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 2008-11-25

00:00:56kugelwell I don't really agree
00:01:08ZagorI know :)
00:01:11 Quit ender` (" Microsoft is like a case of herpes. You can't kill it off entirely, just force it to go into a dormant stage for a while.")
00:02:50*Unhelpful is inclined to think that the gain is zero... so, even if we assume the cost to be zero, what's the point?
00:02:55kugelunless rbutil downloads custom bootloader no ipod user ever is gonna install one configured in that way (the risk he gets such a build is higher imho, going by the loads of custom builds compared to (is there even any) "un"loads of custom bootloaders)
00:03:28kugeland the complexity is very controllable. This should be a 3-liner and no-brainer
00:04:10saratogawas it ever determined if the fuze can be unbricked?
00:04:19kugelsaratoga: not that I know of
00:05:01Zagorkugel: you still haven't told me what good this would do anyone. do you also want a config parameter for the 'viewers' directory? how about 'codepages'? they are even less complex to change.
00:05:28kugelZagor: I tell you what my personal gain is: while the fuze port I'm always here and there trying to boot from the microsd (which is indeed doable) in order to test sd code. And I'd rather pass a configure option instead of changing the tree
00:05:33saratogakugel: then I better be careful, are there instructions for the fuze install somewhere?
00:05:39 Quit bmbl ("Woah!")
00:06:00kugelsaratoga: svn bootloader works. I've build it today
00:06:49kugelsaratoga: mkamsboot has a small help text on how to pass file. The patched firmware it creates is to be put on the fuze's root dir
00:07:29kugelZagor: especially since the microsd is correctly recognized, unlike the internal memory
00:09:02saratoga"See comments in mkamsboot.c and dualboot.S for more information." quite the help file
00:09:31kugelsaratoga: did you type make in the mkamsboot dir?
00:10:09Zagorkugel: I think your case is a typical one-off and the proper solution is to change the tree. so, I won't do it. but I won't revert either it if you manage to convince someone else...
00:10:12kugelif you then execute mkamsboot it should say "./mkamsboot <firmware> <bootloader> <patched firmware>" or something like that
00:10:22shotofaddsgevaerts: I just tried it again on XP, which got as far as recognizing a UMS device, but then nothing. On linux the USB ids didn't even show up in lsusb. Unfortunately I haven't the faintest idea where to look for clues..
00:11:22saratogakugel: I saw that, but the name of the firmware file to patch isn't mentioned
00:12:13kugelsaratoga: the name is unimportant. You can rename firmware files if you like before passing it to mkamsboot
00:12:48kugelI always use the latest fuzef.bin from the sansa forums if that helps
00:12:57saratogaso I can't remove it from my player?
00:13:26kugelremove the firmware from your player?
00:13:46kugelwell, you shouldn't execute mkamsboot directly on the fuze
00:14:26kugeljust patch the firmware in mkamsboot/ and cp the patched firmware onto the fuze with a proper filename (like fuzef.bin or fuzet.bin)
00:14:58 Join tyfoo2 [0] (
00:15:33saratogaI'm trying to get the copy to patch, not patch the version already on the player
00:15:44saratogai don't care where its patched, but I would like a copy
00:16:44kugelyou want to dump it from the fuze? I don't know if that's possible. You can get a firmware file here:
00:16:58saratogaok grabbing that and putting it in the wiki
00:17:12 Quit Zagor ("Client exiting")
00:17:17kugelthey are all the same. the filename decides about features
00:17:28 Join skipper [0] (
00:17:57kugelthere's also an older version on Bagder's site
00:18:52saratogadualboot is dualboot_fuze.arm-o then?
00:20:06 Quit tyfoo (Read error: 145 (Connection timed out))
00:20:20kugelyea, but you don't need to care about that
00:20:52kugelyou just pass the OF file, and the bootloader you build using tools/configure to mkamsboot and will get a complete patched firmware
00:21:08kugel(including dualboot and rockbox loading that is)
00:21:21saratogaso just leave it blank?
00:23:02kugelleave what blank?
00:23:42saratogathe bootloader
00:24:20kugelyou need to files
00:24:24kugelthe of file
00:24:30kugeland the bootloader you built yourself
00:24:58kugelthen do "./mkamsboot of_file.bin bootloader-fuze.sansa patched_of.bin"
00:25:08saratogaso its the rockbox bootloader then?
00:25:31kugelDid I say something different?
00:26:51saratogano i misunderstood
00:31:09kugelsaratoga: I don't there's detailed information.
00:31:29kugelso maybe someone should put that into the wiki
00:31:57saratogaI'll type them up later
00:32:30kugeladd a think somewhere in t
00:32:36kugelthat sentence, btw
00:34:58 Quit mcuelenaere ()
00:36:14 Quit n1s ()
00:39:02 Join sbhsu [0] (
00:40:47 Join shot0fadds [0] (n=rob@rockbox/developer/shotofadds)
00:41:02 Join jhulst [0] (n=jhulst@unaffiliated/jhulst)
00:41:11 Join LambdaCalculus37 [0] (i=4453b1b5@gateway/web/ajax/
00:42:16 Quit jhMikeS (Read error: 110 (Connection timed out))
00:43:33 Quit kugel ("ChatZilla 0.9.84 [Firefox 3.0.4/2008111318]")
00:45:10 Join jhMikeS [50] (n=jethead7@rockbox/developer/jhMikeS)
00:46:21 Nick tyfoo2 is now known as tyfoo (
00:46:28 Quit tyfoo ("Carpe diem")
00:47:01 Quit herrwaldo ("Konversation terminated!")
00:48:27Unhelpfulif we have a doc, besides the src, that describes the greyscale LCD layouts, that'd be a help. the HORIZONTAL_PACKING case seems simple enough, the rest i'm not so sure about.
00:50:07 Quit Nico_P (Remote closed the connection)
00:50:49 Quit shot0fadds ("Leaving")
00:53:07 Quit SUSaiyan ()
00:54:33 Quit sbhsu_ (Read error: 113 (No route to host))
00:56:23 Quit tessarakt ("Client exiting")
00:58:23 Quit shotofadds (Read error: 110 (Connection timed out))
01:08:41 Join SUSaiyan [0] (
01:18:29 Join CaptainKewl [0] (
01:22:40 Part toffe82_
01:23:16 Quit culture (Read error: 110 (Connection timed out))
01:24:08 Part pixelma
01:24:35 Join pixelma2 [0] (n=marianne@rockbox/staff/pixelma)
01:26:27 Quit miepchen^schlaf ()
01:27:25linuxstbUnhelpful: No, I don't think there's a doc. Just ask...
01:29:30 Quit MethoS-- (Remote closed the connection)
01:29:42 Quit Thundercloud (Remote closed the connection)
01:30:39 Quit Lynx_ ("Konversation terminated!")
01:33:37 Join miepchen^schlaf [0] (
01:35:22 Nick Bensawsome is now known as NinJew (n=Bensawso@unaffiliated/bensawsome)
01:38:41 Quit saratoga ("CGI:IRC (EOF)")
01:42:47 Nick NinJew is now known as Bensawsome (n=Bensawso@unaffiliated/bensawsome)
01:46:59 Quit MegafEee ("KVIrc 4.0.0 Insomnia")
01:51:05***Saving seen data "./dancer.seen"
02:02:08 Join Darksair [0] (n=user@
02:08:20 Quit Darksair ("People who are zhuangbility want to show their niubility but only reflect their shability.")
02:08:44 Join BHSPitMonkey [0] (n=stephen@unaffiliated/bhspitmonkey)
02:08:48Unhelpfulvertical packing seems... inconvenient for me, but at least not hard to understand. ;)
02:09:10 Quit jeffdameth (Read error: 60 (Operation timed out))
02:10:11 Join fdinel [0] (
02:13:01 Join Darksair [0] (n=user@
02:13:35 Quit kkurbjun ("Leaving.")
02:15:52 Quit LambdaCalculus37 (" ajax IRC Client")
02:17:38 Join JdGordon [0] (n=jonno@rockbox/developer/JdGordon)
02:18:58 Nick Darksair is now known as Awaysair (n=user@
02:29:44 Join massiveH [0] (
02:36:18 Nick Awaysair is now known as Darksair (n=user@
02:36:53 Nick fxb is now known as fxb__ (
02:39:19 Quit massiveH ("Leaving")
02:41:18 Join massiveH [0] (
02:46:48 Quit Xerion (Read error: 60 (Operation timed out))
02:55:11 Quit jhMikeS (Nick collision from services.)
02:55:17 Join jhMikeS [50] (n=jethead7@rockbox/developer/jhMikeS)
02:56:24 Join toffe82 [0] (
03:47:57 Join reacocard [0] (n=reacocar@WL-135.CINE.HMC.Edu)
03:51:06***Saving seen data "./dancer.seen"
04:07:02 Quit CaptainKewl ("( :: NoNameScript 4.02 :: )")
04:09:30 Quit fdinel ("Miranda IM! Smaller, Faster, Easier.")
04:14:19 Join blkhawk- [0] (
04:21:33 Quit notplus_M (SendQ exceeded)
04:24:05 Quit Bensawsome ("The awsome is gone :(")
04:28:27 Nick Darksair is now known as Awaysair (n=user@
04:30:47 Quit blkhawk (Read error: 110 (Connection timed out))
04:31:17 Nick blkhawk- is now known as blkhawk (
04:46:27 Join aarcane [0] (
04:47:01 Join igorr [0] (
04:50:17igorrA quick question −− how does one know if a specific rockbox font supports unicode?
04:56:09 Quit reacocard (".")
04:56:51 Join reacocard [0] (n=reacocar@
04:57:10JdGordonigorr: I think has a listing of which glyphs are in which font
04:57:42 Quit aarcane (Client Quit)">
04:58:07 Join miepchen^schlaf_ [0] (
04:58:27 Join webguest94 [0] (n=3aac9a01@gateway/web/cgi-irc/
04:58:39igorrJdGordon: this looks really nice, thank you (basically, I was wondering why my cyrillic and utf-8 encoded file names looked like garbage on my iRiver)
04:59:21 Quit webguest94 (Client Quit)
05:02:07 Nick Awaysair is now known as Darksair (n=user@
05:09:35Unhelpfulooh. that's good to have.
05:12:59 Quit miepchen^schlaf (Read error: 110 (Connection timed out))
05:23:36 Part igorr
05:33:20jhMikeSDo iPod 1G/2G, Mini and Mini 2G _actually_ have a WM8721 codec?
05:37:28 Nick Darksair is now known as Awaysair (n=user@
05:43:23 Quit Horscht ("electromagnetic radiation from satellite debris")
05:51:09***Saving seen data "./dancer.seen"
05:53:11 Quit Seed ("cu, Andre")
05:56:45 Quit massiveH ("Leaving")
05:57:18 Quit JdGordon (Remote closed the connection)
05:58:48 Join JdGordon [0] (
06:26:57 Quit DataGhost (Read error: 110 (Connection timed out))
06:37:03 Join kharo [0] (
06:44:30 Join DataGhost [0] (i=dataghos@unaffiliated/dataghost)
07:07:46 Part toffe82
07:17:27 Join einhirn [0] (
07:30:22 Quit skipper (Read error: 110 (Connection timed out))
07:37:33 Quit einhirn (Read error: 60 (Operation timed out))
07:38:04 Nick Awaysair is now known as Darksair (n=user@
07:38:29 Join skipper [0] (
07:40:18 Nick fxb__ is now known as fxb (
07:41:54 Join ap0 [0] (
07:42:00 Quit ap0 (Client Quit)
07:45:02 Nick fxb is now known as fxb__ (
07:51:11***Saving seen data "./dancer.seen"
07:55:07 Join Xerion [0] (
08:04:48 Quit miepchen^schlaf_ ()
08:08:04 Join omgun [0] (n=79b45a2d@gateway/web/cgi-irc/
08:11:02 Quit omgun (Client Quit)
08:13:20 Quit skipper (Read error: 60 (Operation timed out))
08:14:02 Quit BigBambi_ (Read error: 104 (Connection reset by peer))
08:18:28 Join n1s [0] (n=nils@rockbox/developer/n1s)
08:18:54 Join Rob2222 [0] (
08:21:00 Quit pixelma2 ("-")
08:21:13 Join pixelma [50] (i=pixelma@rockbox/staff/pixelma)
08:25:01 Join lasser [0] (
08:27:12 Join kugel [0] (n=chatzill@unaffiliated/kugel)
08:28:05kugeljhMikeS: ping
08:28:18 Join stoffel_ [0] (
08:33:19 Join Bagderr [241] (n=daniel@rockbox/developer/bagder)
08:33:46 Nick Bagderr is now known as B4gder (n=daniel@rockbox/developer/bagder)
08:35:10 Part XavierGr
08:35:21 Quit BHSPitMonkey ("Ex-Chat")
08:36:08 Quit Rob2223 (Read error: 110 (Connection timed out))
08:37:52 Quit lasser (Remote closed the connection)
08:38:38 Join ender` [0] (
08:38:52 Quit ender` (Read error: 104 (Connection reset by peer))
08:39:02 Join ender` [0] (
08:39:42 Quit ender` (Client Quit)
08:40:26 Join ender` [0] (
08:46:37 Join lasser [0] (
08:53:21 Quit n17ikh|Lappy ()
09:01:02 Join petur [50] (n=petur@rockbox/developer/petur)
09:03:39jhMikeSkugel: pong
09:04:09kugeljhMikeS: have you seen my backlight fade commit candidate?
09:04:36kugelseems to work fine on all affected targets.
09:04:48 Join nuonguy [0] (
09:06:45kugeljhMikeS: but I'm still wondering about MartinR's comment
09:08:50jhMikeSkugel: I'll go look.
09:13:29jhMikeSkugel: did you solve the issue of when the brightness is 1, the bl fails to work correctly?
09:14:47 Quit GodEater (" ajax IRC Client")
09:14:58jhMikeSMartinR's comment makes no sense to me. Posting messages will not improve accuracy since thread scheduling will be done the same way.
09:16:05jhMikeSThe way to improve accuracy and give fading an advantage is to increase the thread priority when fading so it runs before lower priority threads.
09:18:20 Join GodEater [0] (i=c2cbc962@gateway/web/ajax/
09:19:37kugeljhMikeS: yea, makes sense
09:21:39kugelI just thought that backlight_tick might already start incrementing the counter while backlight_thread is doing the fade step
09:22:46kugelwhich would make sure the interval is always the same, and with timeout the interval is the timeout + the time the fading step takes
09:26:37jhMikeSdoesn't it just reverse the fade from wherever it is if the timeout happens?
09:28:14 Join jfc^3 [0] (
09:32:13kugeljhMikeS: I don't understand. I'm back in 20minutes
09:32:17 Quit kugel (Remote closed the connection)
09:34:21 Quit n1s (Remote closed the connection)
09:37:31 Join Guest_ [0] (
09:37:53 Quit Guest_ (Client Quit)
09:38:40 Join n1s [0] (n=nils@rockbox/developer/n1s)
09:39:02 Quit n1s (Remote closed the connection)
09:39:33 Join LinusN [0] (n=linus@rockbox/developer/LinusN)
09:39:38 Join n1s [0] (n=nils@rockbox/developer/n1s)
09:42:40 Join Thundercloud [0] (
09:43:08 Quit jfc (Read error: 110 (Connection timed out))
09:49:21LinusNLlorean: how about this? ;-)
09:50:06 Join Guest_ [0] (
09:50:16 Nick Guest_ is now known as MartinR (
09:50:17 Join kugel [0] (n=chatzill@unaffiliated/kugel)
09:50:31kugeljhMikeS: I'm back
09:50:59*MartinR saw his name popping up...
09:51:15***Saving seen data "./dancer.seen"
09:51:18MartinRjhMikeS: My point was just that the execution time of the _backlight_fade_* functions will add to the FADE_DELAY.
09:52:23MartinRjhMikeS: As these functions may be mutexed, fading may become irregular.
09:52:52 Quit Dhraakellian ("Reconnecting")
09:53:16MartinRjhMikeS: But it's merely visible, even at high CPU load, so...
09:54:37kugelMartinR: (in case you didn't read his answer)
09:55:34jhMikeSMartinR: mutexed? uncontended mutexes are almost a noop and I'd expect any mutexes to be so most of the time.
09:56:25 Join beast__ [0] (
09:57:09jhMikeSmost delay will come from threads executing ahead of the backlight thread. scheduling and i2c communication is microsecond level stuff.
09:58:02 Join The3_14ed|r [0] (
09:58:07 Quit The3_14ed|r (Client Quit)
10:03:36 Quit beast_ (Read error: 145 (Connection timed out))
10:05:31MartinRjhMikeS: Damn Phone. Ok, then let's stay with it, I was just kind of concerned. :)
10:06:42MartinRkugel: Good work on this!
10:07:27kugelMartinR: thanks :) But hardly possible without your work before :)
10:08:31MartinRkugel: ...and all the others before me. :)
10:08:46kugelnah, they just failed :P
10:08:58*kugel was joking
10:09:34kugeljhMikeS: so, if there's no further objections, I think it can be committed
10:11:55jhMikeSkugel: I have none
10:16:22LloreanLinusN: Well, yeah, but it's still not a bug. We don't have a Feature Request tracker any more. :)
10:17:04 Join omgun [0] (n=79b45a2d@gateway/web/cgi-irc/
10:19:36kugelLlorean: are you opposed to having this feature? Afaics Nico_P's patch implements it with minimal costs
10:19:46LinusNi would surely like it
10:20:06LinusNi'd like to double-check the license though
10:20:56LinusNlooks like it is gpl compatible
10:21:18B4gderit is
10:21:44LinusNthen i think we should commit it
10:22:05kugelsure! I'd love it
10:22:21pixelmasome people are opposed I think, haven't thought about it myself
10:24:04markunwhich patch are we talking about?
10:24:18kugeljhMikeS: logf showed the fading steps don't even need 1 tick each, except for the last one (which calls _backlight_off) which needs 13
10:24:25kugelso I think we're fine
10:26:53markunlooks interesting, although I don't see why we wouldn't have this as an option
10:26:54LloreanThe patch says it doesn't require the setting
10:27:02LloreanI'd really hate for it to do it transparently (or by default)
10:27:06LinusNkugel: i noticed one interesting thing with the backlight fading patch, the backlight stays on when playing SID tunes
10:27:57LinusNhmmm, maybe i should check without the patch as well...
10:28:25markunLlorean: I would suggest we rework it a bit a add another setting
10:28:36JdGordonadding more settings?
10:28:42JdGordonfor natural file sorting?!
10:28:50*B4gder agrees
10:28:58Lloreanmarkun: Yeah, I wouldn't mind it as a setting but I think enforcing filetree sorting to *not* respect the actual strings is a bad idea.
10:29:12pixelmaLlorean: I don't understand your last statement (the transparently)
10:29:15kugelLinusN: really? I can't really imagine that it's caused by my patch
10:29:17markunLlorean: that's what I tink as well
10:29:28Lloreanpixelma: Without any prior notification to the user.
10:29:35 Quit Thundercloud (Remote closed the connection)
10:30:21*kugel doesn't see a need for a setting
10:30:33markunkugel: that's why you don't have commit rights :)
10:30:37B4gderthen imagine file names with hex numbers in them
10:30:44B4gderand what that sort will do to them
10:30:57 Join ap0 [0] (
10:31:32kugelThere've been a plenty of "bug reports" which said "bug no natural sorting of numbers". I'd be very surprised if there was a bug report like "bug it sorts naturally"
10:31:33linuxstbB4gder: You have files with hex numbers in (on s DAP)? ;)
10:31:51*JdGordon names his files in octal...
10:31:55B4gderI can just think of situations where the "natural" sort may be a bit surprising
10:32:17kugelmarkun: I think that's not the only reason
10:32:24markunindeed :)
10:32:37B4gderkugel: I'm quite sure we'll see them too if we don't have it as a setting
10:32:38*markun goes to work
10:33:22kugelmarkun: but I don't contribute to get commit rights, but to make rockbox better, so I still state my opinion about stuff
10:33:24fluxthe only way I see people wanting pure string-based comparison would be that natural sorting doesn't work
10:33:38fluxbut I do wonder if it can work always, given the bizarre naming schemes people can have :)
10:33:52B4gderflux: which is when the names don't feature "natural numbers" I guess
10:34:02markunkugel: and I think that's a good thing
10:34:31B4gderbut I'm not feeling strongly about this topic really
10:34:34fluxyeah, perhaps someone has a directory with tunes like 59113.mp3 and 99.mp3 and really wants 9 to be after 5
10:34:44 Join pondlife [50] (n=Steve@rockbox/developer/pondlife)
10:34:44markunsometimes I see commits and wonder why it hasn't been discussed before (or if I missed the discussion)
10:34:53Lloreankugel: How's it "better" to cause files to unexpectedly sort in ways that don't reflect their filenames?
10:34:55*rasher can't imagine any non-contrived example where natural sorting would be wrong/unexpected
10:35:08LloreanAt least with "literal" sorting you can justify it. "Natural" sorting has to be incredibly intelligent to never make errors.
10:35:40LloreanAnd it's pretty trivial to have filenames that will always work with literal sorting anyway. If there's not an option, it more or less should just stay literal.
10:35:57markunmaybe we can make it the default after we've had it as a setting and everyone agrees it's better?
10:36:01kugelLlorean: and how's it better when files are unexpectedly sorted by purley their first digit when they have the track number leading?
10:36:25Lloreankugel: It's better because at least there's a 100% consistent sorting method that people can plan for.
10:36:28pondlifeIf you want to sort by track number, use tags, IMHO. Filenames might, or might not, start with a track number.
10:36:39kugelanyway, the settings submenu for sorting stuff is already there, another setting wouldn't hurt too much
10:39:18*linuxstb wonders if people will understand the term "natural sorting" though...
10:39:52B4gder"naturally I want them sorted" ;-)
10:40:02pixelmawith literal sorting you will know a way how to get the names sorted as you like it - with this natural sorting it might not be that obvious how to get them there if some files are not in the place you want it
10:40:43LinusNpixelma: could you give me an example?
10:40:51pondlifeIt's similar to the "The" prefix issue... tags can solve it anyway
10:40:58 Quit n1s (Remote closed the connection)
10:41:24 Quit jhulst (Read error: 113 (No route to host))
10:41:29kugellinuxstb: if they don't they should rtfm ;p
10:41:30LinusNwe don't want to read the tags to sort file names
10:41:38pixelmaLinusN: no, it also depends on the "cleverness" of the algorithm
10:42:00pondlifeLinusN: I mean - if you don't want to rename your files, use the database
10:42:09LloreanLinusN: I think his point was more "the database exists for sorting based on other attributes than the actual filename string"
10:42:30rasherpixelma: You might know how to get the names sorted, but Joe User won't. He'll just see "it's sorting my files wrong! 10 is larger than 4!" and moan a bit.
10:42:45pondlifeJoe the User ;)
10:42:47Lloreanrasher: most users like that use the database anyway.
10:42:51LinusNas i see it, only engineers think that ascii sorting makes sense
10:43:13*kugel agrees
10:43:18B4gder"banana22" and "c10k" , sorts in what order?
10:43:31rasherUnsurprisingly, I agree with LinusN on this as well
10:43:40rasherB4gder: b before c
10:43:51B4gderok, so it needs the same prefix?
10:44:05B4gderz10 an zz9 ?
10:44:18B4gderwhich comes first?
10:44:27 Join culture [0] (
10:44:32pondlife"54 Cymru Beats.mp3" and "50%.mp3"...
10:44:42pixelmadoes this sorting only affect numerical values or characters as well?
10:44:47 Join n1s [0] (n=nils@rockbox/developer/n1s)
10:45:05pixelmaok, should have read the title better :P
10:45:13rasherB4gder: z10. Numbers sort before letters afaik. So you're comparing "10" and "z9"
10:45:35LinusNkugel: that backlight stays on with plain svn too
10:45:46pondlifeThe change is to assume that leading numerics are a track number, no?
10:45:51LloreanWhat comes first, 007 - James Bond, or "01 - Pilot"
10:46:04LloreanSometimes leading zeros are intentional
10:46:26LinusNLlorean: he is not named 007 to be sorted in a specific order
10:46:37LinusNit's because he has a license to kill
10:46:43kugelLinusN: now idea why. I haven't any of this rare audio formats, but I'd guess it's limited to sid
10:47:11LloreanLinusN: Yes, but if I had folders named after various things, I'd REALLY liked 007 to show up at the top, instead of between "4 Rooms" and "9 and 1/2 Weeks"
10:47:20kugelonly chuck norris would sorted before 007
10:47:28pondlifeLinusN: I would not want Rockbox to assume that "59 Lyndhurst Grove.mp3" referred to track 59, or "2 + 2 = 5.mp3" was track 2..
10:47:33LinusNLlorean: that's because you are used to ascii sorting
10:48:04LinusNpondlife: then turn off natural sorting in the settings
10:48:20 Quit omgun ("CGI:IRC (Ping timeout)")
10:48:28LloreanLinusN: I think we're more trying to justify why it _must_ be a setting, which I think is agreed anyway.
10:48:33rasherpondlife: It's not making any guesses that they are track numbers. It's simply sorting using more human-friendly comparisons
10:48:45LloreanJust clarifying to some people who think it's not necessary for it to be a setting why it'd interfere with normal use for some of us
10:48:48pondlifecomparisons or assumptions...
10:48:48JdGordonthe ONLY way to do it properly is to have a .direcottyr file which could cancel natural sorting
10:48:58kugelpondlife: the patch adds nothing to the interpretation of a leading number as a track number
10:49:08 Join omgun [0] (n=79b45a2d@gateway/web/cgi-irc/
10:49:26rasherI'll dare you to find any non-engineer type that thinks ascii sorting makes sense
10:49:35LinusNrasher: +1
10:50:00fluxpondlife, you would expect to find the 2 + 2 after 59, right?
10:50:03Lloreanrasher: You'll just classify anyone I find as either "an engineer type" or "someone who's gotten used to it"
10:50:05LloreanIt's a pointless challenge.
10:50:16fluxpondlife, would you expect 2 + 2 before or after 100 + 100 ?
10:50:17LloreanBesides, why the prejudice against engineer types?
10:50:24LloreanThey're the most likely people to be installing our software anyway
10:51:02kugelJoe is also likely to use rockbox
10:51:04LinusNexactly our point
10:51:05pondlifeflux: I'd use the tracknumber ;)
10:51:17kugelonce he heard of it
10:51:25kugeland then he files a bug report about sorting
10:51:26JdGordonpeople that want natural sorting should be using the database anyway
10:51:35JdGordontrack number is the only reason to want it
10:51:36pondlifeAs long as there's a setting, I have no objections
10:51:43LloreanSo, what else do we add to sorting
10:51:52LloreanDo we interpret words like "nine" as a numerical "9"?
10:52:00LloreanI mean, how far do we support "parsing" strings in filenames?
10:52:06LinusNLlorean: that's a silly argument
10:52:15pondlifeLlorean: No, just treat leading numerics as numbers.
10:52:19LloreanLinusN: I guarantee we get a feature request for something in that line eventually
10:52:36pondlifeWell, we'd respond that it was a silly request
10:52:36LinusNLlorean: sure, but that doesn't make sense. this does.
10:52:50Lloreanpondlife: This isn't just about leading numericals. At least the article mentioned in the other channel covers sorting z1 z2 z10 rather than z1 z10 z2
10:52:51amiconnkugel: I would open such a bug report at the same instant natural sorting (without an option to disable it) would go in
10:53:05LloreanLinusN: It might make just as much sense to "Joe Average"
10:53:05pondlifeLlorean: Other channel?
10:53:10Lloreanpondlife: Community
10:53:28pondlifeIf the article is relevant, post a link here?
10:53:31rasherIt's exactly not about leading numbers. It's about treating numbers as a unit when sorting, and comparing them numerically
10:53:55amiconnLinusN: Do you have backlight on track change enabled? (re the SID problem)
10:54:12pondlifeWe could start with just leading numerics though.
10:54:15LloreanLinusN: Do we evaluate fractions?
10:54:22pondlifeThat would catch most cases of track number
10:54:25LloreanLinusN: Is 1/2 between 0 and 1, or 1 and 2?
10:54:33LloreanIf we're going by what "Joe Average" expects...\
10:54:40LinusNamiconn: aha, yes i do
10:54:48kugelafaik / ain't allowed in filenames
10:55:03LinusNLlorean: now you are being unnecessarily argumentative
10:55:05pondlife1/2 probably means 1 of 2
10:55:09pixelmaah, the "caption backlight" doesn't work correctly with .mod too because it doesn't/can't report playback times
10:55:28LloreanLinusN: My point is more that it's not wholly predictable unless you publish a detailed description of its decision making process.
10:55:40pondlifeWhat about accented characters - while we're at it..
10:55:59LinusNLlorean: it might not be, but in the typical case (leading track numbers), it is
10:56:11rasherLlorean: no one ever claimed it would match everyone's assumptions, but it sure is closer than asciisort
10:56:33LloreanAscii sorting is at least predictably by anyone simply.
10:56:34LinusNpondlife: now that's a different matter
10:56:46LloreanYou don't need to try to "guess" since you just compare characters one column at a time.
10:57:00rasherLlorean: Where "anyone" means "someone who knows these details"
10:57:13pondlifeLinusN: Yes, I guess that needs rather more .lang-style support :/
10:57:16Lloreanrasher: We have how many thousands of users, and we've had maybe 3 bug reports on it?
10:57:29kugelLlorean: to anyone who knows the ascii table well
10:57:35LloreanIt's not like it's arcane knowledge hidden in a tome at the bottom of a crypt.
10:57:42pondlifeI'd go for a simple, leading-numerics-only, version to start with.
10:57:44kugelwhich doesn't apply for most of the people
10:57:55*amiconn would rather want proper unicode collation than this so called "natural" sortin
10:58:07rasherLlorean: Doesn't mean people haven't been annoyed, and it further doesn't mean it's right, just because we've shoved something poor down the throat of the users
10:58:07Lloreankugel: That's only applicable in cases where "natural" sorting wouldn't change things anyway. Invalid point.
10:58:13LinusNamiconn: that won't help in this case (track numbers)
10:58:35 Quit omgun ("CGI:IRC (Ping timeout)")
10:58:37 Join omgun [0] (n=79b45a2d@gateway/web/cgi-irc/
10:58:51LloreanObscure characters in the ascii table would still sort in their obscure order, "natural" sorting or not.
10:59:12amiconnLinusN: Of course not
10:59:30rasheramiconn: I don't see how we could ever hope to implement proper sorting though. It's a huge task
10:59:34rasherWe might as well just give up
10:59:58kugelso, I think the consensus is clear: It can go with a setting, so that engineers can stick to their logical and always predictable asciisort
11:00:38markunand sorting is language dependant
11:01:03B4gderremember the turkish! ;-)
11:01:54LinusNbut the flamewar is not over, now we have to decide what should be the default! :-)
11:01:57markunand German and some Scandinavian language(s) have different sorting for some characters if I remember correctly
11:02:13pixelmayes, umlauts are sorted differently
11:02:28pondlifeHence my lang comment..
11:02:45rashermarkun: Indeed, and I expect this to crop up for tons of languages. I don't think it's in the scope of a DAP firmware to get this right
11:03:13kugelpixelma: and ascii sort will handle umlauts better than natural sort?
11:03:45LloreanLinusN: While I agree "natural" is intuitive for a lot of users, when it sorts 1, 10, 100, 2 you can *see* what's wrong and look for a setting. When 007 is stuck between 4 and 9, I think it's actually more confusing about what "went wrong" with where it shows up.
11:04:03kugelI don't see how this contributes to the discussion. Language specific issues will always be a problem with asciisort as well as natural sort
11:04:07LloreanAs can be evidenced by bug reports so far, they at least know what's going on when it happens.
11:04:48pixelmakugel: huh? It's not related because that natural sorting applies to numbers - at least following the patch's description. And I didn't want to state it as related...
11:06:30rasherLlorean: Only the ones who file bug reports
11:06:43pixelmakugel: hence *I* didn't mention it, just confirmed the fact markun stated ;)
11:07:35kugelOk, it should've been directed to markun as well
11:07:50Lloreanrasher: If they don't file a bug report, they probably don't think it's a bug, and so *also* understand what's going on (at least to the degree that they'll look for sort options)
11:08:01rasherLlorean: That's quite an assumption
11:08:07 Quit linuxstb (Read error: 60 (Operation timed out))
11:08:22LloreanMy point, more or less, is that I think unexpected natural sorting will look like an actual "bug" to the unexpected, while unexpected ascii sorting will just look like "stupidity" and be something they'll look for a setting about
11:08:33rasherI think the more likely scenario is "it's sorting my files weird... oh well"
11:08:42rasherThe average user doesn't file bugs. It simply doesn't happen
11:09:13 Quit culture (Read error: 110 (Connection timed out))
11:09:26LloreanSo do we want to inspire bug reports, or "ah wells"?
11:09:37rasherI don't see how this does either
11:10:00LloreanI think for someone expecting ascii sorting, natural sorting can look very buggy in edge cases.
11:10:23rasherSomeone expecting ascii sorting should know better
11:10:29kugeland who is ever expecting ascii sort when he's new to something?
11:10:46Lloreankugel: People who install third party firmwares on their MP3 players because they're FOSS geeks?
11:11:05kugelso 1% of the human population, ok
11:11:06LloreanOur primary audience here is the "enginerd", guys.
11:11:30rasherLlorean: Those people are far more likely to understand what's going on than the reverse situation
11:11:34 Quit ap0 ("Baļ")
11:11:55rasher"Oh it's doing natural sorting" is far more likely to happen than "Oh, it's doing asciisorting"
11:11:56 Quit omgun ("CGI:IRC (EOF)")
11:12:20 Join omgun [0] (n=79b45a2d@gateway/web/cgi-irc/
11:12:23LinusNalso, natural sorting is kind of common these days
11:12:24Lloreanrasher: Only if you have evidence to show you it's "natural sorting"
11:12:29markunI would suggest we just implement it as a setting and then worry about the default later
11:12:38LloreanIf your files are numbered safely for asciisort, and then you get ridiculous single out of place files, you don't have a clue.
11:12:54pixelmaso if I have my filenames already in a way "ascii sorting" sorts them correctly, the "natural sorting" would drop the leading zeros and then start sorting? Sounds complicated </just saying>
11:13:28rasherpixelma: it'd reach the same conclusion though, assuming you've set up your files to sort logically
11:13:30LinusNpixelma: and that would give the exact same result, wouldn't it?
11:13:42Lloreanrasher: Not necessarily
11:14:28pixelmayes, but with an overhead in the firmware (I just remember other occasions where it was argued that "this is what you have your PC for")
11:15:23rasherpixelma: I sincerely doubt any significant amount of time is spent sorting
11:15:47Lloreanrasher: If i have "002 - Filename" and "01 - Filename" their sort order would change depending on natural or ascii.
11:15:56pondlifeBesides, you can always set ASCII sorting...
11:16:07Llorean"natural" to me says "the DAP knows better than you do, what order to sort your files in"
11:16:12LinusNLlorean: yes, but would you name them like this and claim it is the "correct" sort order?
11:16:14LloreanWhich is irritating if you've already picked your own sort order.
11:16:28LloreanLinusN: In the 007 and 4 Rooms situation, yes.
11:16:44pixelmaand if it comes across differently, these are only thoughts for developing my own opinion... rasher: but bin size ;) - sorry to overstress this argument
11:16:59Lloreanpixelma: It's rather tiny without a setting, less than 1k.
11:17:17LinusNdiscussions like these make me tired
11:17:35LloreanIs there anywhere else in the firmware where we assume we know better than users about their own data?
11:18:19markungevaerts: I was just talking to omgun and it looks like he managed to enter DFU mode of his Samsung T9
11:18:37LinusNLlorean: this is not a case of "knowing better", it's about helping out when the user doesn't have leading zeroes in the track numbers
11:19:09LinusNit's a nice feature
11:19:23LloreanLinusN: Except we're on the point of default enabling now.
11:19:37LloreanNobody's denied it's a nice feature, just whether it's nice to override their file naming by default.
11:19:53LloreanWell, at least *I* haven't denied it's nice for some.
11:20:11LinusNmy point is that the "clueless" user should have the least resistance from the firmware
11:20:46LloreanLinusN: We've never worked on that policy before. We've always more or less exposed the core, gritty details of what's going on, and then let them do with it what they will.
11:20:47*rasher agrees
11:21:12LloreanI mean, "lease resistance" would be hiding all the advanced options until someone enabled viewing them, so they don't feel overwhelmed, etc.
11:21:24LinusNdid i say that?
11:21:32LloreanSorry, you said "least resistance"
11:21:34LloreanI typoed it.
11:21:47LinusNno, i mean about hiding options
11:22:08LloreanNo, my point is that it's the same situation, in a way
11:22:24Llorean"lease resistance" for casual users, or "technical for advanced users"
11:22:24LinusNi disagree
11:23:14LinusNif we enable natural sorting by default, i bet most users won't even notice
11:23:51pixelmaLlorean: even less than 1k could mean my voice file(s) won't work anymore on my Ondio and that number was retrieved from a Gigabeat build, there could be differences depending on architecture, but I actually didn't want to go into that argument :(
11:24:09LloreanLinusN: Do we think "most users" notice it's not available now? I'd hazard the majority of people who expect it use the database anyway.
11:24:23LloreanLinusN: It's really surprising how often people don't realize you can even get to your music through the filetree these days.
11:25:29LinusNtyp :_)
11:25:33*n1s has nothing against the feature, even if enabled by default but *please* don't call it "natural" sorting in a setting, the term does not mean _anything_ and is AFAIK not commonly used (in user interfaces) either
11:25:59LinusNn1s: that's probably because it isn't an option in other softwares ;-)
11:26:06LloreanAnyway, I'm just saying, I think it's *more* likely to inspire bug reports over time if enabled by default.
11:26:14LinusNlike Windows Explorer or XBMC
11:26:41LloreanDoes "natural" sorting apply only to numbers, or will it include things such as "the" ignoring, etc?
11:26:48LinusNonly numbers
11:27:14pondlifeI agree - call it "numeric sorting" or similar..
11:27:14n1sLinusN: probably, so if we go with a setting i'd like us to use a descriptive name
11:27:27LloreanOr just "Ignore leading zeros"
11:27:32LinusNignoring "the" is a different matter (and something i have no objection to)
11:27:33LloreanUnder the sorting options somewhere.
11:27:37fluxllorean, what would not be accurate
11:27:52Lloreanflux: "Assume infinite leading zeros for all numbers"?
11:28:02fluxllorean, well that's just awkward don't you think?-)
11:28:11LloreanLinusN: Ignoring "the" is vastly more problematical than this, I think.
11:28:25LinusNLlorean: not at all, if you stick to english :-)
11:28:29n1satomic number sorting maybe? :)
11:28:40pondlifeheh, nuclear digits
11:28:42LloreanLinusN: Only if you want ALL "The"s ignore.
11:28:51LinusNLlorean: of course
11:30:33*LinusN goes to lunch
11:31:45LloreanLinusN: Another (more radical) idea (in general). If Rockbox is booted without a config.cfg present (first boot), take them through several options they may wish to disable/enable and ask them what to set them at.
11:32:03LloreanThis way options like fade on stop-pause, where people may just think it's annoying but not optional, get shown to the user and set.
11:32:20LloreanThis would also well suit the numeric sorting question
11:33:04n1sLlorean: i like that idea, many newer phones etc do it on first start up. As long as it's easy to escape the guide
11:33:39Lloreann1s: I imagine it could also be done as a plugin that's run when config.cfg isn't found, so that there's very minimal bin cost, and if the plugin's missing, we just behave like we do currently and assume defaults.
11:34:19B4gderI think that's a good idea
11:34:22n1sLlorean: my thought exactly, but as always this would then depend on that plugin localization patch...
11:34:37n1sB4gder: any news on that?
11:35:12 Join IanS [0] (
11:35:33Lloreann1s: Well, we'd be using the localization strings from the core binary anyway.
11:35:57LloreanMaybe that could be made available in a simpler way?
11:36:51n1sLlorean: i was thinking of displaying a breif explanation for each setting in the guide, which could easily add up to a lot of kB
11:37:17LloreanWe have a manual for that anyway.
11:37:37LloreanWe could add a summary of the "wizard" plugin's options to the getting started section of the manual
11:37:53LloreanAnd maybe make it available as its own PDF if we ever get that quickstart guide idea going.
11:38:05 Nick Darksair is now known as Awaysair (n=user@
11:38:09IanShello there
11:38:25n1sthen i don't see the point, if they need to read the manual anyway the could just use the regular menu
11:38:43IanSI just wanted to say how excellent Rockbox was on my Sansa - finally I can play some decent audio formats!
11:38:48Lloreann1s: The point is to throw the most significant UI settings at them in advance.
11:38:55 Quit kugel (Remote closed the connection)
11:39:02IanSunfortunately, that infamous mainboard-breaking code caused me to drop my player and smash the screen
11:39:05 Join kugel [0] (n=chatzill@unaffiliated/kugel)
11:39:17IanSi just wondered what you think is the best hardware to run rockbox on, as i am considering purchasing another device
11:39:18Lloreann1s: Fade on stop-pause is a major one, we get bug reports on it from time to time. Sorting would be a major one, because bad assumptions there could also inspire bug reports.
11:39:44LloreanBasically, any time we would normally make an assumption about user preference by turning something on by default instead of defaulting to off and letting them enable it, instead prompt them for it.
11:39:58LloreanIanS: Depends on your needs. There's a page in the wiki called BuyersGuide that can help some
11:40:06n1sLlorean: i think it would be way more helpful to display a brief explanation in the plugin for this handful of UI settings
11:40:24Lloreann1s: Possibly *more* helpful, but I don't think it's necessary before the plugin can be useful.
11:40:41LloreanI wouldn't say "what's the point" if we can't add localized descriptions yet.
11:40:45IanSgreat i'll have a look, thanks
11:48:30 Join JdGordon_ [0] (n=jonno@rockbox/developer/JdGordon)
11:51:19***Saving seen data "./dancer.seen"
11:57:24 Quit JdGordon_ ("Miranda IM! Smaller, Faster, Easier.")
12:04:41 Join linuxstb [0] (n=linuxstb@rockbox/developer/linuxstb)
12:10:08 Quit omgun ("CGI:IRC")
12:10:27 Join omgun [0] (n=79b45a2d@gateway/web/cgi-irc/
12:11:12 Nick fxb__ is now known as fxb (
12:14:20 Join robin0800 [0] (
12:14:29 Nick Awaysair is now known as Darksair (n=user@
12:15:02 Join thegeek_ [0] (
12:15:38 Quit nuonguy ("This computer has gone to sleep")
12:16:28 Quit MartinR ("ChatZilla 0.9.84 [Firefox 3.0.4/2008102920]")
12:18:32 Quit thegeek (
12:18:32 Quit Kopfgeldjaeger (
12:18:32 Quit J-23 (
12:18:32 Quit neddy (
12:18:32 Quit Unhelpful (
12:18:52NJointhegeek [0] (
12:18:52NJoinKopfgeldjaeger [0] (
12:18:52NJoinJ-23 [0] (
12:18:52NJoinneddy [0] (n=john@nat/sun/x-214d213f74b4d691)
12:18:52NJoinUnhelpful [0] (
12:20:52 Join skipper [0] (n=skipper@
12:27:02 Part pondlife
12:29:22 Quit robin0800 (Read error: 104 (Connection reset by peer))
12:30:16 Join Llorean1 [0] (
12:30:46 Quit Llorean (Nick collision from services.)
12:30:51 Nick Llorean1 is now known as Llorean (
12:34:24markungevaerts: I download the T9 firmware and it's *not* encrypted
12:35:31 Quit thegeek (Connection timed out)
12:41:27 Join Nico_P [50] (n=nicolas@rockbox/developer/NicoP)
12:42:41 Join moos [0] (
12:44:01 Join Seed [0] (
12:45:54markundoes anyone have some info about Samsung's Whimory FTL?
12:46:55markundenes_: I found out that the Samsung T9 (also S5L8700 based) uses the Whimory FTL as well
12:48:56kugelNico_P: hi. there was a discussion about your (rotting) natural sorting patch
12:49:12Nico_Pkugel: oh, interesting. I'll read the logs
12:50:08 Quit omgun ("CGI:IRC (Ping timeout)")
12:58:25 Quit japc (Read error: 110 (Connection timed out))
12:58:41 Quit kugel (Remote closed the connection)
12:59:37Nico_PIMHO very few "normal users" would prefer ASCII sorting over natural sorting
13:00:25Nico_Pto me, the correct question is whether rockbox attracts (or should attract) these "normal users"
13:04:18LinusNi believe i would prefer natural sorting
13:05:09LinusNmost of the time i wouldn't notice a difference, and when i do find tracks without leading zeroes in the filename, i woudl appreciate it when it sorted them numerically
13:05:32Nico_Pit's pretty much the same for me
13:09:40LinusNbesides, i think some features may be unnecessary because there exists a workaround, but it is imho much nicer when it "just works"
13:09:56LinusNthis is one of those cases
13:10:03LinusNembedded album art is another
13:12:07GodEateras a non "normal user" I think EAA is the biggest waste of disk space ever :(
13:13:15LloreanYeah, but supporting it wouldn't waste your disk space at least.
13:13:43LloreanI imagine the issue is more "jpeg in the core" than EAA anyway. I can't imagine reading the metadata out of one more tag is huge overhead compared to jpeg decoding.
13:15:23GodEaterNo I suspect not
13:15:35GodEaterI'm just not in a hurry to go embedding the same jpeg into 18 files
13:15:52GodEaterso long as we keep our existing AA solution (add jpeg support to that would be nice), I'll be happy
13:16:32LloreanI can't imagine removing our existing AA solution, since it also beautifully allows fallback images for Artists if an album image isn't there, and further fallback for genre, etc, right?
13:16:40LloreanSince it looks in parent folders?
13:17:05GodEaterI just don't want the bin-size crowd asking for one solution or the other - I'll cry :)
13:17:57LloreanBeing occasionally part of the "binsize" crowd, my view on binsize is "if you'd be willing to remove it for something else later, don't add it in the first place"
13:18:07Nico_PI agree EAA is a waste of space for albums, but it seems to have become quite a de-facto standard and it's quite practical for single tracks
13:18:07 Quit jhMikeS (Read error: 104 (Connection reset by peer))
13:18:24LloreanSo, we shouldn't be removing functionality (which this would) for the sake of binsize, and we shouldn't add something we'd be willing to sacrifice for something else later (assuming that sacrifice wasn't transparent to the users)
13:18:39LinusNand it's a thing that would make rockbox shine
13:18:50GodEaterNico_P: if people are going to use it, it's the best argument for single MP3s and cuesheets imo.
13:20:19 Join jhMikeS [50] (n=jethead7@rockbox/developer/jhMikeS)
13:20:34Nico_Pchaptered MP4 files are rather close to that
13:21:26Nico_PI would really like to have nice support for that. with pics and all
13:22:37Nico_Pthis type of debate makes me realize that we don't know much about our typical users
13:22:54*GodEater doesn't think we have "typical" users
13:23:09pixelmado we have to?
13:23:33Nico_PI think we should
13:23:43*gevaerts grabs his lamp and goes searching for a typical user
13:24:07B4gderlamp.rock ?
13:24:13Nico_Pi.e. we could perhaps think about running a survey to get an idea of what people value the most
13:24:35*GodEater thinks this qualifies as "a can of worms", or "pandora's box"
13:24:50gevaertsHow? How many forum users do we have compared to total downloads?
13:24:57gevaertsB4gder: of course :)
13:25:17 Join Bensawsome [0] (n=Bensawso@unaffiliated/bensawsome)
13:25:32 Join robin0800 [0] (
13:25:35Nico_PGodEater: we wouldn't have to let it define the future directions. it would just give us a clearer view of things
13:25:59 Quit robin0800 (Remote closed the connection)
13:26:03*gevaerts thinks that assuming that forum users are typical users is probably just as unfounded as assuming that developers are
13:26:06pixelmaNico_P: and what would you do with the outcome? Code something else, even if it is something you are not interested in?
13:26:18 Join robin0800 [0] (
13:26:53 Quit robin0800 (Remote closed the connection)
13:27:11Nico_Ppixelma: seem my statement above. I would probably use the outcome to weigh and order my priorities
13:27:59 Join robin0800 [0] (
13:28:30 Quit robin0800 (Remote closed the connection)
13:31:47n1snow that we have a freeze coming up in a couple of weeks (we do, right?) would it not be a good time for a "tracker cleanup" or at least a focused look at open bugs and patches?
13:31:58linuxstbNico_P: You could hide some spyware in Rockbox that reports the user's behaviour to rbutil...
13:32:35B4gdern1s: you're right, it would be a good time...
13:32:46LinusNbtw, i like the wizard plugin idea
13:32:52*linuxstb wasn't sure if a freeze-date had been agreed
13:33:29Nico_Plinuxstb: haha, awesome idea :)
13:33:58Nico_Psome projects actually do that btw
13:34:34linuxstbYes, I can imagine... In fact, rbutil could report some info without any changes to Rockbox - e.g. the config.cfg and statistics on the types of audio files on the user's device.
13:35:13B4gder"would you like rbutil to anonymously report about your config to the rockbox developers?"
13:35:37linuxstbOr rather, would you like rbutil to uplaod the following file to the Rockbox develoeprs? (showing the actual data being sent)
13:35:48B4gderthat might be even better, yes
13:36:09LloreanHow much would that really tell us?
13:36:14LloreanOr rather, how much could that information affect?
13:36:32linuxstbI don't know until we have that info...
13:36:45LloreanBut, I mean, what question are we trying to answer?
13:37:07LloreanIsn't it features we *don't* have that are more interesting?
13:37:12n1sok, so i'll send a mail to the list urging people to look at the tracker, confirm, fix, close, commit, set "due in version" on important ones etc. next week ok?
13:37:35B4gdern1s: sounds fine to me!
13:37:52n1sbtw, _did_ we decide on a freeze date?
13:37:59 Join tyfoo [0] (
13:38:31linuxstbLlorean: I'm just curious about how Rockbox is used. It won't necessarily change anything.
13:38:40 Quit shodanX (Read error: 110 (Connection timed out))
13:39:09 Join funman [0] (
13:39:13LloreanAh, well that's okay then.
13:39:33linuxstbn1s: I don't know, I think the discussion was split between 3 months (i.e. just before Christmas) and 4 months (end of January) for the release date.
13:40:24n1slinuxstb: oh, so when christmas comes up we'll just say "oh, we never froze, lets release in january" :/
13:40:28funmanhi, I found a bit more details about how the SD driver fails in the sansa ams bootlaoder
13:40:40Nico_Phaving information about what settings are used *could* help us set good defaults
13:41:36linuxstbNico_P: Yes, but it's hard to say if someone uses the current default because he/she prefers it, or simply because he/she didn't know it could be changed.
13:41:40funmanthe first sector is well read, the 2nd and 3rd sectors are read (2*127 sectors) too far in the file; the 4th to 9th sectors are read (1*127) sectors too far away in the file
13:42:01LloreanI think most of the defaults are "okay" in terms of "they're features that should be left off unless someone wants that specific feature" like equalizer, etc.
13:42:16funmanit could (well, must) be related to the fact that we want to read maximum 127 sectors at a time
13:42:21LloreanThere's a few that are matters of taste/desire (like sorting would be) that would be better served by the wizard than trying to guess whether to turn something on or not.
13:42:46LloreanEspecially given that people will often assume there's *not* a setting if software appears to do something they don't think is a sane default.
13:43:18pixelmacould the wizard idea be extended for RBUtil? I.e. during the first install RBUtil could already ask the user about some settings and write the config.cfg accordingly
13:44:20LloreanIf it didn't execute unless config.cfg was present, it could be in either or both without real problems.
13:50:07 Join kugel [0] (n=chatzill@unaffiliated/kugel)
13:51:23***Saving seen data "./dancer.seen"
13:52:07 Join kushal_12_27_200 [0] (n=kushal@
13:55:34 Join shodanX [0] (
13:56:19 Join Lynx_ [0] (
13:58:29kugelfunman: the button driver isn't quite reliable. I sometimes get just a black screen (i.e. the clear of the lcd worked) without text (i.e. the printfs after the clear did not work)
13:58:56kugelin the bootloader that is. As for the main, going by e200v2 testers, buttons do not work at all
13:59:09funmankugel: i'm not sure what that means; how do you know this problem is caused by buttons ?
13:59:36kugelfunman: because I only get the black screen when I press a button
13:59:48kugelalso, buttons and lcd are connected in someway on the e200v2/fuze
14:00:17kugelremember the afsel has to reset after to make the lcd work again
14:00:21funmankugel: perhaps try to disable irq before modifing gpio*_afsel, and enable them when you restored the registers ?
14:00:42kugelI could try that, yea.
14:01:47funman"going by e200v2 testers" : does that mean you didn't try to use buttons in the main build ?
14:01:59kugelAlso, I'm not entirely sure, but I think the black screen issue appears a bit more often when I remove the gpio*dir stuff. Maybe be randomly happen as well though
14:02:23kugelfunman: no not yet, since I yesterday only tried to find some more buttons
14:02:44 Join nplus [0] (n=nplus@141.25.Globcom.Net)
14:03:08funmani bet if they didn't find them on e200 you won't find them either; perhaps the other buttons & wheel are read through DBOP like fdinel suggested
14:03:09kugeland I used the bootloader which doesn't work so I never came to the main build (which wasn't bad for that purpose)
14:04:03kugelcan one see that by the traces on the pcb? the scrollwheel and the buttons both go in the zif connector
14:04:29kugelthat's why I wouldn't think they're read totally different, but who knows
14:05:18funmanI fixed the bootloader \o/
14:05:36kugelfunman: I tested some GPIOA, but got a white screen only by reading them, can you explain that?
14:05:58funmankugel: no i can't. perhaps those pins you tested have a special meaning ?
14:06:30kugelthe A is mostly for that other i2c bus, not sure how that's related to the white screen
14:06:59kugelgreat job, are you going to commit the dma stuff soon'ish now?
14:07:52funmanyes of course ! :) the problem was that i was not incrementing the destination (SDRAM) pointer, so the data was overwritten if we were transfering more than 127 sectors; which only the bootloader appears to do
14:08:52kugelthe bootloader appears to some other stuff different as well, given that the buttons don't work in the main build but in the bootloader
14:09:26funmankugel: gpio_*dir should be set in button_init_device() and not touched I think
14:09:37kugelwell, the buttons could be due to quite frequent lcd operations (e.g. updating the statusbar)
14:09:45funmanfor example
14:11:45kugelhaven't you said gpio*dir isn't not needed at all?
14:12:19funmanyou need to set the proper direction once; you don't need to change it if the pins do not have another usage
14:12:54funmanIf the *gpio* pins have no other usage. gpio*_afsel selects if the *physical* pin is bound to be a gpio pin or a dbop pin
14:13:05 Quit reacocard (".")
14:13:30kugelah I understand
14:13:55funmanat least that is my understanding .. :)
14:14:13 Join reacocard [0] (n=reacocar@WL-135.CINE.HMC.Edu)
14:15:03 Join beast [0] (
14:15:05kugelfunman: why do we even save the afsel? Wouldn't it be enough to set them up for buttons, and after the reading, set them up for lcd again?
14:15:35 Join massiveH [0] (
14:15:48funmansure, but imo this method is cleaner; because you don't have to know the LCD settings in the button code
14:16:43funmangogogo sansa ams supported in rockbox 3.1 !
14:16:58kugelassuming we leave them for lcd after button_int should be safe imho
14:17:56kugelwe only need to have them set up for the buttons in button_int, otherwise it's basically always need to set up for lcd
14:18:15 Quit gevaerts (Nick collision from services.)
14:18:24 Join gevaerts [0] (n=fg@rockbox/developer/gevaerts)
14:18:57kugelmeh, I'm gonna dig into the data sheet for that once more
14:19:47kugelgtg, will see you later with hopefully committed dma code
14:19:53 Quit kugel ("ChatZilla 0.9.84 [Firefox 3.0.4/2008111318]")
14:28:44 Quit tvelocity (Read error: 110 (Connection timed out))
14:29:15 Join tvelocity [0] (
14:30:39 Quit beast__ (Read error: 110 (Connection timed out))
14:33:31 Part B4gder
14:36:42 Quit kharo ("Leaving.")
14:37:17funmanmake zip shows "<stdin>:1:20: error : config.h: No such file or directory"
14:46:11 Join LambdaCalculus37 [0] (i=44a04303@gateway/web/ajax/
14:47:15funmankugel: (for the logs) I committed the SD/DMA code. Don't forget that you could need to set some pins to DBOP if the other buttons are to be read on this port.
14:47:24 Join {phoenix} [0] (
14:51:16 Quit amiconn (Nick collision from services.)
14:51:22 Join amiconn_ [50] (n=jens@rockbox/developer/amiconn)
14:53:04funmanI'm not sure what's the bottom right icon in WPS : if it's PLAY, does it mean it's PLAYing, or it is going to be PLAYing if I press the play button ?
14:53:22 Quit massiveH ("Leaving")
14:57:34n1sif it shows play it should be playing
14:59:42funmanLambdaCalculus37: you can update your Clip to last firmware to have flac support also
15:00:12funmansince you have a 2GB model, can you check which size is reported by the internal SD ?
15:00:16 Nick jfc^3 is now known as jfc (
15:00:23LambdaCalculus37funman: Can I still mess around with custom code even with 01.01.30?
15:00:35funmanLambdaCalculus37: of course, all versions are supported
15:00:47LambdaCalculus37funman: Cool.
15:01:16funmanwhich means 1.01.{17,18,20,29,30} for the Clip
15:01:55funmanI have pretty advanced reverse engineering of the SD code in the Clip firmware, if your 2GB is reported as 1GB you could be the one that fix the SD driver for >1GB access on the internal storage!
15:01:56LambdaCalculus37funman: Memory is reported as 1949MB.
15:02:03funmanLambdaCalculus37: by rockbox?
15:02:10LambdaCalculus37funman: By the OF.
15:02:18JdGordonfunman: you got dma working properly?
15:02:29LambdaCalculus37I can't do anything Rockbox related while my Mac is on the blink.
15:02:39funmanJdGordon: yes, I was overwriting the transferred data ^^
15:02:46JdGordon:) well done
15:02:53funmanLambdaCalculus37: no hurry, I can code fine with my 1GB Clip ^^
15:04:59*funman checks his list: sound, fm, recording, usb
15:04:59LambdaCalculus37funman: Not unless you want to send me a build zip and a bootloader so I can mess around with Rockbox on it for the time being.
15:06:22funmanLambdaCalculus37: sure, just give me your email in private
15:07:38 Join japc [0] (n=japc@
15:07:58 Quit shodanX (Remote closed the connection)
15:08:56 Join shodanX [0] (
15:10:43 Join saratoga [0] (n=9803c6dd@gateway/web/cgi-irc/
15:12:22 Join Schmogel [0] (
15:13:06kushal_12_27_200Is rockbox freezing in recording mode for Sansa c250? I used to have the latest builds on my sansa and I thought it was rockbox, but I downgraded to Rockbox stable and the issue remains. The player just plain freezes whenever I go to recording mode via the rec button on the right of hold button. Do you know what's happening? Thanks.
15:13:09 Quit funman ("leaving")
15:13:18 Join jeffdameth [0] (
15:13:32 Join blahrus [0] (n=blahrus@
15:18:15n1skushal_12_27_200: do you have the "keyclick" feature enabled?
15:34:25 Join lasser_ [0] (
15:39:08*LambdaCalculus37 has Rockbox running on his Sansa Clip
15:40:19 Quit Llorean ("Leaving.")
15:45:36JdGordonn1s: marking all the tasks as "due in 3.1" is probably silly ebcause alot of them wont happen which means more work next time... cant they be marked "next release" instead?
15:45:54 Quit Lynx_ (Remote closed the connection)
15:46:24n1sJdGordon: i meant that you should mark the ones you feel are important for 3.1 not everything :)
15:46:58JdGordonI was talking more about the ones you changed from 3.0 -> 3.1 becuase they wern't closed
15:47:08 Join Lynx_ [0] (
15:48:24 Join Llorean [0] (
15:48:27n1sJdGordon: ah, well, sure they can be changed to something else it's allready done though and someone felt they were important for 3.0
15:49:09n1salso a generic "Next release" is basically the same as the unmarked bugs as that would just be a "would be nice to get fixed"
15:49:59n1sBut if you feel like you want to change them, feel free :)
15:50:40JdGordonwell "next release" means that someone feels strongly enough that it needs some sort of priority... but really the whole thing is useless because its obvious "we" dont really care about the bug count
15:51:02kushal_12_27_200nls, keyclick is set to moderate.
15:51:09 Quit lasser (Read error: 113 (No route to host))
15:51:22n1skushal_12_27_200: it needs to be disabled when recording
15:51:28***Saving seen data "./dancer.seen"
15:51:41kushal_12_27_200thanks, got it
15:53:57n1sJdGordon: there is a priority field too
15:59:40JdGordonis there a bug report on that keyclick issue?
16:02:09 Join pondlife [50] (n=Steve@rockbox/developer/pondlife)
16:04:24 Join kugel [0] (n=chatzill@unaffiliated/kugel)
16:07:50 Join Strife89 [0] (i=a810ebc0@gateway/web/ajax/
16:09:41markundenes_: did you take a look at ?
16:10:47 Join robin0800 [0] (
16:11:34kugelLambdaCalculus37: did you receive a build from funman already?
16:12:12kugelwell, i guess you'd also need a svn bootloader
16:12:47LambdaCalculus37kugel: Got a build and a bootloader.
16:13:09kugelI'm interesting in the result, i.e. will rockbox detect 2 gb
16:17:09LambdaCalculus37kugel: Rockbox says the Clip's storage is 1.90GB and 1.86GB free.
16:18:45LambdaCalculus37Ha! I have 0KB of buffer! :P
16:19:06 Join funman [0] (
16:19:20 Quit kushal_12_27_200 ("Leaving")
16:21:08 Quit JdGordon (Success)
16:22:42funmanLambdaCalculus37: then we will probably need a 4GB Clip to reproduce this problem, I doubt it's specific to Fuze and e200v2
16:24:22kugelLambdaCalculus37: so it at least detects full memory
16:24:51LambdaCalculus37funman: There is one thing, though... in the debug menu, viewing the partition info shows me four partitions of 0MB each.
16:25:03kugelfunman: yea, the 2GB probably don't tell much, since non-SDHC goes upto 2GB
16:25:25LambdaCalculus37kugel: Now I have 288KB of buffer free after a restart.
16:25:38kugelLambdaCalculus37: well, fdisk will look similar. The partition layout is pretty messed up and hacky
16:26:21funmanLambdaCalculus37: perhaps you have a superfloppy and no partitin table ?
16:26:45 Quit kugel (Remote closed the connection)
16:26:48LambdaCalculus37funman: Perhaps, but right now I have no Unix or Linux near me to work with to find out.
16:27:02funmanby default my Clip had no partition table
16:27:16funmanand windows xp at least has a volume manager in "admin tools" iirfc
16:28:48 Quit Strife89 (" ajax IRC Client")
16:33:02 Join kugel [0] (n=chatzill@unaffiliated/kugel)
16:33:21 Quit Schmogel ("Miranda IM! Smaller, Faster, Easier.")
16:33:34 Join Schmogel [0] (
16:34:35kugelfunman: time to implement a proper shut down. holding power 10s is still annoying ;)
16:35:22kugelbut first fix the buttons in the main ofc (indeed, they neither work on the fuze)
16:36:04funmankugel: it's done already.
16:37:32kugelfunman: haha buttons can't work
16:38:01kugelbutton_int call is within #ifdef BOOTLOADER
16:38:33funmanby the way is button_int() a representative name ? it makes me think of _int_errupts
16:39:05funmanin my opinion all the copy pasted code from e200v1 should go away. It's still in e200v1 code if we need to look at it.
16:39:06 Join toffe82 [0] (
16:39:48kugelfunman: is the whole asm code only irq enable/disable?
16:39:55kugelthat one in the bootloader
16:40:17kugelif yes, it should be in a function so that other parts and the core can use the asm verison
16:40:58funmanit's already a (generic to all ARM targets) function. I changed it in my tree but didn't commit yet. I guess I am waiting to include it in the next bootloader diff ;)
16:41:40funmanI think the bootloader is finished now
16:42:23kugelfunman: eh, yea, in the pastebin'd patch it was en-/disable_irq()
16:47:43kugelfunman: finished bootloader sounds good
16:48:27funmanindeed, good progress has been made
16:50:28kugelin fact, finished alone sounds already good :p
16:51:27 Join webguest44 [0] (n=407c962a@gateway/web/cgi-irc/
16:53:09 Quit webguest44 (Client Quit)
16:55:05 Quit Schmogel (Read error: 110 (Connection timed out))
16:56:25kugelfunman: disabling irq in button_int ain't good
16:57:42funmankugel: you're confused. A lot of code has been copied from e200v1, and doesn't work on e200v2. The e200v2 doesn't use interrupts at all to read the buttons
16:59:22 Quit jhMikeS (Read error: 54 (Connection reset by peer))
16:59:29kugelfunman: you suggested to disable interupts while button reading to make it more reliable
17:00:15funmanOh right, I suggested to disable interrupts when the gpio*_afsel registers are being modified
17:00:55 Join jhMikeS [50] (n=jethead7@rockbox/developer/jhMikeS)
17:00:57kugelok. I now just put disable at the very beginning and enable at the very end
17:00:58funman"ain't good" = still leaves you with blank screens ?
17:01:07kugelit entirely messes everything up
17:01:31kugelthe display flickers horribly for a few seconds, then the screen turns white
17:01:51kugeleventually I see the rockbox menu for a split second, sometimes it's even mirrored
17:02:00kugelanyway, gonna be back in a few minutes
17:02:05 Quit kugel (Remote closed the connection)
17:03:00LambdaCalculus37funman: The LCD display on the Clip is rather shaky when scrolling through menus.
17:03:44 Join bmbl [0] (n=Miranda@unaffiliated/bmbl)
17:04:49funmanLambdaCalculus37: I didn't meet any problem, perhaps we need to insert some delays in the lcd code?
17:11:03 Join mib_v2yq05 [0] (i=2669ecc2@gateway/web/ajax/
17:11:20 Nick mib_v2yq05 is now known as CaptainKewl (i=2669ecc2@gateway/web/ajax/
17:16:52 Quit reacocard (".")
17:19:31 Part pondlife
17:23:41 Quit linuxstb (Read error: 110 (Connection timed out))
17:24:41 Quit perrikwp (" ajax IRC Client")
17:24:54 Join nibbler [0] (
17:25:25Unhelpfulall right, scaling works in greyscale ipod sim. :D
17:26:24nibblerlittle ot, can the harddisk of archos 5 be easily replaced?
17:32:41Lloreannibbler: Please, respect the fact that this channel is for Rockbox discussion. Off-topic things can be asked elsewhere, and should. Thanks.
17:33:06 Join herrwaldo [0] (
17:33:44 Join kugel [0] (n=chatzill@unaffiliated/kugel)
17:33:50 Quit robin0800 (Connection timed out)
17:34:08 Join domonoky [0] (n=Domonoky@rockbox/developer/domonoky)
17:35:10 Quit saratoga ("CGI:IRC (EOF)")
17:38:52kugelfunman: hmm..disabling irq in bootloader doesn't seem to hurt
17:39:37funmankugel: they are enabled in system-xx as well I think
17:39:43 Quit nibbler ("Ex-Chat")
17:39:52Unhelpfulipod4g bin/ram/total sizes: vanilla: 521484/918304/1439788, "old" resize-on-load: 524220/931088/1455308, "new" resize-on-load: 523788/920624/1444412
17:40:06kugelfunman: I mean the disabling in button int
17:40:20funmankugel: does it do any good ?
17:40:27kugelnot sure why the same code doesn't work in main
17:41:20kugelwell no, I still get the black screen
17:41:28kugelso just forget about that
17:42:23*funman will do
17:44:01kugelUnhelpful: what is the second number? difference between bin and ram?
17:44:13Unhelpfuloh, wait.
17:44:21Unhelpfulsecond is ram, third is total.
17:44:49Unhelpfulso, up ~5KiB vs vanilla, but down ~10KiB vs old-resize
17:46:55kugelehh, the whole button_int doesn't work even without irq disabling
17:47:59LambdaCalculus37funman: LCD line updates are a little weird looking. The lines seem to shift about 2 pixels to the right, and a random pixel appears to the very left.
17:48:16LambdaCalculus37Also, the runtime clock is a bit fast.
17:49:34lasser_ok, yesterday fml asked for runtime comparison on e200 between r19494 and before. Yesterday r19195 ran 7:13h with remaining 62% battery level. Today r19173 drained the battery under the same conditions to 55% after 7:13h - or the other way r19173 ran 6:09 to reach 62% batt level. Conclusion: There is an improvement about 15%.
17:49:35 Quit DataGhost (Nick collision from services.)
17:49:43 Join DataGhost [0] (i=dataghos@unaffiliated/dataghost)
17:50:05funmanLambdaCalculus37: which runtime clock? the kernel ticks ?
17:50:15LambdaCalculus37funman: No, the "Running time" clock.
17:50:22LambdaCalculus37In System > Running Time.
17:50:26 Join pondlife [50] (n=Steve@rockbox/developer/pondlife)
17:50:30 Quit pondlife ("Leaving.")
17:51:21funmanif it's derived from kernel tick, yes it's too fast. One need to read the timer doc and fix kernel-as3525.c tick_start() accordingly
17:51:24Unhelpfuland the newest patch is on FS #9458. testers appreciated, should work on color targets, and greyscale ipods
17:51:31***Saving seen data "./dancer.seen"
17:51:56pixelmano other greyscale targets?
17:55:51Unhelpfulpixelma: not yet, but they'll come more quickly now that the framework of the nearest-neighbor scaler is done
17:56:27pixelmaI see
17:57:41Unhelpfulthe only thing that should need to change is the innermost loop where pixels are stored to the destination bitmap. the rest is just stepping over input and output rows and columns by the correct increments
17:58:59kugelcan I block lcd operations while reading buttons?
17:58:59Unhelpful"ipods" is gleaned from the old-scaler comments. if there are other 2bpp greyscale targets using horizontal packing, they should work, too.
17:59:43 Join pondlife [50] (n=Steve@rockbox/developer/pondlife)
17:59:55 Join faemir [0] (
18:01:58 Quit Lynx_ ("Konversation terminated!")
18:03:48kugelpondlife: hey. I think my backlight fade patch is ready to go in
18:04:33kugeljhMikeS has no objections.
18:08:20LambdaCalculus37kugel: At long last? :)
18:08:27 Join linuxstb [0] (n=linuxstb@rockbox/developer/linuxstb)
18:08:34kugelLambdaCalculus37: huh?
18:09:02LambdaCalculus37kugel: That your patch is finally ready to go in.
18:09:06kugellinuxstb: hi. I've got a problem with the fuze's lcd and buttons
18:09:21*kugel didn't know that phrase
18:10:14kugelI'd rather not call it officially "my" patch (if I did, only because it's shorter). Some more people contributed to it
18:11:08kugellinuxstb: the lcd goes crazy when I enable it for the main (i.e. remove the #ifdef BOOTLOADER around the button_int call)
18:12:25 Quit funman ("leaving")
18:12:26kugelI assume there's lcd operations so frequently, that the the safe and reset of afsel isn't quick enough
18:14:04 Quit stoffel_ ("Lost terminal")
18:15:46 Join MethoS- [0] (
18:15:52 Quit petur ("*plop*")
18:17:04 Quit skipper ("Leaving")
18:23:10pondlifekugel: Do you have commit access? Or are you looking for a volunteer? ;)
18:24:32 Join dany_21a_ [0] (
18:25:27 Quit IanS ("Ex-Chat")
18:26:38 Quit ender` (" I spilled Spot Remover on my dog... Now he's gone.")
18:26:56kugelpondlife: I don't have commit access
18:27:20pondlifeI'd be happy to commit it, but probably not until the morning
18:27:29kugellinuxstb: hm, pressing up inverts the lcd colors
18:27:35pondlifeMaybe someone else would like to pop it in first though
18:28:00kugelpondlife: No worries, I can wait until tomorrow :)
18:28:15kugellinuxstb: and does not what it's supposed to be
18:28:46kugelright button seems to initiate lcd_enable :S
18:29:02kugelweird stuff
18:31:08 Quit Darksair ("Zzz...")
18:31:11 Join nuonguy [0] (
18:40:29 Part dany_21a_
18:41:26kugelwhat are all these errors?
18:41:48kugelI just #included system-target.h
18:43:09kugelprobably missing implementations..
18:43:22kugelI am talking about btw
18:46:40amiconn_Unhelpful: There are 3 different pixel packings for 2bpp
18:47:44 Join Schmogel [0] (
18:48:29amiconn_iPods use horizontal packing (4 pixels per byte), iriver H1x0 and iAudio M5 use vertical packing (4 pixels per byte), and the iAudio remote (which is the main display for iAudio M3) uses a different vertical packing (8 pixels in 2 bytes, with the corresponding bits in the byte pair belonging to the same pixel)
18:48:36 Nick amiconn_ is now known as amiconn (n=jens@rockbox/developer/amiconn)
18:48:56 Join ender` [0] (
18:49:12 Nick fxb is now known as fxb__ (
18:49:53linuxstbkugel: Try including system.h, not system-target.h
18:51:17kugellinuxstb: thanks that worked
18:51:37kugellinuxstb: do you have an idea regarding what I wrote you?
18:53:55 Quit ender` (" The Web is a procrastination apparatus: It can absorb as much time as is required to ensure that you won't get any real wor")
18:54:16 Quit {phoenix} (Read error: 104 (Connection reset by peer))
18:54:25 Part LinusN
18:56:31 Join hannesd [0] (
18:57:18 Join miepchen^schlaf [0] (
18:57:24 Join Rondom [0] (
18:58:35 Join el_sneako [0] (n=54bd4f5e@gateway/web/cgi-irc/
19:00:19 Quit el_sneako (Client Quit)
19:00:51 Part hannesd
19:02:55 Join bertrik [0] (
19:05:23 Join {phoenix} [0] (
19:05:26 Join Thundercloud [0] (
19:05:26 Quit moos ("Rockbox rules the DAP world")
19:05:52 Quit LambdaCalculus37 (" ajax IRC Client")
19:07:02 Join ender` [0] (
19:08:18 Join massiveH [0] (
19:11:20 Join culture [0] (
19:15:27 Join MegafEee [0] (n=Linux@unaffiliated/megaf)
19:19:27 Quit miepchen^schlaf ()
19:20:14 Join jhulst [0] (n=jhulst@unaffiliated/jhulst)
19:25:12 Join Horscht [0] (n=Horscht@xbmc/user/horscht)
19:26:48 Join LambdaCalculus37 [0] (i=44a04303@gateway/web/ajax/
19:29:45linuxstbkugel: No, I don't.
19:30:11kugellinuxstb: it seems the afsel stuff isn't working properly, since I "control" the display with the buttons
19:31:05 Join BigBambi [0] (n=Alex@rockbox/staff/BigBambi)
19:31:35 Join Horschti [0] (
19:32:21 Quit Horscht (Nick collision from services.)
19:34:16 Join miepchen^schlaf [0] (
19:36:00 Nick fxb__ is now known as fxb (
19:37:53kugelbut why is it working in the bootloader, and not in the main :S
19:50:11bertrikkugel, are you able yet to browse the menus on your fuze?
19:50:13 Join XavierGr [0] (n=xavier@rockbox/staff/XavierGr)
19:50:48 Join aarcane [0] (
19:50:52kugelbertrik: no, if I enable buttons for main, the display goes crazy
19:51:08bertrikhow reliable is the new sd dma code for as3525? I see write support was enabled, so I'm a little scared of the risk of bricking my clip
19:51:16kugeland I rather control the display (i.e. up button inverts the display colors)
19:51:34***Saving seen data "./dancer.seen"
19:51:49kugelbertrik: working so far. I obviously haven't done significant write operations yet
19:52:28domonokybertrik: seems its pretty reliable.. havent seen it doing something wrong yet, but of course its not much tested...
19:53:47kugelI guess we need proper interrupt handling for e200v2 and fuze buttons
19:53:55domonokyand bricking risk should be minimal, sd-code doesnt allow access to the firmware space. so it can only mess up the fat partition, which should be fixable..
19:55:18kugelon the fuze the firmware even booted with a partition table destroyed by gparted
19:55:41bertrikdomonoky, ok
19:56:02bertrikkugel, why would we need interrupt handling?
19:56:30kugelbecause lcd and buttons are sharing some gpio pins.
19:57:15kugelthe buttons aren't gonna work if the lcd is constantly changing the gpios used for the buttons
19:58:15 Quit japc (Read error: 145 (Connection timed out))
19:58:59 Join funman [0] (
20:00:07*gevaerts wonders if his d2 problems are due to NAND issues or to something else entirely
20:00:49bertrikkugel, ok I understand but I don't understand what interrupts have to do with this
20:01:57kugelthe button driver shouldn't be interrupted between setting up the gpio for reading, reading itself and resetting the gpios again
20:02:17funmanperhaps interrupts on gpio pins changing can't happen when the physical pins are mapped to DBOP
20:02:24kugelsimply disable_irq doesn't work though
20:02:58domonokykugel: the button driver can only be interrupted with interrups, (cooperative multitasking) so you could just disable the ints, while reading the buttons..
20:03:22kugeldomonoky: as i just said, I tried that. Result is that rockbox doesn't load
20:03:38funmandomonoky: yet the kernel tick interrupts does "things", I don't know if these "things" include lcd access
20:03:39kugelas of now, I somehow control the display with the buttons instead of browsing or something
20:03:56domonokykugel: you disabled at the start of button_read, and enabled at the end again ?
20:04:33kugelthis, and only dis/enabling when setting the afsel
20:04:38kugelboth didn't work
20:05:07domonokyshortly disabling the interrupts should not prevent booting, so i think you did something wrong...
20:05:18funmanshouldn't the dependencies file make.dep depend on the SOURCES files ?
20:05:59kugeldomonoky: what can i do wrong when just typing disable_irq(); at the beginning and enable_irq(); and the end of the function?
20:06:26kugelI also tried some_int = disable_irq_save(); and restoring later. no success
20:06:47domonokykugel: i dont reall know. Wild guess: the interrupt enable was never called..
20:07:43kugelwell, pressing the buttons works in the bootloader. after *only* adding *_irq rockbox didn't boot
20:07:57funmanwhere did it stop ?
20:08:05kugelat the bootlogo
20:10:00domonokykugel: maybe pastebin your modifications, so we can see if you missed something :-)
20:10:39domonokyand if you get complete hangups, panicf() is very helpfull to findout where it hangs.. :-)
20:11:25kugelwell, I'm almost entirely sure the hang happens at either enable_ or disable_irq
20:11:36 Join jhulst_ [0] (n=jhulst@unaffiliated/jhulst)
20:12:03funmando you have the button led on the fuze ? you could use it for debugging purposes
20:17:22 Join shotofadds [0] (n=rob@rockbox/developer/shotofadds)
20:17:58shotofaddshi gevaerts. what d2 problems are you having? freezing during ape playback, by any chance?
20:18:39gevaertsshotofadds: freezing in test_codec for ape to be precise
20:18:53shotofaddsyes, same here. I get the same freeze in the WPS too, though
20:19:00shotofaddsFLAC and MP3 seem fine
20:19:27shotofaddsI've never tested APE before, so don't know if this is a new problem
20:20:15gevaertsThe weird thing is that the freeze seems to be always at the same point in the file, until I overwrite it (with the same original). Then it moves to a new freeze spot
20:20:16kugelape is just too cpu intensive. It overheats the cpu
20:20:19kugeljust a guess ;)
20:20:35gevaertsThat's why I still somewhat suspect nand
20:20:58*gevaerts will try the md5sum plugin next
20:21:06amiconngevaerts: That sounds weird. You obviously did a successful ape test before...
20:21:13 Quit MethoS- (Remote closed the connection)
20:21:32shotofaddssounds like it's filesystem realted. does the md5sum plugin display the sum on screen or just write to a file?
20:21:48*amiconn wonders about the status of the other telechips ports
20:21:49shotofaddsI didn't claim the new NAND driver was perfect ;)
20:22:06kugelfunman: it just read " at 00000000" on the display
20:22:27funmankugel: weird
20:22:30kugelah, now the full version: undefined instruction at 00000000"
20:23:07gevaertsshotofadds: it looks like it writes to a file, but there does seem to be md5sum checking functionality as well
20:23:20kugelindeed weird
20:23:35shotofaddssounds like someone should put it to use ;)
20:24:18 Quit jhulst (Read error: 110 (Connection timed out))
20:24:21 Join Zagor [242] (
20:24:54gevaertschecking still writes..
20:25:00kugelit's so randomly what the display shows after boot without disabling irq
20:25:19funmankugel: forget about it and do something else ;)
20:25:32kugelfunman: what should I do? Buttons are essential
20:25:39funmanjust like sound
20:25:51kugelyou're not going to play any file without buttons
20:26:01kugeleven if you had sound
20:26:09funmanyou're not going to debug that if it drives you crazy either ;)
20:27:20kugelit doesn't drive me crazy
20:27:53kugelwell, if anything, than these epilepsy inducting flickering
20:32:33kugelfunman: for a split second I see a "divide by zero" error
20:33:50funmannote the address, and use gdb on the bootloader.elf
20:34:01funmanor objdump rather
20:34:39kugelfunman: it's in the firmware
20:34:51funmanuse it on rockbox.elf then
20:35:02kugelnot in the bootloader. Also, I can't note the address if I only see it for a split second
20:35:19funmanweird since the player should freeze on this message
20:35:59kugelI get that undefined instruction at 00000000 like instantly after
20:37:41 Join perrikwp [0] (i=18ac0c41@gateway/web/ajax/
20:38:19*gevaerts wonders how long this md5sum should take
20:38:38 Quit Schmogel (Read error: 110 (Connection timed out))
20:41:02LambdaCalculus37funman: There's no proper keymap defined for the Clip virtual keyboard, is there?
20:41:34funmanLambdaCalculus37: if proper means something else than 'compilable'; then yes. bertrik did some work to refine it
20:41:51funman.. then yes, there is no proper keymap
20:41:59kugelI get no message at all when switching the buttonlight on before irq_disable
20:42:22LambdaCalculus37That explains why I can't move up and down in the character selection, then. :)
20:42:52funmanI also have trouble jumping in the time/date settings: I always jump 2 settings
20:43:54 Join bluebrother [0] (n=dom@rockbox/staff/bluebrother)
20:44:21LambdaCalculus37Also the Vol Up/Down buttons don't seem to do anything; I figure they're not defined, correct?
20:45:05funmanthey work in the WPS at least
20:45:11domonokyfunman: did you do already some attempts at sound on ams-sansas ? i am still struggeling to understand how the dma should work in this case...
20:45:14funmanand in the Solitaire plugin :-)
20:45:27funmandomonoky: yes i'm trying to do that, but I have the same problem than you..
20:45:44*gevaerts can't get md5sum to work
20:45:50funmanI understand that it's the i2sout peripheral role to request a DMA transfer
20:46:19gevaertsshotofadds: these are 20MB files, so I guess that if there are issues, the chance to get one is pretty big
20:46:22LambdaCalculus37md5sum freezes on the Clip when I run it.
20:46:23funmanand the fact that "DMAC is the data flow controller" means that we set a size of data to be transferred, and stop after that.
20:46:45domonokyah, the DMAC is the flowcontroller in this case ?
20:46:54bertrikfunman, jumping two places in the date/time settings should be fixed
20:47:07funmanyes, because the i2sout peri can't tell "ok we don't want data anymore"
20:47:48gevaerts(not the standard test_codec format, as there is no write support yet)
20:47:49funmanbertrik: ah right, thank you !
20:47:51bertrikI'm only familiar with a few buttons actually, no idea how the virtual keyboard works for example
20:48:47domonokyfunman: so the lenght in the DMAC should be one fifo full of the i2s interface ?
20:49:53funmandomonoky: no, the size/width should match the size of the FIFO (or half of it like in SD)
20:50:12domonokyah. so one burst per fifo (or half of it)...
20:50:17funmanand the transfer size should be the number of times we can transfer a (half of) FIFO size to the i2sout
20:50:25 Join Lear [0] (i=chatzill@rockbox/developer/lear)
20:50:31funmanso it should reflect the available data in pcm buffer
20:51:35domonokyso its the requested size in pcm_play_dma_start divided to burst size, correct ?
20:51:43 Join AndyIL [0] (i=AndyI@
20:51:54funmanI thinkyes
20:52:06funmando you want my draft code?
20:52:18domonokywould be nice, to make some tries... :-)
20:53:11*bertrik 's RTC on his clip still runs without getting reset
20:53:40kugelinteresting. With disable_irq only it boots
20:54:20funmankugel: perhaps the function is entered with irq disabled
20:54:47domonokykugel: maybe at this time, interrrupts are already disabled, then you would enable them at a bad time... :-)
20:54:48shotofaddsgevaerts: and they're probably recently-written files too, which will make a difference. However, the size issue should also apply to FLAC...
20:55:34 Join dabujo [0] (
20:57:09amiconngevaerts: Thanks. Very interesting results....
20:57:15gevaertsshotofadds: I got it to work after a few more tries (copying again and again...)
20:57:34gevaertsamiconn: have fun with then :)
20:57:42amiconnThe (higher level) filters are significantly faster than on ARMv4. They're almost on par with codfire, which is expected
20:58:00shotofaddsgevaerts: bah. back to the drawing board on the flash driver, then :/
20:58:17amiconnBut for some reason -c1000 is (a bit) less efficient than on gigabeat F (ARM9, v4)
20:58:35gevaertsshotofadds: you're on the right track though. It always boots :)
20:58:43kugeldomonoky, funman: unless some other *_init() function doesn't propery enable irq, that shouldn't be the case. button_init() does not disable irq
20:59:08amiconnYou didn't do the entroypy-only and entropy+prediction tests, I presume?
20:59:39shotofaddsgevaerts: unfortunately I don't really have any more ideas at this point. back to the disassembly methinks...
20:59:46gevaertsamiconn: No. I can still do them though
21:00:16*Zagor can't decide if today is just the right time for a flyspray upgrade, or a very bad time.
21:00:19amiconnDoes the D2 have IRAM? If so, how much, and do we use it?
21:00:57gevaertsZagor: the right time is just before you leave for a week :)
21:01:29shotofaddsamiconn: there's 96k IRAM but we don't currently use it as performance was significantly worse (ie. about 30% slower) when using IRAM. Needs more investigation...
21:01:38kugelno, the right time would be if he leaves for a week AND the tracker cleanup week begins
21:01:47amiconnOh? That sounds odd...
21:02:15shotofaddsIt'd be nice to figure out what the problem was..
21:02:38shotofaddswhat's the reason we don't use IRAM on the Gigabeat F? does it have any?
21:02:59 Join frinkazoid [0] (
21:03:05 Quit AndyI (Read error: 110 (Connection timed out))
21:04:44ZagorIt's a bad time because with the tracker cleanup just announced, people are working in the tracker. And the upgrade is a big one, where I'll have to rewrite many of the rockbox customizations we have.
21:05:05bluebrotherwe have a tracker cleanup?
21:05:16funmanZagor: what is the benefit of the upgrade?
21:05:39LambdaCalculus37shotofadds: markun would know.
21:06:19amiconngcc is weird...
21:06:25*bluebrother guesses for less bugs
21:06:29ZagorIt's a good time because otherwise I'll have to wait a long time with it, possibly after the release, for activity to cool down. And the new version brings some nice new features such as custom task fields (svn revision for example).
21:06:44 Nick jhulst_ is now known as jhulst (n=jhulst@unaffiliated/jhulst)
21:07:51bluebrotherZagor: how much work is the upgrade?
21:08:19Zagorbluebrother: I honestly don't know...
21:08:30BigBambiI'd say go for it, but I'm one of those people that upgrade because something is newer and not necessarily better :)
21:08:36bertrikLambdaCalculus37, the Clip probably looks very much like other plays w.r.t. buttons, so probably we could re-use parts of other keymaps (keeping an eye on what makes sense of course)
21:08:50Zagorthe flyspray team is about as keen on writing development docs as we are :-)
21:09:08kugelinteresting. The crazyness of the display reduces when I insert udelays between GPIO*AFSEL and *DIR stuff
21:09:10bluebrotherhehe ;-)
21:09:36bluebrothercan't you just do the upgrade "offline" on a copy and then move the new FS to the current location?
21:09:44bluebrotheror does it require database changes?
21:09:45gevaertsZagor: you mean they have a nice manual that nobody reads? ;)
21:10:31Zagorbluebrother: yes, it requires database changes. I intend to try an upgrade on a copy of the database first though.
21:11:01Zagorbut running everything offline/duplicated takes extra work
21:11:27LambdaCalculus37bertrik: If I can get my Mac back to working order again, I'll take a look at the keymap.
21:11:33bluebrothertrue, especially in that case :/
21:12:04funmanbertrik: LambdaCalculus37 I'm looking at keymaps in plugins and I'm using the C200 case
21:13:19LambdaCalculus37funman: That makes sense on the Clip.
21:13:46*bertrik agrees
21:14:30domonokyfunman: pcm_play_dma_start get never called for me on m200v2..
21:14:41bertrikI wouldn't mind a someone playing the role of "keymap police" or something like that, to keep keymaps sensible and compatible between players where possible
21:16:26gevaertsHow many people apart from me would like to have Mr Someone write a disk formatter plugin?
21:16:37funmandomonoky: oops, I should have checked that ..
21:16:52 Quit herrwaldo (Read error: 110 (Connection timed out))
21:17:00funmanbertrik: someone or Someone ?
21:17:09linuxstbgevaerts: There's some code in ipodpatcher for formatting (512-byte sectors only)... But yes, it's been mentioned before as a useful idea.
21:17:27gevaertsI mainly want it for formatting ramdisks
21:17:40Zagorgevaerts: how many "are you sure" would be in there? :)
21:18:40gevaertsZagor: I'm not sure :)
21:19:42funmangevaerts: if you do that don't forget to set a common define for targets which have the OF in the first sectors (especially the ones which can't be unbricked like most sansa ams)
21:19:47amiconngevaerts: A formatting plugin would be useful for a few users, e.g. when dealing with an lba48 enabled rockbox on old archos. It should also be able to handle partitioning
21:20:33funmanI wonder how rockbox would behave if it's partition was suddenly deleted
21:21:05LambdaCalculus37funman: Would it panic? :)
21:21:08Zagorfunman: it wouldn't know. iirc the partition is only looked at when mounting the disk.
21:21:28funmanaren't some files loaded from /.rockbox after the firmware itself?
21:21:32gevaertsfunman: if you keep hiding those first sectors in the driver, that's not a problem
21:21:37bertrikfunman, someone. Pixelma has been working on the c200 keymap. Maybe the clip keymap should be more like that patch than the current c200 keymap.
21:22:00Zagorfunman: sure but deleting the partition information doesn't remove the data from the disk. so reading files is still possible.
21:22:11funmanbertrik: if the plugin doesn't use BUTTON_REC, I simply add elif(SANSA_CLIP_PAD)
21:22:36funmanZagor: ok, I was thinking of bzero(), not only deleting the partition table
21:22:38gevaertsamiconn: right. Those can indeed use it as well
21:22:46LambdaCalculus37funman: If you're going to follow the c200 keymap on the Clip, then the keymap for the WPS has to be fixed a bit.
21:22:48*gevaerts adds fdisk to the list
21:22:53Zagorfunman: ah. yeah that will annoy the fat driver pretty quick.
21:22:58 Join raky [0] (
21:23:27funmanLambdaCalculus37: how can I fix it if I never used the WPS? :)
21:23:30bertrikis there any kind of keymap guideline?
21:24:07 Quit raky (Client Quit)
21:24:29gevaertsMy next issue is of course that I need FAT16 (for 4MB or 8MB ramdisks. FAT16 is already a stretch), and ipodpatcher doesn't do that
21:25:00amiconngevaerts: That's in fact a quite old idea. Have a plugin that (1) checks the partition table, to make sure the active fat32 partitions ends before the lba28 barrier, (2) creates a second, hidden fat32 partition after the first one and (3) formats it
21:25:34shotofaddsamiconn: presumably the use of -mlong-calls in ARM builds is to cater for calling functions in IRAM? though it seems to always generate long calls, regardless of the target section..
21:26:48 Quit shotofadds (Read error: 54 (Connection reset by peer))
21:26:51 Join raky [0] (
21:26:55 Join shotofadds [0] (n=rob@rockbox/developer/shotofadds)
21:27:27 Quit raky (Client Quit)
21:27:54amiconnAnother item on this wish list would be a faster file copy in rockbox, maybe as a plugin, that uses the audio buffer for buffering data
21:29:12bertrikamiconn, last time I looked at the file copy, there was already an option to use a big buffer IIRC
21:29:26bertrikalso I wondered if the progress bar might be slowing down the copy
21:30:18*bertrik tries to remember the source file
21:31:01amiconnshotofadds: Yes and yes
21:31:47shotofaddsso it might be worth diabling that while not using IRAM..?
21:31:50amiconngcc is pretty limited when it comes to handling arm long calls. -mlong-calls will cause all calls to non-static functions to be long calls, and calls to static functions to be short calls
21:32:18Unhelpfulamiconn: thanks for the summary! the first one is implemented, the second one doesn't look hard, the third sounds positively nasty :/
21:32:41amiconnThis is the reason for the STATICIRAM macro. It's a workaround for a gcc bug
21:33:09amiconnIt causes all ICODE_ATTR functions to lose their 'static' attribute on arm
21:33:23amiconn(if ICODE_ATTR is non-empty, of course)
21:33:44shotofaddsah, that explains it. I'm just wondering if that is causing a performance penalty when not using IRAM
21:34:04amiconnUnhelpful: The 3rd isn't much different from the 2nd. Just somewhat different bit shuffling, and a twice as big pixel block
21:34:15shotofaddsalthough performance should really be the least of my worries..
21:34:31bertrikthe clipboard_pastefile in onplay.c tries to use the plugin buffer, if not available it uses a 512 byte buffer on the stack
21:34:54Unhelpfulthe first is at least easily optimized for line-at-a-time writing. the others pretty much can't be, and i'll end up having to write each byte four times :/
21:35:10amiconnIf you're not using IRAM, or otherwise all code sections can be reached from each other using short calls (that's what tomal did on the iFP - mapping DRAM and IRAM close enough), -mlong-calls isn't needed
21:35:29 Join puzzles [0] (n=dan@xmms2/developer/puzzles)
21:35:48amiconnUnhelpful: That's unavoidable. The current BMP loader also has to do that
21:36:17Unhelpfulamiconn: i could buffer 4 or 8 lines of pixels values, but that's almost certainly worse :/
21:36:46Unhelpfulthe horizontal-packed case will get just as bad if we ever need to load a column-first format
21:37:53amiconnOf course.
21:39:31*amiconn knows those pixel formats pretty well, as he wrote large parts of the optimised lcd drawing functions for the various pixel formats, and developed the greylib, which has to handle those pixel formats too
21:41:00Unhelpfuli'll pester you if i have more trouble - as it turned out, none of the troubles with the H-packed case had anything to do with my understanding of the format, though. all stupid-programmer bugs.
21:41:10amiconnI wonder why so many arm targets specify -mlong-calls. On PP it is necessary, but afaik we don't use IRAM on gigabeat F, so why -mlong-calls?
21:41:50LambdaCalculus37amiconn: Speaking of the iFP... what is the ultimate fate of the port? We haven't seen any activity on it for a long time, and while sound does work, IIRC it was WAV only.
21:42:21amiconnIt played vorbis and mp3 as well, afaik
21:42:25BagderLambdaCalculus37: it's just another one of the stalled ports..
21:42:50LambdaCalculus37amiconn: Hmmm...
21:43:04LambdaCalculus37Bagder: Wonder if Mr. Someone would ever pick it back up. ;)
21:43:24Bagderfor such an outdated and tiny storage device, I find that very unlikely
21:44:00Bagdertomal did some good optimizations though that we all benefit from
21:45:23kugeleh, I think rockbox is looping
21:45:46kugelwhat could cause rockbox to reboot?
21:45:50shotofaddssheesh.. removing -mlong-calls from the D2 build saves 27KB (~5%), I wonder if it has a similar effect on performance...
21:46:28gevaertsDoes that include codecs?
21:47:27shotofaddswell, that figure was binsize, but I assume codecs are also built with that flag too
21:48:04amiconnYes, they have to (on targets where we use IRAM)
21:48:15 Quit jhulst (Read error: 60 (Operation timed out))
21:48:26*gevaerts could test ape right away
21:48:44amiconnUsually the gain from using IRAM is greater than the loss due to long calls
21:49:10amiconngevaerts: I would be surprised if you get a measurable difference
21:49:37amiconnMost performance critical code in libdemac uses static inline functions
21:50:28amiconnshotofadds: Btw, this is one more reason to make all private functions static - it saves them from being long-called on arm
21:51:07n1samiconn: iafaik we don't use iram on the beast either, i think it has only 16kB though
21:51:27 Join notplus_M [0] (
21:51:39***Saving seen data "./dancer.seen"
21:52:26domonokydoes somebody have an idea why pcmbuf_play_start could be called with pcmbuf_unplayed_bytes==0 and pcmbuf_read == 0 ??
21:53:20funmandecoder doesn't work correctly ?
21:54:16 Join japc [0] (
21:54:38domonokyit looks more like problems with memory (the left-over buffer is very small).. maybe not all needed things fit, and error handling is non-existant (like always with mem-alloc :-) )
21:55:42amiconnlinuxstb: ping
21:55:44gevaertsamiconn: the difference _is_ measurable. I got 390.19% for -c1000 (was 384.81% for a -mlong-calls build)
21:56:01 Quit nuonguy ("This computer has gone to sleep")
21:57:10amiconnPerhaps the code size reduction improves caching
21:57:18shotofaddsgevaerts: I also see about a 1% difference in MP3
21:57:44 Join jhulst [0] (n=jhulst@unaffiliated/jhulst)
21:58:52*gevaerts continues testing
22:00:15shotofaddsbtw. I deliberately haven't added any d2 codec performance figures to the wiki yet, since I can't be sure the dodgy flash driver isn't skewing the results :/
22:00:52gevaertsInterestingly, c3000 is slower without -mlong-calls
22:03:10gevaertsDoes storage performance actually influence test_codec? I'd assume that if it gets loaded correctly, the numbers will be ok
22:04:34Unhelpfulfirst one that compiles works. i'm not interested in installing a coldfire toolchain just to get binsizes, though. i'll have the patch up in a minute, if somebody w/ an iriver h1x0 or iaudio m5 wants to try it.
22:05:32shotofaddsgevaerts: I meant bad data being read, rather than performance
22:06:00gevaertsshotofadds: for ape it seems to freeze in those cases anyway :)
22:06:45shotofaddsit's weird, I haven't managed to complete a single APE test run yet. but every other file I try seems file. very strange
22:06:52 Quit martian67 (Remote closed the connection)
22:06:57gevaertsBut yes, those numbers need to be treated as possibly suspicious
22:07:24linuxstbamiconn: pong
22:07:39shotofaddsI think if you can play the file without glitches the numbers should be fine. But I wasn't about to play the whole test set through ;-)
22:07:59gevaertsI haven't actually listened to this test track at all yet :)
22:08:12 Join martian67 [0] (i=lol3izer@about/linux/regular/martian67)
22:08:36shotofaddsgevaerts: do you have any idea where to start looking at USB on 78x? is that something you intend to look at one day or should I get reading?
22:09:31bertrikfunman, I see that one the clip there currently is an option to put the display upside down, but it only puts the buttons upside down.
22:09:33gevaertsshotofadds: I have no real idea right now. I know I've had issues with endpoint allocation
22:09:52bertrikalso I think the OLED brightness/contrast can be changed in the OF, but not in rockbox yet
22:10:00gevaertsBut I don't remember if that was on the D2 with UMS or only on the DAX with serial
22:10:15bertrikI know it's not really important yet, but how about I have a look at that?
22:10:53gevaertsAlso, while USB would no doubt be interesting for development, without write support there isn't that much you can actally do with it
22:10:56 Join funman_ [0] (
22:11:15 Quit funman (Nick collision from services.)
22:11:20 Nick funman_ is now known as funman (
22:12:24funmanbertrik: true lcd_set_flip() is still marked as TODO
22:12:38funmancontrast can be changed, not brightness
22:12:49funmanrockbox lowers the contrast when shutting down, but i couldn't find a setting for it
22:12:58gevaertsI don't actually know very much about this controller either to make it even more fun
22:13:10shotofaddsgevaerts: debugging read support is my first concern. Being able to dump data from the d2 would make things a tad easier
22:13:27gevaertsI know
22:14:05shotofaddsI had hoped the data from the m200 would help, but it turns out the FTL scheme is slightly different between the two
22:14:09gevaertsI need to find some time for it and work out what goes wrong. Does it actually compile with usb enabled in svn?
22:14:12kugelfunman: you probably need to #define HAVE_LCD_CONTRAST in config-clip.h
22:14:41funmankugel: correction, bertrik probably needs to do so ;)
22:14:49shotofaddsgevaerts: you need to include usb-tcc77x.c in SOURCES, but apart from that it does compile, yes
22:14:59funmankugel: thanks for the hint
22:15:06gevaertsok. I'll try to have a look at it soon
22:15:25shotofaddsIn XP I can get it to recognise a generic UMS device, and then freeze. In linux it just freezes..
22:15:40*gevaerts also wants to get it to work on the meizus. Would be fun, usb before lcd...
22:15:52gevaertsDoes dmesg say anything?
22:16:16shotofaddsnothing - the last messages were from inserting in usb-boot mode
22:16:36 Join m0f0x [0] (
22:17:28gevaertsWhat did you use instead of charger_inserted()? Just always USB_INSERTED?
22:18:14shotofaddsI used charger_inserted() ;)
22:19:00gevaertsOK. That makes it easy :)
22:21:03bertrikhow many steps would be sensible for contrast? the controller appears to offer 256 steps, which seems a bit much IMO? how about 16 steps?
22:21:28shotofaddsgevaerts: the logic isn't quite correct, since that will also be true when the AC adapter is plugged in (regardless of USB). but for testing it should be fine
22:22:19 Join EspeonEefi [0] (i=eefi@STRATTON-TWO-EIGHTY.MIT.EDU)
22:22:23*shotofadds gets back into some more disassembly
22:22:47*bertrik experiments
22:23:40pixelmabertrik: the c200's controller is also _supposed_ to do 256 steps too and Rockbox supports that (altough in real, I only have 128 steps which and it repeats itself...)
22:24:10pixelmabroken sentence
22:24:16bertrikthe clip doesn't have a wheel, so scrolling through 256 steps might take a while
22:24:35pixelmathe c200 has no wheel either
22:25:57funmanc200 and clip seems to have the exact same buttons. Safe for REC, and HOME (Clip only ?)
22:26:49pixelmac200 has a Rec button - and Power button (instead of the Home one?)
22:27:06kugelit just gets weirder
22:27:26funmanpixelma: the keys are not at the same places though
22:27:43gevaertsamiconn, shotofadds: the ape numbers without -mlong-calls are at
22:27:55pixelmafunman: I imagine
22:27:57funmanpower (clip) is on the side like rec (c200), and home (clip) is on the front like power (c200)
22:28:04 Quit {phoenix} (Remote closed the connection)
22:28:21funmanI made most plugins build, now I have to look at plugins which depend on screen dimensions
22:28:26 Quit LambdaCalculus37 (" ajax IRC Client")
22:28:27kugelfunman: I inserted a splash now in the root menu which splashes when power is pressed. And I only read power in a seperate function. That splash comes without pressing something and doesn't go away
22:28:33bertrikfull brightness on clip doesn't look very healthy
22:28:33Unhelpfulis the M3 the *only* thing that uses the vertical-interleaved format?
22:29:03pixelmafunman: ah, so it has a Power and a Home button and also Volume Up/Down?
22:29:03funmankugel: then perhaps your reading of power button is wrong
22:29:43kugelit's mostly copied from svn, which works for the bootloader
22:29:44funmanpixelma: yep, look at sdl/UI-clip.bmp : volume +/- is on the right side and power/hold on the left side (invisible on the bmp)
22:30:18bertrikupside down works now, but looks a bit weird because of the yellow top part of the screen which becomes the bottom part of the screen in upside down mode
22:30:33pixelmaunless that changed recently, I couldn't recognise any side buttons in the sim bmp
22:30:35funmanthat screen will look bad everywhere .. :/
22:31:05pixelmafunman: where's "Rec" then?
22:31:07funmanpixelma: the side buttons aren't on the BMP, perhaps we should mark all the buttons with overlaid text on the picture
22:31:10funmanpixelma: there's no rec
22:31:18bertrikdo we really want upside down mode for the clip?
22:31:27funmanbertrik: why not ?
22:31:49pixelmafunman: aha, misunderstood earlier
22:34:04bertrikfunman, because it looks weird and I don't see much use for it (personally)
22:34:32funmando you use lcd flip on other targets?
22:34:50Lloreanfunman: I think the point is more that we can't do anything about the yellow patch being at the bottom of the screen, and offset by two pixels.
22:35:05LloreanIt does have a blank line separating, right?
22:35:07funmanLlorean: i understand, but that's a problem in a lot of other areas
22:35:13funman2 black lines
22:35:33LloreanBut it's more of a problem if it's "part of the list" rather than "the status bar or title"
22:36:16LloreanI'd almost say it needs an option to display the status bar in the yellow area even when flipped to really not be too weird.
22:36:18funmani meant it's a problem in most plugins (except solitaire), wasn't thinking of the screens with status bar specifically
22:38:56funmanmake: *** No rule to make target `/home/fun/clip-fw/pluginbitmaps/bubbles_bubble.h', needed by `/home/fun/clip-fw/apps/plugins/bubbles.o'.
22:40:44 Quit bmbl ("Woah!")
22:41:16pixelmaif you take the Archos graphics for the Clip as a start - remember that some of them will look stretched (where it was possible, to compensate for the rectangular pixels on the Archos display
22:41:55funmanif plugins can be built and fixed later that's not a problem
22:43:06funmanoh I found the SOURCES file
22:43:12Zagorfunman: bingo :)
22:43:16 Quit neddy (Connection timed out)
22:44:06funmanI'm a bit confusedthat some bitmaps use the target screen size in the name, and some other the bitmap size
22:45:27pixelmayes, they are named inconsistently...
22:45:56funmanwhich inconsistency should I be consistent with ? ;)
22:46:24gevaertsshotofadds: if I build a logf build and watch the logf output live, it gets a lot further
22:47:27gevaertsI guess it gets as far as you got it in windows then
22:47:31 Quit massiveH ("Leaving")
22:49:21 Quit perrikwp (" ajax IRC Client")
22:57:28 Join Acikers [0] (n=acikers@
22:57:43 Quit shotofadds ("Leaving")
22:57:43Acikershi, i'm need to help
22:58:23funmanhello Acikers, what happens?
22:58:29n1sAcikers: it's a lot easier to help you if we know what the problem is
22:58:35 Quit Lear ("ChatZilla 0.9.84 [Firefox 3.1b2pre/20081125035317]")
23:00:16Acikerswhat needed for playing avi in rockbox
23:00:52AcikersSorry my bad English...
23:01:09linuxstbAcikers: Rockbox doesn't play any avi formats - you need to convert videos to the format Rockbox supports -
23:01:27AcikersI'll know that...
23:03:52AcikersAnd, how i can open files *.rock? Or where i can find source?
23:04:31MegafEeeAcikers: i think you can find all that infromation on rockbox site
23:04:36MegafEeeon the FAQ
23:05:05pixelmaAcikers: or the manual
23:05:12 Join neddy [0] (n=john@nat/sun/x-896b804bd2843ed6)
23:05:34n1sAcikers: the source is in our SVN repository see this page
23:05:59Unhelpfulif i'm reading this right, there's at least one player that uses one grayscale format on the display, and a different one on the remote?
23:06:06funmanwhat's the difference between bitmaps/mono and bitmaps/native for a mono target?
23:06:17n1s.rock files are rockbox plugins which are compiled from c code, we have documentation in the wiki for setting up compilers etc
23:06:52n1sUnhelpful: that would be the iaudio m5 i think, same remote as m3 and x5
23:06:59BagderUnhelpful: yes, the h100 has 2bit greys on the main and 1bit on the remote
23:07:55 Part Acikers
23:09:06Unhelpfuli'm starting to think that the inside of the nearest-neigbhor scaler should have a case statement for the different formats, and build whichever cases apply for a particular target or its remote
23:09:22Unhelpfulit seems the only way not to copy-paste the entire outer loop
23:10:53Unhelpfulaside from that, there are already several tests inside the inner loop for things like dithering, done on a per-pixel basis - so case'ing the formats that might be encountered wouldn't make things much worse, i'd think.
23:10:53 Quit pondlife ("Leaving.")
23:11:14 Quit martian67 ("gone")
23:13:47pixelmaUnhelpful: the M5 which itself is greyscale can use the Iadio remote which is the M3 display
23:13:48 Quit jhulst (Read error: 110 (Connection timed out))
23:14:00pixelmathe remote is greyscale too
23:14:15 Quit domonoky (Read error: 104 (Connection reset by peer))
23:14:40pixelmawhat n1s said... /me too slow
23:16:12 Join fml [0] (n=4fd3c72c@gateway/web/cgi-irc/
23:16:36fmljhMikeS: ping
23:17:53Unhelpfulpixelma: the M5 is v-packed, and the M3 remote is v-interleaved, i believe?
23:18:38fmljhMikeS: (for the log) I have an impression that after r19214 there are more noises at startup. But less on shutdown (Sansa e200 v1).
23:18:42 Quit dabujo ("( :: NoNameScript 4.2 :: )")
23:18:56pixelmaUnhelpful: that's what amiconn said earlier
23:19:22 Join jhulst [0] (n=jhulst@unaffiliated/jhulst)
23:20:56 Quit nplus (Remote closed the connection)
23:21:03Unhelpfuland some of the greyscale targets have mono remotes. i think there should be exactly one resize_nearest, to save duplicating the outer-loop code... probably the whole lot could go with the color scaler into a resize.c, instead of this resize-common, resize-gray, and resize-color
23:22:29bertrikI think I found the datasheet for the as3525 USB sub-system but it's behind some kind of registration form
23:22:58funmanbertrik: free registration?
23:23:16bertrikhave a look:
23:25:07 Quit aarcane (Read error: 104 (Connection reset by peer))
23:25:47funmanbertrik: just put anything you want including wrong email, and you'll get the (2 pages) download
23:26:26 Quit fml ("CGI:IRC 0.5.9 (2006/06/06)")
23:26:43funmanoh the "full product doc" is protected by a real login
23:29:28 Join shotofadds [0] (n=rob@rockbox/developer/shotofadds)
23:36:25 Join MethoS- [0] (
23:36:33 Join skipper [0] (
23:41:50 Quit Rondom ("Ex-Chat")
23:42:49 Join aarcane [0] (
23:51:41***Saving seen data "./dancer.seen"
23:52:27kugellinuxstb: I have the strong feeling the lcd driver isn't as working as initially thought
23:53:08kugelbesides that it makes it hardly possible to use buttons, I also have problems with lcd_update(), which just crashes

Previous day | Next day