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

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

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

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

#rockbox log for 2010-12-16

00:02:28eWilln1s:you said, "runtime estimate will never be very accurate". Care to ballpark a margin of error?
00:03:32eWillanyone? Wild guess?
00:04:31LloreaneWill: It's impossible to ballpark since it depends on your use patterns, including filetypes, as well as the condition of your battery and even things like the ambient temperature.
00:04:41LloreanThe best person to ballpark the margin of error for a given device is its owner after experience.
00:04:55 Join jgarvey [0] (
00:05:41 Quit jgarvey (Client Quit)
00:10:30 Join krabador [0] (
00:12:12 Quit sideral (Quit: Leaving.)
00:12:58***Saving seen data "./dancer.seen"
00:14:28 Nick MortalScan is now known as MortalScan_Troll (~mortalsca@
00:15:20 Join Keripo [0] (
00:20:04 Quit dfkt (Quit: -= SysReset 2.53=- Sic gorgiamus allos subjectatos nunc.)
00:21:10 Quit wodz (Quit: Leaving)
00:25:49 Join T44 [0] (
00:26:53eWillIs this true: If I modify a source file, and a commit is made that changes that source file, my modified one won't be touched if I "svn update".
00:27:43eWillI should have said, "updated" instead of "touched"
00:28:36 Quit Topy44 (Ping timeout: 240 seconds)
00:28:39n1sno, svn will try to merge the two files
00:28:49n1sif it fails it will create a conflict
00:30:57eWilland notify me of the conflict?
00:31:31n1syes, it prints a "C" for conflict, "G" for successful merge and "U" for regular updates
00:37:04 Quit n1s (Quit: Lämnar)
00:37:25eWillI want to do a fast battery bench (so i can build the voltage/percent table). Here's my plan: repeat a file encoded to a format that RB decodes slow; turn backlight to "ON"; raise brightness to max; use the EQ. Any other suggestions?
00:38:51 Quit bertrik (Ping timeout: 272 seconds)
00:39:20Bushmillslow impedance headphone with volume set to max?
00:40:37Bushmillsthough i think, running f.e. a fractal zoom consumes more CPU than decoding a media file
00:41:16CIA-45New commit by 03pixelma (r28840): Manual - Sansa e200 player images: update the labels for scrolling ('back' and 'fwd' instead of 'up' and 'down') to match the c code and naming in the ...
00:43:13pixelmais this worth a backport?
00:43:24CIA-45r28840 build result: All green
00:45:49 Quit Staphylo (Quit: Bye les gens =))
00:45:56eWillwouldn't playing an audio file that my player can't decode in realtime max-out the CPU?
00:46:37eWilllike a fully compressed APE?
01:06:22 Join BHSPitMonkey [0] (~stephen@unaffiliated/bhspitmonkey)
01:06:58 Join DerPapst [0] (
01:13:19[Saint]Gah....bloody iPod Colour data aborts trying to init database :/
01:22:39[Saint]Need to wait until I can build my own binary to see what its choking on...It'd be really nice if the .map file was packaged with the daily builds/all builds.
01:22:47[Saint]could save some time.
01:38:20 Join simonrvn [0] (
01:42:52 Join telliott [0] (
02:01:48 Quit Kitar|st (Ping timeout: 255 seconds)
02:02:46 Quit tchan1 (Quit: WeeChat 0.3.3-dev)
02:07:57 Join Kitar|st [0] (
02:08:02 Join Zarggg_ [0] (
02:12:03 Quit Zarggg (Ping timeout: 265 seconds)
02:12:59***Saving seen data "./dancer.seen"
02:15:07 Join edboyer93 [0] (
02:16:06 Join Jeshikah [0] (
02:16:59Jeshikahi know this might seem like stupid question, but i came across a plugin mentioned in the manual that i don't have on my device
02:18:00 Quit krabador (Ping timeout: 255 seconds)
02:18:32 Quit chattr (Quit: gone)
02:18:33 Quit Jeshikah (Client Quit)
02:18:48 Join Jeshikah [0] (
02:19:13Jeshikahwhere can i get the rockboy plugin? the gb/gbc emulator?
02:21:18krazykitread the manual more closely. rockboy is a viewer plugin. just click on a rom.
02:22:17Jeshikahah. so basically i just put the roms and it opens it like it would a video file?
02:22:45Jeshikahgreat, thank you very much ^^
02:24:39 Join chattr [0] (
02:29:00 Quit kugel (Remote host closed the connection)
02:30:07 Join krabador [0] (
02:37:04Jeshikahoh one question: can the rockboy handle gba roms too? has it been tested with gba roms? i ask cuz i know many emulators that can play gbc roms can also
02:37:22Jeshikahexample visualboy advance)
02:46:38 Join madalu [0] (~user@unaffiliated/madalu)
02:54:58 Quit Jeshikah (Quit: CGI:IRC 0.5.9 (2006/06/06))
03:00:43 Join JdGordon [0] (~jonno@rockbox/developer/JdGordon)
03:04:51 Quit telliott (Ping timeout: 276 seconds)
03:08:35 Quit madalu (Remote host closed the connection)
03:08:38soapscorche, you said you could view the moderation log. Who deleted Hendricks266's post?
03:12:41soapLlorean, I see the directory approach as giving 80+% of users a simple path to exactly what they need, and the "all files" toggle approach as giving 100% of users kinda what they need.
03:13:01Lloreansoap: I'm not understanding.
03:13:13LloreanWhat does the directory approach offer that the all files one doesn't?
03:13:41soapesay segmentation. One behavior for one class of files and one behavior for the other class of files.
03:13:51LloreanBut is that really useful/necessary?
03:14:32LloreanIs it particularly harmful if files auto-resume when selected? An "insert shuffled" wouldn't auto-resume, or an "insert" or even selecting a playlist, but clicking on a specific file would.
03:14:35soapI believe it is simply because it is the approach seen elsewhere in the "marketplace" and I feel it is an extension of many people's mental groupings.
03:15:12LloreanI don't know that "it's how they do it on the Clip+" is really that big of a reason to do it.
03:15:19soapand apple
03:15:31LloreanAnd it's also not a reason not to take it a step further.
03:15:32 Quit thegeek (Read error: Connection reset by peer)
03:15:34soap(of course it isn't directory based there, but the pathway to it is similar)
03:15:59LloreanI mean, nobody's really said why they think it wouldn't work well for all files. I don't get the "100% of users kinda what they need" statement.
03:16:08soapI wouldn't characterize the "all files" toggle as "a step further" but rather an entirely different approach.
03:16:15LloreanWhat toggle?
03:16:55soapAm I mistaken in believing your "F" is a selectable option? Or are you saying do it universally?
03:17:10soaplet me reread
03:17:14LloreanAuto-resume all files, but have very explicit situations in which a file is auto resumed.
03:17:59soapie on a new class of insertion?
03:18:27LloreanA new class of insertion, or if a file is launched by single-clicking "select" (or equivalent) on it.
03:18:42LloreanDo not auto-resume if the file is reached by directory change, next track in the playlist, or any other automatic means.
03:19:12LloreanThe new class of insertion is, to me, a way to bring some of the "auto resume mid-playlist" functionality in, and not a key part of what I'm suggesting.
03:19:27LloreanJust a way to bridge the gap between no mid-playlist resumes and having them.
03:19:34soapWhich means explicit insertion of all. "Directory" methodology means implicit new behavior.
03:19:46soap*which means explicit insertion of all is required.
03:19:56LloreanExplicit insertion *if* you want mid-playlist resumes.
03:20:03LloreanAs in, resume file 1, then resume file 2 then resume file 3
03:20:14LloreanI'm against automatic mid-playlist resume even with your segmentation proposal
03:20:22LloreanIt still doesn't resolve any of the problems with it i've brought up.
03:20:27soapAs I said, it's how other do it.
03:20:36LloreanDo they?
03:20:52soapyes, Apple treats podcasts and audiobooks "special".
03:20:59LloreanIf I have three chapters of a book on my ipod, queued up in a playlist, seek in 3 minutes of chapter two, then play chapter 1, it'll be 3 minutes into chapter two when it loads that file?
03:21:20soapthey "segment" music away. Apparently Sansa is playing around with something similar, but file based instead of database based.
03:21:33Lloreansoap: I know the iPod auto-resumes podcasts.
03:21:39LloreanBut does it really auto-resume mid playlist, and with books too?
03:21:52soapThat final query I can't say. I do ask that you limit the questions to one at a time.
03:22:30soapone sec, gotta check on something unrelated.
03:22:48LloreanI still think mid-playlist resuming is much more of a fringe case than people just wanting to select a podcast and have it pick up where they left off.
03:22:57LloreanAnd not think about having to train their syncing program to put it in special folders or whatnot
03:23:56LloreanQueuing a file isn't difficult at all, and I'd say asking people who want mid-playlist resume to queue is a lot more "user friendly" than asking people to restrict themselves to certain organizational structure and to be careful because mid-playlist resume will happen whether they want it or not.
03:24:34saratogasoap: I deleted it, I didn't bother PM'ing him
03:25:47soapLlorean, let's throw it back at the OP / patch man. Personally I think the debate is well framed as "explicit" vs "implicit" methodology.
03:26:25soapsaratoga, I'll pm him. I know it was a fucking idiotic noob statement, but I always try to steer them once.
03:27:06Lloreansoap: Well, I think it's also a case of ease of use. I feel my way it a lot easier to use over all, since it doesn't require your files be organized any specific way *in advance* or anything. While your way is vacation-doable you have to use the file browser's copy and paste if you've done anything wrong.
03:27:35saratogai usually just PM if if move the post, but if its stupid/offtopic/whatever i don't bother since then people feel like they can bug me with more questions
03:27:43LloreanAlso, my way allows access to the functionality of your way, while your way doesn't allow access to the functionality of my way which means I'm stuck with a feature working a way I don't want and no way to resolve that.
03:28:17Lloreansoap: I don't know what you mean by "throw it back at the OP"? He's made it clear his use case is for playlists with multiple resume points within them, so I don't think he's going to be in favour of reducing the capability for that.
03:28:55soapsaratoga, I hear that. I have no problem taking on the role of "someone to take the time to explain the explicit rules" while you do real work. ;)
03:30:04saratogaFred Bauer's new test_codec results for the fuzev1 after his clocking changes are really interesting
03:30:10soapLlorean, on your second point I really don't care, no offense, what either of our expectations are of his reaction. I still think an explicit outline of the counter proposal is polite.
03:30:36Lloreansoap: I don't understand.
03:30:40saratogathey're pretty close to the fuzev2
03:31:08saratogalooks like adding that unboosted option to test codec really encouraged people to think about clocking more :)
03:31:19soapLlorean, I'm not sure where I could clarify, but I'm willing to try.
03:31:27Lloreansoap: What are we going back to him for?
03:31:55LloreanThe question is whether this is the way we want such a feature implemented. No offense to him, but does his opinion matter in that question beyond what his preferred implementation is (which he's demonstrated)?
03:32:37LloreanThe feature breaks down to "which files", "should it be optional", "should how it behaves be configurable", and lastly "which configuration options should be offered?"
03:33:18LloreanYour description has been "all files in a specific location", "yes by physical location of file", "no, behave the same for all files in the folder" and "none" from what I understand.
03:33:26LloreanThough the last two I've inferred and may be quite wrong about
03:34:18LloreanHis patch is "any files", "it is, based on a variety of criteria", "yes", and "criteria for resuming, when to resume, parameters for automatic resuming decisions, parameters for when to store a resume point, etc"
03:36:18soaphis opinion, no. He offered, you countered. I think it is polite to explicitly counter and allow him a known to "counter-counter" against. An explicit opportunity to address concerns either through code or argument. As it is now there is no clear consensus statement in the thread.
03:36:19 Quit designate72 (Read error: Operation timed out)
03:36:36Lloreansoap: I've pretty explicitly described my position in the emails, I thought
03:37:06soapand I don't believe your position is posted as "the answer". That's all I'm asking for.
03:38:09LloreanI have no idea what you mean by that...
03:41:59LloreanDo you want a post that is simply framed as my counter proposal? I did include it in one of my posts, but I haven't posted it explicitly.
03:42:21LloreanI've mostly focused on where I feel there are drawbacks and how I would address them in the hopes that other people might see the difficulties and attempt to address them in other ways.
03:42:26 Quit Keripo (Quit: Leaving.)
03:43:08soap<Llorean> Do you want a post that is simply framed as my counter proposal? I did include it in one of my posts, but I haven't posted it explicitly.
03:43:18soapyes, that is all I'm asking. The thread is a mess IMHO.
03:43:38LloreanBut I don't think we should be focusing on proposals, but "what we want from this feature"
03:44:07LloreanI don't want it to descend into his idea vs my idea, but rather look at what his idea is trying to address and try to decide if that's what we want to address (if it is, his idea's fine) and if not, then start working on how to address it.
03:44:18LloreanJust throwing out proposals doesn't work well if we haven't yet agreed on what problem to solve.
03:44:42saratogashould i close bug reports about ipod accessories and point them at the wiki or do we take bug reports on those now?
03:45:08soapsaratoga, I thought we took them in the one IAP task only.
03:45:46soapsince they aren't supposed to be "bugs" but rather dumps of debug information gathered with accessories not owned by the developers.
03:46:15soap(unless there is a regression, which is a situation I haven't seen addressed)
03:47:20soapLlorean, as I see it, it is not just "throwing out proposals", it is steering the debate.
03:48:07 Quit sasquatch (Quit: WeeChat 0.3.2)
03:48:32 Join sasquatch [0] (
03:51:40Lloreansoap: Okay, I've tried to address what I could.
03:52:12LloreanI'd say a regression is a bug.
03:55:25soapyea, I don't know the specifics of what saratoga is asking about. I just recall previously a desire stated to consolidate all IAP info into the IAP task. Lack of support for an previously unknown device can't really be a bug. I was saying that regression, if it popped up, could be though.
03:55:54LloreanYeah, I don't know if that's how we do it now, but collecting them in one task sounds like a "how we should do it" kind of thing
03:56:22saratogathe one task tells you how to collect debug info, so i guess it makes more sense anyway
03:56:27soapLlorean, I think that's a wonderful email. It consolidates all the points into a single "target" as opposed to responding to bits and pieces here and there.
03:56:28saratogaotherwise the bug reports are kind of useless
03:57:18Lloreansoap: See, from my perspective it's the bits and pieces that should be addresses. Whether or not it should be configurable is to an extent an independent point from which files should use it, unless it's decided that which files should be configurable, etc.
03:57:18soapthough you might as well called me David Hall since that's the email account I had been using previously. ;)
03:57:23LloreanIt's really several features.
03:57:38LloreanAuto resume, auto selection of auto resume files, intelligent resume decisions while playing back.
03:57:51soapyou don't see a value in consolidating all those subthreads?
03:58:05LloreanConsolidating them seems to make it a package deal.
03:58:14LloreanI'm not sure it should be treated as a single feature that's all or nothing
03:58:35LloreanI'd almost say we need to explicitly divide it into separate threads after deciding what the actual features are.
03:58:38soapI don't think a single email denotes a single "take it or leave it" feature.
03:58:47soapI just think it organizes the debate.
03:59:21LloreanPossibly. I think it frames the debate. "We're talking about a whole thing here"
03:59:31soapand, yea, let it split up once the "terms of debate" are agreed upon.
04:00:05soapI see it as "regrouping and refocusing" after a fragmented conversation.
04:00:13LloreanThat makes sense.
04:00:34LloreanI think what it *is* depends on how the other people in the discussion respond to it though. :-P
04:03:02 Quit InsDel (Read error: Connection reset by peer)
04:07:30 Quit MethoS- (Remote host closed the connection)
04:09:28soapnow we can wait and see. ;)
04:13:00***Saving seen data "./dancer.seen"
04:13:48 Join DerPapst1 [0] (
04:14:51 Quit DerPapst (Ping timeout: 240 seconds)
04:20:17 Quit krabador (Quit: Sto andando via)
04:22:28 Join Llorean1 [0] (
04:22:32scorchesoap: when was it supposedly deleted?
04:22:33 Quit Llorean1 (Client Quit)
04:22:39 Join Llorean1 [0] (
04:22:54soapsaratoga, answered the question. Thanks, though.
04:23:26 Quit Llorean (Disconnected by services)
04:23:31 Nick Llorean1 is now known as Llorean (
04:23:34 Quit Llorean (Changing host)
04:23:34 Join Llorean [0] (~DarkkOne@rockbox/user/Llorean)
04:25:30 Quit crwl (Ping timeout: 240 seconds)
04:27:24 Quit amiconn (Disconnected by services)
04:27:24 Join amiconn_ [0] (quassel@rockbox/developer/amiconn)
04:27:24 Quit pixelma (Disconnected by services)
04:27:26 Join pixelma_ [0] (quassel@rockbox/staff/pixelma)
04:27:29 Nick pixelma_ is now known as pixelma (quassel@rockbox/staff/pixelma)
04:27:42 Nick amiconn_ is now known as amiconn (quassel@rockbox/developer/amiconn)
04:30:58 Join factor_ [0] (
04:31:04 Quit TheSeven (Ping timeout: 250 seconds)
04:31:33 Quit factor_ (Remote host closed the connection)
04:32:12 Join factor__ [0] (
04:33:13 Quit factor__ (Read error: Connection reset by peer)
04:33:24 Join factor_ [0] (
04:35:46 Quit factor_ (Read error: Connection reset by peer)
04:35:55 Join icarusfactor [0] (
04:35:57 Join TheSeven [0] (~TheSeven@rockbox/developer/TheSeven)
04:37:00 Quit icarusfactor (Remote host closed the connection)
04:37:32 Join Barahir [0] (
04:38:12 Join icarusfactor [0] (
04:39:53 Part icarusfactor
04:40:36 Quit Barahir_ (Ping timeout: 250 seconds)
04:45:31 Quit MortalScan_Troll (Ping timeout: 240 seconds)
04:50:03 Join icarusfactor [0] (
04:51:01 Quit Judas_PhD (Quit: This is a quitting message)
04:51:05 Join factor [0] (
04:51:30 Quit icarusfactor (Client Quit)
04:51:53 Quit panni_ (Read error: Connection reset by peer)
05:04:59 Quit DerPapst1 (Quit: Leaving.)
05:15:57 Join CaptainKewl [0] (
05:23:25 Join LambdaCalculus37 [0] (
05:23:27 Quit LambdaCalculus37 (Changing host)
05:23:27 Join LambdaCalculus37 [0] (~rmenes@rockbox/staff/LambdaCalculus37)
05:27:03LambdaCalculus37Would anyone have any objections if I lowered the default contrast for the iPod 3G down from 50 to 35? It's very hard to see the LCD straight on with that high of a contrast setting by default.
05:27:25JdGordonnot at this time of day
05:28:26LambdaCalculus37JdGordon: I know, most everyone's asleep by now.
05:32:38saratogamake sure theres only 1 kind of screen, IIRC some of the defaults are chosen as a compromise between screen types on the ipods
05:34:09LloreanLambdaCalculus37: Too dark or too light?
05:34:18LloreanI can never remember if higher numbers on mono means more or less black
05:34:19LambdaCalculus37Llorean: Too dark.
05:34:39LloreanIt may be one of the compromises saratoga mentioned then
05:37:23 Quit Rob2222 (Ping timeout: 250 seconds)
05:38:12LambdaCalculus37Llorean: in firmware/target/arm/ipod/lcd-gray.c, the greyscale iPods have #defines for each target (start from line 73).
05:38:52LloreanLambdaCalculus37: Yeah, but I seem to recall some people with one of the grayscale iPods (and I *really* want to say 3G) having the screen completely white and unreadable
05:39:05LloreanI'd check when that contrast value was set and see if there commit message clarifies things
05:39:38LloreanLambdaCalculus37: Some of the iPods have multiple screen types that aren't distinguished in code or necessarily distinguishable, I think
05:41:22LambdaCalculus37The last revision to the file was r26208 by torne, and it was to increase the default contrast of the iPod 3G.
05:42:31LloreanIt'd probably be nice to know how apple gets the contrast right between versions
05:46:11saratogatheres some way to check which screen different ipods have and then adjust the defaults
05:46:27saratogaIIRC some of the older ipods do this, as does the nano2g
05:47:56LambdaCalculus37At least torne mentioned this wasn't the best fix in the comment; I quote: "It seems there is some significant variance between different greyscale ipods of the same model, so it's unlikely that *any* value will be great for everyone. I'm hoping that this is just high enough that the lightest screens aren't totally blank, but not dark enough that the darkest screens are entirely unreadably black ;)
05:48:22LambdaCalculus37Except that I got an iPod 3G where the default contrast was just too dark.
05:58:19saratogai guess you could find someone with a different LCD contrast setting and ask them for their hardware revision number
05:59:03LambdaCalculus37saratoga: Ahh, good call.
05:59:29*LambdaCalculus37 checks his
06:00:00LambdaCalculus37HW rev: 0x00030001; PP version: PP5002D
06:00:44LloreanLambdaCalculus37: If you could tilt it and read it, I think that's not "entirely unreadably black" (in the sense that my impression of the white screens was that some couldn't be seen at all without use of the voice interface or a hand written .cfg to fix them)
06:04:47saratogasame revision i have, mine needs a contrast around 35 to look good
06:05:36LambdaCalculus37saratoga: I set mine to 35 as well.
06:06:18saratogaIIRC amiconn had the other kind
06:06:53LambdaCalculus37amiconn: (for the logs) Can you check your iPod 3G and tell me what the HW revision of it is?
06:10:19 Quit JesusFreak316 (Ping timeout: 240 seconds)
06:13:02***Saving seen data "./dancer.seen"
06:15:28LambdaCalculus37saratoga: In the meantime, I'm going to post a patch to change the default contrast back to 35 for the iPod 3G on Flyspray.
06:16:19 Quit the_Kyle (Ping timeout: 240 seconds)
06:19:31saratogai don't think you need to post a patch for that
06:20:37*LambdaCalculus37 cancels his post
06:20:43saratogayou could probably just look at FS #11034 if you somehow forgot what the number was
06:26:16 Join crwl [0] (
06:32:51 Join the_Kyle [0] (~kyle@
06:34:14 Quit [Saint] (Disconnected by services)
06:34:16 Join S_a_i_n_t [0] (S_a_i_n_t@
06:48:07 Join Curulan [0] (
06:49:32CIA-45New commit by 03saratoga (r28841): Commit FS #11810 by Alexander Meshcheryakov. Boosts the CPU and limits LCD update rate while recursively scanning files in the properties plugin, ...
06:51:38 Quit Zarggg_ (Ping timeout: 260 seconds)
06:51:51CIA-45r28841 build result: All green
06:54:28 Quit saratoga (Quit: Page closed)
06:55:08 Join tchan [0] (~tchan@lunar-linux/developer/tchan)
06:57:18 Quit amee2k (Ping timeout: 255 seconds)
06:58:57 Join amee2k [0] (
07:00:53 Quit ReimuHakurei_ (Quit: BRB FBI)
07:01:08 Join ReimuHakurei [0] (~reimu@
07:44:28 Quit factor (Ping timeout: 276 seconds)
07:52:24 Join Horscht [0] (
07:52:24 Quit Horscht (Changing host)
07:52:24 Join Horscht [0] (~Horschti@xbmc/user/horscht)
07:54:43 Quit Horschti (Ping timeout: 240 seconds)
07:57:06 Join bmbl [0] (~bmbl@unaffiliated/bmbl)
07:57:36 Join factor [0] (
07:59:36amiconnLlorean: SDXC is probably backwards compatible hardware wise, but the sd* standards also specify the file system to use (which is silly, imo)
08:00:25amiconnSDXC specifies exFAT - which rockbox doesn't support, and neither does linux afaik
08:02:41 Join esperegu [0] (~quassel@
08:06:21eWillI think there's something wrong with the manual (pdf format). Searching for the word "shuffled" returns zero results; Copy (from the manual) the word "shuffled", then paste and the word "shued" is pasted; search for "shued" returns all instances of "shuffled".
08:07:09eWillI'm using Foxit pdf reader.
08:08:07 Join Keripo [0] (
08:10:33amiconnLambdaCalculus37: I don't have a 3G, only a 1G and a 2G
08:10:41eWillcould someone just try searching for the word "shuffled" and see if it works for them?
08:11:02 Quit LambdaCalculus37 (Quit: This computer has gone to sleep)
08:11:14amiconnDefault contrast on the greyscale ipods is a real problem, as different hw revisions require different contrast values
08:13:03amiconnThese revisions are not only dependent on the hw revision value; at least for some models gpio checks are necessary. Unfortunately I lost the respective rom routine code :\
08:13:06***Saving seen data "./dancer.seen"
08:13:30amiconnI still have the table, but not the code that accesses it
08:13:57 Quit Keripo (Quit: Leaving.)
08:15:09eWillFile browser Hotkey is not working right (I think). If set to "Insert", it inserts, but moves you to the WPS −− this isn't meant to be is it?
08:17:57the_KyleI believe if you're inserting a single file, you are moved to the WPS, but if you already have a file playing, you stay in the file browser when you insert another.
08:18:28eWillthen it's broken
08:19:10the_KyleThis is the same functionality I get if I select insert from the playlist menu if nothing is playing.
08:19:40eWilland it does the same thing if a song _is_ playing −− right?
08:20:18the_KyleNo. I believe it keeps you in the file browser if a song is playing. I'm checking just to be sure.
08:22:13 Join Keripo [0] (
08:22:35 Join Judas_PhD [0] (
08:22:48the_KyleConfirmed. If nothing is playing and I select insert from the playlist menu, I am taken to the WPS and the song begins to play. While that song is playing, if I go to the file browser and repeat the same steps, the next song is inserted and I am still in the file browser. Hotkey is the same.
08:23:34eWillwhat target?
08:24:13the_KyleBut I suspect this functionality would be the same on all targets.
08:26:06eWillthe_Kyle: it's even doing this on the sim
08:26:43the_KyleI think it's the same on all targets. Try a different sim to confirm.
08:27:10the_KyleThis is core, so shouldn't change on a different target as far as I know.
08:27:49 Quit Keripo (Quit: Leaving.)
08:28:17the_KyleIt would seem that the desired behavior would leave us in the file browser unless we hit the WPS key, even if nothing is playing.
08:28:18 Join sideral [0] (~sideral@unaffiliated/sideral)
08:28:19 Quit BHSPitMonkey (Read error: Connection reset by peer)
08:29:31the_KyleAlso, stopping playback with the power button while in the file browser shouldn't back us up to the parent directory also, but that's probably unrelated.
08:35:04eWillI just build the Ipod Video. I don't think it has a "Home" key. Do most targets have one?
08:36:56the_KyleI'm not sure about ipod video. I think most targets do, but I'm only familiar with clip+ because it's the player I have.
08:37:44the_KyleKeys on different targets will most likely map differently.
08:39:20the_KyleFor example, the clips have wps and browser hotkeys mapped, but the define is commented out. I had to customize my build to get the hotkeys.
08:39:23eWillthe_Kyle: I tried all the buttons. So the clip+ is running. I don't see a Hotkey setting in the menu....?
08:39:53eWillyou just answered me
08:40:24 Quit Judas_PhD (Remote host closed the connection)
08:40:42the_Kyleuncomment #define HAVE_HOTKEY and rebuild. WPS hotkey is mapped to down repeat and browser hotkey is unmapped if I recall.
08:40:54 Quit esperegu (Remote host closed the connection)
08:41:07 Quit liar (Ping timeout: 240 seconds)
08:41:19the_KyleI remapped both hotkeys to home+up. Works quite nicely here, and is finger friendly.
08:41:37 Join liar [0] (
08:42:04the_KyleKeymap is in apps/keymaps/keymap-clip.c
08:42:53 Join Judas_PhD [0] (
08:42:55the_KyleI'm not sure if remapping the hotkeys or uncommenting the #define will work on the sim though. I only tried it on a real clip+.
08:44:48*the_Kyle doesn't know why the hotkeys are left undefined in the clips. All I've gotten is guesses about running out of buttons so far, but I'm using unmapped combinations.
08:46:42 Join kevku [0] (~kevku@2001:7d0:0:f000::135d)
08:47:40*the_Kyle forgot to mention that the #define that needs to be uncommented is the last line of firmware/export/config/sansaclipplus.h.
08:47:50eWillthe_Kyle: would you mind uploading your keymap-clip.c?
08:48:20the_KyleeWill: Not a problem. Where should I upload?
08:50:25 Join Guest83802 [0] (~bjst@rockbox/developer/Zagor)
08:51:11eWillor anywhere
08:52:08 Nick Guest83802 is now known as Zagor (~bjst@rockbox/developer/Zagor)
08:52:32the_KyleIt says it filtered 6 characters. something about being outsie of ascii text guidelines or something. Here's hoping it works.
08:53:35*the_Kyle is waiting for the link.
08:53:55the_KyleBrowser is stuck.
08:54:30the_Kyleit tole me the connection was reset. Trying again.
08:54:32 Join bertrik [0] (
08:54:32 Quit bertrik (Changing host)
08:54:32 Join bertrik [0] (~bertrik@rockbox/developer/bertrik)
08:56:22*the_Kyle hopes it got the whole file.
09:02:48 Quit Judas_PhD (Quit: This is a quitting message)
09:03:56the_KyleeWill: Be sure to uncomment #define HAVE_HOTKEY in firmware/export/config/sansaclipplus.h. If this line stays commented, you still won't see a hotkey menu in your sim.
09:04:08eWilli did. thanks
09:06:30eWillthe clip+ BMP doesn't have lables on the keys
09:07:23eWilli didn't mean to type your name with no text after it
09:07:32the_KyleNot a problem
09:07:47*the_Kyle is visually impaired, so was unaware of bmp issues.
09:08:35*the_Kyle uses voice functions and knows where buttons are by touch.
09:08:52eWillwhat did you say you mapped the file browser hotkey to?
09:09:12the_KyleBoth file browser and wps are mapped to home+up.
09:09:23the_Kyleor home+play.
09:10:49eWillwhat does your home key do on WPS?
09:10:54eWillby itself
09:11:37the_KyleIt takes you to the menu. Holding it takes you to the quick screen.
09:16:54eWillthe_Kyle: ok I figured out the keys. 8 = up, * = home. (you have to hold * first to get hotkey to work!). Conclusion: same thing as my real Fuze v2 −− hotkey inserts, but takes me to WPS (even if song is playing).
09:18:07*the_Kyle checks with hotkey on my clip+. I thought it worked the same as using the menu.
09:19:54the_KyleConfirmed. Hotkey is always going to wps here.
09:20:24the_KyleMenu function, however, goes to wps only if no song is playing.
09:22:01*the_Kyle thought both context/playlist/insert and hotkey insert called the same function, but it seems they are slightly different.
09:22:48*eWill is still glad for info he got on how to remap keys −− something he's been wanting to explore
09:23:33 Quit CaptainKewl (Quit: ( :: NoNameScript 4.21 :: ))
09:23:45the_KyleNot a problem. I'm not that moch of a programmer, and I found the keymapping to be very easy to understand.
09:26:02 Join LinusN [0] (~linus@rockbox/developer/LinusN)
09:26:19eWillI'm trying to build the sim in Cygwin: simdisk/.rockbox: No such file or directory at /home/will/rockbox/tools/ line 129
09:27:39the_KyleHmm. I'm not at all sure about that one.
09:28:12the_KyleI'm running mine on Linux. I did make fullinstall for the sim.
09:28:38*the_Kyle isn't sure that made a difference.
09:28:50the_Kylefullinstall doesn't work on a real player.
09:29:31the_KyleBut if I remember correctly, fullinstall is what creates simdisk.
09:29:34eWillyou have to use fullinstall on the sim.
09:30:06the_KyleAnd fullinstall is giving you the buildzip error?
09:30:21 Join wodz [0] (
09:30:25 Quit bertrik (Ping timeout: 276 seconds)
09:31:18the_KyleAre you building inside of your source tree, or did you create a different build directory?
09:31:21eWillyeah. I'm now running Cygwin install to reinstall zip and unzip...?
09:31:30eWillnew build dir
09:31:49the_KyleEverything looks good, unless you have a missing package.
09:32:38pixelmayou have to use "make install" but not "fullinstall". The difference is that "fullinstall" adds the complete list of fonts, "install" just the ones needed by included themes
09:33:38the_Kylepixelma: Thanks. Sounds like that could solve eWill's problem.
09:33:59wodzehh, I just spotted another software i2c implementation in our tree :/
09:34:01pixelmawell, you need to zip and unzip anyway
09:34:10wodzTheSeven: ping
09:34:11pixelmabe able to
09:34:17 Join petur [0] (d408b802@rockbox/developer/petur)
09:34:38the_KyleThis is true. If zip and unzip are missing, nothing will work.
09:35:09 Part sideral
09:35:36eWillI already had zip and unzip, but reinstalled, and just ran "make install" −− same problem. You DO run "make" first −− right?
09:36:00the_KyleeWill: Yes. Make and then make install. line 129: if ((!$app) && $src && (abs_path($bindir) eq abs_path($src))) {
09:39:45*the_Kyle wonders why it seems to be complaining about simdisk/.rockbox when it's supposed to be creating this tree. Stupid question, but are you still in the build dir when you do make and make install?
09:40:51*the_Kyle knows little to nothing about cygwin.
09:41:32*the_Kyle is a Linux fanboy, so is pretty much stumped at this point.
09:42:57the_KyleDo you have a simdisk directory inside your build dir at all?
09:43:07the_KyleDid it get that far?
09:43:54eWillyeah, but I just deleted everything in the build dir and am trying again.
09:44:10the_KyleWhen in doubt...
09:48:26 Join pamaury [0] (
09:48:29 Quit pamaury (Changing host)
09:48:29 Join pamaury [0] (~quassel@rockbox/developer/pamaury)
09:50:42eWillif you configure chose your target, and select (S)imulator. In _Cygwin_ does that build a windows or linux binary?
09:51:10pixelmait should build a Windows binary
09:53:31pixelmaI wonder if you somehow aren't allowed to create a hidden directory (one with a . in front) in your case. Building sims in cygwin definitely worked for me but I'll try with an updated source tree now, will take some time
09:53:59eWillInstalling your build in your 'simdisk' dir
09:54:01eWillsimdisk/.rockbox: No such file or directory at /home/will/rockbox/tools/ line 129
09:54:02eWillmake: *** [install] Error 9
09:56:18the_KyleIn Windows, hidden files and directories use a special hidden file attribute. I believe the only potential problem is that files and directories that start with a dot are not hidden unless the hidden attribute is set in the filesystem. The file or directory should be created successfully.
09:56:57the_KyleI could be wrong, however, since the last version of Windows I used was XP. What version are you running?
09:57:33eWillwin7 x64
09:58:26the_KyleThat may explain things somewhat, considering that the filesystem is different on 7 as far as I know. It was supposed to be more flexible though, not less.
09:59:33pixelmamaybe some 64bit thing, I can only test 32bit and am currently running a test on XP so maybe results differ
09:59:48eWillI can run the sim, but (in the sim window) it says, "no .rockbox directory" "installation incomplete", while the console output is:
09:59:51eWillCan't open font: /.rockbox/fonts/12-Adobe-Helvetica.fnt
09:59:52eWillread_bmp_file: can't open '/.rockbox/icons/tango_small.bmp', rc: -1
09:59:54eWillread_bmp_file: can't open '/.rockbox/icons/tango_small_viewers.bmp', rc: -1
09:59:56eWillread_bmp_file: can't open '/.rockbox/backdrops/cabbiev2.bmp', rc: -1
10:00:37the_KyleTry creating the .rockbox directory under simdisk and try make install again. See if Windows itself gives you an error.
10:01:38pixelmathose are not fatal errors, I wonder though why it wants something about backdrops as the monochrome screens can't have them at all
10:01:53eWilloh i'm building the fuze
10:02:12eWilli built the clip+ in a linux VM
10:02:18the_Kylepixelma: The errors are just being generated because there's no .rockbox directory at all.
10:02:40TheSevenwodz: gnip
10:02:42eWillthere _is_ a .rockbox dir, and all three of those files are present
10:03:21eWillthe_Kyle: re-creating the .rockbox dir didn't work.
10:03:51pixelmathen it may be a file permission problem, could you check it on the .rockbox folder?
10:03:53eWillall *4* of those files
10:04:05*the_Kyle is truly stumped now. The assumption was that the directory didn't exist. Since it does, and it says it doesn't we have a very strange problem indeed.
10:04:06pixelmaor the individual files
10:04:51the_KyleIf it's a file permission problem, it should say "permission denied" rather than "no such file or directory."
10:05:07wodzTheSeven: You said nobody took photos of codec die used in ipod classic. I think I have access to equipment needed to do such things (although I never did this before)
10:05:11eWillpixelma: ah. I clicked on "Security" for .rockbox: "The permissions on .rockbox are incorrectly ordered, which may cause some entries to be ineffective."
10:05:35the_KyleWell, I was about to say it is worth a look.
10:05:43wodzI have access to SEM microscope as well as to optical crystalographic one
10:05:56wodzsulfuric acid bath is not the problem either
10:06:26TheSeveni already have someone else who did a few uncaps etc., but who doesn't have a broken ipod board right now
10:06:47TheSeveni'll ask him what kind of equipment he has currently and then get back to one of you
10:07:02TheSeven(i have a mechanically damaged ipod classic board lying around)
10:09:53pixelmaeWill: I don't know why you end up with such unclean permissions, but I do remember troubles with them too in cygwin.
10:10:33wodzpcf50606 drivers - what a mess
10:10:56wodztouching i2c mean touching half of lowlevel stuff in rb
10:11:10 Join ReimuHakurei_ [0] (~reimu@
10:11:13 Quit ReimuHakurei (Read error: Connection reset by peer)
10:13:07***Saving seen data "./dancer.seen"
10:13:25eWillwhat director is being referred to in this: simdisk/.rockbox: No such file or directory at /home/will/rockbox/tools/ line 129
10:16:23the_Kylesimdisk/.rockbox. But it doesn't specify its parent, so it's usually assumed to be .
10:16:45the_KyleShould be your build dir.
10:21:29 Quit mc2739 (Ping timeout: 260 seconds)
10:33:30 Join DerPapst [0] (
10:36:12 Quit JdGordon (Ping timeout: 240 seconds)
10:36:15 Join efyx [0] (
10:37:12 Join JdGordon [0] (~jonno@rockbox/developer/JdGordon)
10:37:32 Quit TheSeven (Ping timeout: 260 seconds)
10:37:43 Join n1s [0] (
10:37:43 Quit n1s (Changing host)
10:37:43 Join n1s [0] (~n1s@rockbox/developer/n1s)
10:46:29eWillguys, I feel like an idiot, and I'm really sorry. The .rockbox dir did exist, but it wasn't in /simdisk. So I deleted everything, ran "make && make install", _moved_ .rockbox into /simdisk, re-ran "make && make install", and it worked. (also it moved "/.rockbox" into "/simdisk".
10:47:24eWillthe_Kyle pixelma
10:48:45eWillthat is: it re-built /.rockbox and moved it into /simdisk.
10:49:20 Quit user890104 (Ping timeout: 272 seconds)
10:49:30the_KyleeWill: Good that you got it working.
10:49:45pixelmacygwin moved it or you did?
10:50:21pixelmabtw. I got the same error now in my test build, looks like there is something wrong
10:51:17eWillpixelma: I tried to make, and it was created at root of build dir. I moved it into /simdisk. Now I ran "make && make install" and durring the process it recreated .rockbox at root of build dir, finally moving it to /simdisk.
10:54:07eWillpixelma: so I deleted everything in /simdisk/.rockbox, but left the dir .rockbox. Builds fine. So I guess all you have to do is create /simdisk/.rockbox before you build, and all will work.
10:54:28eWill*italics unintentional
10:54:29 Join DX3 [0] (~Dre@
10:57:28pixelmano, you shouldn't need to move anything manually and I don't remember seeing the problem before
10:57:38 Quit afk (Ping timeout: 250 seconds)
10:58:19eWillI know it's not normal, I'm just giving temporary way to build it.
11:00:54eWillalso, you don't need to move anything, just create dir "/simdisk/.rockbox" in you build dir before you build.
11:02:27pixelmashouldn't be necessary either, maybe there's a timing problem or so
11:04:25 Join einhirn [0] (
11:05:50n1sor a cygwin problem
11:07:10pixelmait worked before
11:10:22eWillThe manual has a bug. In this screencap I selected and copied the word "shuffled" (in bold near the bottom), then pressed CTRL+F and pasted into the white text box you see at the bottom-left. One problem I know is that you can't search for the word "shuffled".
11:11:39gevaertsThat's annoying, sure, but I'm not sure if I'd call it a bug
11:11:47gevaertsIt's just tex using a nice-looking ligature
11:12:06eWillyou mean it's intentional?
11:12:20n1sno, it's an unfortunate side effect
11:12:42eWillSo don't bother with a bug report?
11:13:22pixelmasome PDF readers deal with it correctly. It's not nice and should be solved as it is hindering a search in the PDF
11:13:38pixelmathere already is a similar bug report for "fi"
11:13:51pixelmaor was
11:14:10eWillDocument Reader in Ubuntu, and Foxit Reader in Windows both fail.
11:14:20 Join Judas_PhD [0] (
11:14:30n1swell, it's more of a bug in the pdf generating tools than in our manual iiuc
11:15:42gevaertsTeX is originally meant for good looking typesetting, not for searching :)
11:15:59 Quit JdGordon (Quit: Leaving.)
11:20:06pixelmaok, the bug report was already closed - maybe time to reopen it?
11:20:16 Join TheSeven [0] (~TheSeven@rockbox/developer/TheSeven)
11:21:33gevaertsNot sure
11:21:46gevaertsThat one was about the html version
11:23:09pixelmahmm, yes, saw that now. Maybe the fix for this broke the other? And I'm pretty sure I saw something about "fi" too but can't find it in the tracker now, it may have been in the forums then
11:24:50 Join JdGordon [0] (~jonno@rockbox/developer/JdGordon)
11:25:02 Join InsDel [0] (
11:25:49pixelmaah yes,,25071.0.html
11:26:12*pixelma wonders if bluebroth3r knows more
11:28:07pixelmaI also still wonder if my commit this night is worth a backport
11:30:28gevaertsI'd say it's not essential, but since there's zero risk of it breaking anything it can go in if someone does the work :)
11:31:23eWilldoes "backport" mean insert the change into a previous release?
11:32:40gevaertsIt means committing the change to the active release branch, to be included in the next point release
11:33:53pixelmachecking in the changes into the release branch for a point release (3.7.X) that should contain bug fixes to 3.7 (in contrast to SVN build which can also contain new features)
11:34:41n1sand new bugs!
11:35:03pixelmanever! ;)
11:37:56wodzDo we have datasheet for pcf5060x series? I found only for pcf50603 but it is incomplete (lacking register map).
11:38:40 Join user890104 [0] (
11:38:41wodzthe comment in pcf5060x.h says v2.2 is available somehow
11:39:54wodzfound :-)
11:44:48pixelmait's possible that the problem with the .rockbox dir in cygwin exists for longer and I never noticed as I always build in the same folders and only do a "make clean" in the worst case and this doesn't touch the simdosk dir structure (and shouldn't)
11:45:51pixelmawell, I do build some other sims (other targets) in new folders from time to time so it can't be soo long ago
11:47:36 Quit kevku (Quit: KVIrc 4.0.2 Insomnia
11:51:11 Nick Sudos_ is now known as Sudos (
11:51:12 Quit cozmic (Read error: Connection reset by peer)
11:51:14 Join cozmic_ [0] (
11:57:12 Quit factor (Quit: Leaving)
12:02:10 Join factor [0] (
12:03:01eWillFile browser hotkey > Insert: takes you immediately to WPS. There is a patch shouldn't (think it needs resynced) it be commited?:
12:08:27TorneLambdaCalculus37: please don't change it back to 35, 35 is totally blank on other ipods, even 45 was unusably pale
12:08:30Torneoh, he's gone
12:08:52Tornewe really, really need to work out how to detect the lcd type
12:09:24Tornethe problem is the bootloader always uses hte default, with no way of changing it
12:09:36Torneso bootloader errors become impossible to read
12:12:43eWillI'm trying to eliminate the "Refreshing Your Media" from my fuze v2. Would a question about that be concidered RB-related?
12:13:11***Saving seen data "./dancer.seen"
12:13:27n1snot really
12:23:20Torneyou can't in any reasonable or sensible way. wait for the rockbox usb stack to be supported.
12:28:46 Quit krazykit (Ping timeout: 272 seconds)
12:30:15 Join krazykit [0] (
12:33:32 Quit bmbl (Quit: Verlassend)
12:34:58eWillTorne: all you have to do is delete the ##PORT# dir and fill the device till there is less than 177MB free space (on a 4GB fuze v2).
12:36:50Torne"all you have to do" :)
12:37:16Tornealternatively you could wait until the issues with usb on rockbox are sorted and then the problem no longer exists :)
12:42:16 Join LambdaCalculus37 [0] (~rmenes@rockbox/staff/LambdaCalculus37)
12:42:26 Quit LambdaCalculus37 (Remote host closed the connection)
12:43:58 Join dfkt [0] (dfkt@unaffiliated/dfkt)
12:44:41eWillTorne: I solved my problem. I needed a way to create a file of arbitrary size instantly. Here is the solution for Windows: fsutil file createnew <filename> <size in bytes> Still looking for a *nix solution.
12:45:09Tornedd if=/dev/zero of=somefile bs=1048576 count=numberofmegabytes
12:47:55 Quit krazykit (Ping timeout: 260 seconds)
12:47:58eWillTorne: that isn't instant.
12:48:04Tornewhy would it be?
12:48:11Torneyou can't write out a file isntantly
12:48:19eWillyou can in Windows
12:48:21Torneno you can't.
12:48:25TorneYou can write out a *sparse* file isntantly
12:48:29Tornebut that also works on unix
12:48:31eWilli just gave you the command
12:48:35Tornesparse files don't exsit on FAT
12:49:14Torneso that will be instant on NTFS, but take exactly as long as dd on FAT
12:49:31 Quit soap (Read error: Connection reset by peer)
12:49:50 Join krazykit [0] (
12:49:53TorneFAT has no concept of a file taking up space unless it contains data
12:51:32TorneIt is physically possible, I guess; you would just have to allocate the blocks in the FAT without writing any data into them, and the resulting file would contain whatever happened to be on those sectors of the disk before
12:51:40Tornebut that's not what fsutil does
12:52:08Torneit explicitly creates a file cotnaining zeros; on a filesystem with sparseness it can do that without allocating any real blocks, but that also won't solve your problem since a sparse file *doesn't take up any space*
12:52:32Tornea 1GB sparse file of zeros doesn't reduce the available space on the filesystem by 1GB :)
12:53:16Tornei should've picked a smaller number to confirm my belief, also, my stupid card reader is now writing out a bazillion zeros and it doesn't respond to ^C
12:55:18 Quit InsDel (Ping timeout: 264 seconds)
12:55:37pixelmareminds me of the good old times when setting the System folder on a c200v1 was the way to prevent the OF's database refresh
12:56:06pixelmaor at least shorten it so it only splashes an error message and aborts
12:56:08eWillTorne: CTRL+Break?
13:01:42 Quit The_Pwny (Quit: Clap on! , Clap off! Clap@#&$NO CARRIER)
13:05:54 Join soap [0] (~soap@rockbox/staff/soap)
13:10:23 Join xavieran [0] (
13:19:25 Join earcar [0] (
13:24:58 Join teru [0] (
13:25:31 Quit n1s (Quit: Lämnar)
13:27:35 Quit TheSeven (Ping timeout: 260 seconds)
13:31:14eWillFile browser hotkey > Insert: takes you immediately to WPS. There is a patch shouldn't (think it needs resynced) it be commited?:
13:33:53 Join panni_ [0] (
13:35:24 Quit BlakeJohnson86 (Read error: Connection reset by peer)
13:37:17 Join TheSeven [0] (~TheSeven@rockbox/developer/TheSeven)
13:37:46 Join BlakeJohnson86 [0] (
13:39:14 Join bluebrother [0] (~dom@rockbox/developer/bluebrother)
13:43:06 Quit bluebroth3r (Ping timeout: 276 seconds)
13:49:48 Join Rob2222 [0] (
13:50:32 Quit TheSeven (Ping timeout: 250 seconds)
13:51:55TorneeWill: looks fairly sensible, but my build machine exploded so I can't easily test/commit anything at present
13:52:06Torneanyone want to check that one out? seems sane
13:52:18Torneer, by that one I mean FS #11344
13:56:30 Join tejas [0] (
13:57:30 Part tejas
14:04:38 Quit petur (Quit: IT is going to 'fix' this PC.. ;))
14:08:00 Join TheSeven [0] (~TheSeven@rockbox/developer/TheSeven)
14:08:45 Quit antil33t (Read error: Connection reset by peer)
14:08:53 Join antil33t [0] (
14:09:00 Quit factor (Read error: Connection reset by peer)
14:12:43 Join factor [0] (
14:13:14***Saving seen data "./dancer.seen"
14:30:35 Quit panni_ (Read error: Connection reset by peer)
14:33:58 Join mortalscan [0] (~mortalsca@
14:35:48 Join bmbl [0] (
14:35:49 Quit bmbl (Changing host)
14:35:49 Join bmbl [0] (~bmbl@unaffiliated/bmbl)
14:37:06 Join sideral [0] (~sideral@unaffiliated/sideral)
14:45:49 Join petur [0] (d408b802@rockbox/developer/petur)
14:46:11 Quit sideral (Remote host closed the connection)
14:46:47 Join sideral [0] (~sideral@unaffiliated/sideral)
14:49:10 Join mc2739 [0] (~mc2739@rockbox/developer/mc2739)
15:07:16 Join komputes [0] (~komputes@ubuntu/member/komputes)
15:39:38 Part sideral
15:52:51 Quit teru (Quit: Quit)
15:54:10 Quit Guest24228 (Read error: Operation timed out)
15:58:27 Join madalu [0] (
15:58:30 Quit madalu (Changing host)
15:58:30 Join madalu [0] (~user@unaffiliated/madalu)
15:58:38 Join Feisar [0] (
15:59:05 Nick Feisar is now known as Guest11901 (
15:59:11 Part LinusN
16:03:21 Quit antil33t (Read error: Connection reset by peer)
16:03:31 Join antil33t [0] (
16:04:37 Quit DX3 (Quit: I was raided by the FBI and all I got to keep was this lousy quit message!)
16:08:39 Join Dreamxtreme [0] (~Dre@
16:13:17***Saving seen data "./dancer.seen"
16:15:26 Quit Guest11901 (Ping timeout: 265 seconds)
16:17:01 Join Guinness` [0] (
16:17:09 Quit Guinness (Read error: No route to host)
16:17:10 Quit madalu (Remote host closed the connection)
16:32:37 Join benedikt93 [0] (~benedikt9@unaffiliated/benedikt93)
16:33:30 Join kevku [0] (
16:33:57 Quit Guinness` (Ping timeout: 240 seconds)
16:39:31 Join saratoga [0] (9803c6dd@gateway/web/freenode/ip.
16:40:07 Quit evilnick_B (Quit: Page closed)
16:42:14saratogapictureflow on the Fuze is pretty cool, but its too hard to quit
16:42:29saratogai got into it and it took me forever to find the rockbox menu's again
16:42:58 Join jgarvey [0] (
16:44:52 Join Guinness [0] (
16:46:15 Part Zagor
17:01:35 Join eWill_ [0] (
17:01:47 Join Keripo [0] (
17:02:45 Quit eWill (Ping timeout: 240 seconds)
17:02:47 Nick eWill_ is now known as eWill (
17:04:40 Join evilnick_B [0] (0c140464@rockbox/staff/evilnick)
17:07:28 Join eWill_ [0] (
17:07:56 Quit eWill_ (Client Quit)
17:08:45 Quit eWill (Ping timeout: 240 seconds)
17:18:00 Join marines [0] (
17:18:01 Quit krazykit (Quit: eat a bag of dicks, at&t)
17:26:02pixelmathis change to buildzip causes the .rockbox folder to be created directly in your build folder (not in the simdisk subfolder) when compiling in a fresh builddir under cygwin, any idea why this is the case?
17:26:28 Quit Keripo (Quit: Leaving.)
17:27:10 Join eWill [0] (
17:27:53eWillI want to do a battery bench, but drain the player as fast as possible. Does the LCD being on effect the voltage reading much?
17:30:17 Join Keripo [0] (
17:30:29LloreaneWill: If you'd started the battery bench yesterday, you probably would've finished it already. Is there a reason it must be the fastest possible?
17:31:24eWillI don't see a reason to drag it out? Also, I'd like to be awake when it shuts off so I can get a charge into it immediately (lipo).
17:32:03LloreanThey don't shutdown at "empty" and it won't take damage if it doesn't get charged for a day after it empties.
17:32:32LloreanThey seriously don't expect people to have to know that an empty battery can get damaged and are designed generally to be safe. If you plan to leave it weeks or months without charging after it turns off, that's different.
17:34:22TorneYeah, as long as you charge it again within about 2-4 weeks it'll be fine.
17:34:41Torneit won't selfdischarge to a dangerous level for probably a month
17:34:47 Join krazykit [0] (
17:34:51eWillI knew they weren't empty, it's good to know i have time. 2-4 weeks. Wow.
17:34:58Torneselfdischarging is *slow*
17:35:12eWillbut doesn't it speed up as the voltage lowers?
17:35:27Tornenot to a particularly fast speed
17:36:01Tornealso the shutoff voltage for a given mp3 player is going to be substantially above the point where the cells are at risk
17:36:13Torneusually ~3.1V or so at least
17:36:47Tornethe manufacturer usually guarantees them down to 3.0V
17:36:54Torneand in practise you can go past that a bit without worrying
17:38:05 Join krazykit` [0] (
17:38:21 Quit krazykit (Disconnected by services)
17:38:23 Nick krazykit` is now known as krazykit (
17:40:38eWillSo do any of you lower your cut-off voltage for your personal player? How much extra time do you get?
17:42:55pixelmaI don't think a lot of people do this at all, if there is one person at all
17:43:19TorneThe cutoff voltage in rockbox is not to protect the cells
17:43:35TorneIt's to make sure there's enough power left to spin up the storage and write out settings and resume position
17:43:44Tornei.e. to ensure a "clean" shutdown instead of a hard poweroff
17:44:07TorneIf you lower the cutoff voltage all that will happen is it'll run for slightly longer then it'll hard-poweroff when the regulators fail to be able to generate eonugh power from the lowered battery voltage
17:44:20Torneand that'll *still* be some distance away from the danger point for the cell
17:44:37Tornea lipoly that's discharged down to 3.08V or whatever can't power a 3.3V buck regulator any more :)
17:45:50TorneI'm not certain that this is true for *every* target, but certainly the vast majority of them will power themsleves off by simple regulator failure in plenty of time to save the batteries, whatever rockbox does or doesn't do.
17:47:34Tornei'm not sure why you are wanting to do a battery bench as fast as possible in the first place, thoguh
17:47:45Tornethe point of the battery bench is to determine what a typical runtime for the player is
17:47:56Tornewhich you'd do by having your normal settings, not ones specifically intended to run it down faster
17:48:29Torneit's not for testing the *battery*, it's for testing the effectiveness of rockbox's power management and the efficiency of our code :)
17:48:39eWillall I need is the voltage levels to build an accurate percentage table for the source code.
17:48:59Torneoh, i see.
17:49:10Tornein which case you want to ideally do something that leaves the battery drain constant
17:49:25Tornei.e. no playback
17:49:32Torneand for maximum efficiency, minimal load as well
17:49:41Torneplayback means buffering and the like, which means load fluctuates
17:49:54Torneer, maximum *accuracy* even
17:50:04Tornebut the load level probably doesn't matter too much
17:50:18Torneunless the battery is ancient it won't change the measured voltage much
17:50:20LloreanI'd go for high efficiency so that you end up with a lot of readings at each percentage
17:50:34TorneYeah, so player idle, with backlight off, etc
17:50:49Tornethat will unfortunately take much longer
17:50:52Tornebut it's more accurate
17:50:54LloreanMaybe idle with backlight on?
17:51:01Tornethat's probably ok too
17:51:11Tornebut yeah, playback is gonna make the load decidedly lumpy
17:51:19Tornesice it'll have to do loads more work than average every time it rebuffers
17:51:24eWillthank you
17:52:23 Join Feisar [0] (
17:52:49 Nick Feisar is now known as Guest62744 (
17:58:51 Quit Guest62744 (Ping timeout: 265 seconds)
18:05:14 Quit petur (Quit: Page closed)
18:13:20***Saving seen data "./dancer.seen"
18:23:03 Quit Horscht (Quit: Verlassend)
18:30:43 Quit antil33t (Read error: Connection reset by peer)
18:30:53 Join antil33t [0] (
18:42:04 Join bertrik [0] (~bertrik@rockbox/developer/bertrik)
18:46:05 Join Strife89TX [0] (~cstrife89@
18:50:30 Join LambdaCalculus37 [0] (~3f74f70d@rockbox/staff/LambdaCalculus37)
18:51:33 Quit eWill (Quit: ChatZilla 0.9.86 [Firefox 3.6.13/20101203075014])
18:57:14saratogawould it be possible to make the DSP code insert a single 0 sample whenever pausing so theres no posisbility of a DC offset in rockbox?
18:58:31LambdaCalculus37b0hoon: (for the logs) Does the GoGear HDD6330 have a car adapter mode? I don't remember off the top of my head of it does.
18:58:56saratogai think all players have it
18:59:21saratogait may not actually work as expected though depending on how the bootloaders react to AC power
18:59:51LambdaCalculus37saratoga: Just wanted to make sure so I can add the option for the manual.
19:02:49pixelmaI doubt that all players have it, e.g. the Ondio doesn't really charge but give you a way to run off of USB power
19:09:36 Quit TheSeven (Ping timeout: 276 seconds)
19:15:44pixelmaLambdaCalculus37: if you are lucky and this is something you can in- or exclude based on a feature, then it may already be correct
19:16:55LambdaCalculus37pixelma: I'll look on my HDD6330 again tonight.
19:19:27 Quit mortalscan (Ping timeout: 250 seconds)
19:21:10 Join JesusFreak316 [0] (
19:21:50 Quit JesusFreak316 (Remote host closed the connection)
19:25:54 Quit efyx (Read error: No route to host)
19:26:07 Quit DerPapst (Quit: Leaving.)
19:26:19 Join efyx [0] (
19:26:32 Quit earcar (Quit: earcar)
19:27:04 Join mortalscan [0] (~mortalsca@
19:29:21 Join freddyb [0] (
19:29:25 Quit benedikt93 (Quit: updating pchat)
19:29:38 Quit LambdaCalculus37 (Quit: we're off to find a shrubbery!)
19:31:07 Quit efyx (Read error: Connection reset by peer)
19:32:59 Join efyx [0] (
19:34:39 Join Jerom [0] (~jerome@
19:36:04 Quit wodz (Ping timeout: 260 seconds)
19:36:30 Quit mortalscan (Ping timeout: 265 seconds)
19:39:18 Join mortalscan [0] (~mortalsca@
19:40:30 Join benedikt93 [0] (~benedikt9@unaffiliated/benedikt93)
19:44:07 Join Horscht [0] (
19:44:07 Quit Horscht (Changing host)
19:44:07 Join Horscht [0] (~Horschti@xbmc/user/horscht)
19:44:29 Quit einhirn (Quit: Miranda IM! Smaller, Faster, Easier.
19:52:45freddybSaratoga: I've been running my Fuze at 23.25/23.25MHz and 61/124MHz and it still performs very well. Even mpegplayer is fast enough for me. Give me some time and I'll try to see how much battery life is affected.
19:54:47freddybI also ran some tests on anti-aliased fonts and it seems that AA fonts render about 50-70% as fast as non-AA fonts.
19:55:50 Join hebz0rl [0] (
19:56:33 Nick Rondom_ is now known as Rondom (
19:58:09 Join DerPapst [0] (
20:02:55Lloreanfreddyb: How fast is "fast enough for me"?
20:03:10 Join TheLemonMan [0] (
20:03:30LloreanThe nice thing about mpegplayer is that there's sample videos on the download server, and you can disable framerate limiting and frame skipping to get relatively clear numbers for how it's performing.
20:04:11saratogampegplayer is really fast on the the fuze, so its probably not an issue
20:04:17saratogaAPE or AAC+ might be though
20:05:08saratogafreddyb: your test results are interesting because they suggest that boosted performance is extremely poor on the fuze, but unboosted performance is quite good
20:05:26Lloreansaratoga: Well, having an actual number can't hurt and only takes a few minutes to test, though.
20:05:47freddybI tried Dune and got 29.5ish I tried Futurama and got rates from 29 down to 22. Elephant's Dream runs about 30.5 unlimited.
20:06:10LloreanThat sounds like barely fast enough then...
20:06:18freddybI'm not sure why Fututama is slower...
20:06:38freddybElephants Dream is really only 24fps, tho.
20:07:05LloreanYeah, but if the upper limit on framerate is about 30fps, that doesn't give much leeway.
20:07:08saratoga124 might be a little low, although with a 3 level boost scheme it would be fine IMO
20:07:21LloreanYeah, with a 3 level boost and 124 being mid-level it's probably okay.
20:07:34saratogajust use higher clock for mpegplayer and ape
20:07:35LloreanI'm just saying, you don't want the maximum speed to be "just barely above realtime"
20:07:40freddybSaratoga: I should have suspected that memory couldn't keep up with high multiples.
20:07:49saratogayeah most likely
20:08:26freddybLlorean: You really don't notice any dropped frames, the audio is unbroken.
20:08:33saratogathis reminds me, mpegplayer on the fuzev2 and other armv5 devices could pretty easily be made faster by just tweaking the armv6 asm code a bit to remove a handful of v6 only instructions
20:10:35Lloreanfreddyb: With your sample videos maybe, but you should always plan for a little extra leeway, rather than hoping that the slowest you can get the CPU while not being below 29.97 on your test files will be good enough.
20:10:37bertrikI thought the fuze needed a high PCLK (62 MHz) for smooth scrollwheel operation
20:11:09Lloreanfreddyb: The skipped frames may not be noticeable in futurama, which is animation anyway, but may be more noticeable in other things depending on the distribution of them.
20:11:18LloreanAlso, was that the 4:3 or the 16:9 elephant's dream?
20:11:48freddybThis is just my opinion but if I can get more battery life, it's worth it even if I drop a few frames on movies unless it gets to the point that you can visually see lagging or jerking.
20:12:29Lloreanfreddyb: But you're forcing that decision on others. Generally if we're matching or outperforming the OF we then shouldn't introduce inconvenience or reduced functionality just for battery life.
20:12:43freddybLlorean: 224x176 469kbps.
20:12:49LloreanWhat's the fuze's screen?
20:12:55LloreanI thought the fuze was 320x240
20:13:15LloreanSo it's the 4:3
20:13:22***Saving seen data "./dancer.seen"
20:14:50freddybLlorean: you know that motion pictures are only 24fps?
20:14:58Lloreanfreddyb: And TV is 29.97
20:15:05gevaertsTV is 25!
20:15:08pixelmafreddyb: I agree that numbers would be better, to me it doesn't sound very comforting
20:15:11LloreanHow about "up to 29.97"
20:15:39LloreanI'm all for better battery life, but not at the cost of reducing offered functionality.
20:16:44saratogaLlorean: i think its premature to discuss this since we don't even know if its desirable to reduce the boost clock ...
20:17:10saratogaif it is we can simply use different clocks for mpegplayer, until then its silly to speculate
20:17:51gevaertsIt's not just mpegplayer though
20:17:57bertriksaratoga, svn r11765 does already reduce the boosted clock, right?
20:18:11Lloreansaratoga: And it's worth bringing up now, since if it's not brought up now what if someone who thinks of this isn't around later?
20:18:27gevaertsIt could mean the difference between being able to to use lots of dsp with high-CPU codecs and having to use high-CPU codecs without any dsp
20:18:47LloreanDo we have a way to test_codec with DSP?
20:19:00LloreanOr do we have an assumption like, "faster than realtime by at least X%" that's generally considered good enough?
20:19:08saratogaLlorean: yes we do
20:19:45saratogabertrik: yes
20:19:56gevaertsThere's a difference between "good enough" and "still better than some other players, but worse than last week"
20:21:18LloreanI just caught the language of "fast enough for me" and thought that it should be discussed what fast enough is for different people.
20:21:29LloreanSince without a discussion of it at some point, it may not come up at a later discussion of the patch.
20:21:32saratogathis has actually been discussed quite a few times
20:21:57saratogagevaerts: yes but its not really an issue
20:22:36gevaertssaratoga: for people who play APE with five EQ bands in use, crossfeed, and pitch adjustment, it is :)
20:22:43saratogano that should still work
20:23:12saratogac3000 with 5 EQ bands should still be fine, and c4000 won't work now anyway
20:23:18LloreanAs long as the ability for mpegplayer to get the speed it needs to playback consistently is in place before a slowdown is implemented that would otherwise make it likely to drop below the framerate of the videos I'm happy.
20:23:25gevaertsok then
20:23:40saratoga5 EQ bands is like 15 MHz
20:23:49LloreanBut as long as the player's physically capable of playing video without dropped frames, I can't see it as being reasonable to make it do so without extremely significant battery life gains (or being in an extremely poor battery life position to begin with)
20:24:00saratogabut really if we're going to have another boost level for mpegplayer might as well have it for ape too
20:24:19LloreanBy "do so" I meant "drop frames" just to clarify
20:24:30gevaertsSwitching to a different boosted level for specific cases shouldn't be too hard I think
20:24:40gevaertsA lot easier than a generic multi-level boost anyway
20:24:57saratogaamiconn had other ideas, but I wanted to just have a codec and plugin API call that increases the boost clock
20:25:06saratogasince its simple
20:25:21LloreanSeems like a good idea.
20:25:32*gevaerts nods
20:25:53saratogae.g. APE calls ci->boost_high(true) which either increases the boost clock or evaluates to a NOP if the device only has 2 levels
20:25:58pixelmado more boost levels really gain this much in everyday use?
20:26:13 Join Feisar [0] (
20:26:22pixelmaI just wonder if it's really worth it
20:26:25saratogapixelma: funny you ask, I suggested we check if that was the case before arguing about this
20:26:40 Nick Feisar is now known as Guest75111 (
20:26:51 Part marines
20:27:10saratogabut basically we've been sitting on this idea for 3 years now because we never had any evidence that it was worthwhile on any hardware
20:27:28saratogaalthough i suspect it will be on the AMS players and possibly the Nano2G
20:27:32Lloreansaratoga: Aren't gains (generally) from decreasing unboosted speed in many cases?
20:27:53saratogain this case they'd be from running the CPU clock at a lower multiple of the memory clock
20:28:07saratogawhere its more efficient (theoretically)
20:29:11freddybbertrik: Scrollwheel seems fine at low clocks.
20:29:38 Join BHSPitMonkey [0] (~stephen@unaffiliated/bhspitmonkey)
20:29:49bertrikoh maybe it was the e200v2, ask kugel
20:30:09saratogaaren't the e200v2 and fuzev1 identical with respect to the scroll wheel?
20:33:24saratogaheh dsp_arm.S isn't scheduled correctly for anything but ARM7
20:34:00 Quit Strife89TX (Quit: Leaving briefly.)
20:34:38saratogaif the EQ can use 16 bit coefficients it can also be massively faster on arm9E
20:34:53saratogabut i guess its already about 2MHz per EQ band so maybe theres not much point in speeding it up more
20:35:38freddybtest_codec is not affected by audio settings, right?
20:35:54saratogadepends which option you choose
20:35:59saratogathe DSP one is, the regular one is not
20:37:16 Quit Keripo (Quit: Leaving.)
20:38:22 Join TheSeven [0] (~TheSeven@rockbox/developer/TheSeven)
20:41:38freddybIs there already a way to do a battery bench for mpegplayer?
20:45:06 Join InsDel [0] (
20:46:03CIA-45New commit by 03saratoga (r28842): Avoid an obvious stall on everything newer then arm7tdmi. Note: this can probably be made a lot faster on newer arm targets.
20:46:19freddybOne reason I was looking at a lower top frequency was that I wanted to improve battery life for video.
20:46:26saratogacan't mpegplayer repeat?
20:46:37 Quit Dreamxtreme (Read error: Connection reset by peer)
20:46:38saratogaFWIW I doubt you'll see much improvement, battery life in video is dominated by the screen
20:47:23 Join Dreamxtreme [0] (~Dre@
20:47:27 Quit Dreamxtreme (Read error: Connection reset by peer)
20:47:56*TheSeven wonders who/what broke the nano2g headphone detection
20:48:03CIA-45r28842 build result: 5 errors, 0 warnings (saratoga committed)
20:48:07saratogadamn it preglow needs to get online so i can ask him how this EQ works
20:48:48saratogagonna guess my change to arm ASM didn't really break the ipod simulator
20:49:28gevaertsIt does seem unlikely
20:49:46saratogayeah just some random build glitch as far as I can tell
20:51:45gevaertsStill weird
20:52:34saratogathis EQ asm makes arm11 sad
20:53:30gevaertsIt's not actually broken...
20:54:00gevaertsIt's just that those note: lines make the build system think that something went wron
20:54:26CIA-45New commit by 03theseven (r28843): Fix headphone detection on iPod Nano 2G being inverted (regression from r28800)
20:56:22CIA-45r28843 build result: 10 errors, 0 warnings (theseven committed)
20:56:52*TheSeven pokes Stummi
20:56:58TheSevenwhat on earth is this compiler doing?
20:57:20*gevaerts is fixing it
20:57:27TheSevenbut why does it output these lines?
20:57:40TheSeveni don't think anyone has asked it where those things are defined...
20:57:41gevaertsgcc 4.3 outputs those apparently
20:57:55*gevaerts reproduced it locally
20:58:09TheSevennotes about symbols being defined, with no apparent reason *why* it's saying that?
20:58:19CIA-45New commit by 03gevaerts (r28844): Ignore note: lines that some gcc versions like to output
20:58:40gevaertsI hope that's correct...
20:59:11*pixelma wonders if now is a good time to repeat the question about r28160 from earlier
20:59:17gevaertsBagder: could you svn up the build server, or at least the file there?
20:59:22 Join Dreamxtreme [0] (~Dre@
21:00:45gevaertsBagder: maybe first review that diff. I'm not *that* good at perl
21:01:46pixelmaif it doesn't make the build table vanish as LambdaCalculus' commit semi-recentlly, you are all set ;)
21:02:27gevaertshm, yes, I could probably commit a change to tools/configure that runs an svn update in the right place, and then revert that :)
21:04:22pixelmaspeaking of perl... the revision I asked about changed something in, want to have a look why this causes the .rockbox dir to be created outside the simdisk subfolder in a fresh builddir under cygwin? ;)
21:04:55 Quit Jerom (Ping timeout: 264 seconds)
21:05:56TheSeventhere is only one device on the ipod classic's first i2c bus
21:06:04TheSeven(which is the pmu)
21:06:17TheSevenand the second bus doesn't work at all => is the codec hiding there?
21:07:12bertrikTheSeven, some codecs are write-only, but I would still expect them to ack an i2c transaction
21:07:28TheSevenhm, i did a trivial read scan
21:07:38TheSevenso you have a point :)
21:07:57*TheSeven should dig deeper into the driver
21:08:09TheSevenand probably fix that issue preventing me from using the second bus
21:17:20 Quit jgarvey (Ping timeout: 240 seconds)
21:21:56 Join Jerom [0] (~jerome@
21:24:37 Quit Jerom (Client Quit)
21:24:53 Join Jerom [0] (~jerome@
21:25:52 Quit Jerom (Client Quit)
21:26:10 Join Jerom [0] (~jerome@
21:26:15 Join n1s [0] (~n1s@rockbox/developer/n1s)
21:26:28 Quit benedikt93 (Quit: Bye ;))
21:26:41 Quit Jerom (Client Quit)
21:27:00 Join Jerom [0] (~jerome@
21:27:07 Quit JdGordon (Ping timeout: 260 seconds)
21:27:22saratogahuh replaygain could be a lot faster on arm9 too
21:28:04 Quit bmbl (Quit: Verlassend)
21:28:15 Join stripwax [0] (
21:29:37saratogastripwax: ping
21:30:28 Quit Jerom (Client Quit)
21:30:29saratogastripwax: how hard would it be to reverse the tricks you added to the mdct lib to save memory for targets that don't have IRAM?
21:30:49saratogai suspect that on arm11 the old way would be quite faster since theres no IRAM and sequential accesses are much faster
21:32:00stripwaxinteresting. probably not too hard. that's related to FS #11268 right?
21:32:23*TheSeven reminds saratoga that there might be ARM11 targets with IRAM one day (iPod Nano 4G)
21:32:25saratogastripwax: not specifically
21:32:25stripwaxoh wait. i'm thinking of libtremor. which tricks did I add to the mdct lib?
21:32:43saratogayou ported the tremor trig table trick to it
21:33:02saratogathe old ffmpeg version used much larger tables with more regular memory access patterns
21:33:06stripwaxcrikey, that old chestnut.
21:33:20saratogabuschel and I noticed that liba52 memory scaling is much different then all the other mdct codecs
21:33:21stripwax'difficult' but not impossible.
21:33:29saratogaand it uses the fft, but not the mdct code
21:33:38saratogainstead it uses its own trig tables arranged like ffmpeg's
21:33:41 Join Trunkz|ZzZ [0] (~Trunkz@
21:34:15saratogaunfortunately our mdctlib is now quite well optimized for arm7/arm9 but in a lot of ways we've made it worse on newer arm targets
21:34:29 Join jgarvey [0] (
21:34:33Trunkz|ZzZI take it the forums is the only place to find out about any possible information on the rockbox-ipod classic effort?
21:34:45stripwaxyeah, it's probably not great on any targets without iram trig tables
21:34:53Trunkz|ZzZthanks saratoga
21:34:56 Nick Trunkz|ZzZ is now known as Trunkz (~Trunkz@
21:34:57 Quit Trunkz (Changing host)
21:34:57 Join Trunkz [0] (~Trunkz@unaffiliated/trunkz)
21:35:14saratogaon arm11 the key is basically using lots of ldm, avoding using registers right after you load or multiply to them, and and avoding random memory access patterns
21:35:22stripwaxffmpeg had three tables, iirc. two for the fft part, and a separate one for the mdct.
21:35:42saratogaIMO its probably enough to just special case the 2048 point transform
21:35:45n1sstripwax: if you have time, could you look at FS #11268 some time?
21:35:52stripwaxone of the tricks (reusing the fft tables for the mdct bit, using interpolation ... which I nicked from the original libtremor) could probably still be kept
21:35:55saratogaand unwind the memory accesses for that one, even if it doubles the table sizes
21:36:11saratogasince its the only performance critical one
21:36:30saratogaon arm11 and to some extent arm9 its worthwhile to waste a lot of ram if it allows for more sequential loads
21:36:37stripwaxn1s - ulp. yep. unfortunately i just don't have the time currently :(
21:37:02n1sstripwax: no, problem
21:37:09n1seh no problem
21:37:10saratogayeah i'm in no hurry to look at this either
21:37:24saratogamp3 is going first, but eventually i want to improve the mdct lib
21:37:31 Join Gareth [0] (
21:37:44saratogapretty soon i think i'll get mp3 faster then vorbis on some targets and thats no good ...
21:38:36saratogaarmv5 and armv6 actually have some interesting instructions, just they're not too easy to use
21:38:47saratogathis mp3 project has been a nightmare
21:38:55n1ssaratoga: i managed to make a function that does the mdct doubling and windowing at the same time for tremor but havent' written the asm yet
21:39:10saratogastripwax and i argued abotu doing that a while ago
21:39:57saratogajust make sure you write asm thats not awful on arm11 if you do
21:40:24stripwaxn1s - the benefit comes from reduced load/store presumably?
21:40:53Trunkzoh goodie
21:41:00Trunkzso atleast we're getting somesort of support on that front.
21:41:01saratogahmm the menus in testcodec are broken
21:41:19n1sstripwax: yes, almost one load and store per sample in the block so about 2 loads and 2 stores per output sample
21:43:00 Quit Trunkz (Quit: ( :: NoNameScript 4.22 :: ))
21:43:48n1sthe second half of the mdct doubling (the in place one) took a while to get right and as i've written it now it requires 8 different pointers so i need to make it somewhat cleverer to have enough regs
21:44:12 Quit preglow (Ping timeout: 250 seconds)
21:44:19 Join preglow [0] (
21:44:29 Join tchan1 [0] (
21:45:12saratogan1s: i think you broke test_codec a week or two ago
21:45:41stripwaxn1s - some of that might be made easier (or harder...) if the imdct_half generates the half output into a different part of the full output array (beginning, middle, end)
21:46:12saratogan1s: when i select speed test with DSP it actually selects the item below that (speed test folder with dsp)
21:46:13stripwaxjust because i made imdct_half generate into the current place doesn't mean that's optimal for your change
21:46:39n1ssaratoga: that change fixed the checksum thing for me that got broken with your earlier change
21:46:41 Quit tchan (Read error: Operation timed out)
21:46:45stripwaxalthough hm, i guess i might not understand what you mean by 'inplace'
21:47:04saratogahmm no maybe i broke it
21:47:34stripwaxyou cache the half-imdct for the next block, and then need to (re)unfold it to do the next block's overlap? so your gain would be halved since you process the same half-imdct twice?
21:48:06n1sstripwax: right now the imdct_half places the output in the upper half of the output buffer, half of this is used to fill in the lower half and half is used to fill in the upper half but it is already in the upper half so it's quite complicated
21:48:17saratogacan't we just use the ffmpeg algorithm for this?
21:48:27saratogaIIRC theres is optimal in terms of mutliplications
21:48:38stripwaxright. i think tremor put it into the middle and then moved things around eithers ide of that
21:48:39saratogai think our wmapro decoder did this correctly
21:48:42n1ssaratoga: the tradeoff is memory iiuc
21:48:49stripwaxyeah, mults is fixed
21:48:54saratogaas in loads and stores?
21:49:02stripwaxyou still have to window every sample on the output side (not on the half-imdct side)
21:49:17saratogathe ffmpeg version uses the imdct half trick and is fully general in that it will work with any transform size
21:49:26saratogamt ported it last summer
21:49:46stripwaxthe challenge is making vorbis do that
21:49:48n1ssaratoga: yes but i think ffmpeg stores the half mdct in a separate buffer and doubles into a new buffer
21:50:10stripwaxand the vorbis window is not symmetric
21:50:17n1sso you need half again as much memory
21:50:25stripwaxdunno about wmapro
21:51:34n1sthe way oi wrote my mdct-double+pwindow function with a special case for the symmetric window case (both overlapped blocks same size9 as this happens more than 90 % of al blocks
21:52:14saratogai don't think the wma ones are symmetric are they?
21:52:23saratogabeen too long
21:53:07n1sthe problem with moving the mdct output in the buffer is that the lower half is used for the input to the mdct
21:55:12n1sbut for non iram challenged targets, doing it with separate buffers would almost certainly be faster
21:55:36saratogathe ffmpeg way seems to use 2048 input samples and some number of output samples i can't easily see because the buffer is shared with a bunch of other things
21:56:04saratogai was actually going to switch wma std over to the wma pro version but never got around to it
21:56:11saratogamaybe theres a better way though
22:02:00 Quit cozmic_ (Ping timeout: 240 seconds)
22:02:02 Quit evilnick_B (Quit: Page closed)
22:06:10n1shmm, now looking at this tremor iram dram stuff again, it seems we overlapp add into the dram buffer, that would be faster if it could be done into the iram buffer, and i dont' see why it couldn't
22:08:20 Join cozmic [0] (
22:12:05 Quit stripwax (Quit:
22:13:26***Saving seen data "./dancer.seen"
22:15:37 Join stripwax [0] (
22:31:04 Quit simonrvn (Read error: No route to host)
22:31:27n1spixelma: i'm closing FS #11811 (it's about a string missing from the german translation, thought you might be interested but it's certainly not a bug)
22:34:33 Join wodz [0] (
22:36:38 Join simonrvn [0] (
22:40:21bertrikI edited that task a bit (changed it from a patch to a bug), the original title was very non-specific
22:42:12n1si still think it should be cloesd
22:42:47n1slangs getting out of sync isnt' a bug
22:43:12n1sbut i was hoping soeone woudl comment
22:43:56gevaertsWell, it suggests a translation, so it's a half-patch in a way
22:44:07pixelmaI really don't have the time to do anything about it today
22:44:13pixelmawith a half name too
22:44:21n1spixelma: do you want it to remain open?
22:44:37wodzlovely - h300 uses 2 distinct i2c software drivers + one coldfire channel, h100 uses one i2c software driver + one coldfire channel
22:44:46pixelmaI'll try to remember it, not sure n1s
22:45:10*gevaerts asks for the patch
22:45:43n1si'll leave it alone then but it's just a missing string it will show up in the next translation update
22:47:30 Quit TheSeven (Read error: Connection reset by peer)
22:48:27 Join TheSeven [0] (~TheSeven@rockbox/developer/TheSeven)
22:51:46 Quit m|c (Ping timeout: 265 seconds)
22:51:46 Quit ack` (Ping timeout: 265 seconds)
22:51:46 Quit AlexP (Ping timeout: 265 seconds)
22:51:58 Quit Bushmills (Ping timeout: 245 seconds)
22:52:15 Quit knittl (Ping timeout: 265 seconds)
22:53:52 Quit ReimuHakurei_ (Read error: Connection reset by peer)
22:55:03 Join AlexP [0] (~alex@rockbox/staff/AlexP)
22:57:23 Quit simonrvn (*.net *.split)
22:57:23 Quit literal (*.net *.split)
22:57:23 Quit saratoga (*.net *.split)
22:57:23 Quit Rob2222 (*.net *.split)
22:57:23 Quit crwl (*.net *.split)
22:57:23 Quit Kitar|st (*.net *.split)
22:57:23 Quit mordocai (*.net *.split)
22:57:24 Quit GodEater (*.net *.split)
22:57:24 Quit Battousai (*.net *.split)
22:57:24 Quit mikroflops (*.net *.split)
22:57:24 Quit preglow (*.net *.split)
22:57:24 Quit Horscht (*.net *.split)
22:57:24 Quit efyx (*.net *.split)
22:57:24 Quit komputes (*.net *.split)
22:57:24 Quit pixelma (*.net *.split)
22:57:24 Quit amiconn (*.net *.split)
22:57:24 Quit Stummi (*.net *.split)
22:57:24 Quit bzed (*.net *.split)
22:57:24 Quit TheLemonMan (*.net *.split)
22:57:24 Quit DerPapst (*.net *.split)
22:57:24 Quit hebz0rl (*.net *.split)
22:57:24 Quit factor (*.net *.split)
22:57:24 Quit BlakeJohnson86 (*.net *.split)
22:57:24 Quit pamaury (*.net *.split)
22:57:24 Quit Curulan (*.net *.split)
22:57:24 Quit guymann (*.net *.split)
22:57:24 Quit pjm0616 (*.net *.split)
22:57:24 Quit kkit|sh (*.net *.split)
22:57:25 Quit yosafbridge (*.net *.split)
22:57:25 Quit Bawitdaba (*.net *.split)
22:57:25 Quit maraz (*.net *.split)
22:57:25 Quit jae_ (*.net *.split)
22:57:25 Quit ranmachan (*.net *.split)
22:57:25 Quit ThomasAH (*.net *.split)
22:57:25 Quit soap (*.net *.split)
22:57:25 Quit Judas_PhD (*.net *.split)
22:57:25 Quit Strife89 (*.net *.split)
22:57:25 Quit aevin (*.net *.split)
22:57:25 Quit merbanan (*.net *.split)
22:57:25 Quit Torne (*.net *.split)
22:57:25 Quit rasher (*.net *.split)
22:57:25 Quit parafin (*.net *.split)
22:57:26 Quit n1s (*.net *.split)
22:57:26 Quit kevku (*.net *.split)
22:57:26 Quit ved_ (*.net *.split)
22:57:26 Quit Rondom (*.net *.split)
22:57:26 Quit Galois (*.net *.split)
22:57:26 Quit Bagder (*.net *.split)
22:57:26 Quit TBCOOL (*.net *.split)
22:57:26 Quit Slasheri (*.net *.split)
22:57:27 Quit edboyer93 (*.net *.split)
22:57:27 Quit Hadaka (*.net *.split)
22:57:27 Quit shai (*.net *.split)
22:57:27 Quit zu_ (*.net *.split)
22:57:27 Quit Utchy (*.net *.split)
22:57:27 Quit 5EXABW64X (*.net *.split)
22:57:27 Quit wodz (*.net *.split)
22:57:27 Quit InsDel (*.net *.split)
22:57:27 Quit mortalscan (*.net *.split)
22:57:27 Quit Guinness (*.net *.split)
22:57:28 Quit dfkt (*.net *.split)
22:57:28 Quit liar (*.net *.split)
22:57:28 Quit the_Kyle (*.net *.split)
22:57:28 Quit linuxstb (*.net *.split)
22:57:28 Quit froggyman (*.net *.split)
22:57:28 Quit mystica555_ (*.net *.split)
22:57:28 Quit Farthen (*.net *.split)
22:57:28 Quit Loto (*.net *.split)
22:57:28 Quit stripwax (*.net *.split)
22:57:28 Quit sasquatch (*.net *.split)
22:57:28 Quit Gareth (*.net *.split)
22:57:28 Quit ender` (*.net *.split)
22:57:28 Quit BHSPitMonkey (*.net *.split)
22:57:28 Quit Guest75111 (*.net *.split)
22:57:29 Quit antil33t (*.net *.split)
22:57:29 Quit user890104 (*.net *.split)
22:57:29 Quit MagusG (*.net *.split)
22:57:29 Quit burn (*.net *.split)
22:57:29 Quit timccc (*.net *.split)
22:57:29 Quit part (*.net *.split)
22:57:29 Quit ps-auxw (*.net *.split)
22:57:59 Join ack [0] (
22:57:59 Join ReimuHakurei_ [0] (~reimu@
22:57:59 Join knittl [0] (
22:57:59 Join miceh [0] (
22:57:59 Join Zarggg_ [0] (
22:57:59 Join simonrvn [0] (
22:57:59 Join wodz [0] (
22:57:59 Join stripwax [0] (
22:57:59 Join preglow [0] (
22:57:59 Join Gareth [0] (
22:57:59 Join n1s [0] (~n1s@rockbox/developer/n1s)
22:57:59 Join InsDel [0] (
22:57:59 Join BHSPitMonkey [0] (~stephen@unaffiliated/bhspitmonkey)
22:57:59 Join Guest75111 [0] (
22:57:59 Join TheLemonMan [0] (
22:57:59 Join DerPapst [0] (
22:57:59 Join hebz0rl [0] (
22:57:59 Join Horscht [0] (~Horschti@xbmc/user/horscht)
22:57:59 Join mortalscan [0] (~mortalsca@
22:57:59 Join efyx [0] (
22:57:59 Join antil33t [0] (
22:57:59 Join Guinness [0] (
22:57:59 Join kevku [0] (
22:57:59 Join komputes [0] (~komputes@ubuntu/member/komputes)
22:57:59 Join factor [0] (
22:57:59 Join Rob2222 [0] (
22:57:59 Join BlakeJohnson86 [0] (
22:57:59 Join soap [0] (~soap@rockbox/staff/soap)
22:57:59 Join dfkt [0] (dfkt@unaffiliated/dfkt)
22:57:59 Join user890104 [0] (
22:57:59 Join Judas_PhD [0] (
22:57:59 Join pamaury [0] (~quassel@rockbox/developer/pamaury)
22:57:59 Join liar [0] (
22:57:59 Join the_Kyle [0] (~kyle@
22:57:59 Join crwl [0] (
22:57:59 Join pixelma [0] (quassel@rockbox/staff/pixelma)
22:57:59 Join amiconn [0] (quassel@rockbox/developer/amiconn)
22:57:59 Join sasquatch [0] (
22:57:59 Join edboyer93 [0] (
22:57:59 Join Kitar|st [0] (
22:57:59 Join Hadaka [0] (
22:57:59 Join MagusG [0] (
22:57:59 Join Stummi [0] (Stummi@rockbox/developer/Stummi)
22:57:59 Join linuxstb [0] (~linuxstb@rockbox/developer/linuxstb)
22:57:59 Join mordocai [0] (~mordocai@
22:57:59 Join Strife89 [0] (
22:57:59 Join burn [0] (
22:57:59 Join ender` [0] (
22:57:59 Join shai [0] (
22:57:59 Join froggyman [0] (~seth@unaffiliated/froggyman)
22:57:59 Join bzed [0] (
22:57:59 Join mystica555_ [0] (
22:57:59 Join literal [0] (
22:57:59 Join Farthen [0] (
22:57:59 Join part [0] (part@
22:57:59 Join timccc [0] (~timccc@
22:57:59 Join zu_ [0] (
22:57:59 Join guymann [0] (~charles@
22:57:59 Join pjm0616 [0] (~user@
22:57:59 Join Loto [0] (~nfs@xbmc/user/Loto)
22:57:59 Join ps-auxw [0] (~arneb@2001:470:c807:0:1532:4e5f:2ad3:4123)
22:57:59 Join Utchy [0] (
22:57:59 Join 5EXABW64X [0] (
22:57:59 Join ved_ [0] (
22:57:59 Join Rondom [0] (
22:57:59 Join Galois [0] (
22:57:59 Join Bagder [0] (~daniel@rockbox/developer/bagder)
22:57:59 Join TBCOOL [0] (
22:57:59 Join Slasheri [0] (miipekk@rockbox/developer/Slasheri)
22:57:59 Join aevin [0] (eivindsy@unaffiliated/aevin)
22:57:59 Join merbanan [0] (
22:57:59 Join Torne [0] (torne@rockbox/developer/Torne)
22:57:59 Join kkit|sh [0] (
22:57:59 Join yosafbridge [0] (
22:57:59 Join Bawitdaba [0] (
22:57:59 Join maraz [0] (
22:57:59 Join jae_ [0] (
22:57:59 Join ranmachan [0] (
22:57:59 Join ThomasAH [0] (
22:57:59 Join rasher [0] (~rasher@rockbox/developer/rasher)
22:57:59 Join parafin [0] (
22:57:59 Join GodEater [0] (~bibble@rockbox/staff/GodEater)
22:57:59 Join Battousai [0] (~bryan@gentoo/developer/battousai)
22:57:59 Join mikroflops [0] (
22:58:05 Quit kevku (Quit: KVIrc 4.0.2 Insomnia
23:04:14 Join JdGordon [0] (~jonno@rockbox/developer/JdGordon)
23:05:32 Quit simonrvn (Read error: No route to host)
23:05:42 Join Bushmills [0] (
23:06:04 Quit jgarvey (Quit: Leaving)
23:07:09CIA-45New commit by 03wodz (r28845): HD300 - hacky workaround which makes USB bridge work in rockbox
23:07:20 Quit stripwax (Ping timeout: 276 seconds)
23:08:19 Quit factor (Ping timeout: 250 seconds)
23:08:59 Quit TheLemonMan (Quit: free(me))
23:09:00CIA-45r28845 build result: 15 errors, 0 warnings (wodz committed)
23:09:52*gevaerts pings Bagder again
23:10:23gevaertswodz: build system vs gcc 4.3 thing
23:10:25BagderI rather not mess with the build server
23:10:34gevaertsThis is a simple change!
23:10:54 Join mulenmar [0] (
23:11:06gevaertsAnd it only changes how points are counted really (it adds a special case for this note: thing)
23:11:28 Join r34per [0] (
23:12:59 Quit r34per (Client Quit)
23:13:07Bagderright, but I still have zagor do that as I'm not sure how he does those things these days
23:13:46gevaertsAs long as people know they aren't to blame...
23:15:07n1sshouldn't we block known bad compilers or known bad clients?
23:15:34gevaertsThis is neither
23:15:51gevaertsThe compiler just outputs some lines that the build system doesn't expect
23:16:01wodzit isn't bad it just throws additional notes
23:16:54gevaertsIf it was a bad client, I'd block it right away
23:16:55n1sso we should grep the lines for "error"
23:17:08n1sor "warning"
23:17:21gevaertsWell, I did commit a fix, it just needs to be updated on the server
23:17:39wodzBTW what exactly this 'note' mean?
23:17:50gevaertsAnd we want Zagor for that, so it will have to be like this until somewhere tomorrow I guess
23:18:46 Join t0rc [0] (~t0rc@unaffiliated/t0rc/x-5233201)
23:18:59n1sgevaerts: aha, sorry for not following :)
23:19:47 Quit mulenmar (Quit: Leaving)
23:19:51gevaertsWe *could* block that client for the night, but I'm not convinced that that's a good idea
23:19:51 Join designate72 [0] (
23:20:37n1sno, not that important
23:21:11*TheSeven wonders why this started happening this suddenly
23:21:14TheSevennew build client?
23:21:23 Join mulenmar [0] (
23:21:39wodzor gcc update
23:21:47gevaertsno, it's gcc 4.3 that does this
23:22:23n1sit's in the new mikmod plugin
23:22:34gevaertsyes, that too
23:23:00gevaertsBut this build client is newer
23:24:04 Join Chesteta [0] (
23:25:29wodzmaybe there is a gcc switch to suppress this 'note' lines?
23:25:39wodzas a temp solution?
23:26:00gevaertsagain, for half a day? Not worth it I think
23:26:22 Quit Chesteta (Client Quit)
23:29:38Dreamxtremehrm i'll ask this in here
23:30:03Dreamxtremehow come when i unplug the ipod right away it says around 90% battery
23:30:06wodzinteresting - google doesn't know much about gcc 'note' warnings
23:30:16Dreamxtremedo i need to do a Batt_bench?
23:30:37wodzor live with what you have
23:31:18wodzDreamxtreme: what ipod? hd based?
23:32:32 Nick ved_ is now known as ved (
23:33:22 Quit efyx (Quit: Quitte)
23:35:00wodzTurn it on and wait a bit - battery level should rise. HD takes quite a bit of current and battery voltage drops. When hd spins down voltage will recover to some degree.
23:35:12wodzThere is nothing we can do about this
23:35:23 Join factor [0] (
23:36:20 Join Luca_S [0] (
23:36:47Dreamxtremeo ok
23:37:16Luca_Swodz: in the latest commit, is it correct #ifndef BOOTLADER ?
23:37:34wodzLuca_S: good catch
23:37:57Luca_S:) that's the proof that I code better when I'm drunk (like now)
23:39:01CIA-45New commit by 03wodz (r28846): fix typo, thanks to Luka_S for catching this
23:41:09CIA-45r28846 build result: 5 errors, 0 warnings (wodz committed)
23:45:17 Quit Luca_S (Quit: CGI:IRC (EOF))
23:50:25 Quit froggyman (Quit: Ex-Chat)
23:50:46 Join froggyman [0] (~seth@
23:50:47 Quit froggyman (Changing host)
23:50:47 Join froggyman [0] (~seth@unaffiliated/froggyman)
23:54:55 Join JdGord [0] (~jonno@

Previous day | Next day