnjhMikeS: Current playlist and resume position are related. If you invalidate the former, you also invalidate the latter
22:20:52jhMikeSpixelma: yeah, it's just a seek time stored through the playlist functions anyway
22:21:09pixelmabut you said that would be discarded?
22:21:24TexasRockbox*kicks self
22:21:28 Quit TexasRockbox (Quit: CGI:IRC)
22:21:31archivatorbluebroth3r: no worries, just wondering. I think what domonoky said was the consensus last time..
22:21:43amiconnfml: Umm, using the database thread?
22:22:13*amiconn only updates the db occasionally... mostly for showing pictureflow to someone
22:22:15jhMikeSit calls playlist_update_resume_info(id3); when the audio is stopped
22:23:24fmlamiconn: so it's possible to have multiple threads on hwcodec? From you reply I thought it's impossible. But if it's possible we could have another thread (or use an existing one) for updating the playlists (adding file names to the current position and around it)
22:23:27pixelmadon't bookmarks still work even when tracks in the folder get deleted?
22:23:29jhMikeSindex and offest and tossed inti global_status
22:23:57amiconnfml: Of course it's possible. What I was saying is that the playlist_cachectrl thread doesn't exist on hwcodec
22:24:03jhMikeSs/offest/offset/, s/and/get/, s/inti/into/
22:24:33fmlamiconn: I didn't know it exists on swcodec :-)
22:24:56n1spixelma: i think they are filename based
22:25:11amiconnIirc it was introduced to solve a problem with trackchange on swcodec (and that doesn't affect hwcodec, obviously), so it would be a waste to have it on hwcodec
22:25:27amiconnAfaik it has nothing to do with playlist creation
22:26:42amiconnThreading itself dates back to the beginnings of rockbox (without priorities and stuff like that)
22:27:20fmlThen I think we could hijack a thread for that. Not every thread is heavily busy all the time.
22:28:13amiconnHwcodec still has thread priorities disabled, since there is no reason for this so far
22:33:26jhMikeSit's there mainly to ward-off audio breaks during user activity like scrolling which I can't see HWCODEC having an issue with since it just sends MP3 data with an ISR
22:36:14 Join Boldfilter [0] (
22:37:15n1swasn't it also to help UI responsivness (?) when playback was running unboosted
22:37:27n1si remember the ipod video being mentioned
22:38:34jhMikeSnot sure how it would unless you jump the UI priority. the first scheduler just ran thought all contexts on each context switch. non-runnable threads actually used CPU time when switching tasks.
22:39:34jhMikeSgo back in svn just before slasheri's initial change to a genuine blocking scheduler
22:40:46jhMikeSI think amiconn used to measure some 3400 task switches/sec during playback on archos, now it's about 12 per second
22:41:41amiconnNah, certainly not that many if all threads were sleeping
22:41:59domonokyarchivator: i am just trying you latest patch for rbutil, and got some problems: No progress info while voicing and the log contains a line like : "can not create children for parent in different thread". All on Windows and with sapi as voice.
22:42:08amiconnIirc that was the possible maximum (with the old scheduler) in a yield() test
22:42:11jhMikeSamiconn: you did give me a number in the thousands, I'm sure it's logged somewhere
22:42:55amiconnWhen all threads have nothing to do and call sleep(), there cannot be more than number_of_threads*HZ context switches
22:43:04amiconn..per second
22:44:22archivatordomonoky: interesting. I'll admit I haven't tested it on windows.
22:44:49domonokyarchivator: strangly progress info works for encoding.. maybe its because sapi doesnt run in paralell ?
22:44:50jhMikeSamiconn: of course for idle. but then I misunderstood, or misremember. it was a couple years back or so.
22:45:39archivatordomonoky: yeah, but it's marked as such and thus there should be only one thread. Unless.. are you getting that error from SAPI or from Qt?
22:46:03archivatorI think the problem's that SAPI is initialized in one thread and then used in another.
22:46:09 Join Darkknight512 [0] (
22:46:21domonokyarchivator: that message about parent in wrong thread is from Qt.
22:47:48jhMikeSamiconn: last time I watched the numbers accumulate on archos during buffering, it seemed to rack them up more intensely than a few thousand/sec.
22:48:44 Join droidcore [0] (
22:48:49jhMikeSit was probably my response to your mention of that other figure :)
22:50:31 Quit evilnick_B (Quit: Page closed)
22:50:54 Join mc2739 [0] (~mc2739@rockbox/developer/mc2739)
22:51:11archivatordomonoky: any line number? that really should either happen across all platforms or not at all :-/
22:51:17amiconnHehe, my first yield() round-trip-time test dates back to June 2004
22:51:25archivatorJust tested with both eSpeak and festival, works fine.
22:51:41domonokyarchivator: jup, looks like its a problem with ttssapi using its own thread for the script. If i set the max thread count to 2, progress info works again, but still looks strange.. (probably because ttssapi cant really be called in paralell.
22:53:37domonokythe message is: QObject: Cannot create children for a parent that is in a different thread. Unfortunatly no line number.
22:53:44 Quit efyx (Quit: Quitte)
22:53:55domonokyParent is QProcess(0x14be798), parent's thread is QThread(0x3f4e20), current thread is QThread(0x1575400
22:55:04archivatordomonoky: I know how to solve it quick'n'dirty. It might introduce a performance penalty though.
22:56:22domonokydepends on how dirty it is :-)
22:56:50domonokydoes progressinfo work for you if you use a non-paralell tts ?
22:57:30archivatordomonoky: both non-parallel tts are also non-linux :) So, haven't really tested them.
22:57:51archivatorI think I know why it doesn't work in your case though. Event processing stops after that error message.
22:57:51domonokyarchivator: you could modify the flags for a test :-)
22:59:47 Nick fxb__ is now known as fxb (
23:01:23 Quit mt (Read error: Connection reset by peer)
23:02:30archivatordomonoky: yes, works in both cases.
23:02:40 Join mt [0] (~mtee@rockbox/developer/mt)
23:03:02domonokystrange, so this is a ttssapi problem ?
23:03:09 Quit fml (Quit: ChatZilla 0.9.86 [Firefox 3.6.3/20100401080539])
23:05:29archivatorIt is. What needs to be done is to add "voicescript->moveToThread(QThread::currentThread());" at the beginning of voice()
23:05:48archivatorNot sure if the stream should be moved as well, though.
23:06:34archivatorthe stream can't be moved, actually, since it has a parent (the script). Can you test if that works?
23:09:27 Quit robin0800 (Remote host closed the connection)
23:09:52 Quit jd (Quit: Ω)
23:09:57 Join robin0800 [0] (
23:11:23***Saving seen data "./dancer.seen"
23:13:03 Join arbingordon [0] (~w@unaffiliated/arbingordon)
23:14:15archivatordomonoky: I'll be signing off for the day (had an excruciatingly long journey), please keep me posted on SAPI progress. We might have a similar problem with Carbon (if it doesn't like being called from another thread)...
23:16:49 Join jd [0] (
23:16:49 Quit jd (Changing host)
23:16:49 Join jd [0] (~jd@Wikipedia/HellDragon)
23:22:55 Quit Strife89 (Quit: Leaving work.)
23:26:33 Join M3DLG [0] (~M3DLG@
23:27:03 Quit robin0800 (Remote host closed the connection)
23:27:28 Quit Luca_S (Quit: CGI:IRC (EOF))
23:27:31 Join robin0800 [0] (
23:33:34 Quit mc2739 (Ping timeout: 276 seconds)
23:34:49amiconnjhMikeS: 20070823 :)
23:35:04 Join mc2739 [0] (~mc2739@rockbox/developer/mc2739)
23:35:50 Quit M3DLG (Ping timeout: 260 seconds)
23:36:09 Quit liar (Ping timeout: 258 seconds)
23:37:50jhMikeSamiconn: heh, will have a peek. :)
23:39:09 Quit jgarvey (Quit: Leaving)
23:39:12 Quit Darkknight512 (Quit: ChatZilla 0.9.86 [Firefox 3.5.9/20100315083431])
23:39:45 Quit n1s (Quit: Lmnar)
23:40:18 Quit ender` (Quit: Thats right, babe, Cochrane chortled. Were not possessors, were just like dimensionally disadvantaged.-- Peter F. Hamilton: Neutronium Alchemist)
23:44:21 Quit captainewkl (Quit: Page closed)
23:45:08jhMikeSamiconn: Aha, 600 round trips/s or 4800 yields. :)
23:45:53amiconnYep, as the possible maximum when not sleeping and without significant load
23:46:04amiconnThat was with the old scheduler
23:50:32 Join BiptoN [0] (
23:51:06 Quit JdGordon (Read error: Operation timed out)
23:51:36 Quit sinthetek (Changing host)
23:51:36 Join sinthetek [0] (~sinthetek@unaffiliated/sinthetek)
23:51:39 Quit CGL (Quit: Saliendo)
23:51:40 Quit petur (Quit: aw that hurts)
23:52:48domonokyarchivator: for the log: if i replace the voicescript->waitForReadReady() call in TTSSapi::voice with a loop which waits till i can read a line, the progress info works again, also the error message is gone. But i noticed a new problem: if i abort while voicing rbutil completly hangs up.
23:53:37 Join M3DLG [0] (~M3DLG@
23:55:31 Quit BiptoN (Quit: CGI:IRC (Ping timeout))

Previous day | Next day